diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,36450 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999995191868488,
+  "eval_steps": 500,
+  "global_step": 51995,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 1.9232526048052467e-05,
+      "grad_norm": 332017.82194713707,
+      "learning_rate": 3.846153846153847e-08,
+      "loss": 12108.0527,
+      "step": 1
+    },
+    {
+      "epoch": 0.00019232526048052465,
+      "grad_norm": 245508.79891245405,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 12036.0373,
+      "step": 10
+    },
+    {
+      "epoch": 0.0003846505209610493,
+      "grad_norm": 115889.53374107159,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 10648.9656,
+      "step": 20
+    },
+    {
+      "epoch": 0.000576975781441574,
+      "grad_norm": 206614.59793923626,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 7566.8242,
+      "step": 30
+    },
+    {
+      "epoch": 0.0007693010419220986,
+      "grad_norm": 160550.51842977668,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 5169.2719,
+      "step": 40
+    },
+    {
+      "epoch": 0.0009616263024026233,
+      "grad_norm": 39134.623275659105,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 3722.1172,
+      "step": 50
+    },
+    {
+      "epoch": 0.001153951562883148,
+      "grad_norm": 11341.981755467272,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 2501.6607,
+      "step": 60
+    },
+    {
+      "epoch": 0.0013462768233636726,
+      "grad_norm": 14727.353496582424,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 1709.3086,
+      "step": 70
+    },
+    {
+      "epoch": 0.0015386020838441972,
+      "grad_norm": 7376.485461270763,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 1362.1637,
+      "step": 80
+    },
+    {
+      "epoch": 0.0017309273443247219,
+      "grad_norm": 4689.786549065178,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 1136.6605,
+      "step": 90
+    },
+    {
+      "epoch": 0.0019232526048052465,
+      "grad_norm": 3136.885512323355,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 1077.0688,
+      "step": 100
+    },
+    {
+      "epoch": 0.0021155778652857714,
+      "grad_norm": 2884.0784288983627,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 976.0121,
+      "step": 110
+    },
+    {
+      "epoch": 0.002307903125766296,
+      "grad_norm": 3280.7089284628487,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 928.786,
+      "step": 120
+    },
+    {
+      "epoch": 0.0025002283862468207,
+      "grad_norm": 2939.4442984784437,
+      "learning_rate": 5e-06,
+      "loss": 851.515,
+      "step": 130
+    },
+    {
+      "epoch": 0.002692553646727345,
+      "grad_norm": 2912.9567570884933,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 817.7365,
+      "step": 140
+    },
+    {
+      "epoch": 0.00288487890720787,
+      "grad_norm": 2067.4095594512014,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 789.5539,
+      "step": 150
+    },
+    {
+      "epoch": 0.0030772041676883945,
+      "grad_norm": 1952.0933429185154,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 757.9412,
+      "step": 160
+    },
+    {
+      "epoch": 0.0032695294281689193,
+      "grad_norm": 1976.9487027681635,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 740.7521,
+      "step": 170
+    },
+    {
+      "epoch": 0.0034618546886494438,
+      "grad_norm": 1786.3723313558642,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 728.6194,
+      "step": 180
+    },
+    {
+      "epoch": 0.0036541799491299687,
+      "grad_norm": 2310.282588762028,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 704.6479,
+      "step": 190
+    },
+    {
+      "epoch": 0.003846505209610493,
+      "grad_norm": 2074.87412983887,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 704.4103,
+      "step": 200
+    },
+    {
+      "epoch": 0.0040388304700910175,
+      "grad_norm": 2300.109138357241,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 664.9248,
+      "step": 210
+    },
+    {
+      "epoch": 0.004231155730571543,
+      "grad_norm": 1911.5281782387024,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 660.9208,
+      "step": 220
+    },
+    {
+      "epoch": 0.004423480991052067,
+      "grad_norm": 1469.8004882525568,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 642.6973,
+      "step": 230
+    },
+    {
+      "epoch": 0.004615806251532592,
+      "grad_norm": 1875.6806394934072,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 642.3829,
+      "step": 240
+    },
+    {
+      "epoch": 0.004808131512013117,
+      "grad_norm": 1954.1006277348115,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 643.9223,
+      "step": 250
+    },
+    {
+      "epoch": 0.0050004567724936414,
+      "grad_norm": 1737.3193764809394,
+      "learning_rate": 1e-05,
+      "loss": 645.1771,
+      "step": 260
+    },
+    {
+      "epoch": 0.005192782032974166,
+      "grad_norm": 1327.6624239260284,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 637.8946,
+      "step": 270
+    },
+    {
+      "epoch": 0.00538510729345469,
+      "grad_norm": 7981.270677692642,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 618.1356,
+      "step": 280
+    },
+    {
+      "epoch": 0.005577432553935216,
+      "grad_norm": 1709.4807102572415,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 630.8181,
+      "step": 290
+    },
+    {
+      "epoch": 0.00576975781441574,
+      "grad_norm": 1644.4225063710128,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 605.8881,
+      "step": 300
+    },
+    {
+      "epoch": 0.0059620830748962645,
+      "grad_norm": 1966.4081016099562,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 604.7532,
+      "step": 310
+    },
+    {
+      "epoch": 0.006154408335376789,
+      "grad_norm": 1491.86389210639,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 609.7085,
+      "step": 320
+    },
+    {
+      "epoch": 0.006346733595857314,
+      "grad_norm": 1485.4453324763408,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 602.9562,
+      "step": 330
+    },
+    {
+      "epoch": 0.006539058856337839,
+      "grad_norm": 1310.387260927013,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 595.0032,
+      "step": 340
+    },
+    {
+      "epoch": 0.006731384116818363,
+      "grad_norm": 2275.2781855062035,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 579.4093,
+      "step": 350
+    },
+    {
+      "epoch": 0.0069237093772988876,
+      "grad_norm": 1849.868176409446,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 585.0145,
+      "step": 360
+    },
+    {
+      "epoch": 0.007116034637779413,
+      "grad_norm": 1615.7507754722587,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 592.8799,
+      "step": 370
+    },
+    {
+      "epoch": 0.007308359898259937,
+      "grad_norm": 1610.504408373516,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 589.673,
+      "step": 380
+    },
+    {
+      "epoch": 0.007500685158740462,
+      "grad_norm": 1704.5476494298723,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 582.5739,
+      "step": 390
+    },
+    {
+      "epoch": 0.007693010419220986,
+      "grad_norm": 1502.1680695792734,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 571.0479,
+      "step": 400
+    },
+    {
+      "epoch": 0.00788533567970151,
+      "grad_norm": 1579.6577195492494,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 563.5073,
+      "step": 410
+    },
+    {
+      "epoch": 0.008077660940182035,
+      "grad_norm": 1785.8398259450537,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 581.3297,
+      "step": 420
+    },
+    {
+      "epoch": 0.008269986200662561,
+      "grad_norm": 1771.438162106497,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 568.3677,
+      "step": 430
+    },
+    {
+      "epoch": 0.008462311461143086,
+      "grad_norm": 1363.7967686916552,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 567.2524,
+      "step": 440
+    },
+    {
+      "epoch": 0.00865463672162361,
+      "grad_norm": 1417.5481457867793,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 592.7938,
+      "step": 450
+    },
+    {
+      "epoch": 0.008846961982104135,
+      "grad_norm": 2912.185460436148,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 558.3062,
+      "step": 460
+    },
+    {
+      "epoch": 0.009039287242584659,
+      "grad_norm": 1419.8660565163475,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 563.9717,
+      "step": 470
+    },
+    {
+      "epoch": 0.009231612503065183,
+      "grad_norm": 1538.9205592163212,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 541.6976,
+      "step": 480
+    },
+    {
+      "epoch": 0.009423937763545708,
+      "grad_norm": 1674.7941560707338,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 562.8638,
+      "step": 490
+    },
+    {
+      "epoch": 0.009616263024026234,
+      "grad_norm": 1184.5116494790307,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 550.3092,
+      "step": 500
+    },
+    {
+      "epoch": 0.009808588284506758,
+      "grad_norm": 1918.2405919716136,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 569.4998,
+      "step": 510
+    },
+    {
+      "epoch": 0.010000913544987283,
+      "grad_norm": 1258.73028504583,
+      "learning_rate": 2e-05,
+      "loss": 562.6359,
+      "step": 520
+    },
+    {
+      "epoch": 0.010193238805467807,
+      "grad_norm": 1957.1855022650734,
+      "learning_rate": 1.999999813758258e-05,
+      "loss": 540.9211,
+      "step": 530
+    },
+    {
+      "epoch": 0.010385564065948332,
+      "grad_norm": 1445.1918558753534,
+      "learning_rate": 1.9999992550331007e-05,
+      "loss": 552.493,
+      "step": 540
+    },
+    {
+      "epoch": 0.010577889326428856,
+      "grad_norm": 1425.9554496443363,
+      "learning_rate": 1.999998323824736e-05,
+      "loss": 551.791,
+      "step": 550
+    },
+    {
+      "epoch": 0.01077021458690938,
+      "grad_norm": 1387.5175764634103,
+      "learning_rate": 1.999997020133512e-05,
+      "loss": 539.2529,
+      "step": 560
+    },
+    {
+      "epoch": 0.010962539847389905,
+      "grad_norm": 1282.1917414212555,
+      "learning_rate": 1.9999953439599132e-05,
+      "loss": 543.0103,
+      "step": 570
+    },
+    {
+      "epoch": 0.011154865107870431,
+      "grad_norm": 1387.5597126230348,
+      "learning_rate": 1.9999932953045638e-05,
+      "loss": 549.0967,
+      "step": 580
+    },
+    {
+      "epoch": 0.011347190368350956,
+      "grad_norm": 1495.9706529903751,
+      "learning_rate": 1.999990874168228e-05,
+      "loss": 542.2873,
+      "step": 590
+    },
+    {
+      "epoch": 0.01153951562883148,
+      "grad_norm": 1366.7575608631,
+      "learning_rate": 1.9999880805518067e-05,
+      "loss": 550.7072,
+      "step": 600
+    },
+    {
+      "epoch": 0.011731840889312005,
+      "grad_norm": 1550.7141727742994,
+      "learning_rate": 1.9999849144563406e-05,
+      "loss": 536.1631,
+      "step": 610
+    },
+    {
+      "epoch": 0.011924166149792529,
+      "grad_norm": 1799.438677775851,
+      "learning_rate": 1.9999813758830092e-05,
+      "loss": 533.2322,
+      "step": 620
+    },
+    {
+      "epoch": 0.012116491410273053,
+      "grad_norm": 1594.2553453970218,
+      "learning_rate": 1.9999774648331307e-05,
+      "loss": 531.8908,
+      "step": 630
+    },
+    {
+      "epoch": 0.012308816670753578,
+      "grad_norm": 1380.3126968166453,
+      "learning_rate": 1.9999731813081616e-05,
+      "loss": 539.0061,
+      "step": 640
+    },
+    {
+      "epoch": 0.012501141931234102,
+      "grad_norm": 1288.9094624087347,
+      "learning_rate": 1.9999685253096975e-05,
+      "loss": 533.8396,
+      "step": 650
+    },
+    {
+      "epoch": 0.012693467191714628,
+      "grad_norm": 1467.6587965930335,
+      "learning_rate": 1.999963496839473e-05,
+      "loss": 511.2933,
+      "step": 660
+    },
+    {
+      "epoch": 0.012885792452195153,
+      "grad_norm": 1144.4252030903722,
+      "learning_rate": 1.9999580958993606e-05,
+      "loss": 518.686,
+      "step": 670
+    },
+    {
+      "epoch": 0.013078117712675677,
+      "grad_norm": 1510.7992006303907,
+      "learning_rate": 1.9999523224913722e-05,
+      "loss": 541.2946,
+      "step": 680
+    },
+    {
+      "epoch": 0.013270442973156202,
+      "grad_norm": 1459.5247736769318,
+      "learning_rate": 1.999946176617659e-05,
+      "loss": 531.0906,
+      "step": 690
+    },
+    {
+      "epoch": 0.013462768233636726,
+      "grad_norm": 1704.7295855200716,
+      "learning_rate": 1.999939658280509e-05,
+      "loss": 515.0679,
+      "step": 700
+    },
+    {
+      "epoch": 0.01365509349411725,
+      "grad_norm": 1369.6286303265301,
+      "learning_rate": 1.9999327674823513e-05,
+      "loss": 529.9729,
+      "step": 710
+    },
+    {
+      "epoch": 0.013847418754597775,
+      "grad_norm": 1171.6944991940852,
+      "learning_rate": 1.9999255042257522e-05,
+      "loss": 523.1766,
+      "step": 720
+    },
+    {
+      "epoch": 0.0140397440150783,
+      "grad_norm": 1733.2115650083967,
+      "learning_rate": 1.999917868513417e-05,
+      "loss": 512.975,
+      "step": 730
+    },
+    {
+      "epoch": 0.014232069275558826,
+      "grad_norm": 1410.6921404307557,
+      "learning_rate": 1.9999098603481895e-05,
+      "loss": 496.8678,
+      "step": 740
+    },
+    {
+      "epoch": 0.01442439453603935,
+      "grad_norm": 1053.8093124978518,
+      "learning_rate": 1.9999014797330536e-05,
+      "loss": 511.1349,
+      "step": 750
+    },
+    {
+      "epoch": 0.014616719796519875,
+      "grad_norm": 1194.4866698136166,
+      "learning_rate": 1.99989272667113e-05,
+      "loss": 514.8527,
+      "step": 760
+    },
+    {
+      "epoch": 0.014809045057000399,
+      "grad_norm": 1126.7421355720714,
+      "learning_rate": 1.99988360116568e-05,
+      "loss": 514.3693,
+      "step": 770
+    },
+    {
+      "epoch": 0.015001370317480923,
+      "grad_norm": 1178.4782527711411,
+      "learning_rate": 1.999874103220102e-05,
+      "loss": 506.5624,
+      "step": 780
+    },
+    {
+      "epoch": 0.015193695577961448,
+      "grad_norm": 1297.0010276174844,
+      "learning_rate": 1.999864232837934e-05,
+      "loss": 519.1336,
+      "step": 790
+    },
+    {
+      "epoch": 0.015386020838441972,
+      "grad_norm": 1069.667364513505,
+      "learning_rate": 1.9998539900228526e-05,
+      "loss": 514.1775,
+      "step": 800
+    },
+    {
+      "epoch": 0.015578346098922499,
+      "grad_norm": 1262.000307745403,
+      "learning_rate": 1.999843374778673e-05,
+      "loss": 511.9821,
+      "step": 810
+    },
+    {
+      "epoch": 0.01577067135940302,
+      "grad_norm": 1500.5916093620833,
+      "learning_rate": 1.999832387109349e-05,
+      "loss": 505.3807,
+      "step": 820
+    },
+    {
+      "epoch": 0.015962996619883547,
+      "grad_norm": 1398.4282733840143,
+      "learning_rate": 1.9998210270189736e-05,
+      "loss": 503.6255,
+      "step": 830
+    },
+    {
+      "epoch": 0.01615532188036407,
+      "grad_norm": 1364.4269370667485,
+      "learning_rate": 1.9998092945117786e-05,
+      "loss": 495.5404,
+      "step": 840
+    },
+    {
+      "epoch": 0.016347647140844596,
+      "grad_norm": 1458.3289472638558,
+      "learning_rate": 1.999797189592134e-05,
+      "loss": 500.7938,
+      "step": 850
+    },
+    {
+      "epoch": 0.016539972401325122,
+      "grad_norm": 35408.406023696065,
+      "learning_rate": 1.999784712264548e-05,
+      "loss": 541.5755,
+      "step": 860
+    },
+    {
+      "epoch": 0.016732297661805645,
+      "grad_norm": 1259.0220001600435,
+      "learning_rate": 1.9997718625336686e-05,
+      "loss": 488.8561,
+      "step": 870
+    },
+    {
+      "epoch": 0.01692462292228617,
+      "grad_norm": 1115.7727735781639,
+      "learning_rate": 1.9997586404042825e-05,
+      "loss": 503.5537,
+      "step": 880
+    },
+    {
+      "epoch": 0.017116948182766694,
+      "grad_norm": 1261.659255730506,
+      "learning_rate": 1.9997450458813142e-05,
+      "loss": 489.9525,
+      "step": 890
+    },
+    {
+      "epoch": 0.01730927344324722,
+      "grad_norm": 1174.6527357294908,
+      "learning_rate": 1.9997310789698276e-05,
+      "loss": 484.3942,
+      "step": 900
+    },
+    {
+      "epoch": 0.017501598703727743,
+      "grad_norm": 1060.893242229332,
+      "learning_rate": 1.999716739675025e-05,
+      "loss": 494.4259,
+      "step": 910
+    },
+    {
+      "epoch": 0.01769392396420827,
+      "grad_norm": 1378.6178778139686,
+      "learning_rate": 1.9997020280022482e-05,
+      "loss": 492.2053,
+      "step": 920
+    },
+    {
+      "epoch": 0.017886249224688795,
+      "grad_norm": 1384.0970365292112,
+      "learning_rate": 1.9996869439569767e-05,
+      "loss": 486.3698,
+      "step": 930
+    },
+    {
+      "epoch": 0.018078574485169318,
+      "grad_norm": 1141.0195763132995,
+      "learning_rate": 1.9996714875448286e-05,
+      "loss": 484.9983,
+      "step": 940
+    },
+    {
+      "epoch": 0.018270899745649844,
+      "grad_norm": 1534.190545857729,
+      "learning_rate": 1.9996556587715617e-05,
+      "loss": 491.6367,
+      "step": 950
+    },
+    {
+      "epoch": 0.018463225006130367,
+      "grad_norm": 1260.3034572615545,
+      "learning_rate": 1.9996394576430716e-05,
+      "loss": 491.5582,
+      "step": 960
+    },
+    {
+      "epoch": 0.018655550266610893,
+      "grad_norm": 1131.6626054510543,
+      "learning_rate": 1.9996228841653932e-05,
+      "loss": 492.3444,
+      "step": 970
+    },
+    {
+      "epoch": 0.018847875527091416,
+      "grad_norm": 1080.0350168322668,
+      "learning_rate": 1.9996059383447e-05,
+      "loss": 481.0039,
+      "step": 980
+    },
+    {
+      "epoch": 0.019040200787571942,
+      "grad_norm": 1186.8576062599248,
+      "learning_rate": 1.999588620187303e-05,
+      "loss": 488.0979,
+      "step": 990
+    },
+    {
+      "epoch": 0.019232526048052468,
+      "grad_norm": 1284.027907525847,
+      "learning_rate": 1.9995709296996545e-05,
+      "loss": 490.5315,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01942485130853299,
+      "grad_norm": 1084.408370907811,
+      "learning_rate": 1.999552866888343e-05,
+      "loss": 494.6817,
+      "step": 1010
+    },
+    {
+      "epoch": 0.019617176569013517,
+      "grad_norm": 1078.4092927240245,
+      "learning_rate": 1.9995344317600965e-05,
+      "loss": 502.1682,
+      "step": 1020
+    },
+    {
+      "epoch": 0.01980950182949404,
+      "grad_norm": 2212.2117524226715,
+      "learning_rate": 1.9995156243217824e-05,
+      "loss": 497.8876,
+      "step": 1030
+    },
+    {
+      "epoch": 0.020001827089974566,
+      "grad_norm": 1450.9938826004914,
+      "learning_rate": 1.9994964445804054e-05,
+      "loss": 500.4225,
+      "step": 1040
+    },
+    {
+      "epoch": 0.02019415235045509,
+      "grad_norm": 1439.5951412209038,
+      "learning_rate": 1.9994768925431104e-05,
+      "loss": 478.5474,
+      "step": 1050
+    },
+    {
+      "epoch": 0.020386477610935615,
+      "grad_norm": 1111.8307979449899,
+      "learning_rate": 1.9994569682171796e-05,
+      "loss": 472.0606,
+      "step": 1060
+    },
+    {
+      "epoch": 0.020578802871416137,
+      "grad_norm": 1130.0088246480677,
+      "learning_rate": 1.9994366716100346e-05,
+      "loss": 479.0884,
+      "step": 1070
+    },
+    {
+      "epoch": 0.020771128131896664,
+      "grad_norm": 1368.426354451509,
+      "learning_rate": 1.999416002729236e-05,
+      "loss": 475.8784,
+      "step": 1080
+    },
+    {
+      "epoch": 0.02096345339237719,
+      "grad_norm": 1471.6865813297497,
+      "learning_rate": 1.999394961582482e-05,
+      "loss": 493.142,
+      "step": 1090
+    },
+    {
+      "epoch": 0.021155778652857712,
+      "grad_norm": 1151.4689769800118,
+      "learning_rate": 1.9993735481776105e-05,
+      "loss": 467.3699,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02134810391333824,
+      "grad_norm": 1103.1388757982427,
+      "learning_rate": 1.9993517625225976e-05,
+      "loss": 468.3865,
+      "step": 1110
+    },
+    {
+      "epoch": 0.02154042917381876,
+      "grad_norm": 915.9358718055254,
+      "learning_rate": 1.9993296046255578e-05,
+      "loss": 462.7445,
+      "step": 1120
+    },
+    {
+      "epoch": 0.021732754434299287,
+      "grad_norm": 997.6094400871649,
+      "learning_rate": 1.9993070744947447e-05,
+      "loss": 475.512,
+      "step": 1130
+    },
+    {
+      "epoch": 0.02192507969477981,
+      "grad_norm": 1202.0248958828354,
+      "learning_rate": 1.9992841721385508e-05,
+      "loss": 464.4746,
+      "step": 1140
+    },
+    {
+      "epoch": 0.022117404955260336,
+      "grad_norm": 11237.009349666056,
+      "learning_rate": 1.999260897565506e-05,
+      "loss": 486.6293,
+      "step": 1150
+    },
+    {
+      "epoch": 0.022309730215740863,
+      "grad_norm": 1032.047329108433,
+      "learning_rate": 1.9992372507842807e-05,
+      "loss": 484.2495,
+      "step": 1160
+    },
+    {
+      "epoch": 0.022502055476221385,
+      "grad_norm": 1290.0903297656175,
+      "learning_rate": 1.9992132318036825e-05,
+      "loss": 481.1325,
+      "step": 1170
+    },
+    {
+      "epoch": 0.02269438073670191,
+      "grad_norm": 1037.6319401213834,
+      "learning_rate": 1.9991888406326575e-05,
+      "loss": 486.5979,
+      "step": 1180
+    },
+    {
+      "epoch": 0.022886705997182434,
+      "grad_norm": 1077.555863211059,
+      "learning_rate": 1.9991640772802916e-05,
+      "loss": 467.6946,
+      "step": 1190
+    },
+    {
+      "epoch": 0.02307903125766296,
+      "grad_norm": 1173.5983183577753,
+      "learning_rate": 1.9991389417558088e-05,
+      "loss": 469.8534,
+      "step": 1200
+    },
+    {
+      "epoch": 0.023271356518143483,
+      "grad_norm": 1047.9155015558533,
+      "learning_rate": 1.9991134340685713e-05,
+      "loss": 463.1124,
+      "step": 1210
+    },
+    {
+      "epoch": 0.02346368177862401,
+      "grad_norm": 1197.2310531685393,
+      "learning_rate": 1.999087554228081e-05,
+      "loss": 463.741,
+      "step": 1220
+    },
+    {
+      "epoch": 0.023656007039104535,
+      "grad_norm": 949.005837498942,
+      "learning_rate": 1.999061302243977e-05,
+      "loss": 470.0569,
+      "step": 1230
+    },
+    {
+      "epoch": 0.023848332299585058,
+      "grad_norm": 1222.0283637424163,
+      "learning_rate": 1.9990346781260378e-05,
+      "loss": 469.6994,
+      "step": 1240
+    },
+    {
+      "epoch": 0.024040657560065584,
+      "grad_norm": 1721.3575140480177,
+      "learning_rate": 1.9990076818841805e-05,
+      "loss": 467.2944,
+      "step": 1250
+    },
+    {
+      "epoch": 0.024232982820546107,
+      "grad_norm": 1169.3409682586346,
+      "learning_rate": 1.998980313528461e-05,
+      "loss": 471.5979,
+      "step": 1260
+    },
+    {
+      "epoch": 0.024425308081026633,
+      "grad_norm": 1358.8689502222835,
+      "learning_rate": 1.9989525730690736e-05,
+      "loss": 467.8751,
+      "step": 1270
+    },
+    {
+      "epoch": 0.024617633341507156,
+      "grad_norm": 1166.25480338363,
+      "learning_rate": 1.998924460516351e-05,
+      "loss": 478.3506,
+      "step": 1280
+    },
+    {
+      "epoch": 0.024809958601987682,
+      "grad_norm": 1048.097700016728,
+      "learning_rate": 1.9988959758807645e-05,
+      "loss": 469.7929,
+      "step": 1290
+    },
+    {
+      "epoch": 0.025002283862468205,
+      "grad_norm": 1041.7763831771526,
+      "learning_rate": 1.9988671191729243e-05,
+      "loss": 478.3628,
+      "step": 1300
+    },
+    {
+      "epoch": 0.02519460912294873,
+      "grad_norm": 1182.0911908058742,
+      "learning_rate": 1.998837890403579e-05,
+      "loss": 470.6274,
+      "step": 1310
+    },
+    {
+      "epoch": 0.025386934383429257,
+      "grad_norm": 1211.438189548962,
+      "learning_rate": 1.998808289583616e-05,
+      "loss": 461.9727,
+      "step": 1320
+    },
+    {
+      "epoch": 0.02557925964390978,
+      "grad_norm": 1013.8385452188595,
+      "learning_rate": 1.998778316724061e-05,
+      "loss": 458.8419,
+      "step": 1330
+    },
+    {
+      "epoch": 0.025771584904390306,
+      "grad_norm": 970.6944019104968,
+      "learning_rate": 1.998747971836078e-05,
+      "loss": 446.9404,
+      "step": 1340
+    },
+    {
+      "epoch": 0.02596391016487083,
+      "grad_norm": 1021.9827701608976,
+      "learning_rate": 1.9987172549309707e-05,
+      "loss": 467.6753,
+      "step": 1350
+    },
+    {
+      "epoch": 0.026156235425351355,
+      "grad_norm": 1501.7882814399545,
+      "learning_rate": 1.9986861660201802e-05,
+      "loss": 460.099,
+      "step": 1360
+    },
+    {
+      "epoch": 0.026348560685831877,
+      "grad_norm": 1119.8102022832766,
+      "learning_rate": 1.998654705115287e-05,
+      "loss": 476.4499,
+      "step": 1370
+    },
+    {
+      "epoch": 0.026540885946312404,
+      "grad_norm": 1607.4317521625792,
+      "learning_rate": 1.9986228722280093e-05,
+      "loss": 479.8248,
+      "step": 1380
+    },
+    {
+      "epoch": 0.02673321120679293,
+      "grad_norm": 1111.9087737119664,
+      "learning_rate": 1.998590667370204e-05,
+      "loss": 454.7312,
+      "step": 1390
+    },
+    {
+      "epoch": 0.026925536467273452,
+      "grad_norm": 1047.659163378557,
+      "learning_rate": 1.998558090553868e-05,
+      "loss": 465.4294,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02711786172775398,
+      "grad_norm": 871.5336873619162,
+      "learning_rate": 1.9985251417911347e-05,
+      "loss": 454.3733,
+      "step": 1410
+    },
+    {
+      "epoch": 0.0273101869882345,
+      "grad_norm": 1003.1759990372688,
+      "learning_rate": 1.9984918210942776e-05,
+      "loss": 449.2252,
+      "step": 1420
+    },
+    {
+      "epoch": 0.027502512248715028,
+      "grad_norm": 896.2012244693171,
+      "learning_rate": 1.9984581284757074e-05,
+      "loss": 458.1591,
+      "step": 1430
+    },
+    {
+      "epoch": 0.02769483750919555,
+      "grad_norm": 909.5865635412036,
+      "learning_rate": 1.9984240639479745e-05,
+      "loss": 453.5459,
+      "step": 1440
+    },
+    {
+      "epoch": 0.027887162769676076,
+      "grad_norm": 1671.0348857078125,
+      "learning_rate": 1.9983896275237677e-05,
+      "loss": 451.2657,
+      "step": 1450
+    },
+    {
+      "epoch": 0.0280794880301566,
+      "grad_norm": 1199.632444130952,
+      "learning_rate": 1.9983548192159132e-05,
+      "loss": 450.7684,
+      "step": 1460
+    },
+    {
+      "epoch": 0.028271813290637125,
+      "grad_norm": 1180.9154545121703,
+      "learning_rate": 1.998319639037377e-05,
+      "loss": 460.3022,
+      "step": 1470
+    },
+    {
+      "epoch": 0.02846413855111765,
+      "grad_norm": 1171.2654855596793,
+      "learning_rate": 1.9982840870012626e-05,
+      "loss": 469.2199,
+      "step": 1480
+    },
+    {
+      "epoch": 0.028656463811598174,
+      "grad_norm": 1350.7356071250701,
+      "learning_rate": 1.9982481631208138e-05,
+      "loss": 447.264,
+      "step": 1490
+    },
+    {
+      "epoch": 0.0288487890720787,
+      "grad_norm": 1340.1830938477137,
+      "learning_rate": 1.9982118674094104e-05,
+      "loss": 453.9175,
+      "step": 1500
+    },
+    {
+      "epoch": 0.029041114332559223,
+      "grad_norm": 1285.6520270172832,
+      "learning_rate": 1.9981751998805725e-05,
+      "loss": 447.0713,
+      "step": 1510
+    },
+    {
+      "epoch": 0.02923343959303975,
+      "grad_norm": 985.5172319049026,
+      "learning_rate": 1.998138160547958e-05,
+      "loss": 450.3087,
+      "step": 1520
+    },
+    {
+      "epoch": 0.029425764853520272,
+      "grad_norm": 1103.5772288640608,
+      "learning_rate": 1.9981007494253638e-05,
+      "loss": 446.5301,
+      "step": 1530
+    },
+    {
+      "epoch": 0.029618090114000798,
+      "grad_norm": 1168.413254974082,
+      "learning_rate": 1.998062966526724e-05,
+      "loss": 454.4448,
+      "step": 1540
+    },
+    {
+      "epoch": 0.029810415374481324,
+      "grad_norm": 1073.2938121474965,
+      "learning_rate": 1.998024811866113e-05,
+      "loss": 454.6996,
+      "step": 1550
+    },
+    {
+      "epoch": 0.030002740634961847,
+      "grad_norm": 1300.1753681145342,
+      "learning_rate": 1.9979862854577427e-05,
+      "loss": 457.9132,
+      "step": 1560
+    },
+    {
+      "epoch": 0.030195065895442373,
+      "grad_norm": 1162.6163973584632,
+      "learning_rate": 1.9979473873159635e-05,
+      "loss": 448.1717,
+      "step": 1570
+    },
+    {
+      "epoch": 0.030387391155922896,
+      "grad_norm": 1074.2415654472136,
+      "learning_rate": 1.9979081174552638e-05,
+      "loss": 454.8047,
+      "step": 1580
+    },
+    {
+      "epoch": 0.030579716416403422,
+      "grad_norm": 919.2474230803733,
+      "learning_rate": 1.997868475890271e-05,
+      "loss": 453.3545,
+      "step": 1590
+    },
+    {
+      "epoch": 0.030772041676883945,
+      "grad_norm": 1204.5616691859047,
+      "learning_rate": 1.997828462635752e-05,
+      "loss": 440.2779,
+      "step": 1600
+    },
+    {
+      "epoch": 0.03096436693736447,
+      "grad_norm": 1043.1071384961128,
+      "learning_rate": 1.99778807770661e-05,
+      "loss": 444.5987,
+      "step": 1610
+    },
+    {
+      "epoch": 0.031156692197844997,
+      "grad_norm": 1259.4732265726643,
+      "learning_rate": 1.9977473211178882e-05,
+      "loss": 476.3025,
+      "step": 1620
+    },
+    {
+      "epoch": 0.03134901745832552,
+      "grad_norm": 1009.046667196821,
+      "learning_rate": 1.9977061928847676e-05,
+      "loss": 455.5854,
+      "step": 1630
+    },
+    {
+      "epoch": 0.03154134271880604,
+      "grad_norm": 939.047196075382,
+      "learning_rate": 1.9976646930225678e-05,
+      "loss": 441.729,
+      "step": 1640
+    },
+    {
+      "epoch": 0.03173366797928657,
+      "grad_norm": 906.5229132871757,
+      "learning_rate": 1.997622821546747e-05,
+      "loss": 441.6997,
+      "step": 1650
+    },
+    {
+      "epoch": 0.031925993239767095,
+      "grad_norm": 993.2324977509386,
+      "learning_rate": 1.9975805784729008e-05,
+      "loss": 445.505,
+      "step": 1660
+    },
+    {
+      "epoch": 0.03211831850024762,
+      "grad_norm": 1093.0301208448795,
+      "learning_rate": 1.9975379638167654e-05,
+      "loss": 462.6874,
+      "step": 1670
+    },
+    {
+      "epoch": 0.03231064376072814,
+      "grad_norm": 1025.2680021460812,
+      "learning_rate": 1.9974949775942134e-05,
+      "loss": 433.165,
+      "step": 1680
+    },
+    {
+      "epoch": 0.032502969021208666,
+      "grad_norm": 972.8447632264914,
+      "learning_rate": 1.997451619821256e-05,
+      "loss": 451.5788,
+      "step": 1690
+    },
+    {
+      "epoch": 0.03269529428168919,
+      "grad_norm": 810.4883911530105,
+      "learning_rate": 1.997407890514044e-05,
+      "loss": 444.4225,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03288761954216972,
+      "grad_norm": 923.0036790051319,
+      "learning_rate": 1.9973637896888652e-05,
+      "loss": 436.9401,
+      "step": 1710
+    },
+    {
+      "epoch": 0.033079944802650245,
+      "grad_norm": 1755.782027319123,
+      "learning_rate": 1.997319317362147e-05,
+      "loss": 440.4286,
+      "step": 1720
+    },
+    {
+      "epoch": 0.033272270063130764,
+      "grad_norm": 979.1780716001408,
+      "learning_rate": 1.9972744735504542e-05,
+      "loss": 428.3424,
+      "step": 1730
+    },
+    {
+      "epoch": 0.03346459532361129,
+      "grad_norm": 907.9041043732557,
+      "learning_rate": 1.9972292582704905e-05,
+      "loss": 448.4176,
+      "step": 1740
+    },
+    {
+      "epoch": 0.033656920584091816,
+      "grad_norm": 1043.591930171376,
+      "learning_rate": 1.997183671539098e-05,
+      "loss": 442.5342,
+      "step": 1750
+    },
+    {
+      "epoch": 0.03384924584457234,
+      "grad_norm": 1079.8265419064423,
+      "learning_rate": 1.9971377133732567e-05,
+      "loss": 439.2012,
+      "step": 1760
+    },
+    {
+      "epoch": 0.03404157110505287,
+      "grad_norm": 823.2231568926577,
+      "learning_rate": 1.997091383790086e-05,
+      "loss": 451.7246,
+      "step": 1770
+    },
+    {
+      "epoch": 0.03423389636553339,
+      "grad_norm": 820.2895789250954,
+      "learning_rate": 1.997044682806842e-05,
+      "loss": 430.022,
+      "step": 1780
+    },
+    {
+      "epoch": 0.034426221626013914,
+      "grad_norm": 907.6838320401715,
+      "learning_rate": 1.9969976104409202e-05,
+      "loss": 440.6493,
+      "step": 1790
+    },
+    {
+      "epoch": 0.03461854688649444,
+      "grad_norm": 1222.668516608212,
+      "learning_rate": 1.9969501667098547e-05,
+      "loss": 448.0128,
+      "step": 1800
+    },
+    {
+      "epoch": 0.03481087214697497,
+      "grad_norm": 1022.7512877024485,
+      "learning_rate": 1.996902351631317e-05,
+      "loss": 431.8569,
+      "step": 1810
+    },
+    {
+      "epoch": 0.035003197407455486,
+      "grad_norm": 976.8999279841343,
+      "learning_rate": 1.996854165223118e-05,
+      "loss": 430.3105,
+      "step": 1820
+    },
+    {
+      "epoch": 0.03519552266793601,
+      "grad_norm": 994.2191083156916,
+      "learning_rate": 1.9968056075032058e-05,
+      "loss": 428.3338,
+      "step": 1830
+    },
+    {
+      "epoch": 0.03538784792841654,
+      "grad_norm": 927.5937101726631,
+      "learning_rate": 1.9967566784896676e-05,
+      "loss": 422.1842,
+      "step": 1840
+    },
+    {
+      "epoch": 0.035580173188897064,
+      "grad_norm": 804.752796715029,
+      "learning_rate": 1.996707378200729e-05,
+      "loss": 420.1022,
+      "step": 1850
+    },
+    {
+      "epoch": 0.03577249844937759,
+      "grad_norm": 875.2326312461167,
+      "learning_rate": 1.9966577066547526e-05,
+      "loss": 428.2064,
+      "step": 1860
+    },
+    {
+      "epoch": 0.03596482370985811,
+      "grad_norm": 892.9573667028674,
+      "learning_rate": 1.9966076638702412e-05,
+      "loss": 437.3851,
+      "step": 1870
+    },
+    {
+      "epoch": 0.036157148970338636,
+      "grad_norm": 792.2376623578725,
+      "learning_rate": 1.9965572498658346e-05,
+      "loss": 433.6019,
+      "step": 1880
+    },
+    {
+      "epoch": 0.03634947423081916,
+      "grad_norm": 952.014487394457,
+      "learning_rate": 1.996506464660311e-05,
+      "loss": 431.6811,
+      "step": 1890
+    },
+    {
+      "epoch": 0.03654179949129969,
+      "grad_norm": 1041.9240608157052,
+      "learning_rate": 1.9964553082725873e-05,
+      "loss": 425.2496,
+      "step": 1900
+    },
+    {
+      "epoch": 0.03673412475178021,
+      "grad_norm": 1226.2781807241984,
+      "learning_rate": 1.996403780721718e-05,
+      "loss": 427.2512,
+      "step": 1910
+    },
+    {
+      "epoch": 0.036926450012260734,
+      "grad_norm": 1252.631493328946,
+      "learning_rate": 1.9963518820268968e-05,
+      "loss": 437.5074,
+      "step": 1920
+    },
+    {
+      "epoch": 0.03711877527274126,
+      "grad_norm": 1019.2683176579892,
+      "learning_rate": 1.996299612207455e-05,
+      "loss": 425.1087,
+      "step": 1930
+    },
+    {
+      "epoch": 0.037311100533221786,
+      "grad_norm": 946.588907288007,
+      "learning_rate": 1.9962469712828613e-05,
+      "loss": 436.6689,
+      "step": 1940
+    },
+    {
+      "epoch": 0.03750342579370231,
+      "grad_norm": 972.8227997856911,
+      "learning_rate": 1.996193959272725e-05,
+      "loss": 413.2797,
+      "step": 1950
+    },
+    {
+      "epoch": 0.03769575105418283,
+      "grad_norm": 850.4193662342182,
+      "learning_rate": 1.9961405761967914e-05,
+      "loss": 424.1742,
+      "step": 1960
+    },
+    {
+      "epoch": 0.03788807631466336,
+      "grad_norm": 1050.2895049038134,
+      "learning_rate": 1.996086822074945e-05,
+      "loss": 419.6982,
+      "step": 1970
+    },
+    {
+      "epoch": 0.038080401575143884,
+      "grad_norm": 1058.8252707233637,
+      "learning_rate": 1.996032696927208e-05,
+      "loss": 428.4798,
+      "step": 1980
+    },
+    {
+      "epoch": 0.03827272683562441,
+      "grad_norm": 831.6684934800379,
+      "learning_rate": 1.9959782007737418e-05,
+      "loss": 433.5806,
+      "step": 1990
+    },
+    {
+      "epoch": 0.038465052096104936,
+      "grad_norm": 955.7162157930247,
+      "learning_rate": 1.9959233336348452e-05,
+      "loss": 413.9336,
+      "step": 2000
+    },
+    {
+      "epoch": 0.038657377356585455,
+      "grad_norm": 1099.209046114368,
+      "learning_rate": 1.9958680955309546e-05,
+      "loss": 428.3293,
+      "step": 2010
+    },
+    {
+      "epoch": 0.03884970261706598,
+      "grad_norm": 1021.2233849816279,
+      "learning_rate": 1.9958124864826457e-05,
+      "loss": 430.3237,
+      "step": 2020
+    },
+    {
+      "epoch": 0.03904202787754651,
+      "grad_norm": 976.7190757307437,
+      "learning_rate": 1.9957565065106318e-05,
+      "loss": 415.4346,
+      "step": 2030
+    },
+    {
+      "epoch": 0.039234353138027034,
+      "grad_norm": 1054.2423629433902,
+      "learning_rate": 1.9957001556357652e-05,
+      "loss": 419.6495,
+      "step": 2040
+    },
+    {
+      "epoch": 0.03942667839850755,
+      "grad_norm": 1070.0019862922163,
+      "learning_rate": 1.995643433879035e-05,
+      "loss": 412.5856,
+      "step": 2050
+    },
+    {
+      "epoch": 0.03961900365898808,
+      "grad_norm": 1482.1736202677944,
+      "learning_rate": 1.9955863412615693e-05,
+      "loss": 428.2741,
+      "step": 2060
+    },
+    {
+      "epoch": 0.039811328919468605,
+      "grad_norm": 1073.4692781718343,
+      "learning_rate": 1.9955288778046338e-05,
+      "loss": 426.4848,
+      "step": 2070
+    },
+    {
+      "epoch": 0.04000365417994913,
+      "grad_norm": 1013.6713168858257,
+      "learning_rate": 1.995471043529633e-05,
+      "loss": 418.8785,
+      "step": 2080
+    },
+    {
+      "epoch": 0.04019597944042966,
+      "grad_norm": 837.5462305927567,
+      "learning_rate": 1.99541283845811e-05,
+      "loss": 418.0007,
+      "step": 2090
+    },
+    {
+      "epoch": 0.04038830470091018,
+      "grad_norm": 857.5303704163729,
+      "learning_rate": 1.9953542626117437e-05,
+      "loss": 422.5934,
+      "step": 2100
+    },
+    {
+      "epoch": 0.0405806299613907,
+      "grad_norm": 869.4962463265555,
+      "learning_rate": 1.9952953160123537e-05,
+      "loss": 426.6799,
+      "step": 2110
+    },
+    {
+      "epoch": 0.04077295522187123,
+      "grad_norm": 949.3244808560535,
+      "learning_rate": 1.995235998681896e-05,
+      "loss": 422.9911,
+      "step": 2120
+    },
+    {
+      "epoch": 0.040965280482351756,
+      "grad_norm": 966.6456323863318,
+      "learning_rate": 1.9951763106424658e-05,
+      "loss": 423.5903,
+      "step": 2130
+    },
+    {
+      "epoch": 0.041157605742832275,
+      "grad_norm": 919.3932010760028,
+      "learning_rate": 1.9951162519162962e-05,
+      "loss": 417.3357,
+      "step": 2140
+    },
+    {
+      "epoch": 0.0413499310033128,
+      "grad_norm": 875.4564352669491,
+      "learning_rate": 1.9950558225257574e-05,
+      "loss": 419.633,
+      "step": 2150
+    },
+    {
+      "epoch": 0.04154225626379333,
+      "grad_norm": 786.7485120612257,
+      "learning_rate": 1.9949950224933584e-05,
+      "loss": 417.6999,
+      "step": 2160
+    },
+    {
+      "epoch": 0.04173458152427385,
+      "grad_norm": 868.8891187995704,
+      "learning_rate": 1.994933851841747e-05,
+      "loss": 416.0338,
+      "step": 2170
+    },
+    {
+      "epoch": 0.04192690678475438,
+      "grad_norm": 892.5026500827428,
+      "learning_rate": 1.994872310593707e-05,
+      "loss": 416.687,
+      "step": 2180
+    },
+    {
+      "epoch": 0.0421192320452349,
+      "grad_norm": 970.6300400036097,
+      "learning_rate": 1.994810398772162e-05,
+      "loss": 418.6071,
+      "step": 2190
+    },
+    {
+      "epoch": 0.042311557305715425,
+      "grad_norm": 838.8530553637123,
+      "learning_rate": 1.994748116400174e-05,
+      "loss": 413.7869,
+      "step": 2200
+    },
+    {
+      "epoch": 0.04250388256619595,
+      "grad_norm": 1024.561778936901,
+      "learning_rate": 1.994685463500941e-05,
+      "loss": 418.3925,
+      "step": 2210
+    },
+    {
+      "epoch": 0.04269620782667648,
+      "grad_norm": 862.8999327027489,
+      "learning_rate": 1.9946224400978006e-05,
+      "loss": 416.8202,
+      "step": 2220
+    },
+    {
+      "epoch": 0.042888533087157,
+      "grad_norm": 864.6845014807649,
+      "learning_rate": 1.994559046214228e-05,
+      "loss": 415.3689,
+      "step": 2230
+    },
+    {
+      "epoch": 0.04308085834763752,
+      "grad_norm": 1207.8281915584857,
+      "learning_rate": 1.9944952818738366e-05,
+      "loss": 424.385,
+      "step": 2240
+    },
+    {
+      "epoch": 0.04327318360811805,
+      "grad_norm": 856.9755440271097,
+      "learning_rate": 1.9944311471003775e-05,
+      "loss": 427.1351,
+      "step": 2250
+    },
+    {
+      "epoch": 0.043465508868598575,
+      "grad_norm": 797.2202312561943,
+      "learning_rate": 1.9943666419177392e-05,
+      "loss": 401.6336,
+      "step": 2260
+    },
+    {
+      "epoch": 0.0436578341290791,
+      "grad_norm": 840.993989742356,
+      "learning_rate": 1.9943017663499497e-05,
+      "loss": 413.9854,
+      "step": 2270
+    },
+    {
+      "epoch": 0.04385015938955962,
+      "grad_norm": 926.8012079644898,
+      "learning_rate": 1.9942365204211734e-05,
+      "loss": 413.2362,
+      "step": 2280
+    },
+    {
+      "epoch": 0.044042484650040147,
+      "grad_norm": 24676.784984066762,
+      "learning_rate": 1.9941709041557134e-05,
+      "loss": 419.889,
+      "step": 2290
+    },
+    {
+      "epoch": 0.04423480991052067,
+      "grad_norm": 946.0603847587324,
+      "learning_rate": 1.994104917578011e-05,
+      "loss": 435.1779,
+      "step": 2300
+    },
+    {
+      "epoch": 0.0444271351710012,
+      "grad_norm": 962.153063049471,
+      "learning_rate": 1.9940385607126456e-05,
+      "loss": 410.552,
+      "step": 2310
+    },
+    {
+      "epoch": 0.044619460431481725,
+      "grad_norm": 965.2673431893036,
+      "learning_rate": 1.9939718335843326e-05,
+      "loss": 420.7652,
+      "step": 2320
+    },
+    {
+      "epoch": 0.044811785691962244,
+      "grad_norm": 1268.9167278481075,
+      "learning_rate": 1.9939047362179283e-05,
+      "loss": 418.1449,
+      "step": 2330
+    },
+    {
+      "epoch": 0.04500411095244277,
+      "grad_norm": 836.5103474682797,
+      "learning_rate": 1.9938372686384244e-05,
+      "loss": 403.4094,
+      "step": 2340
+    },
+    {
+      "epoch": 0.0451964362129233,
+      "grad_norm": 753.7340795731817,
+      "learning_rate": 1.9937694308709514e-05,
+      "loss": 409.9182,
+      "step": 2350
+    },
+    {
+      "epoch": 0.04538876147340382,
+      "grad_norm": 821.820710879436,
+      "learning_rate": 1.9937012229407783e-05,
+      "loss": 419.3094,
+      "step": 2360
+    },
+    {
+      "epoch": 0.04558108673388434,
+      "grad_norm": 851.0589297896992,
+      "learning_rate": 1.9936326448733107e-05,
+      "loss": 416.9893,
+      "step": 2370
+    },
+    {
+      "epoch": 0.04577341199436487,
+      "grad_norm": 1075.1200638865676,
+      "learning_rate": 1.9935636966940938e-05,
+      "loss": 405.4084,
+      "step": 2380
+    },
+    {
+      "epoch": 0.045965737254845394,
+      "grad_norm": 956.2910990830981,
+      "learning_rate": 1.993494378428809e-05,
+      "loss": 416.4961,
+      "step": 2390
+    },
+    {
+      "epoch": 0.04615806251532592,
+      "grad_norm": 815.6936262963642,
+      "learning_rate": 1.9934246901032764e-05,
+      "loss": 402.3952,
+      "step": 2400
+    },
+    {
+      "epoch": 0.04635038777580645,
+      "grad_norm": 2360.759105306595,
+      "learning_rate": 1.9933546317434536e-05,
+      "loss": 413.8724,
+      "step": 2410
+    },
+    {
+      "epoch": 0.046542713036286966,
+      "grad_norm": 943.481299935079,
+      "learning_rate": 1.9932842033754363e-05,
+      "loss": 419.4986,
+      "step": 2420
+    },
+    {
+      "epoch": 0.04673503829676749,
+      "grad_norm": 846.2634230104537,
+      "learning_rate": 1.9932134050254578e-05,
+      "loss": 413.2273,
+      "step": 2430
+    },
+    {
+      "epoch": 0.04692736355724802,
+      "grad_norm": 871.5530094276198,
+      "learning_rate": 1.9931422367198893e-05,
+      "loss": 410.6645,
+      "step": 2440
+    },
+    {
+      "epoch": 0.047119688817728544,
+      "grad_norm": 918.4069177694724,
+      "learning_rate": 1.9930706984852403e-05,
+      "loss": 407.7853,
+      "step": 2450
+    },
+    {
+      "epoch": 0.04731201407820907,
+      "grad_norm": 828.3693733593514,
+      "learning_rate": 1.992998790348157e-05,
+      "loss": 407.0076,
+      "step": 2460
+    },
+    {
+      "epoch": 0.04750433933868959,
+      "grad_norm": 802.7119429227891,
+      "learning_rate": 1.992926512335424e-05,
+      "loss": 417.0812,
+      "step": 2470
+    },
+    {
+      "epoch": 0.047696664599170116,
+      "grad_norm": 783.1281826299662,
+      "learning_rate": 1.992853864473964e-05,
+      "loss": 404.1433,
+      "step": 2480
+    },
+    {
+      "epoch": 0.04788898985965064,
+      "grad_norm": 732.2406768577065,
+      "learning_rate": 1.992780846790837e-05,
+      "loss": 411.2688,
+      "step": 2490
+    },
+    {
+      "epoch": 0.04808131512013117,
+      "grad_norm": 975.6063398004994,
+      "learning_rate": 1.9927074593132412e-05,
+      "loss": 408.5953,
+      "step": 2500
+    },
+    {
+      "epoch": 0.04827364038061169,
+      "grad_norm": 832.3209682983506,
+      "learning_rate": 1.9926337020685114e-05,
+      "loss": 417.095,
+      "step": 2510
+    },
+    {
+      "epoch": 0.048465965641092214,
+      "grad_norm": 1000.2467965932198,
+      "learning_rate": 1.992559575084122e-05,
+      "loss": 415.1272,
+      "step": 2520
+    },
+    {
+      "epoch": 0.04865829090157274,
+      "grad_norm": 843.4947995629191,
+      "learning_rate": 1.9924850783876832e-05,
+      "loss": 409.7536,
+      "step": 2530
+    },
+    {
+      "epoch": 0.048850616162053266,
+      "grad_norm": 909.1986478819955,
+      "learning_rate": 1.992410212006944e-05,
+      "loss": 411.6839,
+      "step": 2540
+    },
+    {
+      "epoch": 0.04904294142253379,
+      "grad_norm": 727.4074209194242,
+      "learning_rate": 1.9923349759697915e-05,
+      "loss": 402.9094,
+      "step": 2550
+    },
+    {
+      "epoch": 0.04923526668301431,
+      "grad_norm": 784.9272579755053,
+      "learning_rate": 1.992259370304249e-05,
+      "loss": 399.2156,
+      "step": 2560
+    },
+    {
+      "epoch": 0.04942759194349484,
+      "grad_norm": 801.5705870825869,
+      "learning_rate": 1.992183395038479e-05,
+      "loss": 408.6303,
+      "step": 2570
+    },
+    {
+      "epoch": 0.049619917203975364,
+      "grad_norm": 776.0226228782816,
+      "learning_rate": 1.9921070502007808e-05,
+      "loss": 406.1699,
+      "step": 2580
+    },
+    {
+      "epoch": 0.04981224246445589,
+      "grad_norm": 791.2287159718755,
+      "learning_rate": 1.9920303358195916e-05,
+      "loss": 401.3249,
+      "step": 2590
+    },
+    {
+      "epoch": 0.05000456772493641,
+      "grad_norm": 859.5715851591746,
+      "learning_rate": 1.9919532519234858e-05,
+      "loss": 406.9079,
+      "step": 2600
+    },
+    {
+      "epoch": 0.050196892985416935,
+      "grad_norm": 950.0375357774553,
+      "learning_rate": 1.9918757985411767e-05,
+      "loss": 398.9475,
+      "step": 2610
+    },
+    {
+      "epoch": 0.05038921824589746,
+      "grad_norm": 3697.0217417772647,
+      "learning_rate": 1.991797975701514e-05,
+      "loss": 413.058,
+      "step": 2620
+    },
+    {
+      "epoch": 0.05058154350637799,
+      "grad_norm": 21484.157037856894,
+      "learning_rate": 1.9917197834334858e-05,
+      "loss": 457.0581,
+      "step": 2630
+    },
+    {
+      "epoch": 0.050773868766858514,
+      "grad_norm": 840.8131994500599,
+      "learning_rate": 1.9916412217662162e-05,
+      "loss": 395.2674,
+      "step": 2640
+    },
+    {
+      "epoch": 0.05096619402733903,
+      "grad_norm": 848.3677713278562,
+      "learning_rate": 1.9915622907289695e-05,
+      "loss": 393.97,
+      "step": 2650
+    },
+    {
+      "epoch": 0.05115851928781956,
+      "grad_norm": 773.3080723966837,
+      "learning_rate": 1.9914829903511458e-05,
+      "loss": 397.9258,
+      "step": 2660
+    },
+    {
+      "epoch": 0.051350844548300086,
+      "grad_norm": 722.654539565636,
+      "learning_rate": 1.9914033206622828e-05,
+      "loss": 408.9118,
+      "step": 2670
+    },
+    {
+      "epoch": 0.05154316980878061,
+      "grad_norm": 766.6787271546868,
+      "learning_rate": 1.9913232816920565e-05,
+      "loss": 405.4553,
+      "step": 2680
+    },
+    {
+      "epoch": 0.05173549506926114,
+      "grad_norm": 820.7846023812327,
+      "learning_rate": 1.99124287347028e-05,
+      "loss": 407.4249,
+      "step": 2690
+    },
+    {
+      "epoch": 0.05192782032974166,
+      "grad_norm": 696.7079214788865,
+      "learning_rate": 1.991162096026904e-05,
+      "loss": 412.3783,
+      "step": 2700
+    },
+    {
+      "epoch": 0.05212014559022218,
+      "grad_norm": 995.1881132131933,
+      "learning_rate": 1.9910809493920172e-05,
+      "loss": 405.0052,
+      "step": 2710
+    },
+    {
+      "epoch": 0.05231247085070271,
+      "grad_norm": 900.7112278783013,
+      "learning_rate": 1.990999433595845e-05,
+      "loss": 406.1632,
+      "step": 2720
+    },
+    {
+      "epoch": 0.052504796111183236,
+      "grad_norm": 882.6939502542203,
+      "learning_rate": 1.99091754866875e-05,
+      "loss": 406.4494,
+      "step": 2730
+    },
+    {
+      "epoch": 0.052697121371663755,
+      "grad_norm": 793.843854468107,
+      "learning_rate": 1.990835294641234e-05,
+      "loss": 403.5715,
+      "step": 2740
+    },
+    {
+      "epoch": 0.05288944663214428,
+      "grad_norm": 768.9838928381889,
+      "learning_rate": 1.990752671543935e-05,
+      "loss": 402.1033,
+      "step": 2750
+    },
+    {
+      "epoch": 0.05308177189262481,
+      "grad_norm": 763.1077293076072,
+      "learning_rate": 1.9906696794076284e-05,
+      "loss": 403.7594,
+      "step": 2760
+    },
+    {
+      "epoch": 0.05327409715310533,
+      "grad_norm": 708.1979175398459,
+      "learning_rate": 1.9905863182632285e-05,
+      "loss": 403.3684,
+      "step": 2770
+    },
+    {
+      "epoch": 0.05346642241358586,
+      "grad_norm": 780.7564832522181,
+      "learning_rate": 1.990502588141784e-05,
+      "loss": 403.7348,
+      "step": 2780
+    },
+    {
+      "epoch": 0.05365874767406638,
+      "grad_norm": 788.2704196253177,
+      "learning_rate": 1.990418489074485e-05,
+      "loss": 390.1319,
+      "step": 2790
+    },
+    {
+      "epoch": 0.053851072934546905,
+      "grad_norm": 777.7139743744849,
+      "learning_rate": 1.9903340210926555e-05,
+      "loss": 391.7088,
+      "step": 2800
+    },
+    {
+      "epoch": 0.05404339819502743,
+      "grad_norm": 779.8392157540648,
+      "learning_rate": 1.9902491842277592e-05,
+      "loss": 409.1859,
+      "step": 2810
+    },
+    {
+      "epoch": 0.05423572345550796,
+      "grad_norm": 939.190832945537,
+      "learning_rate": 1.9901639785113967e-05,
+      "loss": 402.1446,
+      "step": 2820
+    },
+    {
+      "epoch": 0.05442804871598848,
+      "grad_norm": 798.106765825879,
+      "learning_rate": 1.990078403975305e-05,
+      "loss": 411.2067,
+      "step": 2830
+    },
+    {
+      "epoch": 0.054620373976469,
+      "grad_norm": 744.7205098821862,
+      "learning_rate": 1.9899924606513593e-05,
+      "loss": 397.5615,
+      "step": 2840
+    },
+    {
+      "epoch": 0.05481269923694953,
+      "grad_norm": 828.5586351859567,
+      "learning_rate": 1.9899061485715726e-05,
+      "loss": 392.6941,
+      "step": 2850
+    },
+    {
+      "epoch": 0.055005024497430055,
+      "grad_norm": 743.7545184832085,
+      "learning_rate": 1.9898194677680943e-05,
+      "loss": 401.0686,
+      "step": 2860
+    },
+    {
+      "epoch": 0.05519734975791058,
+      "grad_norm": 787.9522241003444,
+      "learning_rate": 1.9897324182732118e-05,
+      "loss": 396.134,
+      "step": 2870
+    },
+    {
+      "epoch": 0.0553896750183911,
+      "grad_norm": 698.5125841213974,
+      "learning_rate": 1.989645000119349e-05,
+      "loss": 393.7889,
+      "step": 2880
+    },
+    {
+      "epoch": 0.05558200027887163,
+      "grad_norm": 738.498781475107,
+      "learning_rate": 1.9895572133390687e-05,
+      "loss": 394.8915,
+      "step": 2890
+    },
+    {
+      "epoch": 0.05577432553935215,
+      "grad_norm": 754.9880133446674,
+      "learning_rate": 1.9894690579650694e-05,
+      "loss": 399.7915,
+      "step": 2900
+    },
+    {
+      "epoch": 0.05596665079983268,
+      "grad_norm": 838.9035752511419,
+      "learning_rate": 1.9893805340301876e-05,
+      "loss": 399.2658,
+      "step": 2910
+    },
+    {
+      "epoch": 0.0561589760603132,
+      "grad_norm": 887.0247498898594,
+      "learning_rate": 1.989291641567397e-05,
+      "loss": 395.2259,
+      "step": 2920
+    },
+    {
+      "epoch": 0.056351301320793724,
+      "grad_norm": 6533.06697295512,
+      "learning_rate": 1.9892023806098083e-05,
+      "loss": 386.8454,
+      "step": 2930
+    },
+    {
+      "epoch": 0.05654362658127425,
+      "grad_norm": 729.544176376949,
+      "learning_rate": 1.9891127511906703e-05,
+      "loss": 391.1382,
+      "step": 2940
+    },
+    {
+      "epoch": 0.05673595184175478,
+      "grad_norm": 836.7696344728582,
+      "learning_rate": 1.9890227533433685e-05,
+      "loss": 392.0289,
+      "step": 2950
+    },
+    {
+      "epoch": 0.0569282771022353,
+      "grad_norm": 949.1072841508442,
+      "learning_rate": 1.988932387101425e-05,
+      "loss": 405.9888,
+      "step": 2960
+    },
+    {
+      "epoch": 0.05712060236271582,
+      "grad_norm": 758.5882201805449,
+      "learning_rate": 1.9888416524985e-05,
+      "loss": 388.5938,
+      "step": 2970
+    },
+    {
+      "epoch": 0.05731292762319635,
+      "grad_norm": 743.6184716836035,
+      "learning_rate": 1.988750549568391e-05,
+      "loss": 399.7922,
+      "step": 2980
+    },
+    {
+      "epoch": 0.057505252883676874,
+      "grad_norm": 717.041287798182,
+      "learning_rate": 1.9886590783450317e-05,
+      "loss": 381.7268,
+      "step": 2990
+    },
+    {
+      "epoch": 0.0576975781441574,
+      "grad_norm": 861.2531062344228,
+      "learning_rate": 1.9885672388624942e-05,
+      "loss": 393.7416,
+      "step": 3000
+    },
+    {
+      "epoch": 0.05788990340463793,
+      "grad_norm": 1594.7740384574417,
+      "learning_rate": 1.9884750311549868e-05,
+      "loss": 408.4041,
+      "step": 3010
+    },
+    {
+      "epoch": 0.058082228665118446,
+      "grad_norm": 983.0430145740795,
+      "learning_rate": 1.9883824552568557e-05,
+      "loss": 398.9454,
+      "step": 3020
+    },
+    {
+      "epoch": 0.05827455392559897,
+      "grad_norm": 683.6360021672427,
+      "learning_rate": 1.9882895112025835e-05,
+      "loss": 404.4664,
+      "step": 3030
+    },
+    {
+      "epoch": 0.0584668791860795,
+      "grad_norm": 1104.3304960459254,
+      "learning_rate": 1.9881961990267906e-05,
+      "loss": 385.0514,
+      "step": 3040
+    },
+    {
+      "epoch": 0.058659204446560025,
+      "grad_norm": 1031.366504909178,
+      "learning_rate": 1.988102518764234e-05,
+      "loss": 397.696,
+      "step": 3050
+    },
+    {
+      "epoch": 0.058851529707040544,
+      "grad_norm": 951.7019853471531,
+      "learning_rate": 1.9880084704498084e-05,
+      "loss": 393.2042,
+      "step": 3060
+    },
+    {
+      "epoch": 0.05904385496752107,
+      "grad_norm": 1037.7212633536412,
+      "learning_rate": 1.987914054118545e-05,
+      "loss": 404.7147,
+      "step": 3070
+    },
+    {
+      "epoch": 0.059236180228001596,
+      "grad_norm": 890.2787056501567,
+      "learning_rate": 1.9878192698056125e-05,
+      "loss": 394.0958,
+      "step": 3080
+    },
+    {
+      "epoch": 0.05942850548848212,
+      "grad_norm": 850.4448760599917,
+      "learning_rate": 1.9877241175463165e-05,
+      "loss": 391.8196,
+      "step": 3090
+    },
+    {
+      "epoch": 0.05962083074896265,
+      "grad_norm": 768.3356728445398,
+      "learning_rate": 1.9876285973760993e-05,
+      "loss": 399.0993,
+      "step": 3100
+    },
+    {
+      "epoch": 0.05981315600944317,
+      "grad_norm": 792.6694642669811,
+      "learning_rate": 1.9875327093305405e-05,
+      "loss": 393.3022,
+      "step": 3110
+    },
+    {
+      "epoch": 0.060005481269923694,
+      "grad_norm": 866.5568098952982,
+      "learning_rate": 1.9874364534453577e-05,
+      "loss": 392.7167,
+      "step": 3120
+    },
+    {
+      "epoch": 0.06019780653040422,
+      "grad_norm": 845.036169611911,
+      "learning_rate": 1.9873398297564036e-05,
+      "loss": 395.2803,
+      "step": 3130
+    },
+    {
+      "epoch": 0.060390131790884746,
+      "grad_norm": 751.8245712608418,
+      "learning_rate": 1.9872428382996697e-05,
+      "loss": 399.1471,
+      "step": 3140
+    },
+    {
+      "epoch": 0.060582457051365265,
+      "grad_norm": 767.234381644173,
+      "learning_rate": 1.987145479111283e-05,
+      "loss": 385.6532,
+      "step": 3150
+    },
+    {
+      "epoch": 0.06077478231184579,
+      "grad_norm": 1181.862494733402,
+      "learning_rate": 1.987047752227509e-05,
+      "loss": 396.884,
+      "step": 3160
+    },
+    {
+      "epoch": 0.06096710757232632,
+      "grad_norm": 685.2155147894781,
+      "learning_rate": 1.9869496576847488e-05,
+      "loss": 382.569,
+      "step": 3170
+    },
+    {
+      "epoch": 0.061159432832806844,
+      "grad_norm": 791.469058721252,
+      "learning_rate": 1.9868511955195407e-05,
+      "loss": 391.4913,
+      "step": 3180
+    },
+    {
+      "epoch": 0.06135175809328737,
+      "grad_norm": 930.5923387819682,
+      "learning_rate": 1.986752365768561e-05,
+      "loss": 373.7412,
+      "step": 3190
+    },
+    {
+      "epoch": 0.06154408335376789,
+      "grad_norm": 949.0032349730969,
+      "learning_rate": 1.986653168468622e-05,
+      "loss": 391.077,
+      "step": 3200
+    },
+    {
+      "epoch": 0.061736408614248416,
+      "grad_norm": 934.8037434742106,
+      "learning_rate": 1.9865536036566727e-05,
+      "loss": 389.2291,
+      "step": 3210
+    },
+    {
+      "epoch": 0.06192873387472894,
+      "grad_norm": 826.8212148070362,
+      "learning_rate": 1.9864536713697992e-05,
+      "loss": 402.5538,
+      "step": 3220
+    },
+    {
+      "epoch": 0.06212105913520947,
+      "grad_norm": 858.983103558221,
+      "learning_rate": 1.986353371645225e-05,
+      "loss": 399.0848,
+      "step": 3230
+    },
+    {
+      "epoch": 0.062313384395689994,
+      "grad_norm": 733.6433381123185,
+      "learning_rate": 1.9862527045203105e-05,
+      "loss": 392.7022,
+      "step": 3240
+    },
+    {
+      "epoch": 0.06250570965617051,
+      "grad_norm": 752.8523887221875,
+      "learning_rate": 1.986151670032552e-05,
+      "loss": 388.4725,
+      "step": 3250
+    },
+    {
+      "epoch": 0.06269803491665105,
+      "grad_norm": 1020.4632534326246,
+      "learning_rate": 1.986050268219583e-05,
+      "loss": 386.5425,
+      "step": 3260
+    },
+    {
+      "epoch": 0.06289036017713157,
+      "grad_norm": 896.9055389206998,
+      "learning_rate": 1.9859484991191742e-05,
+      "loss": 397.7183,
+      "step": 3270
+    },
+    {
+      "epoch": 0.06308268543761208,
+      "grad_norm": 795.1442032924809,
+      "learning_rate": 1.985846362769233e-05,
+      "loss": 400.3567,
+      "step": 3280
+    },
+    {
+      "epoch": 0.06327501069809262,
+      "grad_norm": 828.0036465962801,
+      "learning_rate": 1.9857438592078034e-05,
+      "loss": 388.7515,
+      "step": 3290
+    },
+    {
+      "epoch": 0.06346733595857314,
+      "grad_norm": 764.7935820700304,
+      "learning_rate": 1.9856409884730667e-05,
+      "loss": 393.2867,
+      "step": 3300
+    },
+    {
+      "epoch": 0.06365966121905367,
+      "grad_norm": 782.433216331679,
+      "learning_rate": 1.98553775060334e-05,
+      "loss": 384.751,
+      "step": 3310
+    },
+    {
+      "epoch": 0.06385198647953419,
+      "grad_norm": 770.5239118718077,
+      "learning_rate": 1.9854341456370777e-05,
+      "loss": 379.5989,
+      "step": 3320
+    },
+    {
+      "epoch": 0.06404431174001471,
+      "grad_norm": 830.1033045195829,
+      "learning_rate": 1.9853301736128712e-05,
+      "loss": 390.4898,
+      "step": 3330
+    },
+    {
+      "epoch": 0.06423663700049524,
+      "grad_norm": 926.2149843676547,
+      "learning_rate": 1.9852258345694486e-05,
+      "loss": 400.0977,
+      "step": 3340
+    },
+    {
+      "epoch": 0.06442896226097576,
+      "grad_norm": 809.9690346247447,
+      "learning_rate": 1.9851211285456738e-05,
+      "loss": 389.6271,
+      "step": 3350
+    },
+    {
+      "epoch": 0.06462128752145628,
+      "grad_norm": 755.2125565665904,
+      "learning_rate": 1.9850160555805485e-05,
+      "loss": 401.8327,
+      "step": 3360
+    },
+    {
+      "epoch": 0.06481361278193681,
+      "grad_norm": 741.8845538744961,
+      "learning_rate": 1.9849106157132105e-05,
+      "loss": 399.9646,
+      "step": 3370
+    },
+    {
+      "epoch": 0.06500593804241733,
+      "grad_norm": 686.883082073912,
+      "learning_rate": 1.9848048089829347e-05,
+      "loss": 394.8569,
+      "step": 3380
+    },
+    {
+      "epoch": 0.06519826330289787,
+      "grad_norm": 736.2050614077401,
+      "learning_rate": 1.9846986354291324e-05,
+      "loss": 383.8164,
+      "step": 3390
+    },
+    {
+      "epoch": 0.06539058856337839,
+      "grad_norm": 803.9426233603608,
+      "learning_rate": 1.9845920950913506e-05,
+      "loss": 400.4979,
+      "step": 3400
+    },
+    {
+      "epoch": 0.0655829138238589,
+      "grad_norm": 926.4258115884888,
+      "learning_rate": 1.9844851880092748e-05,
+      "loss": 380.4864,
+      "step": 3410
+    },
+    {
+      "epoch": 0.06577523908433944,
+      "grad_norm": 886.5657112848164,
+      "learning_rate": 1.9843779142227258e-05,
+      "loss": 388.6589,
+      "step": 3420
+    },
+    {
+      "epoch": 0.06596756434481996,
+      "grad_norm": 907.7194073038047,
+      "learning_rate": 1.984270273771661e-05,
+      "loss": 388.0827,
+      "step": 3430
+    },
+    {
+      "epoch": 0.06615988960530049,
+      "grad_norm": 874.0977699889922,
+      "learning_rate": 1.9841622666961756e-05,
+      "loss": 394.0443,
+      "step": 3440
+    },
+    {
+      "epoch": 0.06635221486578101,
+      "grad_norm": 744.3752919633399,
+      "learning_rate": 1.9840538930364992e-05,
+      "loss": 391.6839,
+      "step": 3450
+    },
+    {
+      "epoch": 0.06654454012626153,
+      "grad_norm": 748.822685279414,
+      "learning_rate": 1.983945152833e-05,
+      "loss": 378.0743,
+      "step": 3460
+    },
+    {
+      "epoch": 0.06673686538674206,
+      "grad_norm": 877.9787494251104,
+      "learning_rate": 1.9838360461261817e-05,
+      "loss": 390.5966,
+      "step": 3470
+    },
+    {
+      "epoch": 0.06692919064722258,
+      "grad_norm": 670.5103547705536,
+      "learning_rate": 1.9837265729566853e-05,
+      "loss": 378.118,
+      "step": 3480
+    },
+    {
+      "epoch": 0.06712151590770311,
+      "grad_norm": 755.2276168722524,
+      "learning_rate": 1.9836167333652866e-05,
+      "loss": 394.8704,
+      "step": 3490
+    },
+    {
+      "epoch": 0.06731384116818363,
+      "grad_norm": 864.8964584726651,
+      "learning_rate": 1.9835065273929002e-05,
+      "loss": 384.7451,
+      "step": 3500
+    },
+    {
+      "epoch": 0.06750616642866415,
+      "grad_norm": 711.2318778664892,
+      "learning_rate": 1.9833959550805754e-05,
+      "loss": 387.1255,
+      "step": 3510
+    },
+    {
+      "epoch": 0.06769849168914469,
+      "grad_norm": 904.7148051183538,
+      "learning_rate": 1.9832850164694983e-05,
+      "loss": 383.078,
+      "step": 3520
+    },
+    {
+      "epoch": 0.0678908169496252,
+      "grad_norm": 1449.0431850047282,
+      "learning_rate": 1.9831737116009924e-05,
+      "loss": 384.6444,
+      "step": 3530
+    },
+    {
+      "epoch": 0.06808314221010574,
+      "grad_norm": 796.7373246812646,
+      "learning_rate": 1.9830620405165164e-05,
+      "loss": 382.1027,
+      "step": 3540
+    },
+    {
+      "epoch": 0.06827546747058626,
+      "grad_norm": 742.1817234384118,
+      "learning_rate": 1.982950003257666e-05,
+      "loss": 394.5984,
+      "step": 3550
+    },
+    {
+      "epoch": 0.06846779273106678,
+      "grad_norm": 735.6763919798921,
+      "learning_rate": 1.9828375998661738e-05,
+      "loss": 388.0653,
+      "step": 3560
+    },
+    {
+      "epoch": 0.06866011799154731,
+      "grad_norm": 705.0031427898053,
+      "learning_rate": 1.9827248303839073e-05,
+      "loss": 364.7262,
+      "step": 3570
+    },
+    {
+      "epoch": 0.06885244325202783,
+      "grad_norm": 889.1804309525011,
+      "learning_rate": 1.982611694852872e-05,
+      "loss": 379.9477,
+      "step": 3580
+    },
+    {
+      "epoch": 0.06904476851250835,
+      "grad_norm": 841.0288674010438,
+      "learning_rate": 1.9824981933152087e-05,
+      "loss": 373.4608,
+      "step": 3590
+    },
+    {
+      "epoch": 0.06923709377298888,
+      "grad_norm": 730.5043399291195,
+      "learning_rate": 1.9823843258131945e-05,
+      "loss": 381.3542,
+      "step": 3600
+    },
+    {
+      "epoch": 0.0694294190334694,
+      "grad_norm": 967.5864584721487,
+      "learning_rate": 1.9822700923892438e-05,
+      "loss": 398.1269,
+      "step": 3610
+    },
+    {
+      "epoch": 0.06962174429394993,
+      "grad_norm": 845.1950823294573,
+      "learning_rate": 1.9821554930859066e-05,
+      "loss": 385.2426,
+      "step": 3620
+    },
+    {
+      "epoch": 0.06981406955443045,
+      "grad_norm": 782.6785606969343,
+      "learning_rate": 1.982040527945869e-05,
+      "loss": 386.971,
+      "step": 3630
+    },
+    {
+      "epoch": 0.07000639481491097,
+      "grad_norm": 724.8900255479541,
+      "learning_rate": 1.9819251970119534e-05,
+      "loss": 390.1686,
+      "step": 3640
+    },
+    {
+      "epoch": 0.0701987200753915,
+      "grad_norm": 731.9915541707064,
+      "learning_rate": 1.9818095003271193e-05,
+      "loss": 377.8536,
+      "step": 3650
+    },
+    {
+      "epoch": 0.07039104533587202,
+      "grad_norm": 782.9719906016511,
+      "learning_rate": 1.9816934379344613e-05,
+      "loss": 382.285,
+      "step": 3660
+    },
+    {
+      "epoch": 0.07058337059635256,
+      "grad_norm": 779.6638708072722,
+      "learning_rate": 1.9815770098772108e-05,
+      "loss": 378.9326,
+      "step": 3670
+    },
+    {
+      "epoch": 0.07077569585683308,
+      "grad_norm": 775.5533482950776,
+      "learning_rate": 1.9814602161987354e-05,
+      "loss": 375.4497,
+      "step": 3680
+    },
+    {
+      "epoch": 0.0709680211173136,
+      "grad_norm": 825.1492664708319,
+      "learning_rate": 1.981343056942539e-05,
+      "loss": 376.8522,
+      "step": 3690
+    },
+    {
+      "epoch": 0.07116034637779413,
+      "grad_norm": 790.2378369347501,
+      "learning_rate": 1.9812255321522614e-05,
+      "loss": 396.3279,
+      "step": 3700
+    },
+    {
+      "epoch": 0.07135267163827465,
+      "grad_norm": 703.9794990286542,
+      "learning_rate": 1.981107641871678e-05,
+      "loss": 380.933,
+      "step": 3710
+    },
+    {
+      "epoch": 0.07154499689875518,
+      "grad_norm": 738.1836530230871,
+      "learning_rate": 1.980989386144702e-05,
+      "loss": 379.7828,
+      "step": 3720
+    },
+    {
+      "epoch": 0.0717373221592357,
+      "grad_norm": 684.3928445081146,
+      "learning_rate": 1.980870765015381e-05,
+      "loss": 386.5646,
+      "step": 3730
+    },
+    {
+      "epoch": 0.07192964741971622,
+      "grad_norm": 720.6833523018175,
+      "learning_rate": 1.9807517785278997e-05,
+      "loss": 381.8005,
+      "step": 3740
+    },
+    {
+      "epoch": 0.07212197268019675,
+      "grad_norm": 735.5504960392883,
+      "learning_rate": 1.9806324267265786e-05,
+      "loss": 380.8302,
+      "step": 3750
+    },
+    {
+      "epoch": 0.07231429794067727,
+      "grad_norm": 723.363734759545,
+      "learning_rate": 1.9805127096558742e-05,
+      "loss": 380.8795,
+      "step": 3760
+    },
+    {
+      "epoch": 0.0725066232011578,
+      "grad_norm": 779.084986349587,
+      "learning_rate": 1.980392627360379e-05,
+      "loss": 383.6675,
+      "step": 3770
+    },
+    {
+      "epoch": 0.07269894846163832,
+      "grad_norm": 924.4286417371314,
+      "learning_rate": 1.9802721798848225e-05,
+      "loss": 389.7277,
+      "step": 3780
+    },
+    {
+      "epoch": 0.07289127372211884,
+      "grad_norm": 741.5577517729787,
+      "learning_rate": 1.980151367274068e-05,
+      "loss": 374.1295,
+      "step": 3790
+    },
+    {
+      "epoch": 0.07308359898259938,
+      "grad_norm": 763.0281988934574,
+      "learning_rate": 1.9800301895731172e-05,
+      "loss": 388.0132,
+      "step": 3800
+    },
+    {
+      "epoch": 0.0732759242430799,
+      "grad_norm": 723.2651647549225,
+      "learning_rate": 1.9799086468271065e-05,
+      "loss": 382.7571,
+      "step": 3810
+    },
+    {
+      "epoch": 0.07346824950356041,
+      "grad_norm": 751.877215063999,
+      "learning_rate": 1.9797867390813086e-05,
+      "loss": 376.49,
+      "step": 3820
+    },
+    {
+      "epoch": 0.07366057476404095,
+      "grad_norm": 823.0893082654114,
+      "learning_rate": 1.9796644663811318e-05,
+      "loss": 382.0146,
+      "step": 3830
+    },
+    {
+      "epoch": 0.07385290002452147,
+      "grad_norm": 783.869557455776,
+      "learning_rate": 1.9795418287721215e-05,
+      "loss": 379.0613,
+      "step": 3840
+    },
+    {
+      "epoch": 0.074045225285002,
+      "grad_norm": 916.0368052570652,
+      "learning_rate": 1.9794188262999574e-05,
+      "loss": 388.5759,
+      "step": 3850
+    },
+    {
+      "epoch": 0.07423755054548252,
+      "grad_norm": 846.4465131910968,
+      "learning_rate": 1.979295459010456e-05,
+      "loss": 379.8879,
+      "step": 3860
+    },
+    {
+      "epoch": 0.07442987580596304,
+      "grad_norm": 735.5392846914212,
+      "learning_rate": 1.9791717269495698e-05,
+      "loss": 371.2041,
+      "step": 3870
+    },
+    {
+      "epoch": 0.07462220106644357,
+      "grad_norm": 955.2603207902944,
+      "learning_rate": 1.979047630163387e-05,
+      "loss": 374.3904,
+      "step": 3880
+    },
+    {
+      "epoch": 0.07481452632692409,
+      "grad_norm": 754.946047484617,
+      "learning_rate": 1.9789231686981313e-05,
+      "loss": 386.2628,
+      "step": 3890
+    },
+    {
+      "epoch": 0.07500685158740462,
+      "grad_norm": 892.6055943942398,
+      "learning_rate": 1.978798342600163e-05,
+      "loss": 374.2574,
+      "step": 3900
+    },
+    {
+      "epoch": 0.07519917684788514,
+      "grad_norm": 816.3551244875831,
+      "learning_rate": 1.978673151915977e-05,
+      "loss": 383.0144,
+      "step": 3910
+    },
+    {
+      "epoch": 0.07539150210836566,
+      "grad_norm": 671.0823774479612,
+      "learning_rate": 1.9785475966922055e-05,
+      "loss": 382.7799,
+      "step": 3920
+    },
+    {
+      "epoch": 0.0755838273688462,
+      "grad_norm": 792.6868282216401,
+      "learning_rate": 1.9784216769756156e-05,
+      "loss": 374.9068,
+      "step": 3930
+    },
+    {
+      "epoch": 0.07577615262932672,
+      "grad_norm": 722.6049277567629,
+      "learning_rate": 1.97829539281311e-05,
+      "loss": 372.5016,
+      "step": 3940
+    },
+    {
+      "epoch": 0.07596847788980725,
+      "grad_norm": 938.273755253152,
+      "learning_rate": 1.9781687442517278e-05,
+      "loss": 379.5824,
+      "step": 3950
+    },
+    {
+      "epoch": 0.07616080315028777,
+      "grad_norm": 943.205581022564,
+      "learning_rate": 1.9780417313386433e-05,
+      "loss": 385.3431,
+      "step": 3960
+    },
+    {
+      "epoch": 0.07635312841076829,
+      "grad_norm": 838.245722493269,
+      "learning_rate": 1.9779143541211664e-05,
+      "loss": 377.7045,
+      "step": 3970
+    },
+    {
+      "epoch": 0.07654545367124882,
+      "grad_norm": 809.1246764452217,
+      "learning_rate": 1.9777866126467436e-05,
+      "loss": 367.2468,
+      "step": 3980
+    },
+    {
+      "epoch": 0.07673777893172934,
+      "grad_norm": 811.6322890808202,
+      "learning_rate": 1.9776585069629566e-05,
+      "loss": 371.5746,
+      "step": 3990
+    },
+    {
+      "epoch": 0.07693010419220987,
+      "grad_norm": 749.3625528702271,
+      "learning_rate": 1.9775300371175225e-05,
+      "loss": 374.6235,
+      "step": 4000
+    },
+    {
+      "epoch": 0.07712242945269039,
+      "grad_norm": 927.368083067949,
+      "learning_rate": 1.9774012031582935e-05,
+      "loss": 368.9598,
+      "step": 4010
+    },
+    {
+      "epoch": 0.07731475471317091,
+      "grad_norm": 696.3606303210178,
+      "learning_rate": 1.9772720051332585e-05,
+      "loss": 375.025,
+      "step": 4020
+    },
+    {
+      "epoch": 0.07750707997365144,
+      "grad_norm": 1067.7751944482493,
+      "learning_rate": 1.977142443090542e-05,
+      "loss": 373.9631,
+      "step": 4030
+    },
+    {
+      "epoch": 0.07769940523413196,
+      "grad_norm": 796.1897032237234,
+      "learning_rate": 1.9770125170784035e-05,
+      "loss": 377.0201,
+      "step": 4040
+    },
+    {
+      "epoch": 0.07789173049461248,
+      "grad_norm": 912.7540015233076,
+      "learning_rate": 1.9768822271452385e-05,
+      "loss": 386.0567,
+      "step": 4050
+    },
+    {
+      "epoch": 0.07808405575509302,
+      "grad_norm": 804.5646961662917,
+      "learning_rate": 1.9767515733395774e-05,
+      "loss": 375.4905,
+      "step": 4060
+    },
+    {
+      "epoch": 0.07827638101557353,
+      "grad_norm": 910.9621386160426,
+      "learning_rate": 1.976620555710087e-05,
+      "loss": 378.288,
+      "step": 4070
+    },
+    {
+      "epoch": 0.07846870627605407,
+      "grad_norm": 808.0390835997099,
+      "learning_rate": 1.976489174305569e-05,
+      "loss": 376.1768,
+      "step": 4080
+    },
+    {
+      "epoch": 0.07866103153653459,
+      "grad_norm": 694.1471357236522,
+      "learning_rate": 1.9763574291749603e-05,
+      "loss": 365.676,
+      "step": 4090
+    },
+    {
+      "epoch": 0.0788533567970151,
+      "grad_norm": 669.5197035263998,
+      "learning_rate": 1.9762253203673348e-05,
+      "loss": 366.5308,
+      "step": 4100
+    },
+    {
+      "epoch": 0.07904568205749564,
+      "grad_norm": 785.2512576017631,
+      "learning_rate": 1.9760928479319003e-05,
+      "loss": 378.3666,
+      "step": 4110
+    },
+    {
+      "epoch": 0.07923800731797616,
+      "grad_norm": 688.3637184992946,
+      "learning_rate": 1.9759600119180005e-05,
+      "loss": 374.6983,
+      "step": 4120
+    },
+    {
+      "epoch": 0.07943033257845669,
+      "grad_norm": 651.3619326538491,
+      "learning_rate": 1.975826812375115e-05,
+      "loss": 369.3953,
+      "step": 4130
+    },
+    {
+      "epoch": 0.07962265783893721,
+      "grad_norm": 719.4246492086702,
+      "learning_rate": 1.9756932493528583e-05,
+      "loss": 357.6536,
+      "step": 4140
+    },
+    {
+      "epoch": 0.07981498309941773,
+      "grad_norm": 770.6968075867582,
+      "learning_rate": 1.97555932290098e-05,
+      "loss": 418.1854,
+      "step": 4150
+    },
+    {
+      "epoch": 0.08000730835989826,
+      "grad_norm": 749.0590226592501,
+      "learning_rate": 1.9754250330693658e-05,
+      "loss": 372.4671,
+      "step": 4160
+    },
+    {
+      "epoch": 0.08019963362037878,
+      "grad_norm": 609.8059454924893,
+      "learning_rate": 1.9752903799080366e-05,
+      "loss": 383.8606,
+      "step": 4170
+    },
+    {
+      "epoch": 0.08039195888085932,
+      "grad_norm": 649.5123407312859,
+      "learning_rate": 1.9751553634671485e-05,
+      "loss": 363.1326,
+      "step": 4180
+    },
+    {
+      "epoch": 0.08058428414133983,
+      "grad_norm": 682.8437842372026,
+      "learning_rate": 1.9750199837969922e-05,
+      "loss": 367.2999,
+      "step": 4190
+    },
+    {
+      "epoch": 0.08077660940182035,
+      "grad_norm": 672.827466269095,
+      "learning_rate": 1.9748842409479953e-05,
+      "loss": 374.4353,
+      "step": 4200
+    },
+    {
+      "epoch": 0.08096893466230089,
+      "grad_norm": 710.9532577309513,
+      "learning_rate": 1.9747481349707197e-05,
+      "loss": 380.4755,
+      "step": 4210
+    },
+    {
+      "epoch": 0.0811612599227814,
+      "grad_norm": 746.9490197406803,
+      "learning_rate": 1.9746116659158618e-05,
+      "loss": 374.4888,
+      "step": 4220
+    },
+    {
+      "epoch": 0.08135358518326194,
+      "grad_norm": 777.1558146388504,
+      "learning_rate": 1.9744748338342546e-05,
+      "loss": 370.0567,
+      "step": 4230
+    },
+    {
+      "epoch": 0.08154591044374246,
+      "grad_norm": 718.6288805842618,
+      "learning_rate": 1.974337638776866e-05,
+      "loss": 373.2439,
+      "step": 4240
+    },
+    {
+      "epoch": 0.08173823570422298,
+      "grad_norm": 706.8570694706435,
+      "learning_rate": 1.9742000807947986e-05,
+      "loss": 365.4311,
+      "step": 4250
+    },
+    {
+      "epoch": 0.08193056096470351,
+      "grad_norm": 688.9848362082146,
+      "learning_rate": 1.9740621599392907e-05,
+      "loss": 368.1906,
+      "step": 4260
+    },
+    {
+      "epoch": 0.08212288622518403,
+      "grad_norm": 725.5829260231745,
+      "learning_rate": 1.9739238762617155e-05,
+      "loss": 367.0304,
+      "step": 4270
+    },
+    {
+      "epoch": 0.08231521148566455,
+      "grad_norm": 717.2467725821838,
+      "learning_rate": 1.973785229813581e-05,
+      "loss": 375.8666,
+      "step": 4280
+    },
+    {
+      "epoch": 0.08250753674614508,
+      "grad_norm": 745.5834559902293,
+      "learning_rate": 1.973646220646531e-05,
+      "loss": 378.2281,
+      "step": 4290
+    },
+    {
+      "epoch": 0.0826998620066256,
+      "grad_norm": 707.8169237140677,
+      "learning_rate": 1.973506848812344e-05,
+      "loss": 368.5782,
+      "step": 4300
+    },
+    {
+      "epoch": 0.08289218726710613,
+      "grad_norm": 681.9545939357413,
+      "learning_rate": 1.9733671143629342e-05,
+      "loss": 387.3969,
+      "step": 4310
+    },
+    {
+      "epoch": 0.08308451252758665,
+      "grad_norm": 733.1412882987519,
+      "learning_rate": 1.9732270173503493e-05,
+      "loss": 372.3555,
+      "step": 4320
+    },
+    {
+      "epoch": 0.08327683778806717,
+      "grad_norm": 686.8459966312021,
+      "learning_rate": 1.9730865578267745e-05,
+      "loss": 364.6364,
+      "step": 4330
+    },
+    {
+      "epoch": 0.0834691630485477,
+      "grad_norm": 751.0190659936467,
+      "learning_rate": 1.972945735844528e-05,
+      "loss": 370.8999,
+      "step": 4340
+    },
+    {
+      "epoch": 0.08366148830902823,
+      "grad_norm": 678.5074425036346,
+      "learning_rate": 1.972804551456063e-05,
+      "loss": 374.4807,
+      "step": 4350
+    },
+    {
+      "epoch": 0.08385381356950876,
+      "grad_norm": 759.0010394301813,
+      "learning_rate": 1.9726630047139695e-05,
+      "loss": 375.4909,
+      "step": 4360
+    },
+    {
+      "epoch": 0.08404613882998928,
+      "grad_norm": 634.9276459035196,
+      "learning_rate": 1.9725210956709707e-05,
+      "loss": 361.6657,
+      "step": 4370
+    },
+    {
+      "epoch": 0.0842384640904698,
+      "grad_norm": 691.5276471240323,
+      "learning_rate": 1.9723788243799253e-05,
+      "loss": 378.3001,
+      "step": 4380
+    },
+    {
+      "epoch": 0.08443078935095033,
+      "grad_norm": 765.1228459329928,
+      "learning_rate": 1.972236190893827e-05,
+      "loss": 364.8584,
+      "step": 4390
+    },
+    {
+      "epoch": 0.08462311461143085,
+      "grad_norm": 997.4870547867154,
+      "learning_rate": 1.972093195265805e-05,
+      "loss": 380.687,
+      "step": 4400
+    },
+    {
+      "epoch": 0.08481543987191138,
+      "grad_norm": 735.5785138605547,
+      "learning_rate": 1.9719498375491224e-05,
+      "loss": 364.8229,
+      "step": 4410
+    },
+    {
+      "epoch": 0.0850077651323919,
+      "grad_norm": 777.5815048764625,
+      "learning_rate": 1.9718061177971777e-05,
+      "loss": 368.2549,
+      "step": 4420
+    },
+    {
+      "epoch": 0.08520009039287242,
+      "grad_norm": 670.5791581190966,
+      "learning_rate": 1.9716620360635036e-05,
+      "loss": 372.5692,
+      "step": 4430
+    },
+    {
+      "epoch": 0.08539241565335295,
+      "grad_norm": 762.1654045664332,
+      "learning_rate": 1.971517592401769e-05,
+      "loss": 381.9217,
+      "step": 4440
+    },
+    {
+      "epoch": 0.08558474091383347,
+      "grad_norm": 992.1490506568286,
+      "learning_rate": 1.9713727868657764e-05,
+      "loss": 364.6855,
+      "step": 4450
+    },
+    {
+      "epoch": 0.085777066174314,
+      "grad_norm": 840.4760098460681,
+      "learning_rate": 1.971227619509463e-05,
+      "loss": 362.5352,
+      "step": 4460
+    },
+    {
+      "epoch": 0.08596939143479453,
+      "grad_norm": 968.5539944692922,
+      "learning_rate": 1.971082090386902e-05,
+      "loss": 377.6954,
+      "step": 4470
+    },
+    {
+      "epoch": 0.08616171669527505,
+      "grad_norm": 2843.7545849064295,
+      "learning_rate": 1.9709361995523e-05,
+      "loss": 365.9315,
+      "step": 4480
+    },
+    {
+      "epoch": 0.08635404195575558,
+      "grad_norm": 851.154691325466,
+      "learning_rate": 1.9707899470599998e-05,
+      "loss": 379.393,
+      "step": 4490
+    },
+    {
+      "epoch": 0.0865463672162361,
+      "grad_norm": 723.639897000417,
+      "learning_rate": 1.970643332964477e-05,
+      "loss": 363.6261,
+      "step": 4500
+    },
+    {
+      "epoch": 0.08673869247671662,
+      "grad_norm": 707.6678333878551,
+      "learning_rate": 1.9704963573203435e-05,
+      "loss": 365.22,
+      "step": 4510
+    },
+    {
+      "epoch": 0.08693101773719715,
+      "grad_norm": 1093.9096422715052,
+      "learning_rate": 1.970349020182345e-05,
+      "loss": 375.2929,
+      "step": 4520
+    },
+    {
+      "epoch": 0.08712334299767767,
+      "grad_norm": 781.9974730618354,
+      "learning_rate": 1.9702013216053623e-05,
+      "loss": 358.5067,
+      "step": 4530
+    },
+    {
+      "epoch": 0.0873156682581582,
+      "grad_norm": 708.7466764529239,
+      "learning_rate": 1.9700532616444114e-05,
+      "loss": 372.7084,
+      "step": 4540
+    },
+    {
+      "epoch": 0.08750799351863872,
+      "grad_norm": 1063.5562486170961,
+      "learning_rate": 1.969904840354641e-05,
+      "loss": 369.8918,
+      "step": 4550
+    },
+    {
+      "epoch": 0.08770031877911924,
+      "grad_norm": 693.4258589690572,
+      "learning_rate": 1.9697560577913358e-05,
+      "loss": 382.4882,
+      "step": 4560
+    },
+    {
+      "epoch": 0.08789264403959977,
+      "grad_norm": 791.7809123061226,
+      "learning_rate": 1.9696069140099152e-05,
+      "loss": 363.5707,
+      "step": 4570
+    },
+    {
+      "epoch": 0.08808496930008029,
+      "grad_norm": 877.4785235656661,
+      "learning_rate": 1.969457409065933e-05,
+      "loss": 364.0813,
+      "step": 4580
+    },
+    {
+      "epoch": 0.08827729456056083,
+      "grad_norm": 751.3586040127548,
+      "learning_rate": 1.969307543015077e-05,
+      "loss": 364.6259,
+      "step": 4590
+    },
+    {
+      "epoch": 0.08846961982104135,
+      "grad_norm": 829.6726739476259,
+      "learning_rate": 1.9691573159131696e-05,
+      "loss": 366.9448,
+      "step": 4600
+    },
+    {
+      "epoch": 0.08866194508152186,
+      "grad_norm": 876.1646962391158,
+      "learning_rate": 1.9690067278161686e-05,
+      "loss": 367.6636,
+      "step": 4610
+    },
+    {
+      "epoch": 0.0888542703420024,
+      "grad_norm": 899.4193387993457,
+      "learning_rate": 1.9688557787801647e-05,
+      "loss": 367.4333,
+      "step": 4620
+    },
+    {
+      "epoch": 0.08904659560248292,
+      "grad_norm": 757.6314580666606,
+      "learning_rate": 1.968704468861385e-05,
+      "loss": 368.2446,
+      "step": 4630
+    },
+    {
+      "epoch": 0.08923892086296345,
+      "grad_norm": 738.5256711356122,
+      "learning_rate": 1.968552798116189e-05,
+      "loss": 351.7744,
+      "step": 4640
+    },
+    {
+      "epoch": 0.08943124612344397,
+      "grad_norm": 775.1896435013546,
+      "learning_rate": 1.9684007666010716e-05,
+      "loss": 361.262,
+      "step": 4650
+    },
+    {
+      "epoch": 0.08962357138392449,
+      "grad_norm": 641.0132364280885,
+      "learning_rate": 1.9682483743726624e-05,
+      "loss": 370.4693,
+      "step": 4660
+    },
+    {
+      "epoch": 0.08981589664440502,
+      "grad_norm": 752.5796369691224,
+      "learning_rate": 1.968095621487725e-05,
+      "loss": 363.7262,
+      "step": 4670
+    },
+    {
+      "epoch": 0.09000822190488554,
+      "grad_norm": 752.5472495333028,
+      "learning_rate": 1.9679425080031574e-05,
+      "loss": 360.372,
+      "step": 4680
+    },
+    {
+      "epoch": 0.09020054716536607,
+      "grad_norm": 837.3847840187569,
+      "learning_rate": 1.9677890339759914e-05,
+      "loss": 369.5652,
+      "step": 4690
+    },
+    {
+      "epoch": 0.0903928724258466,
+      "grad_norm": 775.2121393653373,
+      "learning_rate": 1.967635199463394e-05,
+      "loss": 362.2298,
+      "step": 4700
+    },
+    {
+      "epoch": 0.09058519768632711,
+      "grad_norm": 778.091423998472,
+      "learning_rate": 1.9674810045226658e-05,
+      "loss": 366.6832,
+      "step": 4710
+    },
+    {
+      "epoch": 0.09077752294680765,
+      "grad_norm": 676.5260672928845,
+      "learning_rate": 1.967326449211242e-05,
+      "loss": 358.6009,
+      "step": 4720
+    },
+    {
+      "epoch": 0.09096984820728816,
+      "grad_norm": 787.7257095915079,
+      "learning_rate": 1.9671715335866915e-05,
+      "loss": 359.7101,
+      "step": 4730
+    },
+    {
+      "epoch": 0.09116217346776868,
+      "grad_norm": 726.7840286038974,
+      "learning_rate": 1.9670162577067182e-05,
+      "loss": 378.2719,
+      "step": 4740
+    },
+    {
+      "epoch": 0.09135449872824922,
+      "grad_norm": 860.1818991378439,
+      "learning_rate": 1.9668606216291598e-05,
+      "loss": 363.3302,
+      "step": 4750
+    },
+    {
+      "epoch": 0.09154682398872974,
+      "grad_norm": 813.8855071102088,
+      "learning_rate": 1.9667046254119878e-05,
+      "loss": 356.8751,
+      "step": 4760
+    },
+    {
+      "epoch": 0.09173914924921027,
+      "grad_norm": 699.9496488839774,
+      "learning_rate": 1.966548269113309e-05,
+      "loss": 364.8364,
+      "step": 4770
+    },
+    {
+      "epoch": 0.09193147450969079,
+      "grad_norm": 689.1103427101364,
+      "learning_rate": 1.9663915527913628e-05,
+      "loss": 364.6322,
+      "step": 4780
+    },
+    {
+      "epoch": 0.09212379977017131,
+      "grad_norm": 743.5321878825619,
+      "learning_rate": 1.9662344765045237e-05,
+      "loss": 372.8812,
+      "step": 4790
+    },
+    {
+      "epoch": 0.09231612503065184,
+      "grad_norm": 638.1040062486387,
+      "learning_rate": 1.9660770403112996e-05,
+      "loss": 362.1715,
+      "step": 4800
+    },
+    {
+      "epoch": 0.09250845029113236,
+      "grad_norm": 807.1382883930204,
+      "learning_rate": 1.9659192442703336e-05,
+      "loss": 365.8542,
+      "step": 4810
+    },
+    {
+      "epoch": 0.0927007755516129,
+      "grad_norm": 752.1295341406421,
+      "learning_rate": 1.965761088440402e-05,
+      "loss": 367.9731,
+      "step": 4820
+    },
+    {
+      "epoch": 0.09289310081209341,
+      "grad_norm": 687.0910114607802,
+      "learning_rate": 1.9656025728804147e-05,
+      "loss": 364.8634,
+      "step": 4830
+    },
+    {
+      "epoch": 0.09308542607257393,
+      "grad_norm": 632.6157545516736,
+      "learning_rate": 1.9654436976494165e-05,
+      "loss": 359.9041,
+      "step": 4840
+    },
+    {
+      "epoch": 0.09327775133305446,
+      "grad_norm": 661.1788050399986,
+      "learning_rate": 1.9652844628065857e-05,
+      "loss": 355.8173,
+      "step": 4850
+    },
+    {
+      "epoch": 0.09347007659353498,
+      "grad_norm": 702.4271151111567,
+      "learning_rate": 1.965124868411235e-05,
+      "loss": 371.1025,
+      "step": 4860
+    },
+    {
+      "epoch": 0.09366240185401552,
+      "grad_norm": 735.7176251611235,
+      "learning_rate": 1.96496491452281e-05,
+      "loss": 358.993,
+      "step": 4870
+    },
+    {
+      "epoch": 0.09385472711449604,
+      "grad_norm": 701.9085546602543,
+      "learning_rate": 1.9648046012008916e-05,
+      "loss": 355.9271,
+      "step": 4880
+    },
+    {
+      "epoch": 0.09404705237497656,
+      "grad_norm": 668.1042470286606,
+      "learning_rate": 1.9646439285051936e-05,
+      "loss": 366.6708,
+      "step": 4890
+    },
+    {
+      "epoch": 0.09423937763545709,
+      "grad_norm": 1329.9780620536628,
+      "learning_rate": 1.9644828964955633e-05,
+      "loss": 372.2765,
+      "step": 4900
+    },
+    {
+      "epoch": 0.09443170289593761,
+      "grad_norm": 654.2950590246775,
+      "learning_rate": 1.9643215052319836e-05,
+      "loss": 373.0559,
+      "step": 4910
+    },
+    {
+      "epoch": 0.09462402815641814,
+      "grad_norm": 702.124295888291,
+      "learning_rate": 1.9641597547745694e-05,
+      "loss": 375.7776,
+      "step": 4920
+    },
+    {
+      "epoch": 0.09481635341689866,
+      "grad_norm": 747.2709381896793,
+      "learning_rate": 1.9639976451835698e-05,
+      "loss": 376.6263,
+      "step": 4930
+    },
+    {
+      "epoch": 0.09500867867737918,
+      "grad_norm": 749.2638300155742,
+      "learning_rate": 1.9638351765193685e-05,
+      "loss": 371.4045,
+      "step": 4940
+    },
+    {
+      "epoch": 0.09520100393785971,
+      "grad_norm": 798.0008789179675,
+      "learning_rate": 1.9636723488424823e-05,
+      "loss": 350.803,
+      "step": 4950
+    },
+    {
+      "epoch": 0.09539332919834023,
+      "grad_norm": 652.989139825633,
+      "learning_rate": 1.9635091622135616e-05,
+      "loss": 367.1494,
+      "step": 4960
+    },
+    {
+      "epoch": 0.09558565445882075,
+      "grad_norm": 627.0662273778905,
+      "learning_rate": 1.963345616693391e-05,
+      "loss": 359.2938,
+      "step": 4970
+    },
+    {
+      "epoch": 0.09577797971930128,
+      "grad_norm": 697.185717014197,
+      "learning_rate": 1.9631817123428883e-05,
+      "loss": 366.4442,
+      "step": 4980
+    },
+    {
+      "epoch": 0.0959703049797818,
+      "grad_norm": 711.409397793336,
+      "learning_rate": 1.9630174492231052e-05,
+      "loss": 358.1978,
+      "step": 4990
+    },
+    {
+      "epoch": 0.09616263024026234,
+      "grad_norm": 666.5162533135881,
+      "learning_rate": 1.962852827395227e-05,
+      "loss": 375.9189,
+      "step": 5000
+    },
+    {
+      "epoch": 0.09635495550074286,
+      "grad_norm": 642.8915335248588,
+      "learning_rate": 1.962687846920573e-05,
+      "loss": 360.1929,
+      "step": 5010
+    },
+    {
+      "epoch": 0.09654728076122338,
+      "grad_norm": 785.8465530759893,
+      "learning_rate": 1.9625225078605946e-05,
+      "loss": 358.44,
+      "step": 5020
+    },
+    {
+      "epoch": 0.09673960602170391,
+      "grad_norm": 678.1645670579533,
+      "learning_rate": 1.9623568102768792e-05,
+      "loss": 364.0865,
+      "step": 5030
+    },
+    {
+      "epoch": 0.09693193128218443,
+      "grad_norm": 695.7291546323536,
+      "learning_rate": 1.9621907542311457e-05,
+      "loss": 364.631,
+      "step": 5040
+    },
+    {
+      "epoch": 0.09712425654266496,
+      "grad_norm": 829.7757885688217,
+      "learning_rate": 1.9620243397852473e-05,
+      "loss": 349.2464,
+      "step": 5050
+    },
+    {
+      "epoch": 0.09731658180314548,
+      "grad_norm": 688.8738361253692,
+      "learning_rate": 1.9618575670011705e-05,
+      "loss": 348.4353,
+      "step": 5060
+    },
+    {
+      "epoch": 0.097508907063626,
+      "grad_norm": 940.3285039849533,
+      "learning_rate": 1.9616904359410357e-05,
+      "loss": 355.8785,
+      "step": 5070
+    },
+    {
+      "epoch": 0.09770123232410653,
+      "grad_norm": 684.5118133090992,
+      "learning_rate": 1.9615229466670963e-05,
+      "loss": 358.3975,
+      "step": 5080
+    },
+    {
+      "epoch": 0.09789355758458705,
+      "grad_norm": 672.031305833946,
+      "learning_rate": 1.9613550992417396e-05,
+      "loss": 357.6419,
+      "step": 5090
+    },
+    {
+      "epoch": 0.09808588284506758,
+      "grad_norm": 645.4269998111222,
+      "learning_rate": 1.961186893727486e-05,
+      "loss": 357.7922,
+      "step": 5100
+    },
+    {
+      "epoch": 0.0982782081055481,
+      "grad_norm": 694.6823162953259,
+      "learning_rate": 1.9610183301869882e-05,
+      "loss": 347.938,
+      "step": 5110
+    },
+    {
+      "epoch": 0.09847053336602862,
+      "grad_norm": 757.6132463114961,
+      "learning_rate": 1.9608494086830348e-05,
+      "loss": 353.2071,
+      "step": 5120
+    },
+    {
+      "epoch": 0.09866285862650916,
+      "grad_norm": 691.0368103490606,
+      "learning_rate": 1.9606801292785452e-05,
+      "loss": 360.3291,
+      "step": 5130
+    },
+    {
+      "epoch": 0.09885518388698968,
+      "grad_norm": 632.6504849744924,
+      "learning_rate": 1.960510492036574e-05,
+      "loss": 360.4072,
+      "step": 5140
+    },
+    {
+      "epoch": 0.09904750914747021,
+      "grad_norm": 692.0316509173975,
+      "learning_rate": 1.9603404970203078e-05,
+      "loss": 363.7523,
+      "step": 5150
+    },
+    {
+      "epoch": 0.09923983440795073,
+      "grad_norm": 777.8436046072212,
+      "learning_rate": 1.9601701442930667e-05,
+      "loss": 371.8547,
+      "step": 5160
+    },
+    {
+      "epoch": 0.09943215966843125,
+      "grad_norm": 710.4627615354874,
+      "learning_rate": 1.9599994339183047e-05,
+      "loss": 363.1294,
+      "step": 5170
+    },
+    {
+      "epoch": 0.09962448492891178,
+      "grad_norm": 692.4692790921382,
+      "learning_rate": 1.9598283659596084e-05,
+      "loss": 352.3487,
+      "step": 5180
+    },
+    {
+      "epoch": 0.0998168101893923,
+      "grad_norm": 1150.4552627716428,
+      "learning_rate": 1.9596569404806983e-05,
+      "loss": 354.4399,
+      "step": 5190
+    },
+    {
+      "epoch": 0.10000913544987282,
+      "grad_norm": 695.4429732404097,
+      "learning_rate": 1.9594851575454266e-05,
+      "loss": 352.4259,
+      "step": 5200
+    },
+    {
+      "epoch": 0.10020146071035335,
+      "grad_norm": 677.9990103567972,
+      "learning_rate": 1.9593130172177806e-05,
+      "loss": 369.2995,
+      "step": 5210
+    },
+    {
+      "epoch": 0.10039378597083387,
+      "grad_norm": 710.1974302890869,
+      "learning_rate": 1.959140519561879e-05,
+      "loss": 359.1806,
+      "step": 5220
+    },
+    {
+      "epoch": 0.1005861112313144,
+      "grad_norm": 686.7699233718275,
+      "learning_rate": 1.9589676646419744e-05,
+      "loss": 360.2993,
+      "step": 5230
+    },
+    {
+      "epoch": 0.10077843649179492,
+      "grad_norm": 709.0860221895744,
+      "learning_rate": 1.958794452522453e-05,
+      "loss": 365.8043,
+      "step": 5240
+    },
+    {
+      "epoch": 0.10097076175227544,
+      "grad_norm": 618.6174095299081,
+      "learning_rate": 1.9586208832678328e-05,
+      "loss": 355.2659,
+      "step": 5250
+    },
+    {
+      "epoch": 0.10116308701275598,
+      "grad_norm": 636.9513429109644,
+      "learning_rate": 1.958446956942766e-05,
+      "loss": 362.7319,
+      "step": 5260
+    },
+    {
+      "epoch": 0.1013554122732365,
+      "grad_norm": 634.1443462789841,
+      "learning_rate": 1.9582726736120365e-05,
+      "loss": 354.1151,
+      "step": 5270
+    },
+    {
+      "epoch": 0.10154773753371703,
+      "grad_norm": 694.5956914001258,
+      "learning_rate": 1.958098033340563e-05,
+      "loss": 358.8986,
+      "step": 5280
+    },
+    {
+      "epoch": 0.10174006279419755,
+      "grad_norm": 722.9248946007976,
+      "learning_rate": 1.9579230361933952e-05,
+      "loss": 371.4438,
+      "step": 5290
+    },
+    {
+      "epoch": 0.10193238805467807,
+      "grad_norm": 724.2618423668106,
+      "learning_rate": 1.9577476822357174e-05,
+      "loss": 349.7284,
+      "step": 5300
+    },
+    {
+      "epoch": 0.1021247133151586,
+      "grad_norm": 792.4144490762708,
+      "learning_rate": 1.9575719715328457e-05,
+      "loss": 367.7335,
+      "step": 5310
+    },
+    {
+      "epoch": 0.10231703857563912,
+      "grad_norm": 622.6828197063883,
+      "learning_rate": 1.957395904150229e-05,
+      "loss": 348.4879,
+      "step": 5320
+    },
+    {
+      "epoch": 0.10250936383611965,
+      "grad_norm": 717.6866249410048,
+      "learning_rate": 1.9572194801534504e-05,
+      "loss": 359.715,
+      "step": 5330
+    },
+    {
+      "epoch": 0.10270168909660017,
+      "grad_norm": 668.9250455978879,
+      "learning_rate": 1.9570426996082238e-05,
+      "loss": 358.7419,
+      "step": 5340
+    },
+    {
+      "epoch": 0.10289401435708069,
+      "grad_norm": 819.6418580879654,
+      "learning_rate": 1.9568655625803982e-05,
+      "loss": 350.3772,
+      "step": 5350
+    },
+    {
+      "epoch": 0.10308633961756122,
+      "grad_norm": 631.5286785486417,
+      "learning_rate": 1.956688069135954e-05,
+      "loss": 360.9678,
+      "step": 5360
+    },
+    {
+      "epoch": 0.10327866487804174,
+      "grad_norm": 590.7187007886677,
+      "learning_rate": 1.9565102193410035e-05,
+      "loss": 351.1492,
+      "step": 5370
+    },
+    {
+      "epoch": 0.10347099013852228,
+      "grad_norm": 639.5551477174666,
+      "learning_rate": 1.956332013261794e-05,
+      "loss": 355.9312,
+      "step": 5380
+    },
+    {
+      "epoch": 0.1036633153990028,
+      "grad_norm": 713.853127241327,
+      "learning_rate": 1.9561534509647038e-05,
+      "loss": 346.8531,
+      "step": 5390
+    },
+    {
+      "epoch": 0.10385564065948331,
+      "grad_norm": 678.2709414907933,
+      "learning_rate": 1.9559745325162445e-05,
+      "loss": 365.8308,
+      "step": 5400
+    },
+    {
+      "epoch": 0.10404796591996385,
+      "grad_norm": 696.1133632246981,
+      "learning_rate": 1.9557952579830604e-05,
+      "loss": 358.5058,
+      "step": 5410
+    },
+    {
+      "epoch": 0.10424029118044437,
+      "grad_norm": 609.1631720205271,
+      "learning_rate": 1.955615627431928e-05,
+      "loss": 360.643,
+      "step": 5420
+    },
+    {
+      "epoch": 0.10443261644092489,
+      "grad_norm": 588.216185762683,
+      "learning_rate": 1.955435640929757e-05,
+      "loss": 346.53,
+      "step": 5430
+    },
+    {
+      "epoch": 0.10462494170140542,
+      "grad_norm": 687.938172394234,
+      "learning_rate": 1.9552552985435893e-05,
+      "loss": 348.8743,
+      "step": 5440
+    },
+    {
+      "epoch": 0.10481726696188594,
+      "grad_norm": 646.2257032802848,
+      "learning_rate": 1.9550746003405996e-05,
+      "loss": 353.3726,
+      "step": 5450
+    },
+    {
+      "epoch": 0.10500959222236647,
+      "grad_norm": 615.3275930754071,
+      "learning_rate": 1.9548935463880945e-05,
+      "loss": 361.7423,
+      "step": 5460
+    },
+    {
+      "epoch": 0.10520191748284699,
+      "grad_norm": 920.3547594582011,
+      "learning_rate": 1.9547121367535143e-05,
+      "loss": 368.0178,
+      "step": 5470
+    },
+    {
+      "epoch": 0.10539424274332751,
+      "grad_norm": 673.3909171381894,
+      "learning_rate": 1.9545303715044305e-05,
+      "loss": 363.9247,
+      "step": 5480
+    },
+    {
+      "epoch": 0.10558656800380804,
+      "grad_norm": 668.479775303425,
+      "learning_rate": 1.9543482507085484e-05,
+      "loss": 360.2404,
+      "step": 5490
+    },
+    {
+      "epoch": 0.10577889326428856,
+      "grad_norm": 701.994986175009,
+      "learning_rate": 1.9541657744337038e-05,
+      "loss": 349.8952,
+      "step": 5500
+    },
+    {
+      "epoch": 0.1059712185247691,
+      "grad_norm": 705.6009801245955,
+      "learning_rate": 1.9539829427478675e-05,
+      "loss": 362.0219,
+      "step": 5510
+    },
+    {
+      "epoch": 0.10616354378524961,
+      "grad_norm": 661.6704816373879,
+      "learning_rate": 1.95379975571914e-05,
+      "loss": 367.3877,
+      "step": 5520
+    },
+    {
+      "epoch": 0.10635586904573013,
+      "grad_norm": 687.3223282814702,
+      "learning_rate": 1.953616213415756e-05,
+      "loss": 358.3039,
+      "step": 5530
+    },
+    {
+      "epoch": 0.10654819430621067,
+      "grad_norm": 616.9728859653206,
+      "learning_rate": 1.9534323159060824e-05,
+      "loss": 365.5358,
+      "step": 5540
+    },
+    {
+      "epoch": 0.10674051956669119,
+      "grad_norm": 794.7069763948178,
+      "learning_rate": 1.9532480632586175e-05,
+      "loss": 358.5932,
+      "step": 5550
+    },
+    {
+      "epoch": 0.10693284482717172,
+      "grad_norm": 677.6407233525834,
+      "learning_rate": 1.953063455541992e-05,
+      "loss": 365.166,
+      "step": 5560
+    },
+    {
+      "epoch": 0.10712517008765224,
+      "grad_norm": 629.0218444824332,
+      "learning_rate": 1.9528784928249703e-05,
+      "loss": 352.1474,
+      "step": 5570
+    },
+    {
+      "epoch": 0.10731749534813276,
+      "grad_norm": 666.7341636614977,
+      "learning_rate": 1.9526931751764467e-05,
+      "loss": 345.3576,
+      "step": 5580
+    },
+    {
+      "epoch": 0.10750982060861329,
+      "grad_norm": 629.6700672193283,
+      "learning_rate": 1.95250750266545e-05,
+      "loss": 354.5343,
+      "step": 5590
+    },
+    {
+      "epoch": 0.10770214586909381,
+      "grad_norm": 711.6264440434755,
+      "learning_rate": 1.9523214753611398e-05,
+      "loss": 343.4514,
+      "step": 5600
+    },
+    {
+      "epoch": 0.10789447112957434,
+      "grad_norm": 739.432375882402,
+      "learning_rate": 1.952135093332808e-05,
+      "loss": 363.1715,
+      "step": 5610
+    },
+    {
+      "epoch": 0.10808679639005486,
+      "grad_norm": 803.0180117993846,
+      "learning_rate": 1.9519483566498788e-05,
+      "loss": 351.7206,
+      "step": 5620
+    },
+    {
+      "epoch": 0.10827912165053538,
+      "grad_norm": 650.5084174821643,
+      "learning_rate": 1.9517612653819088e-05,
+      "loss": 354.0946,
+      "step": 5630
+    },
+    {
+      "epoch": 0.10847144691101591,
+      "grad_norm": 658.895998477785,
+      "learning_rate": 1.9515738195985868e-05,
+      "loss": 364.1155,
+      "step": 5640
+    },
+    {
+      "epoch": 0.10866377217149643,
+      "grad_norm": 704.3438306020199,
+      "learning_rate": 1.951386019369732e-05,
+      "loss": 349.9277,
+      "step": 5650
+    },
+    {
+      "epoch": 0.10885609743197695,
+      "grad_norm": 641.1259216105735,
+      "learning_rate": 1.9511978647652984e-05,
+      "loss": 346.7582,
+      "step": 5660
+    },
+    {
+      "epoch": 0.10904842269245749,
+      "grad_norm": 644.5491587588273,
+      "learning_rate": 1.9510093558553687e-05,
+      "loss": 351.1167,
+      "step": 5670
+    },
+    {
+      "epoch": 0.109240747952938,
+      "grad_norm": 620.6813360337377,
+      "learning_rate": 1.950820492710161e-05,
+      "loss": 348.5378,
+      "step": 5680
+    },
+    {
+      "epoch": 0.10943307321341854,
+      "grad_norm": 644.5416008845856,
+      "learning_rate": 1.9506312754000235e-05,
+      "loss": 351.0252,
+      "step": 5690
+    },
+    {
+      "epoch": 0.10962539847389906,
+      "grad_norm": 654.059831443735,
+      "learning_rate": 1.9504417039954357e-05,
+      "loss": 348.4304,
+      "step": 5700
+    },
+    {
+      "epoch": 0.10981772373437958,
+      "grad_norm": 644.1505153246875,
+      "learning_rate": 1.9502517785670098e-05,
+      "loss": 359.1387,
+      "step": 5710
+    },
+    {
+      "epoch": 0.11001004899486011,
+      "grad_norm": 769.20654559226,
+      "learning_rate": 1.950061499185491e-05,
+      "loss": 359.4678,
+      "step": 5720
+    },
+    {
+      "epoch": 0.11020237425534063,
+      "grad_norm": 679.3344717558374,
+      "learning_rate": 1.9498708659217542e-05,
+      "loss": 350.7424,
+      "step": 5730
+    },
+    {
+      "epoch": 0.11039469951582116,
+      "grad_norm": 657.8142033708348,
+      "learning_rate": 1.9496798788468077e-05,
+      "loss": 342.7845,
+      "step": 5740
+    },
+    {
+      "epoch": 0.11058702477630168,
+      "grad_norm": 741.24155744283,
+      "learning_rate": 1.9494885380317906e-05,
+      "loss": 346.9994,
+      "step": 5750
+    },
+    {
+      "epoch": 0.1107793500367822,
+      "grad_norm": 742.9321197573132,
+      "learning_rate": 1.9492968435479744e-05,
+      "loss": 345.0701,
+      "step": 5760
+    },
+    {
+      "epoch": 0.11097167529726273,
+      "grad_norm": 639.619633875009,
+      "learning_rate": 1.949104795466762e-05,
+      "loss": 346.2675,
+      "step": 5770
+    },
+    {
+      "epoch": 0.11116400055774325,
+      "grad_norm": 635.3939443905493,
+      "learning_rate": 1.9489123938596886e-05,
+      "loss": 338.2012,
+      "step": 5780
+    },
+    {
+      "epoch": 0.11135632581822379,
+      "grad_norm": 606.3943827545633,
+      "learning_rate": 1.94871963879842e-05,
+      "loss": 351.4652,
+      "step": 5790
+    },
+    {
+      "epoch": 0.1115486510787043,
+      "grad_norm": 854.4458033211714,
+      "learning_rate": 1.9485265303547547e-05,
+      "loss": 347.1553,
+      "step": 5800
+    },
+    {
+      "epoch": 0.11174097633918482,
+      "grad_norm": 732.4929729662073,
+      "learning_rate": 1.9483330686006223e-05,
+      "loss": 353.6084,
+      "step": 5810
+    },
+    {
+      "epoch": 0.11193330159966536,
+      "grad_norm": 662.4497031342546,
+      "learning_rate": 1.948139253608084e-05,
+      "loss": 372.1042,
+      "step": 5820
+    },
+    {
+      "epoch": 0.11212562686014588,
+      "grad_norm": 766.0917879572403,
+      "learning_rate": 1.9479450854493327e-05,
+      "loss": 348.1985,
+      "step": 5830
+    },
+    {
+      "epoch": 0.1123179521206264,
+      "grad_norm": 750.8279452274128,
+      "learning_rate": 1.9477505641966933e-05,
+      "loss": 344.6354,
+      "step": 5840
+    },
+    {
+      "epoch": 0.11251027738110693,
+      "grad_norm": 735.8955775076766,
+      "learning_rate": 1.9475556899226213e-05,
+      "loss": 350.4105,
+      "step": 5850
+    },
+    {
+      "epoch": 0.11270260264158745,
+      "grad_norm": 623.2418608358216,
+      "learning_rate": 1.9473604626997037e-05,
+      "loss": 342.1461,
+      "step": 5860
+    },
+    {
+      "epoch": 0.11289492790206798,
+      "grad_norm": 617.0366422133485,
+      "learning_rate": 1.94716488260066e-05,
+      "loss": 347.9125,
+      "step": 5870
+    },
+    {
+      "epoch": 0.1130872531625485,
+      "grad_norm": 711.3109167472553,
+      "learning_rate": 1.946968949698341e-05,
+      "loss": 348.1037,
+      "step": 5880
+    },
+    {
+      "epoch": 0.11327957842302902,
+      "grad_norm": 705.609924571381,
+      "learning_rate": 1.9467726640657277e-05,
+      "loss": 343.4828,
+      "step": 5890
+    },
+    {
+      "epoch": 0.11347190368350955,
+      "grad_norm": 736.4183661571752,
+      "learning_rate": 1.9465760257759336e-05,
+      "loss": 348.7198,
+      "step": 5900
+    },
+    {
+      "epoch": 0.11366422894399007,
+      "grad_norm": 611.3064252789177,
+      "learning_rate": 1.9463790349022027e-05,
+      "loss": 338.8815,
+      "step": 5910
+    },
+    {
+      "epoch": 0.1138565542044706,
+      "grad_norm": 603.676672088241,
+      "learning_rate": 1.9461816915179117e-05,
+      "loss": 346.2747,
+      "step": 5920
+    },
+    {
+      "epoch": 0.11404887946495113,
+      "grad_norm": 713.939519252679,
+      "learning_rate": 1.945983995696567e-05,
+      "loss": 352.6707,
+      "step": 5930
+    },
+    {
+      "epoch": 0.11424120472543164,
+      "grad_norm": 624.5364853448472,
+      "learning_rate": 1.9457859475118077e-05,
+      "loss": 345.4377,
+      "step": 5940
+    },
+    {
+      "epoch": 0.11443352998591218,
+      "grad_norm": 694.9688179948644,
+      "learning_rate": 1.9455875470374027e-05,
+      "loss": 361.0383,
+      "step": 5950
+    },
+    {
+      "epoch": 0.1146258552463927,
+      "grad_norm": 714.9516773147197,
+      "learning_rate": 1.9453887943472532e-05,
+      "loss": 336.5125,
+      "step": 5960
+    },
+    {
+      "epoch": 0.11481818050687323,
+      "grad_norm": 734.8832746480035,
+      "learning_rate": 1.945189689515392e-05,
+      "loss": 359.7186,
+      "step": 5970
+    },
+    {
+      "epoch": 0.11501050576735375,
+      "grad_norm": 646.6308216216573,
+      "learning_rate": 1.9449902326159815e-05,
+      "loss": 338.3782,
+      "step": 5980
+    },
+    {
+      "epoch": 0.11520283102783427,
+      "grad_norm": 697.1561776823747,
+      "learning_rate": 1.9447904237233164e-05,
+      "loss": 349.8259,
+      "step": 5990
+    },
+    {
+      "epoch": 0.1153951562883148,
+      "grad_norm": 669.4929816486336,
+      "learning_rate": 1.9445902629118223e-05,
+      "loss": 342.1568,
+      "step": 6000
+    },
+    {
+      "epoch": 0.11558748154879532,
+      "grad_norm": 602.6243015218835,
+      "learning_rate": 1.9443897502560555e-05,
+      "loss": 354.5575,
+      "step": 6010
+    },
+    {
+      "epoch": 0.11577980680927585,
+      "grad_norm": 635.3106979930712,
+      "learning_rate": 1.9441888858307042e-05,
+      "loss": 364.4904,
+      "step": 6020
+    },
+    {
+      "epoch": 0.11597213206975637,
+      "grad_norm": 605.2859705768255,
+      "learning_rate": 1.943987669710586e-05,
+      "loss": 340.5251,
+      "step": 6030
+    },
+    {
+      "epoch": 0.11616445733023689,
+      "grad_norm": 633.4166407738802,
+      "learning_rate": 1.9437861019706522e-05,
+      "loss": 342.5962,
+      "step": 6040
+    },
+    {
+      "epoch": 0.11635678259071743,
+      "grad_norm": 645.851267654988,
+      "learning_rate": 1.943584182685982e-05,
+      "loss": 354.7195,
+      "step": 6050
+    },
+    {
+      "epoch": 0.11654910785119794,
+      "grad_norm": 846.2031756676997,
+      "learning_rate": 1.9433819119317878e-05,
+      "loss": 348.1721,
+      "step": 6060
+    },
+    {
+      "epoch": 0.11674143311167846,
+      "grad_norm": 654.0855910798667,
+      "learning_rate": 1.9431792897834115e-05,
+      "loss": 343.3366,
+      "step": 6070
+    },
+    {
+      "epoch": 0.116933758372159,
+      "grad_norm": 851.0349887230075,
+      "learning_rate": 1.9429763163163273e-05,
+      "loss": 356.5918,
+      "step": 6080
+    },
+    {
+      "epoch": 0.11712608363263952,
+      "grad_norm": 695.9728498156604,
+      "learning_rate": 1.942772991606139e-05,
+      "loss": 338.5591,
+      "step": 6090
+    },
+    {
+      "epoch": 0.11731840889312005,
+      "grad_norm": 796.0992941708482,
+      "learning_rate": 1.9425693157285816e-05,
+      "loss": 347.8489,
+      "step": 6100
+    },
+    {
+      "epoch": 0.11751073415360057,
+      "grad_norm": 738.4894558855128,
+      "learning_rate": 1.942365288759521e-05,
+      "loss": 338.323,
+      "step": 6110
+    },
+    {
+      "epoch": 0.11770305941408109,
+      "grad_norm": 647.0316876095417,
+      "learning_rate": 1.9421609107749542e-05,
+      "loss": 354.4408,
+      "step": 6120
+    },
+    {
+      "epoch": 0.11789538467456162,
+      "grad_norm": 714.2423151992201,
+      "learning_rate": 1.9419561818510085e-05,
+      "loss": 347.4304,
+      "step": 6130
+    },
+    {
+      "epoch": 0.11808770993504214,
+      "grad_norm": 662.1491125046128,
+      "learning_rate": 1.9417511020639416e-05,
+      "loss": 345.5496,
+      "step": 6140
+    },
+    {
+      "epoch": 0.11828003519552267,
+      "grad_norm": 615.1461389522196,
+      "learning_rate": 1.9415456714901432e-05,
+      "loss": 350.8063,
+      "step": 6150
+    },
+    {
+      "epoch": 0.11847236045600319,
+      "grad_norm": 606.9754745270544,
+      "learning_rate": 1.941339890206132e-05,
+      "loss": 352.5946,
+      "step": 6160
+    },
+    {
+      "epoch": 0.11866468571648371,
+      "grad_norm": 634.6995729232873,
+      "learning_rate": 1.9411337582885587e-05,
+      "loss": 344.4547,
+      "step": 6170
+    },
+    {
+      "epoch": 0.11885701097696424,
+      "grad_norm": 710.4985092822184,
+      "learning_rate": 1.9409272758142034e-05,
+      "loss": 338.1419,
+      "step": 6180
+    },
+    {
+      "epoch": 0.11904933623744476,
+      "grad_norm": 687.113663528976,
+      "learning_rate": 1.940720442859978e-05,
+      "loss": 347.3788,
+      "step": 6190
+    },
+    {
+      "epoch": 0.1192416614979253,
+      "grad_norm": 604.0974446875828,
+      "learning_rate": 1.940513259502924e-05,
+      "loss": 347.2637,
+      "step": 6200
+    },
+    {
+      "epoch": 0.11943398675840582,
+      "grad_norm": 625.1258844503341,
+      "learning_rate": 1.9403057258202144e-05,
+      "loss": 346.1806,
+      "step": 6210
+    },
+    {
+      "epoch": 0.11962631201888634,
+      "grad_norm": 618.2532754834317,
+      "learning_rate": 1.940097841889151e-05,
+      "loss": 343.0665,
+      "step": 6220
+    },
+    {
+      "epoch": 0.11981863727936687,
+      "grad_norm": 599.6548476871179,
+      "learning_rate": 1.939889607787168e-05,
+      "loss": 352.7839,
+      "step": 6230
+    },
+    {
+      "epoch": 0.12001096253984739,
+      "grad_norm": 666.041125551419,
+      "learning_rate": 1.9396810235918287e-05,
+      "loss": 348.3168,
+      "step": 6240
+    },
+    {
+      "epoch": 0.12020328780032792,
+      "grad_norm": 669.6349246964463,
+      "learning_rate": 1.939472089380828e-05,
+      "loss": 343.3021,
+      "step": 6250
+    },
+    {
+      "epoch": 0.12039561306080844,
+      "grad_norm": 664.6639247045922,
+      "learning_rate": 1.9392628052319895e-05,
+      "loss": 348.0812,
+      "step": 6260
+    },
+    {
+      "epoch": 0.12058793832128896,
+      "grad_norm": 578.9207598443023,
+      "learning_rate": 1.9390531712232687e-05,
+      "loss": 347.8523,
+      "step": 6270
+    },
+    {
+      "epoch": 0.12078026358176949,
+      "grad_norm": 727.9456926484588,
+      "learning_rate": 1.9388431874327504e-05,
+      "loss": 358.9287,
+      "step": 6280
+    },
+    {
+      "epoch": 0.12097258884225001,
+      "grad_norm": 841.7329823391808,
+      "learning_rate": 1.9386328539386502e-05,
+      "loss": 350.1853,
+      "step": 6290
+    },
+    {
+      "epoch": 0.12116491410273053,
+      "grad_norm": 863.8318261379947,
+      "learning_rate": 1.938422170819314e-05,
+      "loss": 349.7141,
+      "step": 6300
+    },
+    {
+      "epoch": 0.12135723936321106,
+      "grad_norm": 741.1597664964307,
+      "learning_rate": 1.938211138153218e-05,
+      "loss": 337.517,
+      "step": 6310
+    },
+    {
+      "epoch": 0.12154956462369158,
+      "grad_norm": 710.6308803333314,
+      "learning_rate": 1.9379997560189677e-05,
+      "loss": 342.7385,
+      "step": 6320
+    },
+    {
+      "epoch": 0.12174188988417212,
+      "grad_norm": 587.5917142179576,
+      "learning_rate": 1.9377880244953e-05,
+      "loss": 338.0116,
+      "step": 6330
+    },
+    {
+      "epoch": 0.12193421514465264,
+      "grad_norm": 708.5960196524559,
+      "learning_rate": 1.9375759436610813e-05,
+      "loss": 344.2697,
+      "step": 6340
+    },
+    {
+      "epoch": 0.12212654040513315,
+      "grad_norm": 818.7838214632518,
+      "learning_rate": 1.937363513595308e-05,
+      "loss": 346.9611,
+      "step": 6350
+    },
+    {
+      "epoch": 0.12231886566561369,
+      "grad_norm": 743.013756708669,
+      "learning_rate": 1.937150734377107e-05,
+      "loss": 332.9242,
+      "step": 6360
+    },
+    {
+      "epoch": 0.12251119092609421,
+      "grad_norm": 695.2887217131536,
+      "learning_rate": 1.9369376060857354e-05,
+      "loss": 332.5873,
+      "step": 6370
+    },
+    {
+      "epoch": 0.12270351618657474,
+      "grad_norm": 724.3502879672002,
+      "learning_rate": 1.936724128800579e-05,
+      "loss": 348.6979,
+      "step": 6380
+    },
+    {
+      "epoch": 0.12289584144705526,
+      "grad_norm": 607.9127161403849,
+      "learning_rate": 1.9365103026011555e-05,
+      "loss": 344.3744,
+      "step": 6390
+    },
+    {
+      "epoch": 0.12308816670753578,
+      "grad_norm": 717.833106513343,
+      "learning_rate": 1.9362961275671112e-05,
+      "loss": 347.6986,
+      "step": 6400
+    },
+    {
+      "epoch": 0.12328049196801631,
+      "grad_norm": 647.4911508762264,
+      "learning_rate": 1.936081603778223e-05,
+      "loss": 357.9332,
+      "step": 6410
+    },
+    {
+      "epoch": 0.12347281722849683,
+      "grad_norm": 622.8712891042829,
+      "learning_rate": 1.9358667313143972e-05,
+      "loss": 345.6661,
+      "step": 6420
+    },
+    {
+      "epoch": 0.12366514248897736,
+      "grad_norm": 761.8234931548434,
+      "learning_rate": 1.93565151025567e-05,
+      "loss": 346.5215,
+      "step": 6430
+    },
+    {
+      "epoch": 0.12385746774945788,
+      "grad_norm": 623.572070801288,
+      "learning_rate": 1.9354359406822084e-05,
+      "loss": 348.8408,
+      "step": 6440
+    },
+    {
+      "epoch": 0.1240497930099384,
+      "grad_norm": 581.4091063182992,
+      "learning_rate": 1.935220022674308e-05,
+      "loss": 337.5883,
+      "step": 6450
+    },
+    {
+      "epoch": 0.12424211827041894,
+      "grad_norm": 616.7900679575442,
+      "learning_rate": 1.9350037563123947e-05,
+      "loss": 331.99,
+      "step": 6460
+    },
+    {
+      "epoch": 0.12443444353089946,
+      "grad_norm": 599.017758292423,
+      "learning_rate": 1.9347871416770245e-05,
+      "loss": 348.6796,
+      "step": 6470
+    },
+    {
+      "epoch": 0.12462676879137999,
+      "grad_norm": 634.0349459514991,
+      "learning_rate": 1.9345701788488825e-05,
+      "loss": 336.4672,
+      "step": 6480
+    },
+    {
+      "epoch": 0.12481909405186051,
+      "grad_norm": 610.0361773632778,
+      "learning_rate": 1.9343528679087837e-05,
+      "loss": 345.7589,
+      "step": 6490
+    },
+    {
+      "epoch": 0.12501141931234103,
+      "grad_norm": 617.2343731916664,
+      "learning_rate": 1.934135208937673e-05,
+      "loss": 342.7926,
+      "step": 6500
+    },
+    {
+      "epoch": 0.12520374457282155,
+      "grad_norm": 616.367124603368,
+      "learning_rate": 1.9339172020166245e-05,
+      "loss": 351.4255,
+      "step": 6510
+    },
+    {
+      "epoch": 0.1253960698333021,
+      "grad_norm": 632.0282420968152,
+      "learning_rate": 1.933698847226843e-05,
+      "loss": 346.1836,
+      "step": 6520
+    },
+    {
+      "epoch": 0.1255883950937826,
+      "grad_norm": 687.1816525001773,
+      "learning_rate": 1.9334801446496606e-05,
+      "loss": 339.5251,
+      "step": 6530
+    },
+    {
+      "epoch": 0.12578072035426313,
+      "grad_norm": 597.5403840054404,
+      "learning_rate": 1.933261094366542e-05,
+      "loss": 343.9508,
+      "step": 6540
+    },
+    {
+      "epoch": 0.12597304561474365,
+      "grad_norm": 676.3999630134922,
+      "learning_rate": 1.9330416964590787e-05,
+      "loss": 331.0079,
+      "step": 6550
+    },
+    {
+      "epoch": 0.12616537087522417,
+      "grad_norm": 701.7101529357468,
+      "learning_rate": 1.932821951008993e-05,
+      "loss": 352.7236,
+      "step": 6560
+    },
+    {
+      "epoch": 0.12635769613570472,
+      "grad_norm": 630.7615641777062,
+      "learning_rate": 1.9326018580981373e-05,
+      "loss": 341.4861,
+      "step": 6570
+    },
+    {
+      "epoch": 0.12655002139618524,
+      "grad_norm": 752.0100341378687,
+      "learning_rate": 1.9323814178084914e-05,
+      "loss": 336.0607,
+      "step": 6580
+    },
+    {
+      "epoch": 0.12674234665666576,
+      "grad_norm": 625.633523638851,
+      "learning_rate": 1.9321606302221662e-05,
+      "loss": 346.4885,
+      "step": 6590
+    },
+    {
+      "epoch": 0.12693467191714627,
+      "grad_norm": 602.5569802194531,
+      "learning_rate": 1.9319394954214013e-05,
+      "loss": 347.0677,
+      "step": 6600
+    },
+    {
+      "epoch": 0.1271269971776268,
+      "grad_norm": 640.8650204211425,
+      "learning_rate": 1.931718013488566e-05,
+      "loss": 343.656,
+      "step": 6610
+    },
+    {
+      "epoch": 0.12731932243810734,
+      "grad_norm": 659.4461697183357,
+      "learning_rate": 1.9314961845061584e-05,
+      "loss": 357.7287,
+      "step": 6620
+    },
+    {
+      "epoch": 0.12751164769858786,
+      "grad_norm": 902.6648312801847,
+      "learning_rate": 1.9312740085568063e-05,
+      "loss": 354.6147,
+      "step": 6630
+    },
+    {
+      "epoch": 0.12770397295906838,
+      "grad_norm": 718.5158139674871,
+      "learning_rate": 1.9310514857232666e-05,
+      "loss": 335.9984,
+      "step": 6640
+    },
+    {
+      "epoch": 0.1278962982195489,
+      "grad_norm": 633.9312661204737,
+      "learning_rate": 1.930828616088425e-05,
+      "loss": 330.3928,
+      "step": 6650
+    },
+    {
+      "epoch": 0.12808862348002942,
+      "grad_norm": 600.8964809056338,
+      "learning_rate": 1.9306053997352973e-05,
+      "loss": 336.6688,
+      "step": 6660
+    },
+    {
+      "epoch": 0.12828094874050994,
+      "grad_norm": 596.5066188530008,
+      "learning_rate": 1.9303818367470274e-05,
+      "loss": 338.1062,
+      "step": 6670
+    },
+    {
+      "epoch": 0.12847327400099048,
+      "grad_norm": 776.3918205095136,
+      "learning_rate": 1.9301579272068894e-05,
+      "loss": 340.7341,
+      "step": 6680
+    },
+    {
+      "epoch": 0.128665599261471,
+      "grad_norm": 623.1809290465961,
+      "learning_rate": 1.9299336711982853e-05,
+      "loss": 342.0376,
+      "step": 6690
+    },
+    {
+      "epoch": 0.12885792452195152,
+      "grad_norm": 596.649370007816,
+      "learning_rate": 1.9297090688047473e-05,
+      "loss": 342.6335,
+      "step": 6700
+    },
+    {
+      "epoch": 0.12905024978243204,
+      "grad_norm": 629.7070535141022,
+      "learning_rate": 1.929484120109936e-05,
+      "loss": 338.046,
+      "step": 6710
+    },
+    {
+      "epoch": 0.12924257504291256,
+      "grad_norm": 634.0553810209492,
+      "learning_rate": 1.9292588251976404e-05,
+      "loss": 334.4757,
+      "step": 6720
+    },
+    {
+      "epoch": 0.1294349003033931,
+      "grad_norm": 710.2582775948679,
+      "learning_rate": 1.92903318415178e-05,
+      "loss": 342.4301,
+      "step": 6730
+    },
+    {
+      "epoch": 0.12962722556387363,
+      "grad_norm": 715.7001005567705,
+      "learning_rate": 1.9288071970564015e-05,
+      "loss": 350.7262,
+      "step": 6740
+    },
+    {
+      "epoch": 0.12981955082435415,
+      "grad_norm": 621.9873318526683,
+      "learning_rate": 1.9285808639956823e-05,
+      "loss": 343.0189,
+      "step": 6750
+    },
+    {
+      "epoch": 0.13001187608483467,
+      "grad_norm": 610.5442544220016,
+      "learning_rate": 1.9283541850539272e-05,
+      "loss": 340.7758,
+      "step": 6760
+    },
+    {
+      "epoch": 0.13020420134531518,
+      "grad_norm": 687.8958445182545,
+      "learning_rate": 1.92812716031557e-05,
+      "loss": 333.1836,
+      "step": 6770
+    },
+    {
+      "epoch": 0.13039652660579573,
+      "grad_norm": 649.6750384549912,
+      "learning_rate": 1.9278997898651746e-05,
+      "loss": 342.2303,
+      "step": 6780
+    },
+    {
+      "epoch": 0.13058885186627625,
+      "grad_norm": 604.4225607392067,
+      "learning_rate": 1.9276720737874327e-05,
+      "loss": 340.7247,
+      "step": 6790
+    },
+    {
+      "epoch": 0.13078117712675677,
+      "grad_norm": 602.3709031839262,
+      "learning_rate": 1.9274440121671637e-05,
+      "loss": 332.4386,
+      "step": 6800
+    },
+    {
+      "epoch": 0.1309735023872373,
+      "grad_norm": 617.7365851893695,
+      "learning_rate": 1.9272156050893173e-05,
+      "loss": 340.4061,
+      "step": 6810
+    },
+    {
+      "epoch": 0.1311658276477178,
+      "grad_norm": 794.7884869155839,
+      "learning_rate": 1.926986852638972e-05,
+      "loss": 341.3555,
+      "step": 6820
+    },
+    {
+      "epoch": 0.13135815290819836,
+      "grad_norm": 830.0588303420416,
+      "learning_rate": 1.926757754901333e-05,
+      "loss": 336.2551,
+      "step": 6830
+    },
+    {
+      "epoch": 0.13155047816867887,
+      "grad_norm": 710.8954264814854,
+      "learning_rate": 1.926528311961737e-05,
+      "loss": 342.6202,
+      "step": 6840
+    },
+    {
+      "epoch": 0.1317428034291594,
+      "grad_norm": 617.4477314684151,
+      "learning_rate": 1.9262985239056463e-05,
+      "loss": 345.4343,
+      "step": 6850
+    },
+    {
+      "epoch": 0.1319351286896399,
+      "grad_norm": 675.1644760008425,
+      "learning_rate": 1.9260683908186544e-05,
+      "loss": 336.6608,
+      "step": 6860
+    },
+    {
+      "epoch": 0.13212745395012043,
+      "grad_norm": 604.4334772531856,
+      "learning_rate": 1.9258379127864808e-05,
+      "loss": 334.0639,
+      "step": 6870
+    },
+    {
+      "epoch": 0.13231977921060098,
+      "grad_norm": 774.9197996782158,
+      "learning_rate": 1.925607089894976e-05,
+      "loss": 334.4737,
+      "step": 6880
+    },
+    {
+      "epoch": 0.1325121044710815,
+      "grad_norm": 642.9739454617887,
+      "learning_rate": 1.9253759222301168e-05,
+      "loss": 338.4757,
+      "step": 6890
+    },
+    {
+      "epoch": 0.13270442973156202,
+      "grad_norm": 641.4484476637223,
+      "learning_rate": 1.9251444098780095e-05,
+      "loss": 338.1292,
+      "step": 6900
+    },
+    {
+      "epoch": 0.13289675499204254,
+      "grad_norm": 591.0390873304048,
+      "learning_rate": 1.924912552924889e-05,
+      "loss": 336.706,
+      "step": 6910
+    },
+    {
+      "epoch": 0.13308908025252306,
+      "grad_norm": 736.4347602075876,
+      "learning_rate": 1.924680351457118e-05,
+      "loss": 340.7009,
+      "step": 6920
+    },
+    {
+      "epoch": 0.1332814055130036,
+      "grad_norm": 640.9332097910735,
+      "learning_rate": 1.9244478055611875e-05,
+      "loss": 331.8275,
+      "step": 6930
+    },
+    {
+      "epoch": 0.13347373077348412,
+      "grad_norm": 615.3553578255887,
+      "learning_rate": 1.9242149153237175e-05,
+      "loss": 346.1782,
+      "step": 6940
+    },
+    {
+      "epoch": 0.13366605603396464,
+      "grad_norm": 666.5546352548604,
+      "learning_rate": 1.923981680831455e-05,
+      "loss": 329.8955,
+      "step": 6950
+    },
+    {
+      "epoch": 0.13385838129444516,
+      "grad_norm": 578.9245733843179,
+      "learning_rate": 1.923748102171277e-05,
+      "loss": 342.6567,
+      "step": 6960
+    },
+    {
+      "epoch": 0.13405070655492568,
+      "grad_norm": 684.1602269539654,
+      "learning_rate": 1.9235141794301867e-05,
+      "loss": 339.303,
+      "step": 6970
+    },
+    {
+      "epoch": 0.13424303181540623,
+      "grad_norm": 772.5479196530189,
+      "learning_rate": 1.9232799126953173e-05,
+      "loss": 341.7923,
+      "step": 6980
+    },
+    {
+      "epoch": 0.13443535707588675,
+      "grad_norm": 780.8375673557058,
+      "learning_rate": 1.9230453020539285e-05,
+      "loss": 340.7893,
+      "step": 6990
+    },
+    {
+      "epoch": 0.13462768233636727,
+      "grad_norm": 680.6012761445076,
+      "learning_rate": 1.9228103475934096e-05,
+      "loss": 330.813,
+      "step": 7000
+    },
+    {
+      "epoch": 0.13482000759684779,
+      "grad_norm": 620.675619049162,
+      "learning_rate": 1.9225750494012767e-05,
+      "loss": 344.353,
+      "step": 7010
+    },
+    {
+      "epoch": 0.1350123328573283,
+      "grad_norm": 585.4078426498832,
+      "learning_rate": 1.9223394075651748e-05,
+      "loss": 342.1984,
+      "step": 7020
+    },
+    {
+      "epoch": 0.13520465811780885,
+      "grad_norm": 638.19237560778,
+      "learning_rate": 1.9221034221728764e-05,
+      "loss": 335.0086,
+      "step": 7030
+    },
+    {
+      "epoch": 0.13539698337828937,
+      "grad_norm": 642.7386042549266,
+      "learning_rate": 1.9218670933122826e-05,
+      "loss": 339.8461,
+      "step": 7040
+    },
+    {
+      "epoch": 0.1355893086387699,
+      "grad_norm": 681.7536763670116,
+      "learning_rate": 1.9216304210714213e-05,
+      "loss": 330.6961,
+      "step": 7050
+    },
+    {
+      "epoch": 0.1357816338992504,
+      "grad_norm": 620.7059218852446,
+      "learning_rate": 1.9213934055384498e-05,
+      "loss": 351.9697,
+      "step": 7060
+    },
+    {
+      "epoch": 0.13597395915973093,
+      "grad_norm": 576.6468376808148,
+      "learning_rate": 1.9211560468016516e-05,
+      "loss": 343.2772,
+      "step": 7070
+    },
+    {
+      "epoch": 0.13616628442021148,
+      "grad_norm": 617.909082245975,
+      "learning_rate": 1.9209183449494397e-05,
+      "loss": 331.628,
+      "step": 7080
+    },
+    {
+      "epoch": 0.136358609680692,
+      "grad_norm": 705.1055113964394,
+      "learning_rate": 1.9206803000703534e-05,
+      "loss": 340.7463,
+      "step": 7090
+    },
+    {
+      "epoch": 0.1365509349411725,
+      "grad_norm": 675.4917083850122,
+      "learning_rate": 1.9204419122530614e-05,
+      "loss": 337.46,
+      "step": 7100
+    },
+    {
+      "epoch": 0.13674326020165303,
+      "grad_norm": 562.871960310014,
+      "learning_rate": 1.9202031815863583e-05,
+      "loss": 336.0697,
+      "step": 7110
+    },
+    {
+      "epoch": 0.13693558546213355,
+      "grad_norm": 667.0141217498331,
+      "learning_rate": 1.919964108159168e-05,
+      "loss": 337.3682,
+      "step": 7120
+    },
+    {
+      "epoch": 0.13712791072261407,
+      "grad_norm": 641.5830951888042,
+      "learning_rate": 1.9197246920605408e-05,
+      "loss": 328.0353,
+      "step": 7130
+    },
+    {
+      "epoch": 0.13732023598309462,
+      "grad_norm": 710.6090854819964,
+      "learning_rate": 1.9194849333796557e-05,
+      "loss": 333.5805,
+      "step": 7140
+    },
+    {
+      "epoch": 0.13751256124357514,
+      "grad_norm": 705.2696170093016,
+      "learning_rate": 1.9192448322058187e-05,
+      "loss": 340.3354,
+      "step": 7150
+    },
+    {
+      "epoch": 0.13770488650405566,
+      "grad_norm": 707.3442795351257,
+      "learning_rate": 1.9190043886284635e-05,
+      "loss": 336.0162,
+      "step": 7160
+    },
+    {
+      "epoch": 0.13789721176453618,
+      "grad_norm": 606.1381628215864,
+      "learning_rate": 1.9187636027371518e-05,
+      "loss": 328.333,
+      "step": 7170
+    },
+    {
+      "epoch": 0.1380895370250167,
+      "grad_norm": 766.2171264805419,
+      "learning_rate": 1.9185224746215714e-05,
+      "loss": 346.3918,
+      "step": 7180
+    },
+    {
+      "epoch": 0.13828186228549724,
+      "grad_norm": 709.0230045615772,
+      "learning_rate": 1.9182810043715394e-05,
+      "loss": 333.1865,
+      "step": 7190
+    },
+    {
+      "epoch": 0.13847418754597776,
+      "grad_norm": 620.7512744942927,
+      "learning_rate": 1.9180391920769993e-05,
+      "loss": 333.1366,
+      "step": 7200
+    },
+    {
+      "epoch": 0.13866651280645828,
+      "grad_norm": 665.6664450047932,
+      "learning_rate": 1.9177970378280215e-05,
+      "loss": 335.7814,
+      "step": 7210
+    },
+    {
+      "epoch": 0.1388588380669388,
+      "grad_norm": 611.292760408907,
+      "learning_rate": 1.9175545417148056e-05,
+      "loss": 343.0603,
+      "step": 7220
+    },
+    {
+      "epoch": 0.13905116332741932,
+      "grad_norm": 734.063965140063,
+      "learning_rate": 1.9173117038276766e-05,
+      "loss": 340.2624,
+      "step": 7230
+    },
+    {
+      "epoch": 0.13924348858789987,
+      "grad_norm": 720.0361974054264,
+      "learning_rate": 1.9170685242570878e-05,
+      "loss": 343.442,
+      "step": 7240
+    },
+    {
+      "epoch": 0.13943581384838039,
+      "grad_norm": 594.3230591421019,
+      "learning_rate": 1.9168250030936195e-05,
+      "loss": 331.0506,
+      "step": 7250
+    },
+    {
+      "epoch": 0.1396281391088609,
+      "grad_norm": 664.3136613492181,
+      "learning_rate": 1.91658114042798e-05,
+      "loss": 337.0721,
+      "step": 7260
+    },
+    {
+      "epoch": 0.13982046436934142,
+      "grad_norm": 607.1773913861623,
+      "learning_rate": 1.9163369363510026e-05,
+      "loss": 330.6804,
+      "step": 7270
+    },
+    {
+      "epoch": 0.14001278962982194,
+      "grad_norm": 599.1714129942029,
+      "learning_rate": 1.916092390953651e-05,
+      "loss": 335.4198,
+      "step": 7280
+    },
+    {
+      "epoch": 0.1402051148903025,
+      "grad_norm": 632.3992832483767,
+      "learning_rate": 1.915847504327013e-05,
+      "loss": 334.0288,
+      "step": 7290
+    },
+    {
+      "epoch": 0.140397440150783,
+      "grad_norm": 674.1409222683313,
+      "learning_rate": 1.9156022765623057e-05,
+      "loss": 334.1434,
+      "step": 7300
+    },
+    {
+      "epoch": 0.14058976541126353,
+      "grad_norm": 669.1862374108495,
+      "learning_rate": 1.9153567077508718e-05,
+      "loss": 329.2249,
+      "step": 7310
+    },
+    {
+      "epoch": 0.14078209067174405,
+      "grad_norm": 638.9009796322381,
+      "learning_rate": 1.9151107979841824e-05,
+      "loss": 332.2039,
+      "step": 7320
+    },
+    {
+      "epoch": 0.14097441593222457,
+      "grad_norm": 595.6871552404173,
+      "learning_rate": 1.9148645473538338e-05,
+      "loss": 328.8862,
+      "step": 7330
+    },
+    {
+      "epoch": 0.14116674119270511,
+      "grad_norm": 591.4480288783271,
+      "learning_rate": 1.9146179559515507e-05,
+      "loss": 329.3257,
+      "step": 7340
+    },
+    {
+      "epoch": 0.14135906645318563,
+      "grad_norm": 619.5815660735237,
+      "learning_rate": 1.9143710238691847e-05,
+      "loss": 331.3541,
+      "step": 7350
+    },
+    {
+      "epoch": 0.14155139171366615,
+      "grad_norm": 588.7563903968278,
+      "learning_rate": 1.9141237511987137e-05,
+      "loss": 329.303,
+      "step": 7360
+    },
+    {
+      "epoch": 0.14174371697414667,
+      "grad_norm": 666.945728559143,
+      "learning_rate": 1.9138761380322425e-05,
+      "loss": 340.8805,
+      "step": 7370
+    },
+    {
+      "epoch": 0.1419360422346272,
+      "grad_norm": 655.9890050856072,
+      "learning_rate": 1.913628184462003e-05,
+      "loss": 332.1415,
+      "step": 7380
+    },
+    {
+      "epoch": 0.14212836749510774,
+      "grad_norm": 658.4454242978924,
+      "learning_rate": 1.913379890580354e-05,
+      "loss": 336.3184,
+      "step": 7390
+    },
+    {
+      "epoch": 0.14232069275558826,
+      "grad_norm": 665.714018548699,
+      "learning_rate": 1.9131312564797805e-05,
+      "loss": 333.4196,
+      "step": 7400
+    },
+    {
+      "epoch": 0.14251301801606878,
+      "grad_norm": 686.1007421990183,
+      "learning_rate": 1.912882282252895e-05,
+      "loss": 323.0851,
+      "step": 7410
+    },
+    {
+      "epoch": 0.1427053432765493,
+      "grad_norm": 608.6239734024941,
+      "learning_rate": 1.9126329679924364e-05,
+      "loss": 337.3194,
+      "step": 7420
+    },
+    {
+      "epoch": 0.14289766853702981,
+      "grad_norm": 604.2857848168327,
+      "learning_rate": 1.9123833137912693e-05,
+      "loss": 334.6706,
+      "step": 7430
+    },
+    {
+      "epoch": 0.14308999379751036,
+      "grad_norm": 657.9875766448478,
+      "learning_rate": 1.9121333197423867e-05,
+      "loss": 350.4117,
+      "step": 7440
+    },
+    {
+      "epoch": 0.14328231905799088,
+      "grad_norm": 774.1198900493707,
+      "learning_rate": 1.9118829859389067e-05,
+      "loss": 344.4096,
+      "step": 7450
+    },
+    {
+      "epoch": 0.1434746443184714,
+      "grad_norm": 698.4358990600406,
+      "learning_rate": 1.9116323124740748e-05,
+      "loss": 331.8275,
+      "step": 7460
+    },
+    {
+      "epoch": 0.14366696957895192,
+      "grad_norm": 716.0417295608296,
+      "learning_rate": 1.9113812994412627e-05,
+      "loss": 334.3449,
+      "step": 7470
+    },
+    {
+      "epoch": 0.14385929483943244,
+      "grad_norm": 674.3569202341058,
+      "learning_rate": 1.911129946933968e-05,
+      "loss": 335.0988,
+      "step": 7480
+    },
+    {
+      "epoch": 0.14405162009991299,
+      "grad_norm": 577.068949304984,
+      "learning_rate": 1.9108782550458164e-05,
+      "loss": 345.8265,
+      "step": 7490
+    },
+    {
+      "epoch": 0.1442439453603935,
+      "grad_norm": 586.7220912778528,
+      "learning_rate": 1.9106262238705583e-05,
+      "loss": 341.1002,
+      "step": 7500
+    },
+    {
+      "epoch": 0.14443627062087402,
+      "grad_norm": 540.3048891693815,
+      "learning_rate": 1.9103738535020713e-05,
+      "loss": 339.9908,
+      "step": 7510
+    },
+    {
+      "epoch": 0.14462859588135454,
+      "grad_norm": 598.8228092314863,
+      "learning_rate": 1.910121144034359e-05,
+      "loss": 327.1496,
+      "step": 7520
+    },
+    {
+      "epoch": 0.14482092114183506,
+      "grad_norm": 643.0008655568517,
+      "learning_rate": 1.909868095561552e-05,
+      "loss": 333.6696,
+      "step": 7530
+    },
+    {
+      "epoch": 0.1450132464023156,
+      "grad_norm": 558.8109142460015,
+      "learning_rate": 1.9096147081779063e-05,
+      "loss": 341.1401,
+      "step": 7540
+    },
+    {
+      "epoch": 0.14520557166279613,
+      "grad_norm": 636.1715695277825,
+      "learning_rate": 1.9093609819778044e-05,
+      "loss": 334.1333,
+      "step": 7550
+    },
+    {
+      "epoch": 0.14539789692327665,
+      "grad_norm": 584.1294965803066,
+      "learning_rate": 1.9091069170557554e-05,
+      "loss": 340.2397,
+      "step": 7560
+    },
+    {
+      "epoch": 0.14559022218375717,
+      "grad_norm": 588.7759487555222,
+      "learning_rate": 1.9088525135063944e-05,
+      "loss": 334.4876,
+      "step": 7570
+    },
+    {
+      "epoch": 0.1457825474442377,
+      "grad_norm": 586.7412371937382,
+      "learning_rate": 1.9085977714244822e-05,
+      "loss": 331.9889,
+      "step": 7580
+    },
+    {
+      "epoch": 0.1459748727047182,
+      "grad_norm": 623.1962765941188,
+      "learning_rate": 1.908342690904906e-05,
+      "loss": 337.2166,
+      "step": 7590
+    },
+    {
+      "epoch": 0.14616719796519875,
+      "grad_norm": 605.0510054429278,
+      "learning_rate": 1.9080872720426793e-05,
+      "loss": 333.1549,
+      "step": 7600
+    },
+    {
+      "epoch": 0.14635952322567927,
+      "grad_norm": 652.5794801569881,
+      "learning_rate": 1.9078315149329413e-05,
+      "loss": 322.555,
+      "step": 7610
+    },
+    {
+      "epoch": 0.1465518484861598,
+      "grad_norm": 597.5804096642445,
+      "learning_rate": 1.9075754196709574e-05,
+      "loss": 339.9878,
+      "step": 7620
+    },
+    {
+      "epoch": 0.1467441737466403,
+      "grad_norm": 582.7306056941077,
+      "learning_rate": 1.9073189863521184e-05,
+      "loss": 330.2846,
+      "step": 7630
+    },
+    {
+      "epoch": 0.14693649900712083,
+      "grad_norm": 653.4866928502033,
+      "learning_rate": 1.9070622150719423e-05,
+      "loss": 334.186,
+      "step": 7640
+    },
+    {
+      "epoch": 0.14712882426760138,
+      "grad_norm": 656.2065448271608,
+      "learning_rate": 1.9068051059260716e-05,
+      "loss": 333.487,
+      "step": 7650
+    },
+    {
+      "epoch": 0.1473211495280819,
+      "grad_norm": 739.5742222454226,
+      "learning_rate": 1.9065476590102752e-05,
+      "loss": 338.4543,
+      "step": 7660
+    },
+    {
+      "epoch": 0.14751347478856242,
+      "grad_norm": 659.8715115346957,
+      "learning_rate": 1.906289874420448e-05,
+      "loss": 344.8408,
+      "step": 7670
+    },
+    {
+      "epoch": 0.14770580004904293,
+      "grad_norm": 585.4940058032927,
+      "learning_rate": 1.9060317522526105e-05,
+      "loss": 325.9033,
+      "step": 7680
+    },
+    {
+      "epoch": 0.14789812530952345,
+      "grad_norm": 607.4854347187479,
+      "learning_rate": 1.905773292602909e-05,
+      "loss": 335.7657,
+      "step": 7690
+    },
+    {
+      "epoch": 0.148090450570004,
+      "grad_norm": 667.362261826713,
+      "learning_rate": 1.905514495567615e-05,
+      "loss": 329.7076,
+      "step": 7700
+    },
+    {
+      "epoch": 0.14828277583048452,
+      "grad_norm": 536.0276096399463,
+      "learning_rate": 1.9052553612431268e-05,
+      "loss": 345.2904,
+      "step": 7710
+    },
+    {
+      "epoch": 0.14847510109096504,
+      "grad_norm": 614.1735022171694,
+      "learning_rate": 1.9049958897259674e-05,
+      "loss": 327.6999,
+      "step": 7720
+    },
+    {
+      "epoch": 0.14866742635144556,
+      "grad_norm": 565.0650579387441,
+      "learning_rate": 1.904736081112785e-05,
+      "loss": 332.4991,
+      "step": 7730
+    },
+    {
+      "epoch": 0.14885975161192608,
+      "grad_norm": 698.5517829582707,
+      "learning_rate": 1.9044759355003552e-05,
+      "loss": 336.9797,
+      "step": 7740
+    },
+    {
+      "epoch": 0.14905207687240662,
+      "grad_norm": 654.2539263044354,
+      "learning_rate": 1.904215452985577e-05,
+      "loss": 330.1729,
+      "step": 7750
+    },
+    {
+      "epoch": 0.14924440213288714,
+      "grad_norm": 657.149243090159,
+      "learning_rate": 1.9039546336654765e-05,
+      "loss": 340.3054,
+      "step": 7760
+    },
+    {
+      "epoch": 0.14943672739336766,
+      "grad_norm": 595.0120563008803,
+      "learning_rate": 1.903693477637204e-05,
+      "loss": 327.7932,
+      "step": 7770
+    },
+    {
+      "epoch": 0.14962905265384818,
+      "grad_norm": 586.3951566892398,
+      "learning_rate": 1.903431984998036e-05,
+      "loss": 331.5635,
+      "step": 7780
+    },
+    {
+      "epoch": 0.1498213779143287,
+      "grad_norm": 652.1093705478022,
+      "learning_rate": 1.9031701558453747e-05,
+      "loss": 348.5615,
+      "step": 7790
+    },
+    {
+      "epoch": 0.15001370317480925,
+      "grad_norm": 657.4240718892927,
+      "learning_rate": 1.902907990276746e-05,
+      "loss": 321.6034,
+      "step": 7800
+    },
+    {
+      "epoch": 0.15020602843528977,
+      "grad_norm": 695.5448517014397,
+      "learning_rate": 1.9026454883898036e-05,
+      "loss": 338.977,
+      "step": 7810
+    },
+    {
+      "epoch": 0.1503983536957703,
+      "grad_norm": 619.4563587059992,
+      "learning_rate": 1.902382650282324e-05,
+      "loss": 326.9431,
+      "step": 7820
+    },
+    {
+      "epoch": 0.1505906789562508,
+      "grad_norm": 610.2707928349413,
+      "learning_rate": 1.902119476052211e-05,
+      "loss": 322.5898,
+      "step": 7830
+    },
+    {
+      "epoch": 0.15078300421673133,
+      "grad_norm": 689.2975637286215,
+      "learning_rate": 1.9018559657974918e-05,
+      "loss": 332.3525,
+      "step": 7840
+    },
+    {
+      "epoch": 0.15097532947721187,
+      "grad_norm": 720.2792218360873,
+      "learning_rate": 1.90159211961632e-05,
+      "loss": 363.9882,
+      "step": 7850
+    },
+    {
+      "epoch": 0.1511676547376924,
+      "grad_norm": 757.0191041019314,
+      "learning_rate": 1.901327937606974e-05,
+      "loss": 331.6766,
+      "step": 7860
+    },
+    {
+      "epoch": 0.1513599799981729,
+      "grad_norm": 668.9888292883863,
+      "learning_rate": 1.901063419867857e-05,
+      "loss": 337.8688,
+      "step": 7870
+    },
+    {
+      "epoch": 0.15155230525865343,
+      "grad_norm": 609.4311837295198,
+      "learning_rate": 1.900798566497498e-05,
+      "loss": 330.1886,
+      "step": 7880
+    },
+    {
+      "epoch": 0.15174463051913395,
+      "grad_norm": 592.320955820024,
+      "learning_rate": 1.9005333775945496e-05,
+      "loss": 333.4431,
+      "step": 7890
+    },
+    {
+      "epoch": 0.1519369557796145,
+      "grad_norm": 652.1124104405058,
+      "learning_rate": 1.9002678532577915e-05,
+      "loss": 325.7347,
+      "step": 7900
+    },
+    {
+      "epoch": 0.15212928104009502,
+      "grad_norm": 617.7623400505815,
+      "learning_rate": 1.900001993586126e-05,
+      "loss": 337.0063,
+      "step": 7910
+    },
+    {
+      "epoch": 0.15232160630057553,
+      "grad_norm": 601.888597412138,
+      "learning_rate": 1.8997357986785822e-05,
+      "loss": 341.1846,
+      "step": 7920
+    },
+    {
+      "epoch": 0.15251393156105605,
+      "grad_norm": 758.839415022158,
+      "learning_rate": 1.899469268634313e-05,
+      "loss": 338.0605,
+      "step": 7930
+    },
+    {
+      "epoch": 0.15270625682153657,
+      "grad_norm": 611.9683144775163,
+      "learning_rate": 1.8992024035525964e-05,
+      "loss": 325.2646,
+      "step": 7940
+    },
+    {
+      "epoch": 0.15289858208201712,
+      "grad_norm": 607.2554279084135,
+      "learning_rate": 1.8989352035328352e-05,
+      "loss": 335.0041,
+      "step": 7950
+    },
+    {
+      "epoch": 0.15309090734249764,
+      "grad_norm": 641.7327001857283,
+      "learning_rate": 1.8986676686745572e-05,
+      "loss": 333.3883,
+      "step": 7960
+    },
+    {
+      "epoch": 0.15328323260297816,
+      "grad_norm": 654.5740927509847,
+      "learning_rate": 1.8983997990774145e-05,
+      "loss": 332.6246,
+      "step": 7970
+    },
+    {
+      "epoch": 0.15347555786345868,
+      "grad_norm": 589.067280724116,
+      "learning_rate": 1.8981315948411842e-05,
+      "loss": 336.7391,
+      "step": 7980
+    },
+    {
+      "epoch": 0.1536678831239392,
+      "grad_norm": 645.77199995531,
+      "learning_rate": 1.897863056065768e-05,
+      "loss": 337.1307,
+      "step": 7990
+    },
+    {
+      "epoch": 0.15386020838441974,
+      "grad_norm": 587.5748922401691,
+      "learning_rate": 1.8975941828511923e-05,
+      "loss": 329.6803,
+      "step": 8000
+    },
+    {
+      "epoch": 0.15405253364490026,
+      "grad_norm": 624.962479288896,
+      "learning_rate": 1.8973249752976075e-05,
+      "loss": 330.1949,
+      "step": 8010
+    },
+    {
+      "epoch": 0.15424485890538078,
+      "grad_norm": 678.6662294430486,
+      "learning_rate": 1.8970554335052897e-05,
+      "loss": 327.6478,
+      "step": 8020
+    },
+    {
+      "epoch": 0.1544371841658613,
+      "grad_norm": 617.7029085283698,
+      "learning_rate": 1.8967855575746375e-05,
+      "loss": 329.804,
+      "step": 8030
+    },
+    {
+      "epoch": 0.15462950942634182,
+      "grad_norm": 763.7413110459936,
+      "learning_rate": 1.8965153476061763e-05,
+      "loss": 335.0965,
+      "step": 8040
+    },
+    {
+      "epoch": 0.15482183468682234,
+      "grad_norm": 618.8287237521282,
+      "learning_rate": 1.896244803700555e-05,
+      "loss": 330.6702,
+      "step": 8050
+    },
+    {
+      "epoch": 0.1550141599473029,
+      "grad_norm": 625.8824151776017,
+      "learning_rate": 1.8959739259585458e-05,
+      "loss": 329.7756,
+      "step": 8060
+    },
+    {
+      "epoch": 0.1552064852077834,
+      "grad_norm": 630.6961204218625,
+      "learning_rate": 1.895702714481047e-05,
+      "loss": 319.1156,
+      "step": 8070
+    },
+    {
+      "epoch": 0.15539881046826393,
+      "grad_norm": 575.5776074392129,
+      "learning_rate": 1.8954311693690798e-05,
+      "loss": 317.7853,
+      "step": 8080
+    },
+    {
+      "epoch": 0.15559113572874445,
+      "grad_norm": 705.1483458938271,
+      "learning_rate": 1.8951592907237906e-05,
+      "loss": 337.7698,
+      "step": 8090
+    },
+    {
+      "epoch": 0.15578346098922496,
+      "grad_norm": 789.2054488690464,
+      "learning_rate": 1.8948870786464496e-05,
+      "loss": 339.1818,
+      "step": 8100
+    },
+    {
+      "epoch": 0.1559757862497055,
+      "grad_norm": 604.9163142409899,
+      "learning_rate": 1.8946145332384515e-05,
+      "loss": 321.4155,
+      "step": 8110
+    },
+    {
+      "epoch": 0.15616811151018603,
+      "grad_norm": 761.7838080703335,
+      "learning_rate": 1.8943416546013148e-05,
+      "loss": 340.1508,
+      "step": 8120
+    },
+    {
+      "epoch": 0.15636043677066655,
+      "grad_norm": 687.0311909921215,
+      "learning_rate": 1.894068442836682e-05,
+      "loss": 327.8359,
+      "step": 8130
+    },
+    {
+      "epoch": 0.15655276203114707,
+      "grad_norm": 665.0449909303669,
+      "learning_rate": 1.8937948980463207e-05,
+      "loss": 325.6881,
+      "step": 8140
+    },
+    {
+      "epoch": 0.1567450872916276,
+      "grad_norm": 680.2703750609774,
+      "learning_rate": 1.893521020332121e-05,
+      "loss": 339.19,
+      "step": 8150
+    },
+    {
+      "epoch": 0.15693741255210814,
+      "grad_norm": 712.2445486851191,
+      "learning_rate": 1.8932468097960988e-05,
+      "loss": 325.6005,
+      "step": 8160
+    },
+    {
+      "epoch": 0.15712973781258865,
+      "grad_norm": 634.5783082217703,
+      "learning_rate": 1.892972266540392e-05,
+      "loss": 317.995,
+      "step": 8170
+    },
+    {
+      "epoch": 0.15732206307306917,
+      "grad_norm": 590.7140843673521,
+      "learning_rate": 1.8926973906672635e-05,
+      "loss": 321.0524,
+      "step": 8180
+    },
+    {
+      "epoch": 0.1575143883335497,
+      "grad_norm": 622.6352617042718,
+      "learning_rate": 1.892422182279101e-05,
+      "loss": 333.9439,
+      "step": 8190
+    },
+    {
+      "epoch": 0.1577067135940302,
+      "grad_norm": 612.4240165467864,
+      "learning_rate": 1.892146641478414e-05,
+      "loss": 328.3868,
+      "step": 8200
+    },
+    {
+      "epoch": 0.15789903885451076,
+      "grad_norm": 646.1644417693855,
+      "learning_rate": 1.8918707683678376e-05,
+      "loss": 319.4142,
+      "step": 8210
+    },
+    {
+      "epoch": 0.15809136411499128,
+      "grad_norm": 640.908573407253,
+      "learning_rate": 1.8915945630501296e-05,
+      "loss": 326.9317,
+      "step": 8220
+    },
+    {
+      "epoch": 0.1582836893754718,
+      "grad_norm": 692.7289847209751,
+      "learning_rate": 1.8913180256281723e-05,
+      "loss": 328.568,
+      "step": 8230
+    },
+    {
+      "epoch": 0.15847601463595232,
+      "grad_norm": 594.7027713128756,
+      "learning_rate": 1.8910411562049706e-05,
+      "loss": 321.2016,
+      "step": 8240
+    },
+    {
+      "epoch": 0.15866833989643284,
+      "grad_norm": 568.8287473525063,
+      "learning_rate": 1.8907639548836548e-05,
+      "loss": 325.0564,
+      "step": 8250
+    },
+    {
+      "epoch": 0.15886066515691338,
+      "grad_norm": 639.3651153095016,
+      "learning_rate": 1.8904864217674766e-05,
+      "loss": 339.9999,
+      "step": 8260
+    },
+    {
+      "epoch": 0.1590529904173939,
+      "grad_norm": 693.6719533383076,
+      "learning_rate": 1.8902085569598136e-05,
+      "loss": 319.3183,
+      "step": 8270
+    },
+    {
+      "epoch": 0.15924531567787442,
+      "grad_norm": 566.8352443265295,
+      "learning_rate": 1.889930360564165e-05,
+      "loss": 322.6493,
+      "step": 8280
+    },
+    {
+      "epoch": 0.15943764093835494,
+      "grad_norm": 550.8923192127909,
+      "learning_rate": 1.8896518326841554e-05,
+      "loss": 318.822,
+      "step": 8290
+    },
+    {
+      "epoch": 0.15962996619883546,
+      "grad_norm": 636.4844212484624,
+      "learning_rate": 1.889372973423531e-05,
+      "loss": 329.6007,
+      "step": 8300
+    },
+    {
+      "epoch": 0.159822291459316,
+      "grad_norm": 523.7027264193889,
+      "learning_rate": 1.889093782886162e-05,
+      "loss": 325.9975,
+      "step": 8310
+    },
+    {
+      "epoch": 0.16001461671979653,
+      "grad_norm": 628.7320461357699,
+      "learning_rate": 1.8888142611760433e-05,
+      "loss": 324.9394,
+      "step": 8320
+    },
+    {
+      "epoch": 0.16020694198027705,
+      "grad_norm": 632.8876609487307,
+      "learning_rate": 1.8885344083972912e-05,
+      "loss": 321.4737,
+      "step": 8330
+    },
+    {
+      "epoch": 0.16039926724075756,
+      "grad_norm": 672.6796542766039,
+      "learning_rate": 1.8882542246541468e-05,
+      "loss": 341.4587,
+      "step": 8340
+    },
+    {
+      "epoch": 0.16059159250123808,
+      "grad_norm": 685.4735672135316,
+      "learning_rate": 1.887973710050974e-05,
+      "loss": 339.7623,
+      "step": 8350
+    },
+    {
+      "epoch": 0.16078391776171863,
+      "grad_norm": 656.0235413997698,
+      "learning_rate": 1.887692864692259e-05,
+      "loss": 321.4192,
+      "step": 8360
+    },
+    {
+      "epoch": 0.16097624302219915,
+      "grad_norm": 605.1988479424814,
+      "learning_rate": 1.887411688682613e-05,
+      "loss": 331.8671,
+      "step": 8370
+    },
+    {
+      "epoch": 0.16116856828267967,
+      "grad_norm": 607.0216375856667,
+      "learning_rate": 1.887130182126769e-05,
+      "loss": 335.5238,
+      "step": 8380
+    },
+    {
+      "epoch": 0.1613608935431602,
+      "grad_norm": 648.5801451386891,
+      "learning_rate": 1.8868483451295835e-05,
+      "loss": 329.0254,
+      "step": 8390
+    },
+    {
+      "epoch": 0.1615532188036407,
+      "grad_norm": 666.9006726646376,
+      "learning_rate": 1.8865661777960366e-05,
+      "loss": 324.4137,
+      "step": 8400
+    },
+    {
+      "epoch": 0.16174554406412125,
+      "grad_norm": 612.2706066542921,
+      "learning_rate": 1.88628368023123e-05,
+      "loss": 330.1027,
+      "step": 8410
+    },
+    {
+      "epoch": 0.16193786932460177,
+      "grad_norm": 1170.1623289110332,
+      "learning_rate": 1.8860008525403903e-05,
+      "loss": 344.3691,
+      "step": 8420
+    },
+    {
+      "epoch": 0.1621301945850823,
+      "grad_norm": 657.262736303299,
+      "learning_rate": 1.885717694828866e-05,
+      "loss": 328.7068,
+      "step": 8430
+    },
+    {
+      "epoch": 0.1623225198455628,
+      "grad_norm": 582.5110756513626,
+      "learning_rate": 1.8854342072021282e-05,
+      "loss": 318.0018,
+      "step": 8440
+    },
+    {
+      "epoch": 0.16251484510604333,
+      "grad_norm": 633.038781675672,
+      "learning_rate": 1.8851503897657717e-05,
+      "loss": 328.3173,
+      "step": 8450
+    },
+    {
+      "epoch": 0.16270717036652388,
+      "grad_norm": 582.4576397316638,
+      "learning_rate": 1.8848662426255135e-05,
+      "loss": 333.9782,
+      "step": 8460
+    },
+    {
+      "epoch": 0.1628994956270044,
+      "grad_norm": 591.1262231592767,
+      "learning_rate": 1.8845817658871942e-05,
+      "loss": 327.9783,
+      "step": 8470
+    },
+    {
+      "epoch": 0.16309182088748492,
+      "grad_norm": 607.7430738976485,
+      "learning_rate": 1.8842969596567765e-05,
+      "loss": 334.9482,
+      "step": 8480
+    },
+    {
+      "epoch": 0.16328414614796544,
+      "grad_norm": 596.9041121165554,
+      "learning_rate": 1.884011824040346e-05,
+      "loss": 338.8978,
+      "step": 8490
+    },
+    {
+      "epoch": 0.16347647140844596,
+      "grad_norm": 668.8208564240986,
+      "learning_rate": 1.883726359144111e-05,
+      "loss": 330.8432,
+      "step": 8500
+    },
+    {
+      "epoch": 0.16366879666892648,
+      "grad_norm": 566.7436484831923,
+      "learning_rate": 1.8834405650744023e-05,
+      "loss": 325.4643,
+      "step": 8510
+    },
+    {
+      "epoch": 0.16386112192940702,
+      "grad_norm": 650.1926791446019,
+      "learning_rate": 1.883154441937674e-05,
+      "loss": 332.9925,
+      "step": 8520
+    },
+    {
+      "epoch": 0.16405344718988754,
+      "grad_norm": 553.1903423115506,
+      "learning_rate": 1.8828679898405015e-05,
+      "loss": 316.6825,
+      "step": 8530
+    },
+    {
+      "epoch": 0.16424577245036806,
+      "grad_norm": 641.6381934900176,
+      "learning_rate": 1.8825812088895835e-05,
+      "loss": 329.6671,
+      "step": 8540
+    },
+    {
+      "epoch": 0.16443809771084858,
+      "grad_norm": 610.3182236346137,
+      "learning_rate": 1.882294099191742e-05,
+      "loss": 332.7418,
+      "step": 8550
+    },
+    {
+      "epoch": 0.1646304229713291,
+      "grad_norm": 637.8895889859295,
+      "learning_rate": 1.88200666085392e-05,
+      "loss": 324.0332,
+      "step": 8560
+    },
+    {
+      "epoch": 0.16482274823180965,
+      "grad_norm": 621.1644254638762,
+      "learning_rate": 1.8817188939831838e-05,
+      "loss": 334.8323,
+      "step": 8570
+    },
+    {
+      "epoch": 0.16501507349229017,
+      "grad_norm": 620.6175953613363,
+      "learning_rate": 1.8814307986867214e-05,
+      "loss": 318.596,
+      "step": 8580
+    },
+    {
+      "epoch": 0.16520739875277068,
+      "grad_norm": 570.7161465732909,
+      "learning_rate": 1.881142375071844e-05,
+      "loss": 308.5833,
+      "step": 8590
+    },
+    {
+      "epoch": 0.1653997240132512,
+      "grad_norm": 687.5877126737203,
+      "learning_rate": 1.8808536232459844e-05,
+      "loss": 332.4939,
+      "step": 8600
+    },
+    {
+      "epoch": 0.16559204927373172,
+      "grad_norm": 520.6326252944293,
+      "learning_rate": 1.8805645433166976e-05,
+      "loss": 337.6086,
+      "step": 8610
+    },
+    {
+      "epoch": 0.16578437453421227,
+      "grad_norm": 625.2257120664489,
+      "learning_rate": 1.8802751353916618e-05,
+      "loss": 331.5252,
+      "step": 8620
+    },
+    {
+      "epoch": 0.1659766997946928,
+      "grad_norm": 633.4463197650607,
+      "learning_rate": 1.8799853995786763e-05,
+      "loss": 327.5775,
+      "step": 8630
+    },
+    {
+      "epoch": 0.1661690250551733,
+      "grad_norm": 578.659849792283,
+      "learning_rate": 1.8796953359856626e-05,
+      "loss": 310.7596,
+      "step": 8640
+    },
+    {
+      "epoch": 0.16636135031565383,
+      "grad_norm": 635.077697877117,
+      "learning_rate": 1.879404944720665e-05,
+      "loss": 324.8619,
+      "step": 8650
+    },
+    {
+      "epoch": 0.16655367557613435,
+      "grad_norm": 638.3681282519153,
+      "learning_rate": 1.8791142258918496e-05,
+      "loss": 326.9639,
+      "step": 8660
+    },
+    {
+      "epoch": 0.1667460008366149,
+      "grad_norm": 692.6432931836723,
+      "learning_rate": 1.8788231796075037e-05,
+      "loss": 334.5431,
+      "step": 8670
+    },
+    {
+      "epoch": 0.1669383260970954,
+      "grad_norm": 599.4384153302473,
+      "learning_rate": 1.8785318059760384e-05,
+      "loss": 318.0557,
+      "step": 8680
+    },
+    {
+      "epoch": 0.16713065135757593,
+      "grad_norm": 680.5360194556843,
+      "learning_rate": 1.8782401051059838e-05,
+      "loss": 321.4521,
+      "step": 8690
+    },
+    {
+      "epoch": 0.16732297661805645,
+      "grad_norm": 537.1051080229824,
+      "learning_rate": 1.8779480771059954e-05,
+      "loss": 312.5561,
+      "step": 8700
+    },
+    {
+      "epoch": 0.16751530187853697,
+      "grad_norm": 601.6596919036647,
+      "learning_rate": 1.8776557220848477e-05,
+      "loss": 322.7146,
+      "step": 8710
+    },
+    {
+      "epoch": 0.16770762713901752,
+      "grad_norm": 1843.240441019918,
+      "learning_rate": 1.8773630401514388e-05,
+      "loss": 322.8632,
+      "step": 8720
+    },
+    {
+      "epoch": 0.16789995239949804,
+      "grad_norm": 588.6423817884859,
+      "learning_rate": 1.877070031414787e-05,
+      "loss": 327.944,
+      "step": 8730
+    },
+    {
+      "epoch": 0.16809227765997856,
+      "grad_norm": 713.3265183877421,
+      "learning_rate": 1.876776695984034e-05,
+      "loss": 322.0146,
+      "step": 8740
+    },
+    {
+      "epoch": 0.16828460292045908,
+      "grad_norm": 587.0364122055531,
+      "learning_rate": 1.8764830339684426e-05,
+      "loss": 318.2402,
+      "step": 8750
+    },
+    {
+      "epoch": 0.1684769281809396,
+      "grad_norm": 592.3967061258917,
+      "learning_rate": 1.8761890454773965e-05,
+      "loss": 322.9118,
+      "step": 8760
+    },
+    {
+      "epoch": 0.16866925344142014,
+      "grad_norm": 517.6083040950411,
+      "learning_rate": 1.8758947306204012e-05,
+      "loss": 331.4004,
+      "step": 8770
+    },
+    {
+      "epoch": 0.16886157870190066,
+      "grad_norm": 606.6972131828855,
+      "learning_rate": 1.8756000895070854e-05,
+      "loss": 311.8005,
+      "step": 8780
+    },
+    {
+      "epoch": 0.16905390396238118,
+      "grad_norm": 559.9634998313228,
+      "learning_rate": 1.8753051222471968e-05,
+      "loss": 331.0142,
+      "step": 8790
+    },
+    {
+      "epoch": 0.1692462292228617,
+      "grad_norm": 629.1482731144652,
+      "learning_rate": 1.8750098289506066e-05,
+      "loss": 321.3594,
+      "step": 8800
+    },
+    {
+      "epoch": 0.16943855448334222,
+      "grad_norm": 631.8403884212312,
+      "learning_rate": 1.8747142097273057e-05,
+      "loss": 317.4358,
+      "step": 8810
+    },
+    {
+      "epoch": 0.16963087974382277,
+      "grad_norm": 540.2865520805574,
+      "learning_rate": 1.8744182646874085e-05,
+      "loss": 321.2016,
+      "step": 8820
+    },
+    {
+      "epoch": 0.16982320500430328,
+      "grad_norm": 541.4145711976639,
+      "learning_rate": 1.8741219939411494e-05,
+      "loss": 333.4331,
+      "step": 8830
+    },
+    {
+      "epoch": 0.1700155302647838,
+      "grad_norm": 607.313765297245,
+      "learning_rate": 1.873825397598884e-05,
+      "loss": 330.6782,
+      "step": 8840
+    },
+    {
+      "epoch": 0.17020785552526432,
+      "grad_norm": 653.2116764640416,
+      "learning_rate": 1.8735284757710897e-05,
+      "loss": 324.4951,
+      "step": 8850
+    },
+    {
+      "epoch": 0.17040018078574484,
+      "grad_norm": 777.4636895302846,
+      "learning_rate": 1.873231228568365e-05,
+      "loss": 329.9455,
+      "step": 8860
+    },
+    {
+      "epoch": 0.1705925060462254,
+      "grad_norm": 596.4840023893854,
+      "learning_rate": 1.8729336561014294e-05,
+      "loss": 331.0089,
+      "step": 8870
+    },
+    {
+      "epoch": 0.1707848313067059,
+      "grad_norm": 609.0019120766575,
+      "learning_rate": 1.8726357584811242e-05,
+      "loss": 318.682,
+      "step": 8880
+    },
+    {
+      "epoch": 0.17097715656718643,
+      "grad_norm": 556.8876107013019,
+      "learning_rate": 1.8723375358184107e-05,
+      "loss": 329.2189,
+      "step": 8890
+    },
+    {
+      "epoch": 0.17116948182766695,
+      "grad_norm": 586.9369407016659,
+      "learning_rate": 1.8720389882243722e-05,
+      "loss": 319.5048,
+      "step": 8900
+    },
+    {
+      "epoch": 0.17136180708814747,
+      "grad_norm": 586.363607153541,
+      "learning_rate": 1.871740115810213e-05,
+      "loss": 332.6656,
+      "step": 8910
+    },
+    {
+      "epoch": 0.171554132348628,
+      "grad_norm": 612.540106430632,
+      "learning_rate": 1.8714409186872578e-05,
+      "loss": 329.3621,
+      "step": 8920
+    },
+    {
+      "epoch": 0.17174645760910853,
+      "grad_norm": 594.4632984832891,
+      "learning_rate": 1.871141396966953e-05,
+      "loss": 331.4114,
+      "step": 8930
+    },
+    {
+      "epoch": 0.17193878286958905,
+      "grad_norm": 555.042190549085,
+      "learning_rate": 1.8708415507608647e-05,
+      "loss": 318.5437,
+      "step": 8940
+    },
+    {
+      "epoch": 0.17213110813006957,
+      "grad_norm": 604.785455902435,
+      "learning_rate": 1.8705413801806817e-05,
+      "loss": 325.3128,
+      "step": 8950
+    },
+    {
+      "epoch": 0.1723234333905501,
+      "grad_norm": 557.973065606185,
+      "learning_rate": 1.8702408853382114e-05,
+      "loss": 318.5841,
+      "step": 8960
+    },
+    {
+      "epoch": 0.1725157586510306,
+      "grad_norm": 582.253221118452,
+      "learning_rate": 1.8699400663453842e-05,
+      "loss": 319.9609,
+      "step": 8970
+    },
+    {
+      "epoch": 0.17270808391151116,
+      "grad_norm": 610.8792579268109,
+      "learning_rate": 1.8696389233142498e-05,
+      "loss": 319.7314,
+      "step": 8980
+    },
+    {
+      "epoch": 0.17290040917199168,
+      "grad_norm": 634.1242949498051,
+      "learning_rate": 1.8693374563569792e-05,
+      "loss": 319.977,
+      "step": 8990
+    },
+    {
+      "epoch": 0.1730927344324722,
+      "grad_norm": 575.2208012699871,
+      "learning_rate": 1.8690356655858634e-05,
+      "loss": 322.9117,
+      "step": 9000
+    },
+    {
+      "epoch": 0.17328505969295271,
+      "grad_norm": 632.0699801492443,
+      "learning_rate": 1.868733551113315e-05,
+      "loss": 330.6622,
+      "step": 9010
+    },
+    {
+      "epoch": 0.17347738495343323,
+      "grad_norm": 574.6326927324745,
+      "learning_rate": 1.8684311130518663e-05,
+      "loss": 317.9359,
+      "step": 9020
+    },
+    {
+      "epoch": 0.17366971021391378,
+      "grad_norm": 615.0153731040023,
+      "learning_rate": 1.8681283515141705e-05,
+      "loss": 319.9495,
+      "step": 9030
+    },
+    {
+      "epoch": 0.1738620354743943,
+      "grad_norm": 592.5605615329267,
+      "learning_rate": 1.8678252666130016e-05,
+      "loss": 324.2824,
+      "step": 9040
+    },
+    {
+      "epoch": 0.17405436073487482,
+      "grad_norm": 629.4690838948,
+      "learning_rate": 1.8675218584612534e-05,
+      "loss": 322.147,
+      "step": 9050
+    },
+    {
+      "epoch": 0.17424668599535534,
+      "grad_norm": 638.7425197425805,
+      "learning_rate": 1.8672181271719406e-05,
+      "loss": 326.0541,
+      "step": 9060
+    },
+    {
+      "epoch": 0.17443901125583586,
+      "grad_norm": 624.8859833589399,
+      "learning_rate": 1.866914072858198e-05,
+      "loss": 329.8311,
+      "step": 9070
+    },
+    {
+      "epoch": 0.1746313365163164,
+      "grad_norm": 586.6595133477191,
+      "learning_rate": 1.8666096956332805e-05,
+      "loss": 321.3435,
+      "step": 9080
+    },
+    {
+      "epoch": 0.17482366177679692,
+      "grad_norm": 574.6930742523509,
+      "learning_rate": 1.8663049956105642e-05,
+      "loss": 321.0171,
+      "step": 9090
+    },
+    {
+      "epoch": 0.17501598703727744,
+      "grad_norm": 556.3218335808341,
+      "learning_rate": 1.8659999729035445e-05,
+      "loss": 319.513,
+      "step": 9100
+    },
+    {
+      "epoch": 0.17520831229775796,
+      "grad_norm": 573.8952191533285,
+      "learning_rate": 1.8656946276258373e-05,
+      "loss": 320.1525,
+      "step": 9110
+    },
+    {
+      "epoch": 0.17540063755823848,
+      "grad_norm": 591.4819894648808,
+      "learning_rate": 1.8653889598911787e-05,
+      "loss": 328.5546,
+      "step": 9120
+    },
+    {
+      "epoch": 0.17559296281871903,
+      "grad_norm": 612.9698796081294,
+      "learning_rate": 1.865082969813425e-05,
+      "loss": 318.3815,
+      "step": 9130
+    },
+    {
+      "epoch": 0.17578528807919955,
+      "grad_norm": 575.3068076396554,
+      "learning_rate": 1.8647766575065523e-05,
+      "loss": 322.6839,
+      "step": 9140
+    },
+    {
+      "epoch": 0.17597761333968007,
+      "grad_norm": 761.8499357944903,
+      "learning_rate": 1.864470023084657e-05,
+      "loss": 339.5503,
+      "step": 9150
+    },
+    {
+      "epoch": 0.17616993860016059,
+      "grad_norm": 533.6905660311675,
+      "learning_rate": 1.864163066661955e-05,
+      "loss": 323.9123,
+      "step": 9160
+    },
+    {
+      "epoch": 0.1763622638606411,
+      "grad_norm": 558.9204987916131,
+      "learning_rate": 1.8638557883527833e-05,
+      "loss": 318.4725,
+      "step": 9170
+    },
+    {
+      "epoch": 0.17655458912112165,
+      "grad_norm": 568.923807076423,
+      "learning_rate": 1.8635481882715975e-05,
+      "loss": 312.6829,
+      "step": 9180
+    },
+    {
+      "epoch": 0.17674691438160217,
+      "grad_norm": 544.9506885773501,
+      "learning_rate": 1.863240266532973e-05,
+      "loss": 314.0525,
+      "step": 9190
+    },
+    {
+      "epoch": 0.1769392396420827,
+      "grad_norm": 615.8507370787281,
+      "learning_rate": 1.8629320232516063e-05,
+      "loss": 324.0767,
+      "step": 9200
+    },
+    {
+      "epoch": 0.1771315649025632,
+      "grad_norm": 559.5180910964125,
+      "learning_rate": 1.862623458542313e-05,
+      "loss": 321.9804,
+      "step": 9210
+    },
+    {
+      "epoch": 0.17732389016304373,
+      "grad_norm": 551.4510927176239,
+      "learning_rate": 1.862314572520028e-05,
+      "loss": 319.111,
+      "step": 9220
+    },
+    {
+      "epoch": 0.17751621542352428,
+      "grad_norm": 591.3738793871839,
+      "learning_rate": 1.862005365299806e-05,
+      "loss": 313.194,
+      "step": 9230
+    },
+    {
+      "epoch": 0.1777085406840048,
+      "grad_norm": 618.7373130272308,
+      "learning_rate": 1.8616958369968223e-05,
+      "loss": 336.732,
+      "step": 9240
+    },
+    {
+      "epoch": 0.17790086594448531,
+      "grad_norm": 590.6551751931191,
+      "learning_rate": 1.8613859877263708e-05,
+      "loss": 320.7936,
+      "step": 9250
+    },
+    {
+      "epoch": 0.17809319120496583,
+      "grad_norm": 559.5270659580083,
+      "learning_rate": 1.8610758176038647e-05,
+      "loss": 311.5099,
+      "step": 9260
+    },
+    {
+      "epoch": 0.17828551646544635,
+      "grad_norm": 631.9487065025297,
+      "learning_rate": 1.860765326744838e-05,
+      "loss": 316.0143,
+      "step": 9270
+    },
+    {
+      "epoch": 0.1784778417259269,
+      "grad_norm": 632.8101121945842,
+      "learning_rate": 1.8604545152649426e-05,
+      "loss": 324.1879,
+      "step": 9280
+    },
+    {
+      "epoch": 0.17867016698640742,
+      "grad_norm": 551.5142150541681,
+      "learning_rate": 1.860143383279952e-05,
+      "loss": 326.5638,
+      "step": 9290
+    },
+    {
+      "epoch": 0.17886249224688794,
+      "grad_norm": 630.096978179395,
+      "learning_rate": 1.859831930905756e-05,
+      "loss": 315.0706,
+      "step": 9300
+    },
+    {
+      "epoch": 0.17905481750736846,
+      "grad_norm": 607.1346668962717,
+      "learning_rate": 1.8595201582583668e-05,
+      "loss": 320.8782,
+      "step": 9310
+    },
+    {
+      "epoch": 0.17924714276784898,
+      "grad_norm": 573.3499482985779,
+      "learning_rate": 1.859208065453914e-05,
+      "loss": 316.0985,
+      "step": 9320
+    },
+    {
+      "epoch": 0.17943946802832952,
+      "grad_norm": 613.3873449690732,
+      "learning_rate": 1.8588956526086472e-05,
+      "loss": 333.3562,
+      "step": 9330
+    },
+    {
+      "epoch": 0.17963179328881004,
+      "grad_norm": 543.1026188086619,
+      "learning_rate": 1.8585829198389347e-05,
+      "loss": 315.4055,
+      "step": 9340
+    },
+    {
+      "epoch": 0.17982411854929056,
+      "grad_norm": 553.3238573858466,
+      "learning_rate": 1.8582698672612646e-05,
+      "loss": 315.6242,
+      "step": 9350
+    },
+    {
+      "epoch": 0.18001644380977108,
+      "grad_norm": 552.0196877946281,
+      "learning_rate": 1.8579564949922438e-05,
+      "loss": 316.9576,
+      "step": 9360
+    },
+    {
+      "epoch": 0.1802087690702516,
+      "grad_norm": 564.4067501038841,
+      "learning_rate": 1.8576428031485984e-05,
+      "loss": 310.0551,
+      "step": 9370
+    },
+    {
+      "epoch": 0.18040109433073215,
+      "grad_norm": 567.4778220652782,
+      "learning_rate": 1.8573287918471728e-05,
+      "loss": 319.9249,
+      "step": 9380
+    },
+    {
+      "epoch": 0.18059341959121267,
+      "grad_norm": 636.6557564391225,
+      "learning_rate": 1.8570144612049322e-05,
+      "loss": 319.27,
+      "step": 9390
+    },
+    {
+      "epoch": 0.1807857448516932,
+      "grad_norm": 593.7142403903853,
+      "learning_rate": 1.856699811338958e-05,
+      "loss": 318.0193,
+      "step": 9400
+    },
+    {
+      "epoch": 0.1809780701121737,
+      "grad_norm": 712.8194338129573,
+      "learning_rate": 1.8563848423664536e-05,
+      "loss": 326.8473,
+      "step": 9410
+    },
+    {
+      "epoch": 0.18117039537265422,
+      "grad_norm": 620.172842857781,
+      "learning_rate": 1.8560695544047388e-05,
+      "loss": 335.5854,
+      "step": 9420
+    },
+    {
+      "epoch": 0.18136272063313474,
+      "grad_norm": 740.7605188517435,
+      "learning_rate": 1.8557539475712538e-05,
+      "loss": 338.9682,
+      "step": 9430
+    },
+    {
+      "epoch": 0.1815550458936153,
+      "grad_norm": 563.4029680194013,
+      "learning_rate": 1.855438021983556e-05,
+      "loss": 317.2997,
+      "step": 9440
+    },
+    {
+      "epoch": 0.1817473711540958,
+      "grad_norm": 612.5074396086102,
+      "learning_rate": 1.8551217777593233e-05,
+      "loss": 318.5311,
+      "step": 9450
+    },
+    {
+      "epoch": 0.18193969641457633,
+      "grad_norm": 597.0074091244863,
+      "learning_rate": 1.8548052150163514e-05,
+      "loss": 314.7106,
+      "step": 9460
+    },
+    {
+      "epoch": 0.18213202167505685,
+      "grad_norm": 554.7269004078719,
+      "learning_rate": 1.8544883338725544e-05,
+      "loss": 306.6892,
+      "step": 9470
+    },
+    {
+      "epoch": 0.18232434693553737,
+      "grad_norm": 577.8852459584037,
+      "learning_rate": 1.8541711344459652e-05,
+      "loss": 317.9633,
+      "step": 9480
+    },
+    {
+      "epoch": 0.18251667219601792,
+      "grad_norm": 668.2529372949699,
+      "learning_rate": 1.8538536168547353e-05,
+      "loss": 326.373,
+      "step": 9490
+    },
+    {
+      "epoch": 0.18270899745649843,
+      "grad_norm": 706.4714323223383,
+      "learning_rate": 1.8535357812171356e-05,
+      "loss": 330.0399,
+      "step": 9500
+    },
+    {
+      "epoch": 0.18290132271697895,
+      "grad_norm": 630.4043554538246,
+      "learning_rate": 1.8532176276515538e-05,
+      "loss": 316.1966,
+      "step": 9510
+    },
+    {
+      "epoch": 0.18309364797745947,
+      "grad_norm": 606.5355772061898,
+      "learning_rate": 1.8528991562764967e-05,
+      "loss": 318.0851,
+      "step": 9520
+    },
+    {
+      "epoch": 0.18328597323794,
+      "grad_norm": 552.573228905615,
+      "learning_rate": 1.85258036721059e-05,
+      "loss": 313.2862,
+      "step": 9530
+    },
+    {
+      "epoch": 0.18347829849842054,
+      "grad_norm": 544.2464163634174,
+      "learning_rate": 1.8522612605725777e-05,
+      "loss": 323.5065,
+      "step": 9540
+    },
+    {
+      "epoch": 0.18367062375890106,
+      "grad_norm": 628.0910797173485,
+      "learning_rate": 1.8519418364813215e-05,
+      "loss": 318.409,
+      "step": 9550
+    },
+    {
+      "epoch": 0.18386294901938158,
+      "grad_norm": 595.5542255882365,
+      "learning_rate": 1.851622095055801e-05,
+      "loss": 316.1256,
+      "step": 9560
+    },
+    {
+      "epoch": 0.1840552742798621,
+      "grad_norm": 568.4977141118377,
+      "learning_rate": 1.8513020364151155e-05,
+      "loss": 317.5389,
+      "step": 9570
+    },
+    {
+      "epoch": 0.18424759954034262,
+      "grad_norm": 690.7197119756217,
+      "learning_rate": 1.850981660678481e-05,
+      "loss": 315.9089,
+      "step": 9580
+    },
+    {
+      "epoch": 0.18443992480082316,
+      "grad_norm": 617.4316572779147,
+      "learning_rate": 1.8506609679652323e-05,
+      "loss": 313.6196,
+      "step": 9590
+    },
+    {
+      "epoch": 0.18463225006130368,
+      "grad_norm": 615.4863352262362,
+      "learning_rate": 1.8503399583948224e-05,
+      "loss": 325.9829,
+      "step": 9600
+    },
+    {
+      "epoch": 0.1848245753217842,
+      "grad_norm": 560.5758724132193,
+      "learning_rate": 1.8500186320868215e-05,
+      "loss": 308.7002,
+      "step": 9610
+    },
+    {
+      "epoch": 0.18501690058226472,
+      "grad_norm": 567.0365575400731,
+      "learning_rate": 1.8496969891609186e-05,
+      "loss": 329.2433,
+      "step": 9620
+    },
+    {
+      "epoch": 0.18520922584274524,
+      "grad_norm": 568.0367621336482,
+      "learning_rate": 1.8493750297369208e-05,
+      "loss": 325.427,
+      "step": 9630
+    },
+    {
+      "epoch": 0.1854015511032258,
+      "grad_norm": 550.4794735105326,
+      "learning_rate": 1.849052753934752e-05,
+      "loss": 326.7041,
+      "step": 9640
+    },
+    {
+      "epoch": 0.1855938763637063,
+      "grad_norm": 569.5641753096411,
+      "learning_rate": 1.8487301618744552e-05,
+      "loss": 328.1982,
+      "step": 9650
+    },
+    {
+      "epoch": 0.18578620162418683,
+      "grad_norm": 569.9761034507699,
+      "learning_rate": 1.84840725367619e-05,
+      "loss": 310.536,
+      "step": 9660
+    },
+    {
+      "epoch": 0.18597852688466734,
+      "grad_norm": 601.6190225257928,
+      "learning_rate": 1.8480840294602352e-05,
+      "loss": 314.3488,
+      "step": 9670
+    },
+    {
+      "epoch": 0.18617085214514786,
+      "grad_norm": 618.2553619325734,
+      "learning_rate": 1.8477604893469857e-05,
+      "loss": 327.1187,
+      "step": 9680
+    },
+    {
+      "epoch": 0.1863631774056284,
+      "grad_norm": 752.9474167793641,
+      "learning_rate": 1.847436633456955e-05,
+      "loss": 325.2218,
+      "step": 9690
+    },
+    {
+      "epoch": 0.18655550266610893,
+      "grad_norm": 603.0636498303711,
+      "learning_rate": 1.8471124619107744e-05,
+      "loss": 311.5493,
+      "step": 9700
+    },
+    {
+      "epoch": 0.18674782792658945,
+      "grad_norm": 556.6395524569551,
+      "learning_rate": 1.846787974829192e-05,
+      "loss": 313.517,
+      "step": 9710
+    },
+    {
+      "epoch": 0.18694015318706997,
+      "grad_norm": 565.1641454765296,
+      "learning_rate": 1.8464631723330745e-05,
+      "loss": 318.9753,
+      "step": 9720
+    },
+    {
+      "epoch": 0.1871324784475505,
+      "grad_norm": 545.5890638241866,
+      "learning_rate": 1.8461380545434054e-05,
+      "loss": 306.4056,
+      "step": 9730
+    },
+    {
+      "epoch": 0.18732480370803103,
+      "grad_norm": 537.9031786708614,
+      "learning_rate": 1.8458126215812848e-05,
+      "loss": 309.4228,
+      "step": 9740
+    },
+    {
+      "epoch": 0.18751712896851155,
+      "grad_norm": 544.290504443795,
+      "learning_rate": 1.845486873567932e-05,
+      "loss": 313.062,
+      "step": 9750
+    },
+    {
+      "epoch": 0.18770945422899207,
+      "grad_norm": 530.1043127865822,
+      "learning_rate": 1.8451608106246822e-05,
+      "loss": 309.1896,
+      "step": 9760
+    },
+    {
+      "epoch": 0.1879017794894726,
+      "grad_norm": 1974.8696274493948,
+      "learning_rate": 1.8448344328729893e-05,
+      "loss": 323.1416,
+      "step": 9770
+    },
+    {
+      "epoch": 0.1880941047499531,
+      "grad_norm": 587.2253624359105,
+      "learning_rate": 1.8445077404344226e-05,
+      "loss": 309.5772,
+      "step": 9780
+    },
+    {
+      "epoch": 0.18828643001043366,
+      "grad_norm": 609.2034365251025,
+      "learning_rate": 1.8441807334306702e-05,
+      "loss": 313.9833,
+      "step": 9790
+    },
+    {
+      "epoch": 0.18847875527091418,
+      "grad_norm": 536.3689693630588,
+      "learning_rate": 1.8438534119835365e-05,
+      "loss": 308.4715,
+      "step": 9800
+    },
+    {
+      "epoch": 0.1886710805313947,
+      "grad_norm": 564.0070248860729,
+      "learning_rate": 1.8435257762149436e-05,
+      "loss": 323.2902,
+      "step": 9810
+    },
+    {
+      "epoch": 0.18886340579187522,
+      "grad_norm": 541.1094813955598,
+      "learning_rate": 1.8431978262469305e-05,
+      "loss": 324.6447,
+      "step": 9820
+    },
+    {
+      "epoch": 0.18905573105235574,
+      "grad_norm": 574.3562023468872,
+      "learning_rate": 1.8428695622016532e-05,
+      "loss": 310.8277,
+      "step": 9830
+    },
+    {
+      "epoch": 0.18924805631283628,
+      "grad_norm": 573.6950744377439,
+      "learning_rate": 1.8425409842013843e-05,
+      "loss": 322.8399,
+      "step": 9840
+    },
+    {
+      "epoch": 0.1894403815733168,
+      "grad_norm": 524.8989840894691,
+      "learning_rate": 1.8422120923685135e-05,
+      "loss": 307.1694,
+      "step": 9850
+    },
+    {
+      "epoch": 0.18963270683379732,
+      "grad_norm": 571.3089044268454,
+      "learning_rate": 1.8418828868255484e-05,
+      "loss": 312.775,
+      "step": 9860
+    },
+    {
+      "epoch": 0.18982503209427784,
+      "grad_norm": 553.3508873042184,
+      "learning_rate": 1.8415533676951117e-05,
+      "loss": 321.306,
+      "step": 9870
+    },
+    {
+      "epoch": 0.19001735735475836,
+      "grad_norm": 565.7110978394054,
+      "learning_rate": 1.8412235350999444e-05,
+      "loss": 324.4741,
+      "step": 9880
+    },
+    {
+      "epoch": 0.19020968261523888,
+      "grad_norm": 621.1426134637459,
+      "learning_rate": 1.840893389162903e-05,
+      "loss": 346.3106,
+      "step": 9890
+    },
+    {
+      "epoch": 0.19040200787571943,
+      "grad_norm": 535.0733987249391,
+      "learning_rate": 1.8405629300069626e-05,
+      "loss": 320.5854,
+      "step": 9900
+    },
+    {
+      "epoch": 0.19059433313619994,
+      "grad_norm": 670.4813425818097,
+      "learning_rate": 1.840232157755213e-05,
+      "loss": 320.277,
+      "step": 9910
+    },
+    {
+      "epoch": 0.19078665839668046,
+      "grad_norm": 617.809155563652,
+      "learning_rate": 1.8399010725308616e-05,
+      "loss": 311.902,
+      "step": 9920
+    },
+    {
+      "epoch": 0.19097898365716098,
+      "grad_norm": 568.6148701158531,
+      "learning_rate": 1.839569674457232e-05,
+      "loss": 318.7672,
+      "step": 9930
+    },
+    {
+      "epoch": 0.1911713089176415,
+      "grad_norm": 569.9689571022637,
+      "learning_rate": 1.8392379636577647e-05,
+      "loss": 328.9152,
+      "step": 9940
+    },
+    {
+      "epoch": 0.19136363417812205,
+      "grad_norm": 574.9266939103386,
+      "learning_rate": 1.8389059402560165e-05,
+      "loss": 318.1828,
+      "step": 9950
+    },
+    {
+      "epoch": 0.19155595943860257,
+      "grad_norm": 624.1369244831499,
+      "learning_rate": 1.8385736043756605e-05,
+      "loss": 313.1912,
+      "step": 9960
+    },
+    {
+      "epoch": 0.1917482846990831,
+      "grad_norm": 556.790792342805,
+      "learning_rate": 1.838240956140486e-05,
+      "loss": 319.589,
+      "step": 9970
+    },
+    {
+      "epoch": 0.1919406099595636,
+      "grad_norm": 565.7194728615357,
+      "learning_rate": 1.8379079956743996e-05,
+      "loss": 327.9372,
+      "step": 9980
+    },
+    {
+      "epoch": 0.19213293522004413,
+      "grad_norm": 557.3348851818414,
+      "learning_rate": 1.8375747231014233e-05,
+      "loss": 316.3132,
+      "step": 9990
+    },
+    {
+      "epoch": 0.19232526048052467,
+      "grad_norm": 644.8743942262005,
+      "learning_rate": 1.8372411385456956e-05,
+      "loss": 317.5455,
+      "step": 10000
+    },
+    {
+      "epoch": 0.1925175857410052,
+      "grad_norm": 611.0130672157912,
+      "learning_rate": 1.8369072421314717e-05,
+      "loss": 311.7699,
+      "step": 10010
+    },
+    {
+      "epoch": 0.1927099110014857,
+      "grad_norm": 643.2944485951936,
+      "learning_rate": 1.8365730339831212e-05,
+      "loss": 310.4235,
+      "step": 10020
+    },
+    {
+      "epoch": 0.19290223626196623,
+      "grad_norm": 620.125833263131,
+      "learning_rate": 1.8362385142251328e-05,
+      "loss": 317.8711,
+      "step": 10030
+    },
+    {
+      "epoch": 0.19309456152244675,
+      "grad_norm": 642.5480738830829,
+      "learning_rate": 1.8359036829821085e-05,
+      "loss": 311.2124,
+      "step": 10040
+    },
+    {
+      "epoch": 0.1932868867829273,
+      "grad_norm": 542.278779870074,
+      "learning_rate": 1.8355685403787677e-05,
+      "loss": 308.168,
+      "step": 10050
+    },
+    {
+      "epoch": 0.19347921204340782,
+      "grad_norm": 570.9987527543067,
+      "learning_rate": 1.8352330865399457e-05,
+      "loss": 307.4103,
+      "step": 10060
+    },
+    {
+      "epoch": 0.19367153730388834,
+      "grad_norm": 588.362824919098,
+      "learning_rate": 1.834897321590593e-05,
+      "loss": 311.6161,
+      "step": 10070
+    },
+    {
+      "epoch": 0.19386386256436886,
+      "grad_norm": 555.9423039717024,
+      "learning_rate": 1.8345612456557767e-05,
+      "loss": 303.3604,
+      "step": 10080
+    },
+    {
+      "epoch": 0.19405618782484937,
+      "grad_norm": 620.6001646296406,
+      "learning_rate": 1.83422485886068e-05,
+      "loss": 308.114,
+      "step": 10090
+    },
+    {
+      "epoch": 0.19424851308532992,
+      "grad_norm": 599.7921359122605,
+      "learning_rate": 1.833888161330601e-05,
+      "loss": 325.1521,
+      "step": 10100
+    },
+    {
+      "epoch": 0.19444083834581044,
+      "grad_norm": 511.20343896387277,
+      "learning_rate": 1.833551153190954e-05,
+      "loss": 309.2157,
+      "step": 10110
+    },
+    {
+      "epoch": 0.19463316360629096,
+      "grad_norm": 516.257174776799,
+      "learning_rate": 1.8332138345672686e-05,
+      "loss": 309.4907,
+      "step": 10120
+    },
+    {
+      "epoch": 0.19482548886677148,
+      "grad_norm": 563.1608572740786,
+      "learning_rate": 1.832876205585191e-05,
+      "loss": 316.6437,
+      "step": 10130
+    },
+    {
+      "epoch": 0.195017814127252,
+      "grad_norm": 553.3475752491817,
+      "learning_rate": 1.8325382663704826e-05,
+      "loss": 312.1191,
+      "step": 10140
+    },
+    {
+      "epoch": 0.19521013938773255,
+      "grad_norm": 531.5864215766379,
+      "learning_rate": 1.8322000170490194e-05,
+      "loss": 306.8882,
+      "step": 10150
+    },
+    {
+      "epoch": 0.19540246464821306,
+      "grad_norm": 598.789798633388,
+      "learning_rate": 1.831861457746794e-05,
+      "loss": 313.6952,
+      "step": 10160
+    },
+    {
+      "epoch": 0.19559478990869358,
+      "grad_norm": 555.5193570635953,
+      "learning_rate": 1.8315225885899144e-05,
+      "loss": 317.2797,
+      "step": 10170
+    },
+    {
+      "epoch": 0.1957871151691741,
+      "grad_norm": 554.3786845937254,
+      "learning_rate": 1.8311834097046038e-05,
+      "loss": 308.5955,
+      "step": 10180
+    },
+    {
+      "epoch": 0.19597944042965462,
+      "grad_norm": 667.1400002145294,
+      "learning_rate": 1.8308439212172e-05,
+      "loss": 318.206,
+      "step": 10190
+    },
+    {
+      "epoch": 0.19617176569013517,
+      "grad_norm": 563.4817826452128,
+      "learning_rate": 1.830504123254158e-05,
+      "loss": 323.7995,
+      "step": 10200
+    },
+    {
+      "epoch": 0.1963640909506157,
+      "grad_norm": 510.24096084298566,
+      "learning_rate": 1.830164015942046e-05,
+      "loss": 315.2723,
+      "step": 10210
+    },
+    {
+      "epoch": 0.1965564162110962,
+      "grad_norm": 897.7329135630494,
+      "learning_rate": 1.8298235994075488e-05,
+      "loss": 307.7924,
+      "step": 10220
+    },
+    {
+      "epoch": 0.19674874147157673,
+      "grad_norm": 569.6326868493514,
+      "learning_rate": 1.829482873777466e-05,
+      "loss": 322.1561,
+      "step": 10230
+    },
+    {
+      "epoch": 0.19694106673205725,
+      "grad_norm": 14005.331331945481,
+      "learning_rate": 1.8291418391787116e-05,
+      "loss": 336.6291,
+      "step": 10240
+    },
+    {
+      "epoch": 0.1971333919925378,
+      "grad_norm": 580.7591541191108,
+      "learning_rate": 1.8288004957383162e-05,
+      "loss": 313.3958,
+      "step": 10250
+    },
+    {
+      "epoch": 0.1973257172530183,
+      "grad_norm": 589.4898875747193,
+      "learning_rate": 1.8284588435834242e-05,
+      "loss": 314.5349,
+      "step": 10260
+    },
+    {
+      "epoch": 0.19751804251349883,
+      "grad_norm": 596.9254042414015,
+      "learning_rate": 1.828116882841295e-05,
+      "loss": 316.0492,
+      "step": 10270
+    },
+    {
+      "epoch": 0.19771036777397935,
+      "grad_norm": 627.1956505220512,
+      "learning_rate": 1.8277746136393042e-05,
+      "loss": 323.5244,
+      "step": 10280
+    },
+    {
+      "epoch": 0.19790269303445987,
+      "grad_norm": 605.5364380774679,
+      "learning_rate": 1.827432036104941e-05,
+      "loss": 312.2124,
+      "step": 10290
+    },
+    {
+      "epoch": 0.19809501829494042,
+      "grad_norm": 554.8727958777251,
+      "learning_rate": 1.8270891503658096e-05,
+      "loss": 311.6395,
+      "step": 10300
+    },
+    {
+      "epoch": 0.19828734355542094,
+      "grad_norm": 650.6178601497293,
+      "learning_rate": 1.8267459565496298e-05,
+      "loss": 317.9526,
+      "step": 10310
+    },
+    {
+      "epoch": 0.19847966881590146,
+      "grad_norm": 596.3615622397797,
+      "learning_rate": 1.8264024547842346e-05,
+      "loss": 313.5765,
+      "step": 10320
+    },
+    {
+      "epoch": 0.19867199407638197,
+      "grad_norm": 556.4179671335443,
+      "learning_rate": 1.8260586451975745e-05,
+      "loss": 307.384,
+      "step": 10330
+    },
+    {
+      "epoch": 0.1988643193368625,
+      "grad_norm": 630.927531284662,
+      "learning_rate": 1.825714527917711e-05,
+      "loss": 321.1718,
+      "step": 10340
+    },
+    {
+      "epoch": 0.199056644597343,
+      "grad_norm": 589.9843909541497,
+      "learning_rate": 1.8253701030728235e-05,
+      "loss": 316.1164,
+      "step": 10350
+    },
+    {
+      "epoch": 0.19924896985782356,
+      "grad_norm": 606.7908564946719,
+      "learning_rate": 1.8250253707912036e-05,
+      "loss": 317.7305,
+      "step": 10360
+    },
+    {
+      "epoch": 0.19944129511830408,
+      "grad_norm": 545.2147663391177,
+      "learning_rate": 1.8246803312012593e-05,
+      "loss": 303.6862,
+      "step": 10370
+    },
+    {
+      "epoch": 0.1996336203787846,
+      "grad_norm": 541.3311699591419,
+      "learning_rate": 1.8243349844315116e-05,
+      "loss": 304.131,
+      "step": 10380
+    },
+    {
+      "epoch": 0.19982594563926512,
+      "grad_norm": 747.9351319944042,
+      "learning_rate": 1.8239893306105966e-05,
+      "loss": 305.2509,
+      "step": 10390
+    },
+    {
+      "epoch": 0.20001827089974564,
+      "grad_norm": 609.5978708681915,
+      "learning_rate": 1.823643369867264e-05,
+      "loss": 308.1789,
+      "step": 10400
+    },
+    {
+      "epoch": 0.20021059616022618,
+      "grad_norm": 3345.2680882620302,
+      "learning_rate": 1.8232971023303798e-05,
+      "loss": 303.1881,
+      "step": 10410
+    },
+    {
+      "epoch": 0.2004029214207067,
+      "grad_norm": 685.981103328202,
+      "learning_rate": 1.8229505281289216e-05,
+      "loss": 320.6667,
+      "step": 10420
+    },
+    {
+      "epoch": 0.20059524668118722,
+      "grad_norm": 527.8690783067844,
+      "learning_rate": 1.8226036473919836e-05,
+      "loss": 302.881,
+      "step": 10430
+    },
+    {
+      "epoch": 0.20078757194166774,
+      "grad_norm": 554.5850449966307,
+      "learning_rate": 1.8222564602487724e-05,
+      "loss": 311.9064,
+      "step": 10440
+    },
+    {
+      "epoch": 0.20097989720214826,
+      "grad_norm": 588.4499184429349,
+      "learning_rate": 1.82190896682861e-05,
+      "loss": 309.5534,
+      "step": 10450
+    },
+    {
+      "epoch": 0.2011722224626288,
+      "grad_norm": 597.2463973919886,
+      "learning_rate": 1.8215611672609316e-05,
+      "loss": 298.9231,
+      "step": 10460
+    },
+    {
+      "epoch": 0.20136454772310933,
+      "grad_norm": 594.1158512309214,
+      "learning_rate": 1.821213061675287e-05,
+      "loss": 308.9411,
+      "step": 10470
+    },
+    {
+      "epoch": 0.20155687298358985,
+      "grad_norm": 589.4478962209655,
+      "learning_rate": 1.8208646502013395e-05,
+      "loss": 309.6522,
+      "step": 10480
+    },
+    {
+      "epoch": 0.20174919824407037,
+      "grad_norm": 564.6156159595868,
+      "learning_rate": 1.820515932968867e-05,
+      "loss": 310.7461,
+      "step": 10490
+    },
+    {
+      "epoch": 0.20194152350455088,
+      "grad_norm": 609.3176191883252,
+      "learning_rate": 1.8201669101077608e-05,
+      "loss": 310.0776,
+      "step": 10500
+    },
+    {
+      "epoch": 0.20213384876503143,
+      "grad_norm": 529.8941187328708,
+      "learning_rate": 1.819817581748026e-05,
+      "loss": 324.803,
+      "step": 10510
+    },
+    {
+      "epoch": 0.20232617402551195,
+      "grad_norm": 591.0983066934331,
+      "learning_rate": 1.8194679480197817e-05,
+      "loss": 308.572,
+      "step": 10520
+    },
+    {
+      "epoch": 0.20251849928599247,
+      "grad_norm": 563.2375949857469,
+      "learning_rate": 1.8191180090532608e-05,
+      "loss": 313.5502,
+      "step": 10530
+    },
+    {
+      "epoch": 0.202710824546473,
+      "grad_norm": 521.668327254525,
+      "learning_rate": 1.8187677649788097e-05,
+      "loss": 305.6655,
+      "step": 10540
+    },
+    {
+      "epoch": 0.2029031498069535,
+      "grad_norm": 538.8431736144586,
+      "learning_rate": 1.8184172159268884e-05,
+      "loss": 314.5525,
+      "step": 10550
+    },
+    {
+      "epoch": 0.20309547506743406,
+      "grad_norm": 532.0437684267256,
+      "learning_rate": 1.818066362028071e-05,
+      "loss": 312.4004,
+      "step": 10560
+    },
+    {
+      "epoch": 0.20328780032791458,
+      "grad_norm": 571.2943931355435,
+      "learning_rate": 1.8177152034130442e-05,
+      "loss": 321.7054,
+      "step": 10570
+    },
+    {
+      "epoch": 0.2034801255883951,
+      "grad_norm": 527.4672384896986,
+      "learning_rate": 1.8173637402126093e-05,
+      "loss": 322.5302,
+      "step": 10580
+    },
+    {
+      "epoch": 0.2036724508488756,
+      "grad_norm": 555.3133093459373,
+      "learning_rate": 1.8170119725576808e-05,
+      "loss": 313.6648,
+      "step": 10590
+    },
+    {
+      "epoch": 0.20386477610935613,
+      "grad_norm": 550.6922694561971,
+      "learning_rate": 1.8166599005792852e-05,
+      "loss": 310.2605,
+      "step": 10600
+    },
+    {
+      "epoch": 0.20405710136983668,
+      "grad_norm": 574.4122805106684,
+      "learning_rate": 1.816307524408565e-05,
+      "loss": 305.6629,
+      "step": 10610
+    },
+    {
+      "epoch": 0.2042494266303172,
+      "grad_norm": 572.2872382916447,
+      "learning_rate": 1.8159548441767732e-05,
+      "loss": 308.1092,
+      "step": 10620
+    },
+    {
+      "epoch": 0.20444175189079772,
+      "grad_norm": 494.4107401724572,
+      "learning_rate": 1.8156018600152777e-05,
+      "loss": 310.6198,
+      "step": 10630
+    },
+    {
+      "epoch": 0.20463407715127824,
+      "grad_norm": 593.3457969007505,
+      "learning_rate": 1.81524857205556e-05,
+      "loss": 315.9913,
+      "step": 10640
+    },
+    {
+      "epoch": 0.20482640241175876,
+      "grad_norm": 596.9842940297239,
+      "learning_rate": 1.814894980429213e-05,
+      "loss": 327.6448,
+      "step": 10650
+    },
+    {
+      "epoch": 0.2050187276722393,
+      "grad_norm": 620.556813154289,
+      "learning_rate": 1.8145410852679447e-05,
+      "loss": 313.397,
+      "step": 10660
+    },
+    {
+      "epoch": 0.20521105293271982,
+      "grad_norm": 545.4983018426418,
+      "learning_rate": 1.8141868867035745e-05,
+      "loss": 307.4965,
+      "step": 10670
+    },
+    {
+      "epoch": 0.20540337819320034,
+      "grad_norm": 517.2041095178209,
+      "learning_rate": 1.8138323848680354e-05,
+      "loss": 306.488,
+      "step": 10680
+    },
+    {
+      "epoch": 0.20559570345368086,
+      "grad_norm": 542.41660542042,
+      "learning_rate": 1.813477579893374e-05,
+      "loss": 308.95,
+      "step": 10690
+    },
+    {
+      "epoch": 0.20578802871416138,
+      "grad_norm": 623.2558012166824,
+      "learning_rate": 1.8131224719117497e-05,
+      "loss": 309.7012,
+      "step": 10700
+    },
+    {
+      "epoch": 0.20598035397464193,
+      "grad_norm": 568.6246804880005,
+      "learning_rate": 1.8127670610554332e-05,
+      "loss": 310.0226,
+      "step": 10710
+    },
+    {
+      "epoch": 0.20617267923512245,
+      "grad_norm": 583.3577234354464,
+      "learning_rate": 1.81241134745681e-05,
+      "loss": 306.5802,
+      "step": 10720
+    },
+    {
+      "epoch": 0.20636500449560297,
+      "grad_norm": 665.236530462498,
+      "learning_rate": 1.812055331248377e-05,
+      "loss": 315.9872,
+      "step": 10730
+    },
+    {
+      "epoch": 0.20655732975608349,
+      "grad_norm": 548.2231390472602,
+      "learning_rate": 1.811699012562745e-05,
+      "loss": 310.2577,
+      "step": 10740
+    },
+    {
+      "epoch": 0.206749655016564,
+      "grad_norm": 600.6657202388725,
+      "learning_rate": 1.8113423915326362e-05,
+      "loss": 320.0417,
+      "step": 10750
+    },
+    {
+      "epoch": 0.20694198027704455,
+      "grad_norm": 596.3607001974965,
+      "learning_rate": 1.8109854682908864e-05,
+      "loss": 319.0931,
+      "step": 10760
+    },
+    {
+      "epoch": 0.20713430553752507,
+      "grad_norm": 575.9230733072659,
+      "learning_rate": 1.8106282429704436e-05,
+      "loss": 312.4939,
+      "step": 10770
+    },
+    {
+      "epoch": 0.2073266307980056,
+      "grad_norm": 591.2000295821231,
+      "learning_rate": 1.810270715704368e-05,
+      "loss": 319.2311,
+      "step": 10780
+    },
+    {
+      "epoch": 0.2075189560584861,
+      "grad_norm": 587.3434212700096,
+      "learning_rate": 1.809912886625833e-05,
+      "loss": 310.2359,
+      "step": 10790
+    },
+    {
+      "epoch": 0.20771128131896663,
+      "grad_norm": 559.4690945914582,
+      "learning_rate": 1.8095547558681243e-05,
+      "loss": 323.4319,
+      "step": 10800
+    },
+    {
+      "epoch": 0.20790360657944715,
+      "grad_norm": 547.0120576898984,
+      "learning_rate": 1.809196323564639e-05,
+      "loss": 320.4141,
+      "step": 10810
+    },
+    {
+      "epoch": 0.2080959318399277,
+      "grad_norm": 558.1889120357515,
+      "learning_rate": 1.8088375898488873e-05,
+      "loss": 304.7489,
+      "step": 10820
+    },
+    {
+      "epoch": 0.20828825710040821,
+      "grad_norm": 556.2533478506813,
+      "learning_rate": 1.808478554854492e-05,
+      "loss": 313.6703,
+      "step": 10830
+    },
+    {
+      "epoch": 0.20848058236088873,
+      "grad_norm": 563.1230411656593,
+      "learning_rate": 1.8081192187151873e-05,
+      "loss": 309.7366,
+      "step": 10840
+    },
+    {
+      "epoch": 0.20867290762136925,
+      "grad_norm": 591.210113740253,
+      "learning_rate": 1.8077595815648202e-05,
+      "loss": 304.9538,
+      "step": 10850
+    },
+    {
+      "epoch": 0.20886523288184977,
+      "grad_norm": 10459.65265964558,
+      "learning_rate": 1.8073996435373494e-05,
+      "loss": 315.4726,
+      "step": 10860
+    },
+    {
+      "epoch": 0.20905755814233032,
+      "grad_norm": 582.7877514749445,
+      "learning_rate": 1.8070394047668466e-05,
+      "loss": 310.0286,
+      "step": 10870
+    },
+    {
+      "epoch": 0.20924988340281084,
+      "grad_norm": 524.5587946163101,
+      "learning_rate": 1.8066788653874936e-05,
+      "loss": 299.4925,
+      "step": 10880
+    },
+    {
+      "epoch": 0.20944220866329136,
+      "grad_norm": 593.4216247321357,
+      "learning_rate": 1.806318025533586e-05,
+      "loss": 310.3626,
+      "step": 10890
+    },
+    {
+      "epoch": 0.20963453392377188,
+      "grad_norm": 567.0255107174645,
+      "learning_rate": 1.805956885339531e-05,
+      "loss": 319.6213,
+      "step": 10900
+    },
+    {
+      "epoch": 0.2098268591842524,
+      "grad_norm": 668.0588558218071,
+      "learning_rate": 1.8055954449398472e-05,
+      "loss": 320.125,
+      "step": 10910
+    },
+    {
+      "epoch": 0.21001918444473294,
+      "grad_norm": 608.1719502025485,
+      "learning_rate": 1.8052337044691648e-05,
+      "loss": 306.7742,
+      "step": 10920
+    },
+    {
+      "epoch": 0.21021150970521346,
+      "grad_norm": 547.1221358290404,
+      "learning_rate": 1.8048716640622262e-05,
+      "loss": 315.7775,
+      "step": 10930
+    },
+    {
+      "epoch": 0.21040383496569398,
+      "grad_norm": 561.1660562013045,
+      "learning_rate": 1.8045093238538856e-05,
+      "loss": 318.5856,
+      "step": 10940
+    },
+    {
+      "epoch": 0.2105961602261745,
+      "grad_norm": 521.8198888881208,
+      "learning_rate": 1.8041466839791087e-05,
+      "loss": 308.4359,
+      "step": 10950
+    },
+    {
+      "epoch": 0.21078848548665502,
+      "grad_norm": 569.5603157795923,
+      "learning_rate": 1.8037837445729733e-05,
+      "loss": 309.5919,
+      "step": 10960
+    },
+    {
+      "epoch": 0.21098081074713557,
+      "grad_norm": 735.8905510143819,
+      "learning_rate": 1.803420505770668e-05,
+      "loss": 305.6322,
+      "step": 10970
+    },
+    {
+      "epoch": 0.21117313600761609,
+      "grad_norm": 558.9935899233416,
+      "learning_rate": 1.803056967707493e-05,
+      "loss": 306.0594,
+      "step": 10980
+    },
+    {
+      "epoch": 0.2113654612680966,
+      "grad_norm": 560.0566283194227,
+      "learning_rate": 1.8026931305188603e-05,
+      "loss": 315.5964,
+      "step": 10990
+    },
+    {
+      "epoch": 0.21155778652857712,
+      "grad_norm": 565.3773377963382,
+      "learning_rate": 1.802328994340294e-05,
+      "loss": 314.1717,
+      "step": 11000
+    },
+    {
+      "epoch": 0.21175011178905764,
+      "grad_norm": 657.1041754965652,
+      "learning_rate": 1.8019645593074275e-05,
+      "loss": 309.5446,
+      "step": 11010
+    },
+    {
+      "epoch": 0.2119424370495382,
+      "grad_norm": 6572.125609919242,
+      "learning_rate": 1.8015998255560082e-05,
+      "loss": 301.9791,
+      "step": 11020
+    },
+    {
+      "epoch": 0.2121347623100187,
+      "grad_norm": 520.7740018235891,
+      "learning_rate": 1.801234793221892e-05,
+      "loss": 302.8429,
+      "step": 11030
+    },
+    {
+      "epoch": 0.21232708757049923,
+      "grad_norm": 521.0593018524652,
+      "learning_rate": 1.800869462441049e-05,
+      "loss": 291.2706,
+      "step": 11040
+    },
+    {
+      "epoch": 0.21251941283097975,
+      "grad_norm": 588.3576071221225,
+      "learning_rate": 1.8005038333495572e-05,
+      "loss": 311.5564,
+      "step": 11050
+    },
+    {
+      "epoch": 0.21271173809146027,
+      "grad_norm": 569.3722260088634,
+      "learning_rate": 1.8001379060836088e-05,
+      "loss": 310.9369,
+      "step": 11060
+    },
+    {
+      "epoch": 0.21290406335194081,
+      "grad_norm": 522.31194426954,
+      "learning_rate": 1.7997716807795046e-05,
+      "loss": 305.1383,
+      "step": 11070
+    },
+    {
+      "epoch": 0.21309638861242133,
+      "grad_norm": 630.9407644759252,
+      "learning_rate": 1.7994051575736585e-05,
+      "loss": 303.0762,
+      "step": 11080
+    },
+    {
+      "epoch": 0.21328871387290185,
+      "grad_norm": 577.9988210606358,
+      "learning_rate": 1.7990383366025935e-05,
+      "loss": 313.8966,
+      "step": 11090
+    },
+    {
+      "epoch": 0.21348103913338237,
+      "grad_norm": 596.8309543571359,
+      "learning_rate": 1.7986712180029448e-05,
+      "loss": 302.5571,
+      "step": 11100
+    },
+    {
+      "epoch": 0.2136733643938629,
+      "grad_norm": 608.5128534922504,
+      "learning_rate": 1.798303801911458e-05,
+      "loss": 307.915,
+      "step": 11110
+    },
+    {
+      "epoch": 0.21386568965434344,
+      "grad_norm": 507.2044641262266,
+      "learning_rate": 1.7979360884649894e-05,
+      "loss": 305.7907,
+      "step": 11120
+    },
+    {
+      "epoch": 0.21405801491482396,
+      "grad_norm": 618.3844228500747,
+      "learning_rate": 1.7975680778005058e-05,
+      "loss": 303.9178,
+      "step": 11130
+    },
+    {
+      "epoch": 0.21425034017530448,
+      "grad_norm": 531.483007694421,
+      "learning_rate": 1.7971997700550856e-05,
+      "loss": 311.8147,
+      "step": 11140
+    },
+    {
+      "epoch": 0.214442665435785,
+      "grad_norm": 522.6184963030615,
+      "learning_rate": 1.7968311653659177e-05,
+      "loss": 312.576,
+      "step": 11150
+    },
+    {
+      "epoch": 0.21463499069626552,
+      "grad_norm": 576.1439487221288,
+      "learning_rate": 1.7964622638703003e-05,
+      "loss": 314.4616,
+      "step": 11160
+    },
+    {
+      "epoch": 0.21482731595674606,
+      "grad_norm": 658.0935489057877,
+      "learning_rate": 1.796093065705644e-05,
+      "loss": 303.3928,
+      "step": 11170
+    },
+    {
+      "epoch": 0.21501964121722658,
+      "grad_norm": 541.920868642979,
+      "learning_rate": 1.7957235710094686e-05,
+      "loss": 296.9919,
+      "step": 11180
+    },
+    {
+      "epoch": 0.2152119664777071,
+      "grad_norm": 537.1510752699605,
+      "learning_rate": 1.7953537799194042e-05,
+      "loss": 309.1538,
+      "step": 11190
+    },
+    {
+      "epoch": 0.21540429173818762,
+      "grad_norm": 542.4923321336655,
+      "learning_rate": 1.7949836925731934e-05,
+      "loss": 319.2759,
+      "step": 11200
+    },
+    {
+      "epoch": 0.21559661699866814,
+      "grad_norm": 534.3221139524255,
+      "learning_rate": 1.7946133091086858e-05,
+      "loss": 312.7573,
+      "step": 11210
+    },
+    {
+      "epoch": 0.21578894225914869,
+      "grad_norm": 531.0779717398163,
+      "learning_rate": 1.7942426296638447e-05,
+      "loss": 311.4548,
+      "step": 11220
+    },
+    {
+      "epoch": 0.2159812675196292,
+      "grad_norm": 580.0357496194683,
+      "learning_rate": 1.7938716543767412e-05,
+      "loss": 306.2069,
+      "step": 11230
+    },
+    {
+      "epoch": 0.21617359278010972,
+      "grad_norm": 663.0224193800212,
+      "learning_rate": 1.7935003833855576e-05,
+      "loss": 314.1557,
+      "step": 11240
+    },
+    {
+      "epoch": 0.21636591804059024,
+      "grad_norm": 596.3035535643022,
+      "learning_rate": 1.7931288168285863e-05,
+      "loss": 309.4287,
+      "step": 11250
+    },
+    {
+      "epoch": 0.21655824330107076,
+      "grad_norm": 604.8046540891174,
+      "learning_rate": 1.79275695484423e-05,
+      "loss": 306.127,
+      "step": 11260
+    },
+    {
+      "epoch": 0.21675056856155128,
+      "grad_norm": 683.3040649358255,
+      "learning_rate": 1.7923847975710003e-05,
+      "loss": 312.069,
+      "step": 11270
+    },
+    {
+      "epoch": 0.21694289382203183,
+      "grad_norm": 580.2089930819467,
+      "learning_rate": 1.7920123451475203e-05,
+      "loss": 317.7584,
+      "step": 11280
+    },
+    {
+      "epoch": 0.21713521908251235,
+      "grad_norm": 715.5375686891132,
+      "learning_rate": 1.7916395977125227e-05,
+      "loss": 301.8664,
+      "step": 11290
+    },
+    {
+      "epoch": 0.21732754434299287,
+      "grad_norm": 598.8332989322347,
+      "learning_rate": 1.7912665554048486e-05,
+      "loss": 311.0309,
+      "step": 11300
+    },
+    {
+      "epoch": 0.2175198696034734,
+      "grad_norm": 521.3338148124917,
+      "learning_rate": 1.7908932183634515e-05,
+      "loss": 305.5654,
+      "step": 11310
+    },
+    {
+      "epoch": 0.2177121948639539,
+      "grad_norm": 544.6096900019014,
+      "learning_rate": 1.790519586727392e-05,
+      "loss": 316.0747,
+      "step": 11320
+    },
+    {
+      "epoch": 0.21790452012443445,
+      "grad_norm": 594.5251674337779,
+      "learning_rate": 1.790145660635843e-05,
+      "loss": 309.1104,
+      "step": 11330
+    },
+    {
+      "epoch": 0.21809684538491497,
+      "grad_norm": 523.6243561827803,
+      "learning_rate": 1.7897714402280844e-05,
+      "loss": 310.518,
+      "step": 11340
+    },
+    {
+      "epoch": 0.2182891706453955,
+      "grad_norm": 596.961000222273,
+      "learning_rate": 1.789396925643508e-05,
+      "loss": 322.9565,
+      "step": 11350
+    },
+    {
+      "epoch": 0.218481495905876,
+      "grad_norm": 583.2129825484865,
+      "learning_rate": 1.7890221170216144e-05,
+      "loss": 305.7768,
+      "step": 11360
+    },
+    {
+      "epoch": 0.21867382116635653,
+      "grad_norm": 565.6471703498117,
+      "learning_rate": 1.788647014502013e-05,
+      "loss": 302.2495,
+      "step": 11370
+    },
+    {
+      "epoch": 0.21886614642683708,
+      "grad_norm": 547.5601541733321,
+      "learning_rate": 1.7882716182244242e-05,
+      "loss": 300.4357,
+      "step": 11380
+    },
+    {
+      "epoch": 0.2190584716873176,
+      "grad_norm": 549.4726268170266,
+      "learning_rate": 1.7878959283286758e-05,
+      "loss": 305.4491,
+      "step": 11390
+    },
+    {
+      "epoch": 0.21925079694779812,
+      "grad_norm": 559.2964976846993,
+      "learning_rate": 1.787519944954707e-05,
+      "loss": 302.4354,
+      "step": 11400
+    },
+    {
+      "epoch": 0.21944312220827863,
+      "grad_norm": 611.1297539064526,
+      "learning_rate": 1.7871436682425645e-05,
+      "loss": 304.8266,
+      "step": 11410
+    },
+    {
+      "epoch": 0.21963544746875915,
+      "grad_norm": 1160.4401872213225,
+      "learning_rate": 1.786767098332406e-05,
+      "loss": 311.4399,
+      "step": 11420
+    },
+    {
+      "epoch": 0.2198277727292397,
+      "grad_norm": 580.2492717341304,
+      "learning_rate": 1.7863902353644972e-05,
+      "loss": 296.7177,
+      "step": 11430
+    },
+    {
+      "epoch": 0.22002009798972022,
+      "grad_norm": 515.5698565524134,
+      "learning_rate": 1.7860130794792137e-05,
+      "loss": 304.8289,
+      "step": 11440
+    },
+    {
+      "epoch": 0.22021242325020074,
+      "grad_norm": 648.2413118586277,
+      "learning_rate": 1.7856356308170394e-05,
+      "loss": 312.8327,
+      "step": 11450
+    },
+    {
+      "epoch": 0.22040474851068126,
+      "grad_norm": 537.3760888697842,
+      "learning_rate": 1.7852578895185675e-05,
+      "loss": 318.3744,
+      "step": 11460
+    },
+    {
+      "epoch": 0.22059707377116178,
+      "grad_norm": 780.1642022244691,
+      "learning_rate": 1.7848798557245008e-05,
+      "loss": 312.9608,
+      "step": 11470
+    },
+    {
+      "epoch": 0.22078939903164232,
+      "grad_norm": 568.0645992415976,
+      "learning_rate": 1.7845015295756506e-05,
+      "loss": 315.363,
+      "step": 11480
+    },
+    {
+      "epoch": 0.22098172429212284,
+      "grad_norm": 542.4850324846815,
+      "learning_rate": 1.784122911212937e-05,
+      "loss": 306.4943,
+      "step": 11490
+    },
+    {
+      "epoch": 0.22117404955260336,
+      "grad_norm": 570.1950611492051,
+      "learning_rate": 1.7837440007773895e-05,
+      "loss": 301.757,
+      "step": 11500
+    },
+    {
+      "epoch": 0.22136637481308388,
+      "grad_norm": 534.0707811987077,
+      "learning_rate": 1.783364798410146e-05,
+      "loss": 303.0547,
+      "step": 11510
+    },
+    {
+      "epoch": 0.2215587000735644,
+      "grad_norm": 581.8037279006699,
+      "learning_rate": 1.782985304252452e-05,
+      "loss": 305.9409,
+      "step": 11520
+    },
+    {
+      "epoch": 0.22175102533404495,
+      "grad_norm": 612.2218836941645,
+      "learning_rate": 1.7826055184456643e-05,
+      "loss": 312.1448,
+      "step": 11530
+    },
+    {
+      "epoch": 0.22194335059452547,
+      "grad_norm": 585.198505690767,
+      "learning_rate": 1.7822254411312455e-05,
+      "loss": 310.4949,
+      "step": 11540
+    },
+    {
+      "epoch": 0.222135675855006,
+      "grad_norm": 509.0699217340508,
+      "learning_rate": 1.781845072450769e-05,
+      "loss": 302.3834,
+      "step": 11550
+    },
+    {
+      "epoch": 0.2223280011154865,
+      "grad_norm": 1289.932570834128,
+      "learning_rate": 1.7814644125459157e-05,
+      "loss": 322.5668,
+      "step": 11560
+    },
+    {
+      "epoch": 0.22252032637596703,
+      "grad_norm": 572.8635606759849,
+      "learning_rate": 1.781083461558475e-05,
+      "loss": 305.3428,
+      "step": 11570
+    },
+    {
+      "epoch": 0.22271265163644757,
+      "grad_norm": 590.3681787819465,
+      "learning_rate": 1.7807022196303447e-05,
+      "loss": 308.0463,
+      "step": 11580
+    },
+    {
+      "epoch": 0.2229049768969281,
+      "grad_norm": 572.7963502973921,
+      "learning_rate": 1.7803206869035318e-05,
+      "loss": 304.7383,
+      "step": 11590
+    },
+    {
+      "epoch": 0.2230973021574086,
+      "grad_norm": 592.1830499673953,
+      "learning_rate": 1.7799388635201498e-05,
+      "loss": 309.5392,
+      "step": 11600
+    },
+    {
+      "epoch": 0.22328962741788913,
+      "grad_norm": 618.3588998095855,
+      "learning_rate": 1.7795567496224226e-05,
+      "loss": 305.4437,
+      "step": 11610
+    },
+    {
+      "epoch": 0.22348195267836965,
+      "grad_norm": 613.8310598933958,
+      "learning_rate": 1.7791743453526812e-05,
+      "loss": 310.5372,
+      "step": 11620
+    },
+    {
+      "epoch": 0.2236742779388502,
+      "grad_norm": 559.7342391082025,
+      "learning_rate": 1.778791650853364e-05,
+      "loss": 306.2013,
+      "step": 11630
+    },
+    {
+      "epoch": 0.22386660319933072,
+      "grad_norm": 570.6339732172067,
+      "learning_rate": 1.77840866626702e-05,
+      "loss": 303.6731,
+      "step": 11640
+    },
+    {
+      "epoch": 0.22405892845981124,
+      "grad_norm": 608.3936964239032,
+      "learning_rate": 1.7780253917363026e-05,
+      "loss": 307.5745,
+      "step": 11650
+    },
+    {
+      "epoch": 0.22425125372029175,
+      "grad_norm": 561.8810491748844,
+      "learning_rate": 1.7776418274039767e-05,
+      "loss": 305.4122,
+      "step": 11660
+    },
+    {
+      "epoch": 0.22444357898077227,
+      "grad_norm": 601.7923080450054,
+      "learning_rate": 1.7772579734129136e-05,
+      "loss": 308.8326,
+      "step": 11670
+    },
+    {
+      "epoch": 0.2246359042412528,
+      "grad_norm": 510.0679196459565,
+      "learning_rate": 1.776873829906092e-05,
+      "loss": 304.9228,
+      "step": 11680
+    },
+    {
+      "epoch": 0.22482822950173334,
+      "grad_norm": 580.1129868599272,
+      "learning_rate": 1.776489397026599e-05,
+      "loss": 319.2673,
+      "step": 11690
+    },
+    {
+      "epoch": 0.22502055476221386,
+      "grad_norm": 558.125191843454,
+      "learning_rate": 1.7761046749176302e-05,
+      "loss": 301.482,
+      "step": 11700
+    },
+    {
+      "epoch": 0.22521288002269438,
+      "grad_norm": 578.7053744596622,
+      "learning_rate": 1.7757196637224874e-05,
+      "loss": 308.5137,
+      "step": 11710
+    },
+    {
+      "epoch": 0.2254052052831749,
+      "grad_norm": 565.3965773509369,
+      "learning_rate": 1.7753343635845817e-05,
+      "loss": 305.2033,
+      "step": 11720
+    },
+    {
+      "epoch": 0.22559753054365542,
+      "grad_norm": 601.6063592003611,
+      "learning_rate": 1.7749487746474305e-05,
+      "loss": 304.9521,
+      "step": 11730
+    },
+    {
+      "epoch": 0.22578985580413596,
+      "grad_norm": 563.3044930296599,
+      "learning_rate": 1.7745628970546592e-05,
+      "loss": 306.7528,
+      "step": 11740
+    },
+    {
+      "epoch": 0.22598218106461648,
+      "grad_norm": 566.3592920601307,
+      "learning_rate": 1.774176730950001e-05,
+      "loss": 293.4409,
+      "step": 11750
+    },
+    {
+      "epoch": 0.226174506325097,
+      "grad_norm": 559.484689119316,
+      "learning_rate": 1.7737902764772967e-05,
+      "loss": 313.2567,
+      "step": 11760
+    },
+    {
+      "epoch": 0.22636683158557752,
+      "grad_norm": 683.5063044158878,
+      "learning_rate": 1.773403533780494e-05,
+      "loss": 303.2523,
+      "step": 11770
+    },
+    {
+      "epoch": 0.22655915684605804,
+      "grad_norm": 549.2141830491124,
+      "learning_rate": 1.7730165030036482e-05,
+      "loss": 302.4703,
+      "step": 11780
+    },
+    {
+      "epoch": 0.2267514821065386,
+      "grad_norm": 509.9092349300344,
+      "learning_rate": 1.772629184290922e-05,
+      "loss": 302.844,
+      "step": 11790
+    },
+    {
+      "epoch": 0.2269438073670191,
+      "grad_norm": 509.02943504371177,
+      "learning_rate": 1.7722415777865845e-05,
+      "loss": 296.8754,
+      "step": 11800
+    },
+    {
+      "epoch": 0.22713613262749963,
+      "grad_norm": 507.21021105769086,
+      "learning_rate": 1.771853683635014e-05,
+      "loss": 303.3082,
+      "step": 11810
+    },
+    {
+      "epoch": 0.22732845788798015,
+      "grad_norm": 560.8562615770735,
+      "learning_rate": 1.7714655019806932e-05,
+      "loss": 305.1188,
+      "step": 11820
+    },
+    {
+      "epoch": 0.22752078314846066,
+      "grad_norm": 584.0407112283326,
+      "learning_rate": 1.7710770329682145e-05,
+      "loss": 316.5687,
+      "step": 11830
+    },
+    {
+      "epoch": 0.2277131084089412,
+      "grad_norm": 559.5412673922286,
+      "learning_rate": 1.770688276742276e-05,
+      "loss": 307.9617,
+      "step": 11840
+    },
+    {
+      "epoch": 0.22790543366942173,
+      "grad_norm": 555.3979357564721,
+      "learning_rate": 1.770299233447682e-05,
+      "loss": 306.0619,
+      "step": 11850
+    },
+    {
+      "epoch": 0.22809775892990225,
+      "grad_norm": 548.153270289919,
+      "learning_rate": 1.769909903229346e-05,
+      "loss": 299.7813,
+      "step": 11860
+    },
+    {
+      "epoch": 0.22829008419038277,
+      "grad_norm": 847.3025142500062,
+      "learning_rate": 1.7695202862322863e-05,
+      "loss": 303.0395,
+      "step": 11870
+    },
+    {
+      "epoch": 0.2284824094508633,
+      "grad_norm": 562.1608863876545,
+      "learning_rate": 1.769130382601629e-05,
+      "loss": 310.7528,
+      "step": 11880
+    },
+    {
+      "epoch": 0.22867473471134384,
+      "grad_norm": 545.5164611487909,
+      "learning_rate": 1.768740192482607e-05,
+      "loss": 307.0152,
+      "step": 11890
+    },
+    {
+      "epoch": 0.22886705997182435,
+      "grad_norm": 562.817471848758,
+      "learning_rate": 1.7683497160205595e-05,
+      "loss": 308.5016,
+      "step": 11900
+    },
+    {
+      "epoch": 0.22905938523230487,
+      "grad_norm": 565.3214936806978,
+      "learning_rate": 1.7679589533609323e-05,
+      "loss": 310.3926,
+      "step": 11910
+    },
+    {
+      "epoch": 0.2292517104927854,
+      "grad_norm": 521.7265229421238,
+      "learning_rate": 1.767567904649278e-05,
+      "loss": 316.9577,
+      "step": 11920
+    },
+    {
+      "epoch": 0.2294440357532659,
+      "grad_norm": 599.3494243960317,
+      "learning_rate": 1.7671765700312567e-05,
+      "loss": 310.6751,
+      "step": 11930
+    },
+    {
+      "epoch": 0.22963636101374646,
+      "grad_norm": 527.4853462062857,
+      "learning_rate": 1.7667849496526327e-05,
+      "loss": 299.9745,
+      "step": 11940
+    },
+    {
+      "epoch": 0.22982868627422698,
+      "grad_norm": 646.3771999166254,
+      "learning_rate": 1.7663930436592793e-05,
+      "loss": 307.4025,
+      "step": 11950
+    },
+    {
+      "epoch": 0.2300210115347075,
+      "grad_norm": 604.8359393697335,
+      "learning_rate": 1.7660008521971744e-05,
+      "loss": 305.6533,
+      "step": 11960
+    },
+    {
+      "epoch": 0.23021333679518802,
+      "grad_norm": 550.8517134078223,
+      "learning_rate": 1.765608375412403e-05,
+      "loss": 303.4489,
+      "step": 11970
+    },
+    {
+      "epoch": 0.23040566205566854,
+      "grad_norm": 518.0078862451358,
+      "learning_rate": 1.7652156134511563e-05,
+      "loss": 308.9563,
+      "step": 11980
+    },
+    {
+      "epoch": 0.23059798731614908,
+      "grad_norm": 532.9641436854245,
+      "learning_rate": 1.7648225664597314e-05,
+      "loss": 310.4812,
+      "step": 11990
+    },
+    {
+      "epoch": 0.2307903125766296,
+      "grad_norm": 571.0244842024215,
+      "learning_rate": 1.764429234584532e-05,
+      "loss": 307.4173,
+      "step": 12000
+    },
+    {
+      "epoch": 0.23098263783711012,
+      "grad_norm": 501.98426173204206,
+      "learning_rate": 1.7640356179720674e-05,
+      "loss": 290.2313,
+      "step": 12010
+    },
+    {
+      "epoch": 0.23117496309759064,
+      "grad_norm": 561.4131530710366,
+      "learning_rate": 1.7636417167689538e-05,
+      "loss": 311.5085,
+      "step": 12020
+    },
+    {
+      "epoch": 0.23136728835807116,
+      "grad_norm": 516.0838738035736,
+      "learning_rate": 1.7632475311219125e-05,
+      "loss": 297.6313,
+      "step": 12030
+    },
+    {
+      "epoch": 0.2315596136185517,
+      "grad_norm": 521.037984161895,
+      "learning_rate": 1.7628530611777716e-05,
+      "loss": 304.9156,
+      "step": 12040
+    },
+    {
+      "epoch": 0.23175193887903223,
+      "grad_norm": 591.7189014436991,
+      "learning_rate": 1.7624583070834646e-05,
+      "loss": 309.6885,
+      "step": 12050
+    },
+    {
+      "epoch": 0.23194426413951275,
+      "grad_norm": 569.7181944729994,
+      "learning_rate": 1.7620632689860298e-05,
+      "loss": 310.1962,
+      "step": 12060
+    },
+    {
+      "epoch": 0.23213658939999327,
+      "grad_norm": 609.9481837210487,
+      "learning_rate": 1.761667947032614e-05,
+      "loss": 311.4228,
+      "step": 12070
+    },
+    {
+      "epoch": 0.23232891466047378,
+      "grad_norm": 656.8367275661511,
+      "learning_rate": 1.761272341370467e-05,
+      "loss": 302.2965,
+      "step": 12080
+    },
+    {
+      "epoch": 0.23252123992095433,
+      "grad_norm": 523.6629882282335,
+      "learning_rate": 1.7608764521469456e-05,
+      "loss": 296.982,
+      "step": 12090
+    },
+    {
+      "epoch": 0.23271356518143485,
+      "grad_norm": 509.9413106798182,
+      "learning_rate": 1.760480279509512e-05,
+      "loss": 303.0857,
+      "step": 12100
+    },
+    {
+      "epoch": 0.23290589044191537,
+      "grad_norm": 500.69792839245997,
+      "learning_rate": 1.7600838236057342e-05,
+      "loss": 305.0519,
+      "step": 12110
+    },
+    {
+      "epoch": 0.2330982157023959,
+      "grad_norm": 547.4583321583707,
+      "learning_rate": 1.759687084583285e-05,
+      "loss": 301.2934,
+      "step": 12120
+    },
+    {
+      "epoch": 0.2332905409628764,
+      "grad_norm": 591.861994057418,
+      "learning_rate": 1.7592900625899437e-05,
+      "loss": 309.8254,
+      "step": 12130
+    },
+    {
+      "epoch": 0.23348286622335693,
+      "grad_norm": 555.3955269832737,
+      "learning_rate": 1.7588927577735943e-05,
+      "loss": 297.1861,
+      "step": 12140
+    },
+    {
+      "epoch": 0.23367519148383747,
+      "grad_norm": 579.9502284682283,
+      "learning_rate": 1.758495170282226e-05,
+      "loss": 303.1287,
+      "step": 12150
+    },
+    {
+      "epoch": 0.233867516744318,
+      "grad_norm": 525.6276995371792,
+      "learning_rate": 1.7580973002639337e-05,
+      "loss": 308.7534,
+      "step": 12160
+    },
+    {
+      "epoch": 0.2340598420047985,
+      "grad_norm": 539.1975318065454,
+      "learning_rate": 1.7576991478669174e-05,
+      "loss": 304.014,
+      "step": 12170
+    },
+    {
+      "epoch": 0.23425216726527903,
+      "grad_norm": 510.54652607605846,
+      "learning_rate": 1.7573007132394823e-05,
+      "loss": 304.0979,
+      "step": 12180
+    },
+    {
+      "epoch": 0.23444449252575955,
+      "grad_norm": 512.4275361943896,
+      "learning_rate": 1.756901996530039e-05,
+      "loss": 305.9756,
+      "step": 12190
+    },
+    {
+      "epoch": 0.2346368177862401,
+      "grad_norm": 544.3164113279283,
+      "learning_rate": 1.7565029978871025e-05,
+      "loss": 308.1389,
+      "step": 12200
+    },
+    {
+      "epoch": 0.23482914304672062,
+      "grad_norm": 508.8993668217466,
+      "learning_rate": 1.7561037174592933e-05,
+      "loss": 302.3686,
+      "step": 12210
+    },
+    {
+      "epoch": 0.23502146830720114,
+      "grad_norm": 648.9558393513472,
+      "learning_rate": 1.7557041553953368e-05,
+      "loss": 299.2363,
+      "step": 12220
+    },
+    {
+      "epoch": 0.23521379356768166,
+      "grad_norm": 530.6570961756034,
+      "learning_rate": 1.7553043118440634e-05,
+      "loss": 302.2812,
+      "step": 12230
+    },
+    {
+      "epoch": 0.23540611882816218,
+      "grad_norm": 558.0991365727535,
+      "learning_rate": 1.7549041869544077e-05,
+      "loss": 301.6841,
+      "step": 12240
+    },
+    {
+      "epoch": 0.23559844408864272,
+      "grad_norm": 606.5203842871052,
+      "learning_rate": 1.7545037808754105e-05,
+      "loss": 297.3742,
+      "step": 12250
+    },
+    {
+      "epoch": 0.23579076934912324,
+      "grad_norm": 583.2151066740137,
+      "learning_rate": 1.754103093756216e-05,
+      "loss": 292.108,
+      "step": 12260
+    },
+    {
+      "epoch": 0.23598309460960376,
+      "grad_norm": 480.17492009611,
+      "learning_rate": 1.7537021257460732e-05,
+      "loss": 298.9061,
+      "step": 12270
+    },
+    {
+      "epoch": 0.23617541987008428,
+      "grad_norm": 565.2323195979287,
+      "learning_rate": 1.7533008769943366e-05,
+      "loss": 297.8132,
+      "step": 12280
+    },
+    {
+      "epoch": 0.2363677451305648,
+      "grad_norm": 537.2276979691027,
+      "learning_rate": 1.7528993476504644e-05,
+      "loss": 300.3809,
+      "step": 12290
+    },
+    {
+      "epoch": 0.23656007039104535,
+      "grad_norm": 503.6743938791051,
+      "learning_rate": 1.7524975378640198e-05,
+      "loss": 300.62,
+      "step": 12300
+    },
+    {
+      "epoch": 0.23675239565152587,
+      "grad_norm": 515.7159681803287,
+      "learning_rate": 1.75209544778467e-05,
+      "loss": 301.4,
+      "step": 12310
+    },
+    {
+      "epoch": 0.23694472091200638,
+      "grad_norm": 618.2664143332121,
+      "learning_rate": 1.7516930775621873e-05,
+      "loss": 316.3602,
+      "step": 12320
+    },
+    {
+      "epoch": 0.2371370461724869,
+      "grad_norm": 620.7039598249363,
+      "learning_rate": 1.751290427346448e-05,
+      "loss": 303.4737,
+      "step": 12330
+    },
+    {
+      "epoch": 0.23732937143296742,
+      "grad_norm": 482.7419963875554,
+      "learning_rate": 1.7508874972874325e-05,
+      "loss": 306.8661,
+      "step": 12340
+    },
+    {
+      "epoch": 0.23752169669344797,
+      "grad_norm": 582.2451716073994,
+      "learning_rate": 1.7504842875352254e-05,
+      "loss": 301.6255,
+      "step": 12350
+    },
+    {
+      "epoch": 0.2377140219539285,
+      "grad_norm": 598.501108814587,
+      "learning_rate": 1.7500807982400155e-05,
+      "loss": 301.3222,
+      "step": 12360
+    },
+    {
+      "epoch": 0.237906347214409,
+      "grad_norm": 511.4566218165681,
+      "learning_rate": 1.749677029552097e-05,
+      "loss": 298.636,
+      "step": 12370
+    },
+    {
+      "epoch": 0.23809867247488953,
+      "grad_norm": 479.60512775342823,
+      "learning_rate": 1.7492729816218656e-05,
+      "loss": 304.15,
+      "step": 12380
+    },
+    {
+      "epoch": 0.23829099773537005,
+      "grad_norm": 615.06572835905,
+      "learning_rate": 1.7488686545998237e-05,
+      "loss": 295.4768,
+      "step": 12390
+    },
+    {
+      "epoch": 0.2384833229958506,
+      "grad_norm": 569.7022305860479,
+      "learning_rate": 1.7484640486365757e-05,
+      "loss": 301.8458,
+      "step": 12400
+    },
+    {
+      "epoch": 0.2386756482563311,
+      "grad_norm": 597.0175513962797,
+      "learning_rate": 1.748059163882831e-05,
+      "loss": 303.6509,
+      "step": 12410
+    },
+    {
+      "epoch": 0.23886797351681163,
+      "grad_norm": 544.3996289087911,
+      "learning_rate": 1.747654000489402e-05,
+      "loss": 299.9608,
+      "step": 12420
+    },
+    {
+      "epoch": 0.23906029877729215,
+      "grad_norm": 531.5376345597161,
+      "learning_rate": 1.7472485586072062e-05,
+      "loss": 298.2016,
+      "step": 12430
+    },
+    {
+      "epoch": 0.23925262403777267,
+      "grad_norm": 619.75704443402,
+      "learning_rate": 1.746842838387264e-05,
+      "loss": 294.9825,
+      "step": 12440
+    },
+    {
+      "epoch": 0.23944494929825322,
+      "grad_norm": 714.9424099709997,
+      "learning_rate": 1.746436839980698e-05,
+      "loss": 310.8927,
+      "step": 12450
+    },
+    {
+      "epoch": 0.23963727455873374,
+      "grad_norm": 525.0519228337564,
+      "learning_rate": 1.7460305635387372e-05,
+      "loss": 309.1251,
+      "step": 12460
+    },
+    {
+      "epoch": 0.23982959981921426,
+      "grad_norm": 567.3426431180301,
+      "learning_rate": 1.745624009212713e-05,
+      "loss": 304.2225,
+      "step": 12470
+    },
+    {
+      "epoch": 0.24002192507969478,
+      "grad_norm": 656.0862941060705,
+      "learning_rate": 1.7452171771540593e-05,
+      "loss": 306.3733,
+      "step": 12480
+    },
+    {
+      "epoch": 0.2402142503401753,
+      "grad_norm": 594.1934987842768,
+      "learning_rate": 1.744810067514315e-05,
+      "loss": 303.9965,
+      "step": 12490
+    },
+    {
+      "epoch": 0.24040657560065584,
+      "grad_norm": 500.73218516629913,
+      "learning_rate": 1.7444026804451214e-05,
+      "loss": 301.8244,
+      "step": 12500
+    },
+    {
+      "epoch": 0.24059890086113636,
+      "grad_norm": 744.5292787989708,
+      "learning_rate": 1.7439950160982236e-05,
+      "loss": 300.2846,
+      "step": 12510
+    },
+    {
+      "epoch": 0.24079122612161688,
+      "grad_norm": 533.8349242343193,
+      "learning_rate": 1.7435870746254697e-05,
+      "loss": 301.0945,
+      "step": 12520
+    },
+    {
+      "epoch": 0.2409835513820974,
+      "grad_norm": 567.8827021058287,
+      "learning_rate": 1.7431788561788116e-05,
+      "loss": 298.9499,
+      "step": 12530
+    },
+    {
+      "epoch": 0.24117587664257792,
+      "grad_norm": 549.3287660004423,
+      "learning_rate": 1.742770360910303e-05,
+      "loss": 308.0892,
+      "step": 12540
+    },
+    {
+      "epoch": 0.24136820190305847,
+      "grad_norm": 492.5093797278217,
+      "learning_rate": 1.7423615889721027e-05,
+      "loss": 297.5649,
+      "step": 12550
+    },
+    {
+      "epoch": 0.24156052716353899,
+      "grad_norm": 557.3529802738032,
+      "learning_rate": 1.7419525405164705e-05,
+      "loss": 310.8019,
+      "step": 12560
+    },
+    {
+      "epoch": 0.2417528524240195,
+      "grad_norm": 621.6773333995523,
+      "learning_rate": 1.7415432156957713e-05,
+      "loss": 306.2981,
+      "step": 12570
+    },
+    {
+      "epoch": 0.24194517768450002,
+      "grad_norm": 496.63227621071644,
+      "learning_rate": 1.741133614662471e-05,
+      "loss": 310.3273,
+      "step": 12580
+    },
+    {
+      "epoch": 0.24213750294498054,
+      "grad_norm": 617.2419228856705,
+      "learning_rate": 1.7407237375691394e-05,
+      "loss": 314.5568,
+      "step": 12590
+    },
+    {
+      "epoch": 0.24232982820546106,
+      "grad_norm": 602.1228111301185,
+      "learning_rate": 1.740313584568449e-05,
+      "loss": 297.8498,
+      "step": 12600
+    },
+    {
+      "epoch": 0.2425221534659416,
+      "grad_norm": 571.1331473048172,
+      "learning_rate": 1.739903155813175e-05,
+      "loss": 309.2491,
+      "step": 12610
+    },
+    {
+      "epoch": 0.24271447872642213,
+      "grad_norm": 575.4803355379195,
+      "learning_rate": 1.7394924514561955e-05,
+      "loss": 302.9375,
+      "step": 12620
+    },
+    {
+      "epoch": 0.24290680398690265,
+      "grad_norm": 501.7659896461661,
+      "learning_rate": 1.739081471650491e-05,
+      "loss": 295.4767,
+      "step": 12630
+    },
+    {
+      "epoch": 0.24309912924738317,
+      "grad_norm": 502.862777557259,
+      "learning_rate": 1.7386702165491443e-05,
+      "loss": 308.9037,
+      "step": 12640
+    },
+    {
+      "epoch": 0.24329145450786369,
+      "grad_norm": 540.9587849078885,
+      "learning_rate": 1.738258686305342e-05,
+      "loss": 298.2269,
+      "step": 12650
+    },
+    {
+      "epoch": 0.24348377976834423,
+      "grad_norm": 560.3407627782135,
+      "learning_rate": 1.7378468810723713e-05,
+      "loss": 310.2495,
+      "step": 12660
+    },
+    {
+      "epoch": 0.24367610502882475,
+      "grad_norm": 492.6380104274809,
+      "learning_rate": 1.7374348010036235e-05,
+      "loss": 305.7097,
+      "step": 12670
+    },
+    {
+      "epoch": 0.24386843028930527,
+      "grad_norm": 525.378463611019,
+      "learning_rate": 1.7370224462525916e-05,
+      "loss": 302.5172,
+      "step": 12680
+    },
+    {
+      "epoch": 0.2440607555497858,
+      "grad_norm": 755.9612244630933,
+      "learning_rate": 1.736609816972871e-05,
+      "loss": 306.4417,
+      "step": 12690
+    },
+    {
+      "epoch": 0.2442530808102663,
+      "grad_norm": 499.8778357657815,
+      "learning_rate": 1.7361969133181585e-05,
+      "loss": 291.3082,
+      "step": 12700
+    },
+    {
+      "epoch": 0.24444540607074686,
+      "grad_norm": 532.941702434763,
+      "learning_rate": 1.735783735442255e-05,
+      "loss": 303.3684,
+      "step": 12710
+    },
+    {
+      "epoch": 0.24463773133122738,
+      "grad_norm": 555.5077925873111,
+      "learning_rate": 1.7353702834990617e-05,
+      "loss": 322.8935,
+      "step": 12720
+    },
+    {
+      "epoch": 0.2448300565917079,
+      "grad_norm": 674.4471482697896,
+      "learning_rate": 1.7349565576425828e-05,
+      "loss": 299.7307,
+      "step": 12730
+    },
+    {
+      "epoch": 0.24502238185218841,
+      "grad_norm": 512.1654025301481,
+      "learning_rate": 1.7345425580269245e-05,
+      "loss": 304.8884,
+      "step": 12740
+    },
+    {
+      "epoch": 0.24521470711266893,
+      "grad_norm": 504.974209543955,
+      "learning_rate": 1.734128284806295e-05,
+      "loss": 299.1169,
+      "step": 12750
+    },
+    {
+      "epoch": 0.24540703237314948,
+      "grad_norm": 528.5600347707219,
+      "learning_rate": 1.7337137381350033e-05,
+      "loss": 298.6342,
+      "step": 12760
+    },
+    {
+      "epoch": 0.24559935763363,
+      "grad_norm": 510.04817930568515,
+      "learning_rate": 1.7332989181674623e-05,
+      "loss": 291.0542,
+      "step": 12770
+    },
+    {
+      "epoch": 0.24579168289411052,
+      "grad_norm": 535.6167463029005,
+      "learning_rate": 1.7328838250581846e-05,
+      "loss": 300.2511,
+      "step": 12780
+    },
+    {
+      "epoch": 0.24598400815459104,
+      "grad_norm": 520.3089821326004,
+      "learning_rate": 1.7324684589617862e-05,
+      "loss": 296.7046,
+      "step": 12790
+    },
+    {
+      "epoch": 0.24617633341507156,
+      "grad_norm": 596.9425948418302,
+      "learning_rate": 1.7320528200329846e-05,
+      "loss": 304.3464,
+      "step": 12800
+    },
+    {
+      "epoch": 0.2463686586755521,
+      "grad_norm": 581.5105053657546,
+      "learning_rate": 1.7316369084265973e-05,
+      "loss": 292.5362,
+      "step": 12810
+    },
+    {
+      "epoch": 0.24656098393603262,
+      "grad_norm": 640.1870699333763,
+      "learning_rate": 1.731220724297545e-05,
+      "loss": 290.2909,
+      "step": 12820
+    },
+    {
+      "epoch": 0.24675330919651314,
+      "grad_norm": 550.8889959690337,
+      "learning_rate": 1.730804267800849e-05,
+      "loss": 301.2725,
+      "step": 12830
+    },
+    {
+      "epoch": 0.24694563445699366,
+      "grad_norm": 615.6400733063065,
+      "learning_rate": 1.7303875390916338e-05,
+      "loss": 293.0394,
+      "step": 12840
+    },
+    {
+      "epoch": 0.24713795971747418,
+      "grad_norm": 572.3040979754213,
+      "learning_rate": 1.729970538325122e-05,
+      "loss": 293.7845,
+      "step": 12850
+    },
+    {
+      "epoch": 0.24733028497795473,
+      "grad_norm": 573.7230864891717,
+      "learning_rate": 1.7295532656566413e-05,
+      "loss": 305.99,
+      "step": 12860
+    },
+    {
+      "epoch": 0.24752261023843525,
+      "grad_norm": 525.6066716731832,
+      "learning_rate": 1.729135721241618e-05,
+      "loss": 296.0017,
+      "step": 12870
+    },
+    {
+      "epoch": 0.24771493549891577,
+      "grad_norm": 519.4868282111524,
+      "learning_rate": 1.7287179052355803e-05,
+      "loss": 306.5752,
+      "step": 12880
+    },
+    {
+      "epoch": 0.2479072607593963,
+      "grad_norm": 582.4963962470334,
+      "learning_rate": 1.7282998177941586e-05,
+      "loss": 309.5309,
+      "step": 12890
+    },
+    {
+      "epoch": 0.2480995860198768,
+      "grad_norm": 699.9081612715943,
+      "learning_rate": 1.7278814590730826e-05,
+      "loss": 296.2731,
+      "step": 12900
+    },
+    {
+      "epoch": 0.24829191128035735,
+      "grad_norm": 583.8544732208774,
+      "learning_rate": 1.7274628292281846e-05,
+      "loss": 294.651,
+      "step": 12910
+    },
+    {
+      "epoch": 0.24848423654083787,
+      "grad_norm": 516.028225369162,
+      "learning_rate": 1.727043928415397e-05,
+      "loss": 295.5151,
+      "step": 12920
+    },
+    {
+      "epoch": 0.2486765618013184,
+      "grad_norm": 556.5898797919142,
+      "learning_rate": 1.726624756790754e-05,
+      "loss": 301.7446,
+      "step": 12930
+    },
+    {
+      "epoch": 0.2488688870617989,
+      "grad_norm": 569.5119985607591,
+      "learning_rate": 1.7262053145103893e-05,
+      "loss": 304.4961,
+      "step": 12940
+    },
+    {
+      "epoch": 0.24906121232227943,
+      "grad_norm": 634.5322402933849,
+      "learning_rate": 1.7257856017305387e-05,
+      "loss": 299.3397,
+      "step": 12950
+    },
+    {
+      "epoch": 0.24925353758275998,
+      "grad_norm": 597.663695630516,
+      "learning_rate": 1.725365618607538e-05,
+      "loss": 293.8806,
+      "step": 12960
+    },
+    {
+      "epoch": 0.2494458628432405,
+      "grad_norm": 579.0377545012869,
+      "learning_rate": 1.7249453652978244e-05,
+      "loss": 295.638,
+      "step": 12970
+    },
+    {
+      "epoch": 0.24963818810372101,
+      "grad_norm": 495.84654086795075,
+      "learning_rate": 1.7245248419579353e-05,
+      "loss": 302.5208,
+      "step": 12980
+    },
+    {
+      "epoch": 0.24983051336420153,
+      "grad_norm": 527.5841204917268,
+      "learning_rate": 1.7241040487445082e-05,
+      "loss": 295.0076,
+      "step": 12990
+    },
+    {
+      "epoch": 0.25002283862468205,
+      "grad_norm": 495.69539542594634,
+      "learning_rate": 1.723682985814282e-05,
+      "loss": 296.1797,
+      "step": 13000
+    },
+    {
+      "epoch": 0.2502151638851626,
+      "grad_norm": 584.3560064789656,
+      "learning_rate": 1.7232616533240958e-05,
+      "loss": 300.3373,
+      "step": 13010
+    },
+    {
+      "epoch": 0.2504074891456431,
+      "grad_norm": 544.5313476904952,
+      "learning_rate": 1.7228400514308884e-05,
+      "loss": 304.6298,
+      "step": 13020
+    },
+    {
+      "epoch": 0.2505998144061236,
+      "grad_norm": 579.1546103880229,
+      "learning_rate": 1.7224181802917003e-05,
+      "loss": 295.4002,
+      "step": 13030
+    },
+    {
+      "epoch": 0.2507921396666042,
+      "grad_norm": 516.2741820851571,
+      "learning_rate": 1.721996040063671e-05,
+      "loss": 303.0126,
+      "step": 13040
+    },
+    {
+      "epoch": 0.2509844649270847,
+      "grad_norm": 630.6213058366193,
+      "learning_rate": 1.7215736309040408e-05,
+      "loss": 297.3093,
+      "step": 13050
+    },
+    {
+      "epoch": 0.2511767901875652,
+      "grad_norm": 559.9087309239119,
+      "learning_rate": 1.7211509529701507e-05,
+      "loss": 295.7827,
+      "step": 13060
+    },
+    {
+      "epoch": 0.25136911544804574,
+      "grad_norm": 614.6784026929857,
+      "learning_rate": 1.7207280064194403e-05,
+      "loss": 300.8992,
+      "step": 13070
+    },
+    {
+      "epoch": 0.25156144070852626,
+      "grad_norm": 580.7944634711013,
+      "learning_rate": 1.7203047914094514e-05,
+      "loss": 295.1916,
+      "step": 13080
+    },
+    {
+      "epoch": 0.2517537659690068,
+      "grad_norm": 565.5970850023269,
+      "learning_rate": 1.7198813080978235e-05,
+      "loss": 298.4104,
+      "step": 13090
+    },
+    {
+      "epoch": 0.2519460912294873,
+      "grad_norm": 582.0710508609554,
+      "learning_rate": 1.7194575566422975e-05,
+      "loss": 302.5477,
+      "step": 13100
+    },
+    {
+      "epoch": 0.2521384164899678,
+      "grad_norm": 522.6514913090786,
+      "learning_rate": 1.719033537200714e-05,
+      "loss": 290.3607,
+      "step": 13110
+    },
+    {
+      "epoch": 0.25233074175044834,
+      "grad_norm": 544.8798463472215,
+      "learning_rate": 1.7186092499310133e-05,
+      "loss": 300.6581,
+      "step": 13120
+    },
+    {
+      "epoch": 0.25252306701092886,
+      "grad_norm": 509.1442320602432,
+      "learning_rate": 1.7181846949912347e-05,
+      "loss": 295.8213,
+      "step": 13130
+    },
+    {
+      "epoch": 0.25271539227140943,
+      "grad_norm": 477.28183928493957,
+      "learning_rate": 1.717759872539519e-05,
+      "loss": 291.7312,
+      "step": 13140
+    },
+    {
+      "epoch": 0.25290771753188995,
+      "grad_norm": 549.2815820130234,
+      "learning_rate": 1.7173347827341046e-05,
+      "loss": 295.4985,
+      "step": 13150
+    },
+    {
+      "epoch": 0.25310004279237047,
+      "grad_norm": 520.6918095978543,
+      "learning_rate": 1.7169094257333307e-05,
+      "loss": 307.3299,
+      "step": 13160
+    },
+    {
+      "epoch": 0.253292368052851,
+      "grad_norm": 579.072012808588,
+      "learning_rate": 1.716483801695636e-05,
+      "loss": 302.4981,
+      "step": 13170
+    },
+    {
+      "epoch": 0.2534846933133315,
+      "grad_norm": 540.1969180342111,
+      "learning_rate": 1.7160579107795587e-05,
+      "loss": 301.8908,
+      "step": 13180
+    },
+    {
+      "epoch": 0.25367701857381203,
+      "grad_norm": 498.0216408820849,
+      "learning_rate": 1.715631753143735e-05,
+      "loss": 293.6805,
+      "step": 13190
+    },
+    {
+      "epoch": 0.25386934383429255,
+      "grad_norm": 532.8962458202869,
+      "learning_rate": 1.715205328946903e-05,
+      "loss": 304.525,
+      "step": 13200
+    },
+    {
+      "epoch": 0.25406166909477307,
+      "grad_norm": 503.52289463582196,
+      "learning_rate": 1.7147786383478978e-05,
+      "loss": 294.37,
+      "step": 13210
+    },
+    {
+      "epoch": 0.2542539943552536,
+      "grad_norm": 479.37862765508373,
+      "learning_rate": 1.7143516815056545e-05,
+      "loss": 303.5988,
+      "step": 13220
+    },
+    {
+      "epoch": 0.2544463196157341,
+      "grad_norm": 487.75782537001106,
+      "learning_rate": 1.713924458579208e-05,
+      "loss": 312.04,
+      "step": 13230
+    },
+    {
+      "epoch": 0.2546386448762147,
+      "grad_norm": 557.1345963784541,
+      "learning_rate": 1.7134969697276912e-05,
+      "loss": 294.1135,
+      "step": 13240
+    },
+    {
+      "epoch": 0.2548309701366952,
+      "grad_norm": 532.3653339709429,
+      "learning_rate": 1.7130692151103373e-05,
+      "loss": 294.1397,
+      "step": 13250
+    },
+    {
+      "epoch": 0.2550232953971757,
+      "grad_norm": 559.8926544482599,
+      "learning_rate": 1.7126411948864776e-05,
+      "loss": 295.2776,
+      "step": 13260
+    },
+    {
+      "epoch": 0.25521562065765624,
+      "grad_norm": 602.2117571941698,
+      "learning_rate": 1.7122129092155422e-05,
+      "loss": 308.6386,
+      "step": 13270
+    },
+    {
+      "epoch": 0.25540794591813676,
+      "grad_norm": 611.6861783713705,
+      "learning_rate": 1.7117843582570608e-05,
+      "loss": 296.0421,
+      "step": 13280
+    },
+    {
+      "epoch": 0.2556002711786173,
+      "grad_norm": 549.0821252476653,
+      "learning_rate": 1.711355542170661e-05,
+      "loss": 288.2859,
+      "step": 13290
+    },
+    {
+      "epoch": 0.2557925964390978,
+      "grad_norm": 607.3341639249296,
+      "learning_rate": 1.710926461116071e-05,
+      "loss": 288.6771,
+      "step": 13300
+    },
+    {
+      "epoch": 0.2559849216995783,
+      "grad_norm": 477.49183556700274,
+      "learning_rate": 1.710497115253115e-05,
+      "loss": 298.7382,
+      "step": 13310
+    },
+    {
+      "epoch": 0.25617724696005884,
+      "grad_norm": 608.0060738334538,
+      "learning_rate": 1.7100675047417178e-05,
+      "loss": 295.9122,
+      "step": 13320
+    },
+    {
+      "epoch": 0.25636957222053935,
+      "grad_norm": 566.5000757965348,
+      "learning_rate": 1.7096376297419027e-05,
+      "loss": 297.2796,
+      "step": 13330
+    },
+    {
+      "epoch": 0.2565618974810199,
+      "grad_norm": 547.3218261512828,
+      "learning_rate": 1.70920749041379e-05,
+      "loss": 300.8601,
+      "step": 13340
+    },
+    {
+      "epoch": 0.25675422274150045,
+      "grad_norm": 556.6420527426902,
+      "learning_rate": 1.7087770869176005e-05,
+      "loss": 296.1757,
+      "step": 13350
+    },
+    {
+      "epoch": 0.25694654800198097,
+      "grad_norm": 540.4378875181017,
+      "learning_rate": 1.7083464194136517e-05,
+      "loss": 295.9015,
+      "step": 13360
+    },
+    {
+      "epoch": 0.2571388732624615,
+      "grad_norm": 558.8648666429518,
+      "learning_rate": 1.707915488062361e-05,
+      "loss": 299.0723,
+      "step": 13370
+    },
+    {
+      "epoch": 0.257331198522942,
+      "grad_norm": 534.5828983448623,
+      "learning_rate": 1.7074842930242418e-05,
+      "loss": 293.0366,
+      "step": 13380
+    },
+    {
+      "epoch": 0.2575235237834225,
+      "grad_norm": 538.5962490105335,
+      "learning_rate": 1.7070528344599083e-05,
+      "loss": 297.8597,
+      "step": 13390
+    },
+    {
+      "epoch": 0.25771584904390304,
+      "grad_norm": 569.7058279581657,
+      "learning_rate": 1.7066211125300713e-05,
+      "loss": 290.0898,
+      "step": 13400
+    },
+    {
+      "epoch": 0.25790817430438356,
+      "grad_norm": 553.1988829519527,
+      "learning_rate": 1.70618912739554e-05,
+      "loss": 288.9417,
+      "step": 13410
+    },
+    {
+      "epoch": 0.2581004995648641,
+      "grad_norm": 501.10003835685666,
+      "learning_rate": 1.705756879217222e-05,
+      "loss": 294.8937,
+      "step": 13420
+    },
+    {
+      "epoch": 0.2582928248253446,
+      "grad_norm": 603.161560784316,
+      "learning_rate": 1.7053243681561225e-05,
+      "loss": 316.7672,
+      "step": 13430
+    },
+    {
+      "epoch": 0.2584851500858251,
+      "grad_norm": 614.697465602219,
+      "learning_rate": 1.7048915943733444e-05,
+      "loss": 304.4859,
+      "step": 13440
+    },
+    {
+      "epoch": 0.2586774753463057,
+      "grad_norm": 505.5117365134719,
+      "learning_rate": 1.704458558030089e-05,
+      "loss": 293.5885,
+      "step": 13450
+    },
+    {
+      "epoch": 0.2588698006067862,
+      "grad_norm": 531.6401952165609,
+      "learning_rate": 1.704025259287656e-05,
+      "loss": 285.6869,
+      "step": 13460
+    },
+    {
+      "epoch": 0.25906212586726673,
+      "grad_norm": 553.5805254108343,
+      "learning_rate": 1.7035916983074405e-05,
+      "loss": 296.7148,
+      "step": 13470
+    },
+    {
+      "epoch": 0.25925445112774725,
+      "grad_norm": 595.8440058985766,
+      "learning_rate": 1.7031578752509377e-05,
+      "loss": 306.7942,
+      "step": 13480
+    },
+    {
+      "epoch": 0.2594467763882278,
+      "grad_norm": 510.47098422006616,
+      "learning_rate": 1.70272379027974e-05,
+      "loss": 289.0648,
+      "step": 13490
+    },
+    {
+      "epoch": 0.2596391016487083,
+      "grad_norm": 684.6421265091172,
+      "learning_rate": 1.7022894435555356e-05,
+      "loss": 295.8367,
+      "step": 13500
+    },
+    {
+      "epoch": 0.2598314269091888,
+      "grad_norm": 561.8937544173967,
+      "learning_rate": 1.7018548352401123e-05,
+      "loss": 294.5657,
+      "step": 13510
+    },
+    {
+      "epoch": 0.26002375216966933,
+      "grad_norm": 577.4846372742451,
+      "learning_rate": 1.7014199654953543e-05,
+      "loss": 290.9038,
+      "step": 13520
+    },
+    {
+      "epoch": 0.26021607743014985,
+      "grad_norm": 588.4380967028559,
+      "learning_rate": 1.700984834483244e-05,
+      "loss": 293.1427,
+      "step": 13530
+    },
+    {
+      "epoch": 0.26040840269063037,
+      "grad_norm": 534.1212486978449,
+      "learning_rate": 1.7005494423658598e-05,
+      "loss": 296.2717,
+      "step": 13540
+    },
+    {
+      "epoch": 0.26060072795111094,
+      "grad_norm": 502.98505404872185,
+      "learning_rate": 1.7001137893053782e-05,
+      "loss": 288.3195,
+      "step": 13550
+    },
+    {
+      "epoch": 0.26079305321159146,
+      "grad_norm": 540.1714886611936,
+      "learning_rate": 1.6996778754640727e-05,
+      "loss": 292.3107,
+      "step": 13560
+    },
+    {
+      "epoch": 0.260985378472072,
+      "grad_norm": 660.9975055108199,
+      "learning_rate": 1.6992417010043144e-05,
+      "loss": 293.8949,
+      "step": 13570
+    },
+    {
+      "epoch": 0.2611777037325525,
+      "grad_norm": 680.5519555020162,
+      "learning_rate": 1.6988052660885707e-05,
+      "loss": 291.8404,
+      "step": 13580
+    },
+    {
+      "epoch": 0.261370028993033,
+      "grad_norm": 560.9298138071008,
+      "learning_rate": 1.6983685708794064e-05,
+      "loss": 288.3742,
+      "step": 13590
+    },
+    {
+      "epoch": 0.26156235425351354,
+      "grad_norm": 525.2119638390872,
+      "learning_rate": 1.6979316155394834e-05,
+      "loss": 294.1705,
+      "step": 13600
+    },
+    {
+      "epoch": 0.26175467951399406,
+      "grad_norm": 538.9753596662515,
+      "learning_rate": 1.6974944002315605e-05,
+      "loss": 289.604,
+      "step": 13610
+    },
+    {
+      "epoch": 0.2619470047744746,
+      "grad_norm": 525.4341029167142,
+      "learning_rate": 1.697056925118493e-05,
+      "loss": 290.6618,
+      "step": 13620
+    },
+    {
+      "epoch": 0.2621393300349551,
+      "grad_norm": 607.1097781492762,
+      "learning_rate": 1.696619190363233e-05,
+      "loss": 308.0432,
+      "step": 13630
+    },
+    {
+      "epoch": 0.2623316552954356,
+      "grad_norm": 556.9796057232214,
+      "learning_rate": 1.69618119612883e-05,
+      "loss": 289.7933,
+      "step": 13640
+    },
+    {
+      "epoch": 0.2625239805559162,
+      "grad_norm": 525.1332058711408,
+      "learning_rate": 1.695742942578429e-05,
+      "loss": 293.5802,
+      "step": 13650
+    },
+    {
+      "epoch": 0.2627163058163967,
+      "grad_norm": 528.0396211366046,
+      "learning_rate": 1.6953044298752724e-05,
+      "loss": 292.3849,
+      "step": 13660
+    },
+    {
+      "epoch": 0.26290863107687723,
+      "grad_norm": 564.8615567957876,
+      "learning_rate": 1.694865658182699e-05,
+      "loss": 308.7998,
+      "step": 13670
+    },
+    {
+      "epoch": 0.26310095633735775,
+      "grad_norm": 480.5667912058225,
+      "learning_rate": 1.6944266276641442e-05,
+      "loss": 295.2425,
+      "step": 13680
+    },
+    {
+      "epoch": 0.26329328159783827,
+      "grad_norm": 572.5110911428648,
+      "learning_rate": 1.6939873384831394e-05,
+      "loss": 287.6731,
+      "step": 13690
+    },
+    {
+      "epoch": 0.2634856068583188,
+      "grad_norm": 498.2965572713883,
+      "learning_rate": 1.6935477908033124e-05,
+      "loss": 297.5873,
+      "step": 13700
+    },
+    {
+      "epoch": 0.2636779321187993,
+      "grad_norm": 532.8350243166844,
+      "learning_rate": 1.6931079847883877e-05,
+      "loss": 295.7148,
+      "step": 13710
+    },
+    {
+      "epoch": 0.2638702573792798,
+      "grad_norm": 551.2833474544038,
+      "learning_rate": 1.692667920602186e-05,
+      "loss": 296.3375,
+      "step": 13720
+    },
+    {
+      "epoch": 0.26406258263976035,
+      "grad_norm": 572.5843623757257,
+      "learning_rate": 1.6922275984086233e-05,
+      "loss": 303.0679,
+      "step": 13730
+    },
+    {
+      "epoch": 0.26425490790024087,
+      "grad_norm": 628.1976489547187,
+      "learning_rate": 1.691787018371713e-05,
+      "loss": 292.7349,
+      "step": 13740
+    },
+    {
+      "epoch": 0.2644472331607214,
+      "grad_norm": 582.1594080748301,
+      "learning_rate": 1.6913461806555635e-05,
+      "loss": 303.1469,
+      "step": 13750
+    },
+    {
+      "epoch": 0.26463955842120196,
+      "grad_norm": 552.6292632725243,
+      "learning_rate": 1.6909050854243797e-05,
+      "loss": 297.7123,
+      "step": 13760
+    },
+    {
+      "epoch": 0.2648318836816825,
+      "grad_norm": 487.2885543181726,
+      "learning_rate": 1.690463732842462e-05,
+      "loss": 294.969,
+      "step": 13770
+    },
+    {
+      "epoch": 0.265024208942163,
+      "grad_norm": 530.462477027445,
+      "learning_rate": 1.6900221230742073e-05,
+      "loss": 297.3343,
+      "step": 13780
+    },
+    {
+      "epoch": 0.2652165342026435,
+      "grad_norm": 556.9178408469701,
+      "learning_rate": 1.689580256284108e-05,
+      "loss": 304.0897,
+      "step": 13790
+    },
+    {
+      "epoch": 0.26540885946312404,
+      "grad_norm": 566.6930947552812,
+      "learning_rate": 1.689138132636752e-05,
+      "loss": 292.183,
+      "step": 13800
+    },
+    {
+      "epoch": 0.26560118472360456,
+      "grad_norm": 480.87919387494645,
+      "learning_rate": 1.688695752296823e-05,
+      "loss": 288.8656,
+      "step": 13810
+    },
+    {
+      "epoch": 0.2657935099840851,
+      "grad_norm": 563.1264854543442,
+      "learning_rate": 1.6882531154291007e-05,
+      "loss": 293.3224,
+      "step": 13820
+    },
+    {
+      "epoch": 0.2659858352445656,
+      "grad_norm": 616.8299049814059,
+      "learning_rate": 1.6878102221984593e-05,
+      "loss": 303.3631,
+      "step": 13830
+    },
+    {
+      "epoch": 0.2661781605050461,
+      "grad_norm": 526.4508153390387,
+      "learning_rate": 1.6873670727698702e-05,
+      "loss": 304.258,
+      "step": 13840
+    },
+    {
+      "epoch": 0.26637048576552663,
+      "grad_norm": 534.6748436898739,
+      "learning_rate": 1.686923667308398e-05,
+      "loss": 296.1364,
+      "step": 13850
+    },
+    {
+      "epoch": 0.2665628110260072,
+      "grad_norm": 482.08390550350896,
+      "learning_rate": 1.6864800059792057e-05,
+      "loss": 292.8136,
+      "step": 13860
+    },
+    {
+      "epoch": 0.2667551362864877,
+      "grad_norm": 510.38221446309745,
+      "learning_rate": 1.686036088947548e-05,
+      "loss": 311.2935,
+      "step": 13870
+    },
+    {
+      "epoch": 0.26694746154696825,
+      "grad_norm": 568.7434663673843,
+      "learning_rate": 1.6855919163787777e-05,
+      "loss": 301.1041,
+      "step": 13880
+    },
+    {
+      "epoch": 0.26713978680744876,
+      "grad_norm": 527.7892689123219,
+      "learning_rate": 1.6851474884383416e-05,
+      "loss": 289.4896,
+      "step": 13890
+    },
+    {
+      "epoch": 0.2673321120679293,
+      "grad_norm": 514.7795158654546,
+      "learning_rate": 1.6847028052917814e-05,
+      "loss": 294.7565,
+      "step": 13900
+    },
+    {
+      "epoch": 0.2675244373284098,
+      "grad_norm": 549.8434435771735,
+      "learning_rate": 1.6842578671047345e-05,
+      "loss": 290.3513,
+      "step": 13910
+    },
+    {
+      "epoch": 0.2677167625888903,
+      "grad_norm": 532.5415571416761,
+      "learning_rate": 1.683812674042933e-05,
+      "loss": 286.0049,
+      "step": 13920
+    },
+    {
+      "epoch": 0.26790908784937084,
+      "grad_norm": 487.7051289803812,
+      "learning_rate": 1.683367226272204e-05,
+      "loss": 292.6955,
+      "step": 13930
+    },
+    {
+      "epoch": 0.26810141310985136,
+      "grad_norm": 543.9904864676188,
+      "learning_rate": 1.6829215239584695e-05,
+      "loss": 294.8505,
+      "step": 13940
+    },
+    {
+      "epoch": 0.2682937383703319,
+      "grad_norm": 520.4800505776105,
+      "learning_rate": 1.6824755672677458e-05,
+      "loss": 294.392,
+      "step": 13950
+    },
+    {
+      "epoch": 0.26848606363081245,
+      "grad_norm": 642.2395485566753,
+      "learning_rate": 1.682029356366145e-05,
+      "loss": 295.3867,
+      "step": 13960
+    },
+    {
+      "epoch": 0.268678388891293,
+      "grad_norm": 475.886560055544,
+      "learning_rate": 1.6815828914198732e-05,
+      "loss": 281.3129,
+      "step": 13970
+    },
+    {
+      "epoch": 0.2688707141517735,
+      "grad_norm": 541.2896630354487,
+      "learning_rate": 1.6811361725952308e-05,
+      "loss": 303.7885,
+      "step": 13980
+    },
+    {
+      "epoch": 0.269063039412254,
+      "grad_norm": 499.94100153781005,
+      "learning_rate": 1.6806892000586135e-05,
+      "loss": 292.4338,
+      "step": 13990
+    },
+    {
+      "epoch": 0.26925536467273453,
+      "grad_norm": 575.2959380169674,
+      "learning_rate": 1.6802419739765114e-05,
+      "loss": 291.7208,
+      "step": 14000
+    },
+    {
+      "epoch": 0.26944768993321505,
+      "grad_norm": 605.5459299534795,
+      "learning_rate": 1.679794494515508e-05,
+      "loss": 294.1945,
+      "step": 14010
+    },
+    {
+      "epoch": 0.26964001519369557,
+      "grad_norm": 518.5757592090939,
+      "learning_rate": 1.6793467618422828e-05,
+      "loss": 289.982,
+      "step": 14020
+    },
+    {
+      "epoch": 0.2698323404541761,
+      "grad_norm": 550.9170740451434,
+      "learning_rate": 1.6788987761236088e-05,
+      "loss": 296.3643,
+      "step": 14030
+    },
+    {
+      "epoch": 0.2700246657146566,
+      "grad_norm": 542.1601511092452,
+      "learning_rate": 1.6784505375263533e-05,
+      "loss": 291.0375,
+      "step": 14040
+    },
+    {
+      "epoch": 0.27021699097513713,
+      "grad_norm": 516.7184261407725,
+      "learning_rate": 1.678002046217477e-05,
+      "loss": 291.1083,
+      "step": 14050
+    },
+    {
+      "epoch": 0.2704093162356177,
+      "grad_norm": 534.4485602349314,
+      "learning_rate": 1.6775533023640363e-05,
+      "loss": 308.9214,
+      "step": 14060
+    },
+    {
+      "epoch": 0.2706016414960982,
+      "grad_norm": 694.0153284362075,
+      "learning_rate": 1.6771043061331806e-05,
+      "loss": 312.7127,
+      "step": 14070
+    },
+    {
+      "epoch": 0.27079396675657874,
+      "grad_norm": 529.2729478771689,
+      "learning_rate": 1.6766550576921533e-05,
+      "loss": 290.4187,
+      "step": 14080
+    },
+    {
+      "epoch": 0.27098629201705926,
+      "grad_norm": 560.1423028923992,
+      "learning_rate": 1.676205557208293e-05,
+      "loss": 297.6623,
+      "step": 14090
+    },
+    {
+      "epoch": 0.2711786172775398,
+      "grad_norm": 586.9693772440653,
+      "learning_rate": 1.67575580484903e-05,
+      "loss": 281.7389,
+      "step": 14100
+    },
+    {
+      "epoch": 0.2713709425380203,
+      "grad_norm": 515.9894478725496,
+      "learning_rate": 1.6753058007818906e-05,
+      "loss": 303.8303,
+      "step": 14110
+    },
+    {
+      "epoch": 0.2715632677985008,
+      "grad_norm": 492.5634222475541,
+      "learning_rate": 1.674855545174493e-05,
+      "loss": 288.6625,
+      "step": 14120
+    },
+    {
+      "epoch": 0.27175559305898134,
+      "grad_norm": 497.2167725431496,
+      "learning_rate": 1.6744050381945507e-05,
+      "loss": 297.8267,
+      "step": 14130
+    },
+    {
+      "epoch": 0.27194791831946186,
+      "grad_norm": 553.3915355713392,
+      "learning_rate": 1.67395428000987e-05,
+      "loss": 290.1727,
+      "step": 14140
+    },
+    {
+      "epoch": 0.2721402435799424,
+      "grad_norm": 546.5497919918348,
+      "learning_rate": 1.6735032707883502e-05,
+      "loss": 292.7723,
+      "step": 14150
+    },
+    {
+      "epoch": 0.27233256884042295,
+      "grad_norm": 549.7254365469638,
+      "learning_rate": 1.6730520106979855e-05,
+      "loss": 282.8527,
+      "step": 14160
+    },
+    {
+      "epoch": 0.27252489410090347,
+      "grad_norm": 476.7620428067308,
+      "learning_rate": 1.672600499906863e-05,
+      "loss": 292.9566,
+      "step": 14170
+    },
+    {
+      "epoch": 0.272717219361384,
+      "grad_norm": 491.790227535059,
+      "learning_rate": 1.6721487385831622e-05,
+      "loss": 297.068,
+      "step": 14180
+    },
+    {
+      "epoch": 0.2729095446218645,
+      "grad_norm": 503.8566371838667,
+      "learning_rate": 1.6716967268951574e-05,
+      "loss": 294.9846,
+      "step": 14190
+    },
+    {
+      "epoch": 0.273101869882345,
+      "grad_norm": 521.8168280236863,
+      "learning_rate": 1.6712444650112152e-05,
+      "loss": 299.8933,
+      "step": 14200
+    },
+    {
+      "epoch": 0.27329419514282555,
+      "grad_norm": 531.7919913215958,
+      "learning_rate": 1.6707919530997956e-05,
+      "loss": 300.1008,
+      "step": 14210
+    },
+    {
+      "epoch": 0.27348652040330607,
+      "grad_norm": 542.4734961324131,
+      "learning_rate": 1.6703391913294524e-05,
+      "loss": 293.8181,
+      "step": 14220
+    },
+    {
+      "epoch": 0.2736788456637866,
+      "grad_norm": 611.879581939818,
+      "learning_rate": 1.6698861798688312e-05,
+      "loss": 296.7324,
+      "step": 14230
+    },
+    {
+      "epoch": 0.2738711709242671,
+      "grad_norm": 582.7807271420512,
+      "learning_rate": 1.6694329188866717e-05,
+      "loss": 296.1785,
+      "step": 14240
+    },
+    {
+      "epoch": 0.2740634961847476,
+      "grad_norm": 593.1503592417329,
+      "learning_rate": 1.6689794085518057e-05,
+      "loss": 285.3363,
+      "step": 14250
+    },
+    {
+      "epoch": 0.27425582144522814,
+      "grad_norm": 631.3094913957555,
+      "learning_rate": 1.668525649033159e-05,
+      "loss": 301.2793,
+      "step": 14260
+    },
+    {
+      "epoch": 0.2744481467057087,
+      "grad_norm": 531.7179642840727,
+      "learning_rate": 1.6680716404997482e-05,
+      "loss": 291.4309,
+      "step": 14270
+    },
+    {
+      "epoch": 0.27464047196618924,
+      "grad_norm": 586.9634234033227,
+      "learning_rate": 1.667617383120686e-05,
+      "loss": 296.095,
+      "step": 14280
+    },
+    {
+      "epoch": 0.27483279722666976,
+      "grad_norm": 532.4330647997987,
+      "learning_rate": 1.667162877065174e-05,
+      "loss": 293.1977,
+      "step": 14290
+    },
+    {
+      "epoch": 0.2750251224871503,
+      "grad_norm": 467.56773092250654,
+      "learning_rate": 1.6667081225025087e-05,
+      "loss": 289.0699,
+      "step": 14300
+    },
+    {
+      "epoch": 0.2752174477476308,
+      "grad_norm": 561.7862042410331,
+      "learning_rate": 1.666253119602079e-05,
+      "loss": 297.6428,
+      "step": 14310
+    },
+    {
+      "epoch": 0.2754097730081113,
+      "grad_norm": 576.062253058774,
+      "learning_rate": 1.665797868533366e-05,
+      "loss": 286.4333,
+      "step": 14320
+    },
+    {
+      "epoch": 0.27560209826859183,
+      "grad_norm": 548.2936707215093,
+      "learning_rate": 1.6653423694659433e-05,
+      "loss": 302.9739,
+      "step": 14330
+    },
+    {
+      "epoch": 0.27579442352907235,
+      "grad_norm": 519.4515504305616,
+      "learning_rate": 1.6648866225694757e-05,
+      "loss": 293.9797,
+      "step": 14340
+    },
+    {
+      "epoch": 0.27598674878955287,
+      "grad_norm": 558.4126124005084,
+      "learning_rate": 1.6644306280137227e-05,
+      "loss": 290.5317,
+      "step": 14350
+    },
+    {
+      "epoch": 0.2761790740500334,
+      "grad_norm": 515.0605976761065,
+      "learning_rate": 1.6639743859685336e-05,
+      "loss": 297.4429,
+      "step": 14360
+    },
+    {
+      "epoch": 0.27637139931051397,
+      "grad_norm": 523.441858085406,
+      "learning_rate": 1.663517896603852e-05,
+      "loss": 293.4781,
+      "step": 14370
+    },
+    {
+      "epoch": 0.2765637245709945,
+      "grad_norm": 496.2810553965151,
+      "learning_rate": 1.6630611600897126e-05,
+      "loss": 293.7562,
+      "step": 14380
+    },
+    {
+      "epoch": 0.276756049831475,
+      "grad_norm": 533.9125856608609,
+      "learning_rate": 1.6626041765962413e-05,
+      "loss": 300.4407,
+      "step": 14390
+    },
+    {
+      "epoch": 0.2769483750919555,
+      "grad_norm": 489.4734113881238,
+      "learning_rate": 1.662146946293658e-05,
+      "loss": 292.072,
+      "step": 14400
+    },
+    {
+      "epoch": 0.27714070035243604,
+      "grad_norm": 511.10923050644755,
+      "learning_rate": 1.6616894693522727e-05,
+      "loss": 287.8735,
+      "step": 14410
+    },
+    {
+      "epoch": 0.27733302561291656,
+      "grad_norm": 514.2390174675193,
+      "learning_rate": 1.6612317459424884e-05,
+      "loss": 286.7239,
+      "step": 14420
+    },
+    {
+      "epoch": 0.2775253508733971,
+      "grad_norm": 516.3737981947766,
+      "learning_rate": 1.6607737762347987e-05,
+      "loss": 292.7905,
+      "step": 14430
+    },
+    {
+      "epoch": 0.2777176761338776,
+      "grad_norm": 515.7956591308133,
+      "learning_rate": 1.6603155603997908e-05,
+      "loss": 291.3975,
+      "step": 14440
+    },
+    {
+      "epoch": 0.2779100013943581,
+      "grad_norm": 530.2704433262519,
+      "learning_rate": 1.6598570986081424e-05,
+      "loss": 291.6472,
+      "step": 14450
+    },
+    {
+      "epoch": 0.27810232665483864,
+      "grad_norm": 502.7541578391744,
+      "learning_rate": 1.6593983910306225e-05,
+      "loss": 290.5978,
+      "step": 14460
+    },
+    {
+      "epoch": 0.2782946519153192,
+      "grad_norm": 500.9140327890009,
+      "learning_rate": 1.658939437838092e-05,
+      "loss": 285.6688,
+      "step": 14470
+    },
+    {
+      "epoch": 0.27848697717579973,
+      "grad_norm": 506.4368375498071,
+      "learning_rate": 1.658480239201504e-05,
+      "loss": 290.2608,
+      "step": 14480
+    },
+    {
+      "epoch": 0.27867930243628025,
+      "grad_norm": 524.0776124408978,
+      "learning_rate": 1.6580207952919018e-05,
+      "loss": 285.2238,
+      "step": 14490
+    },
+    {
+      "epoch": 0.27887162769676077,
+      "grad_norm": 486.986613601712,
+      "learning_rate": 1.657561106280421e-05,
+      "loss": 289.2197,
+      "step": 14500
+    },
+    {
+      "epoch": 0.2790639529572413,
+      "grad_norm": 527.7534929398079,
+      "learning_rate": 1.6571011723382882e-05,
+      "loss": 288.2382,
+      "step": 14510
+    },
+    {
+      "epoch": 0.2792562782177218,
+      "grad_norm": 523.2433747738955,
+      "learning_rate": 1.6566409936368207e-05,
+      "loss": 294.6493,
+      "step": 14520
+    },
+    {
+      "epoch": 0.27944860347820233,
+      "grad_norm": 506.96334814682155,
+      "learning_rate": 1.6561805703474285e-05,
+      "loss": 284.9292,
+      "step": 14530
+    },
+    {
+      "epoch": 0.27964092873868285,
+      "grad_norm": 517.331582635516,
+      "learning_rate": 1.655719902641611e-05,
+      "loss": 291.482,
+      "step": 14540
+    },
+    {
+      "epoch": 0.27983325399916337,
+      "grad_norm": 526.1254768542298,
+      "learning_rate": 1.6552589906909586e-05,
+      "loss": 285.2927,
+      "step": 14550
+    },
+    {
+      "epoch": 0.2800255792596439,
+      "grad_norm": 505.8240819909218,
+      "learning_rate": 1.654797834667155e-05,
+      "loss": 279.6156,
+      "step": 14560
+    },
+    {
+      "epoch": 0.28021790452012446,
+      "grad_norm": 543.4726265798994,
+      "learning_rate": 1.6543364347419714e-05,
+      "loss": 294.4749,
+      "step": 14570
+    },
+    {
+      "epoch": 0.280410229780605,
+      "grad_norm": 536.3630879008642,
+      "learning_rate": 1.6538747910872733e-05,
+      "loss": 292.2666,
+      "step": 14580
+    },
+    {
+      "epoch": 0.2806025550410855,
+      "grad_norm": 507.5555154178302,
+      "learning_rate": 1.6534129038750145e-05,
+      "loss": 289.061,
+      "step": 14590
+    },
+    {
+      "epoch": 0.280794880301566,
+      "grad_norm": 536.5650200108057,
+      "learning_rate": 1.65295077327724e-05,
+      "loss": 288.3877,
+      "step": 14600
+    },
+    {
+      "epoch": 0.28098720556204654,
+      "grad_norm": 541.7930329558253,
+      "learning_rate": 1.652488399466087e-05,
+      "loss": 281.9151,
+      "step": 14610
+    },
+    {
+      "epoch": 0.28117953082252706,
+      "grad_norm": 537.5973429956484,
+      "learning_rate": 1.6520257826137807e-05,
+      "loss": 294.4806,
+      "step": 14620
+    },
+    {
+      "epoch": 0.2813718560830076,
+      "grad_norm": 582.071556846933,
+      "learning_rate": 1.6515629228926396e-05,
+      "loss": 296.5689,
+      "step": 14630
+    },
+    {
+      "epoch": 0.2815641813434881,
+      "grad_norm": 523.4653139210134,
+      "learning_rate": 1.6510998204750702e-05,
+      "loss": 301.8642,
+      "step": 14640
+    },
+    {
+      "epoch": 0.2817565066039686,
+      "grad_norm": 484.41401024861267,
+      "learning_rate": 1.650636475533571e-05,
+      "loss": 288.9062,
+      "step": 14650
+    },
+    {
+      "epoch": 0.28194883186444913,
+      "grad_norm": 630.3587033045286,
+      "learning_rate": 1.6501728882407305e-05,
+      "loss": 291.7619,
+      "step": 14660
+    },
+    {
+      "epoch": 0.28214115712492965,
+      "grad_norm": 517.8249178111148,
+      "learning_rate": 1.649709058769227e-05,
+      "loss": 283.0053,
+      "step": 14670
+    },
+    {
+      "epoch": 0.28233348238541023,
+      "grad_norm": 558.3279959930346,
+      "learning_rate": 1.6492449872918293e-05,
+      "loss": 282.5668,
+      "step": 14680
+    },
+    {
+      "epoch": 0.28252580764589075,
+      "grad_norm": 496.024160108566,
+      "learning_rate": 1.6487806739813966e-05,
+      "loss": 287.0649,
+      "step": 14690
+    },
+    {
+      "epoch": 0.28271813290637127,
+      "grad_norm": 518.8328203843299,
+      "learning_rate": 1.6483161190108778e-05,
+      "loss": 287.9272,
+      "step": 14700
+    },
+    {
+      "epoch": 0.2829104581668518,
+      "grad_norm": 527.9861405296551,
+      "learning_rate": 1.6478513225533117e-05,
+      "loss": 295.1303,
+      "step": 14710
+    },
+    {
+      "epoch": 0.2831027834273323,
+      "grad_norm": 483.3714306016936,
+      "learning_rate": 1.647386284781828e-05,
+      "loss": 293.6143,
+      "step": 14720
+    },
+    {
+      "epoch": 0.2832951086878128,
+      "grad_norm": 474.9589950497544,
+      "learning_rate": 1.6469210058696448e-05,
+      "loss": 295.0874,
+      "step": 14730
+    },
+    {
+      "epoch": 0.28348743394829334,
+      "grad_norm": 523.3837780029888,
+      "learning_rate": 1.646455485990071e-05,
+      "loss": 285.4751,
+      "step": 14740
+    },
+    {
+      "epoch": 0.28367975920877386,
+      "grad_norm": 492.0379669582084,
+      "learning_rate": 1.645989725316506e-05,
+      "loss": 286.6602,
+      "step": 14750
+    },
+    {
+      "epoch": 0.2838720844692544,
+      "grad_norm": 484.59617082131706,
+      "learning_rate": 1.6455237240224364e-05,
+      "loss": 284.2392,
+      "step": 14760
+    },
+    {
+      "epoch": 0.2840644097297349,
+      "grad_norm": 526.1945236290193,
+      "learning_rate": 1.6450574822814412e-05,
+      "loss": 290.8893,
+      "step": 14770
+    },
+    {
+      "epoch": 0.2842567349902155,
+      "grad_norm": 507.346995927789,
+      "learning_rate": 1.6445910002671872e-05,
+      "loss": 290.3344,
+      "step": 14780
+    },
+    {
+      "epoch": 0.284449060250696,
+      "grad_norm": 485.57558249831214,
+      "learning_rate": 1.644124278153431e-05,
+      "loss": 292.5404,
+      "step": 14790
+    },
+    {
+      "epoch": 0.2846413855111765,
+      "grad_norm": 522.1363442611338,
+      "learning_rate": 1.64365731611402e-05,
+      "loss": 292.2073,
+      "step": 14800
+    },
+    {
+      "epoch": 0.28483371077165703,
+      "grad_norm": 481.477040786785,
+      "learning_rate": 1.6431901143228888e-05,
+      "loss": 283.1542,
+      "step": 14810
+    },
+    {
+      "epoch": 0.28502603603213755,
+      "grad_norm": 546.037380587678,
+      "learning_rate": 1.6427226729540623e-05,
+      "loss": 284.7875,
+      "step": 14820
+    },
+    {
+      "epoch": 0.2852183612926181,
+      "grad_norm": 490.40573605475146,
+      "learning_rate": 1.6422549921816556e-05,
+      "loss": 292.0841,
+      "step": 14830
+    },
+    {
+      "epoch": 0.2854106865530986,
+      "grad_norm": 504.07270542810454,
+      "learning_rate": 1.641787072179871e-05,
+      "loss": 285.0437,
+      "step": 14840
+    },
+    {
+      "epoch": 0.2856030118135791,
+      "grad_norm": 522.3888537112274,
+      "learning_rate": 1.6413189131230022e-05,
+      "loss": 285.5351,
+      "step": 14850
+    },
+    {
+      "epoch": 0.28579533707405963,
+      "grad_norm": 517.7551877217255,
+      "learning_rate": 1.6408505151854292e-05,
+      "loss": 287.3235,
+      "step": 14860
+    },
+    {
+      "epoch": 0.28598766233454015,
+      "grad_norm": 493.6430189679726,
+      "learning_rate": 1.6403818785416236e-05,
+      "loss": 283.4672,
+      "step": 14870
+    },
+    {
+      "epoch": 0.2861799875950207,
+      "grad_norm": 465.872406438963,
+      "learning_rate": 1.6399130033661444e-05,
+      "loss": 298.2659,
+      "step": 14880
+    },
+    {
+      "epoch": 0.28637231285550124,
+      "grad_norm": 521.8523892610701,
+      "learning_rate": 1.6394438898336402e-05,
+      "loss": 288.5079,
+      "step": 14890
+    },
+    {
+      "epoch": 0.28656463811598176,
+      "grad_norm": 577.4684491701477,
+      "learning_rate": 1.6389745381188475e-05,
+      "loss": 295.961,
+      "step": 14900
+    },
+    {
+      "epoch": 0.2867569633764623,
+      "grad_norm": 538.2385550799697,
+      "learning_rate": 1.6385049483965926e-05,
+      "loss": 283.0698,
+      "step": 14910
+    },
+    {
+      "epoch": 0.2869492886369428,
+      "grad_norm": 542.6443433651598,
+      "learning_rate": 1.6380351208417897e-05,
+      "loss": 287.7351,
+      "step": 14920
+    },
+    {
+      "epoch": 0.2871416138974233,
+      "grad_norm": 520.5855883193285,
+      "learning_rate": 1.6375650556294417e-05,
+      "loss": 285.7551,
+      "step": 14930
+    },
+    {
+      "epoch": 0.28733393915790384,
+      "grad_norm": 530.7103527091674,
+      "learning_rate": 1.6370947529346404e-05,
+      "loss": 288.2293,
+      "step": 14940
+    },
+    {
+      "epoch": 0.28752626441838436,
+      "grad_norm": 541.268105811768,
+      "learning_rate": 1.6366242129325652e-05,
+      "loss": 289.2752,
+      "step": 14950
+    },
+    {
+      "epoch": 0.2877185896788649,
+      "grad_norm": 473.16286300124784,
+      "learning_rate": 1.636153435798485e-05,
+      "loss": 292.9462,
+      "step": 14960
+    },
+    {
+      "epoch": 0.2879109149393454,
+      "grad_norm": 503.52311442459086,
+      "learning_rate": 1.6356824217077564e-05,
+      "loss": 278.4941,
+      "step": 14970
+    },
+    {
+      "epoch": 0.28810324019982597,
+      "grad_norm": 516.2886240153437,
+      "learning_rate": 1.6352111708358243e-05,
+      "loss": 287.0422,
+      "step": 14980
+    },
+    {
+      "epoch": 0.2882955654603065,
+      "grad_norm": 540.5420920883236,
+      "learning_rate": 1.6347396833582224e-05,
+      "loss": 291.1956,
+      "step": 14990
+    },
+    {
+      "epoch": 0.288487890720787,
+      "grad_norm": 535.6816182130684,
+      "learning_rate": 1.634267959450571e-05,
+      "loss": 284.8218,
+      "step": 15000
+    },
+    {
+      "epoch": 0.28868021598126753,
+      "grad_norm": 531.0880916730234,
+      "learning_rate": 1.63379599928858e-05,
+      "loss": 297.1833,
+      "step": 15010
+    },
+    {
+      "epoch": 0.28887254124174805,
+      "grad_norm": 452.18802455534757,
+      "learning_rate": 1.6333238030480473e-05,
+      "loss": 290.4448,
+      "step": 15020
+    },
+    {
+      "epoch": 0.28906486650222857,
+      "grad_norm": 551.1378948565866,
+      "learning_rate": 1.6328513709048573e-05,
+      "loss": 281.2265,
+      "step": 15030
+    },
+    {
+      "epoch": 0.2892571917627091,
+      "grad_norm": 531.8185124943761,
+      "learning_rate": 1.6323787030349833e-05,
+      "loss": 287.2374,
+      "step": 15040
+    },
+    {
+      "epoch": 0.2894495170231896,
+      "grad_norm": 544.5212845421312,
+      "learning_rate": 1.6319057996144868e-05,
+      "loss": 290.1431,
+      "step": 15050
+    },
+    {
+      "epoch": 0.2896418422836701,
+      "grad_norm": 488.4442261682339,
+      "learning_rate": 1.631432660819516e-05,
+      "loss": 280.7577,
+      "step": 15060
+    },
+    {
+      "epoch": 0.28983416754415064,
+      "grad_norm": 527.3468863746809,
+      "learning_rate": 1.6309592868263075e-05,
+      "loss": 287.4474,
+      "step": 15070
+    },
+    {
+      "epoch": 0.2900264928046312,
+      "grad_norm": 515.9194998870802,
+      "learning_rate": 1.630485677811185e-05,
+      "loss": 292.0219,
+      "step": 15080
+    },
+    {
+      "epoch": 0.29021881806511174,
+      "grad_norm": 524.2664270558995,
+      "learning_rate": 1.630011833950561e-05,
+      "loss": 288.1353,
+      "step": 15090
+    },
+    {
+      "epoch": 0.29041114332559226,
+      "grad_norm": 511.65504954355316,
+      "learning_rate": 1.6295377554209338e-05,
+      "loss": 290.7549,
+      "step": 15100
+    },
+    {
+      "epoch": 0.2906034685860728,
+      "grad_norm": 560.0968704815492,
+      "learning_rate": 1.6290634423988897e-05,
+      "loss": 291.5172,
+      "step": 15110
+    },
+    {
+      "epoch": 0.2907957938465533,
+      "grad_norm": 562.3453899403073,
+      "learning_rate": 1.6285888950611023e-05,
+      "loss": 293.7454,
+      "step": 15120
+    },
+    {
+      "epoch": 0.2909881191070338,
+      "grad_norm": 581.0782142311772,
+      "learning_rate": 1.6281141135843334e-05,
+      "loss": 284.4832,
+      "step": 15130
+    },
+    {
+      "epoch": 0.29118044436751434,
+      "grad_norm": 507.7186426134632,
+      "learning_rate": 1.6276390981454306e-05,
+      "loss": 288.9794,
+      "step": 15140
+    },
+    {
+      "epoch": 0.29137276962799485,
+      "grad_norm": 540.4939759787093,
+      "learning_rate": 1.6271638489213297e-05,
+      "loss": 295.8042,
+      "step": 15150
+    },
+    {
+      "epoch": 0.2915650948884754,
+      "grad_norm": 569.0475529586984,
+      "learning_rate": 1.6266883660890527e-05,
+      "loss": 294.2423,
+      "step": 15160
+    },
+    {
+      "epoch": 0.2917574201489559,
+      "grad_norm": 557.2936792072236,
+      "learning_rate": 1.6262126498257098e-05,
+      "loss": 287.2764,
+      "step": 15170
+    },
+    {
+      "epoch": 0.2919497454094364,
+      "grad_norm": 508.37719933528643,
+      "learning_rate": 1.625736700308497e-05,
+      "loss": 289.2119,
+      "step": 15180
+    },
+    {
+      "epoch": 0.292142070669917,
+      "grad_norm": 565.1171992592233,
+      "learning_rate": 1.6252605177146978e-05,
+      "loss": 289.3611,
+      "step": 15190
+    },
+    {
+      "epoch": 0.2923343959303975,
+      "grad_norm": 510.1849774791188,
+      "learning_rate": 1.624784102221682e-05,
+      "loss": 297.7609,
+      "step": 15200
+    },
+    {
+      "epoch": 0.292526721190878,
+      "grad_norm": 516.8600679003113,
+      "learning_rate": 1.6243074540069067e-05,
+      "loss": 283.5954,
+      "step": 15210
+    },
+    {
+      "epoch": 0.29271904645135854,
+      "grad_norm": 509.4394451179388,
+      "learning_rate": 1.6238305732479158e-05,
+      "loss": 292.1387,
+      "step": 15220
+    },
+    {
+      "epoch": 0.29291137171183906,
+      "grad_norm": 517.029756206194,
+      "learning_rate": 1.6233534601223396e-05,
+      "loss": 288.12,
+      "step": 15230
+    },
+    {
+      "epoch": 0.2931036969723196,
+      "grad_norm": 477.2522619687717,
+      "learning_rate": 1.6228761148078943e-05,
+      "loss": 290.5635,
+      "step": 15240
+    },
+    {
+      "epoch": 0.2932960222328001,
+      "grad_norm": 507.15852330827,
+      "learning_rate": 1.622398537482383e-05,
+      "loss": 293.5147,
+      "step": 15250
+    },
+    {
+      "epoch": 0.2934883474932806,
+      "grad_norm": 533.1843285064343,
+      "learning_rate": 1.621920728323696e-05,
+      "loss": 293.4276,
+      "step": 15260
+    },
+    {
+      "epoch": 0.29368067275376114,
+      "grad_norm": 514.1786183919115,
+      "learning_rate": 1.621442687509809e-05,
+      "loss": 287.0028,
+      "step": 15270
+    },
+    {
+      "epoch": 0.29387299801424166,
+      "grad_norm": 520.3083702372918,
+      "learning_rate": 1.6209644152187848e-05,
+      "loss": 293.8355,
+      "step": 15280
+    },
+    {
+      "epoch": 0.29406532327472223,
+      "grad_norm": 543.6579665627644,
+      "learning_rate": 1.620485911628771e-05,
+      "loss": 293.7546,
+      "step": 15290
+    },
+    {
+      "epoch": 0.29425764853520275,
+      "grad_norm": 469.0832516060488,
+      "learning_rate": 1.6200071769180026e-05,
+      "loss": 274.7911,
+      "step": 15300
+    },
+    {
+      "epoch": 0.2944499737956833,
+      "grad_norm": 492.4973677012339,
+      "learning_rate": 1.6195282112648007e-05,
+      "loss": 287.7066,
+      "step": 15310
+    },
+    {
+      "epoch": 0.2946422990561638,
+      "grad_norm": 528.688635376134,
+      "learning_rate": 1.6190490148475724e-05,
+      "loss": 283.4887,
+      "step": 15320
+    },
+    {
+      "epoch": 0.2948346243166443,
+      "grad_norm": 490.60267757464464,
+      "learning_rate": 1.6185695878448094e-05,
+      "loss": 290.6048,
+      "step": 15330
+    },
+    {
+      "epoch": 0.29502694957712483,
+      "grad_norm": 555.9835794695705,
+      "learning_rate": 1.6180899304350915e-05,
+      "loss": 277.0562,
+      "step": 15340
+    },
+    {
+      "epoch": 0.29521927483760535,
+      "grad_norm": 524.6071123671601,
+      "learning_rate": 1.6176100427970826e-05,
+      "loss": 283.2577,
+      "step": 15350
+    },
+    {
+      "epoch": 0.29541160009808587,
+      "grad_norm": 524.1591809928741,
+      "learning_rate": 1.6171299251095324e-05,
+      "loss": 285.3737,
+      "step": 15360
+    },
+    {
+      "epoch": 0.2956039253585664,
+      "grad_norm": 574.4022909974907,
+      "learning_rate": 1.6166495775512777e-05,
+      "loss": 296.8317,
+      "step": 15370
+    },
+    {
+      "epoch": 0.2957962506190469,
+      "grad_norm": 531.5310987046762,
+      "learning_rate": 1.6161690003012392e-05,
+      "loss": 291.3665,
+      "step": 15380
+    },
+    {
+      "epoch": 0.2959885758795275,
+      "grad_norm": 499.6099135110918,
+      "learning_rate": 1.615688193538425e-05,
+      "loss": 282.2095,
+      "step": 15390
+    },
+    {
+      "epoch": 0.296180901140008,
+      "grad_norm": 594.1663922426048,
+      "learning_rate": 1.615207157441927e-05,
+      "loss": 290.5614,
+      "step": 15400
+    },
+    {
+      "epoch": 0.2963732264004885,
+      "grad_norm": 512.0859341425991,
+      "learning_rate": 1.6147258921909236e-05,
+      "loss": 283.3074,
+      "step": 15410
+    },
+    {
+      "epoch": 0.29656555166096904,
+      "grad_norm": 545.6213383466101,
+      "learning_rate": 1.6142443979646774e-05,
+      "loss": 299.5071,
+      "step": 15420
+    },
+    {
+      "epoch": 0.29675787692144956,
+      "grad_norm": 487.5165986744006,
+      "learning_rate": 1.6137626749425377e-05,
+      "loss": 282.3447,
+      "step": 15430
+    },
+    {
+      "epoch": 0.2969502021819301,
+      "grad_norm": 542.058428442061,
+      "learning_rate": 1.6132807233039382e-05,
+      "loss": 286.6273,
+      "step": 15440
+    },
+    {
+      "epoch": 0.2971425274424106,
+      "grad_norm": 475.7983625499094,
+      "learning_rate": 1.612798543228398e-05,
+      "loss": 284.6235,
+      "step": 15450
+    },
+    {
+      "epoch": 0.2973348527028911,
+      "grad_norm": 533.1886356499948,
+      "learning_rate": 1.612316134895521e-05,
+      "loss": 289.368,
+      "step": 15460
+    },
+    {
+      "epoch": 0.29752717796337164,
+      "grad_norm": 493.71259909282486,
+      "learning_rate": 1.611833498484997e-05,
+      "loss": 288.4728,
+      "step": 15470
+    },
+    {
+      "epoch": 0.29771950322385216,
+      "grad_norm": 483.3422549506859,
+      "learning_rate": 1.611350634176599e-05,
+      "loss": 289.1593,
+      "step": 15480
+    },
+    {
+      "epoch": 0.29791182848433273,
+      "grad_norm": 550.8641844276592,
+      "learning_rate": 1.6108675421501865e-05,
+      "loss": 291.3389,
+      "step": 15490
+    },
+    {
+      "epoch": 0.29810415374481325,
+      "grad_norm": 486.47445034849903,
+      "learning_rate": 1.610384222585704e-05,
+      "loss": 296.147,
+      "step": 15500
+    },
+    {
+      "epoch": 0.29829647900529377,
+      "grad_norm": 471.2088660255904,
+      "learning_rate": 1.609900675663179e-05,
+      "loss": 281.4358,
+      "step": 15510
+    },
+    {
+      "epoch": 0.2984888042657743,
+      "grad_norm": 518.0627511329293,
+      "learning_rate": 1.609416901562725e-05,
+      "loss": 277.0075,
+      "step": 15520
+    },
+    {
+      "epoch": 0.2986811295262548,
+      "grad_norm": 487.7412202062457,
+      "learning_rate": 1.60893290046454e-05,
+      "loss": 285.6411,
+      "step": 15530
+    },
+    {
+      "epoch": 0.2988734547867353,
+      "grad_norm": 548.9052584462366,
+      "learning_rate": 1.608448672548907e-05,
+      "loss": 289.8053,
+      "step": 15540
+    },
+    {
+      "epoch": 0.29906578004721585,
+      "grad_norm": 498.9607684600205,
+      "learning_rate": 1.6079642179961917e-05,
+      "loss": 276.2588,
+      "step": 15550
+    },
+    {
+      "epoch": 0.29925810530769636,
+      "grad_norm": 501.02411342709263,
+      "learning_rate": 1.6074795369868463e-05,
+      "loss": 290.0694,
+      "step": 15560
+    },
+    {
+      "epoch": 0.2994504305681769,
+      "grad_norm": 485.3502661698066,
+      "learning_rate": 1.6069946297014064e-05,
+      "loss": 284.116,
+      "step": 15570
+    },
+    {
+      "epoch": 0.2996427558286574,
+      "grad_norm": 523.5369260488685,
+      "learning_rate": 1.6065094963204915e-05,
+      "loss": 282.9058,
+      "step": 15580
+    },
+    {
+      "epoch": 0.2998350810891379,
+      "grad_norm": 489.7669709899146,
+      "learning_rate": 1.6060241370248064e-05,
+      "loss": 287.4737,
+      "step": 15590
+    },
+    {
+      "epoch": 0.3000274063496185,
+      "grad_norm": 521.4870696588582,
+      "learning_rate": 1.605538551995139e-05,
+      "loss": 290.3221,
+      "step": 15600
+    },
+    {
+      "epoch": 0.300219731610099,
+      "grad_norm": 491.1485641743671,
+      "learning_rate": 1.6050527414123616e-05,
+      "loss": 284.1289,
+      "step": 15610
+    },
+    {
+      "epoch": 0.30041205687057954,
+      "grad_norm": 529.0215173613176,
+      "learning_rate": 1.604566705457431e-05,
+      "loss": 275.0153,
+      "step": 15620
+    },
+    {
+      "epoch": 0.30060438213106006,
+      "grad_norm": 521.1648195009329,
+      "learning_rate": 1.604080444311387e-05,
+      "loss": 280.3334,
+      "step": 15630
+    },
+    {
+      "epoch": 0.3007967073915406,
+      "grad_norm": 625.2767107928526,
+      "learning_rate": 1.6035939581553543e-05,
+      "loss": 289.4082,
+      "step": 15640
+    },
+    {
+      "epoch": 0.3009890326520211,
+      "grad_norm": 632.0418719421094,
+      "learning_rate": 1.603107247170541e-05,
+      "loss": 293.2012,
+      "step": 15650
+    },
+    {
+      "epoch": 0.3011813579125016,
+      "grad_norm": 587.1309367442407,
+      "learning_rate": 1.6026203115382392e-05,
+      "loss": 294.6388,
+      "step": 15660
+    },
+    {
+      "epoch": 0.30137368317298213,
+      "grad_norm": 541.275040449254,
+      "learning_rate": 1.6021331514398233e-05,
+      "loss": 278.5768,
+      "step": 15670
+    },
+    {
+      "epoch": 0.30156600843346265,
+      "grad_norm": 504.9462380983586,
+      "learning_rate": 1.6016457670567535e-05,
+      "loss": 294.8726,
+      "step": 15680
+    },
+    {
+      "epoch": 0.30175833369394317,
+      "grad_norm": 502.5374819203824,
+      "learning_rate": 1.6011581585705715e-05,
+      "loss": 286.1138,
+      "step": 15690
+    },
+    {
+      "epoch": 0.30195065895442375,
+      "grad_norm": 534.9752537584386,
+      "learning_rate": 1.600670326162904e-05,
+      "loss": 291.6314,
+      "step": 15700
+    },
+    {
+      "epoch": 0.30214298421490426,
+      "grad_norm": 474.3314493571712,
+      "learning_rate": 1.600182270015461e-05,
+      "loss": 280.5248,
+      "step": 15710
+    },
+    {
+      "epoch": 0.3023353094753848,
+      "grad_norm": 503.11554185218836,
+      "learning_rate": 1.5996939903100338e-05,
+      "loss": 278.0978,
+      "step": 15720
+    },
+    {
+      "epoch": 0.3025276347358653,
+      "grad_norm": 504.01519958601074,
+      "learning_rate": 1.5992054872285005e-05,
+      "loss": 281.9744,
+      "step": 15730
+    },
+    {
+      "epoch": 0.3027199599963458,
+      "grad_norm": 604.2841576339626,
+      "learning_rate": 1.5987167609528187e-05,
+      "loss": 288.0925,
+      "step": 15740
+    },
+    {
+      "epoch": 0.30291228525682634,
+      "grad_norm": 588.0113227879084,
+      "learning_rate": 1.598227811665032e-05,
+      "loss": 294.2167,
+      "step": 15750
+    },
+    {
+      "epoch": 0.30310461051730686,
+      "grad_norm": 579.8760044105293,
+      "learning_rate": 1.597738639547265e-05,
+      "loss": 277.3983,
+      "step": 15760
+    },
+    {
+      "epoch": 0.3032969357777874,
+      "grad_norm": 514.2280040722753,
+      "learning_rate": 1.597249244781727e-05,
+      "loss": 287.1545,
+      "step": 15770
+    },
+    {
+      "epoch": 0.3034892610382679,
+      "grad_norm": 488.35231534242575,
+      "learning_rate": 1.5967596275507094e-05,
+      "loss": 289.3087,
+      "step": 15780
+    },
+    {
+      "epoch": 0.3036815862987484,
+      "grad_norm": 491.15281409562715,
+      "learning_rate": 1.5962697880365863e-05,
+      "loss": 278.6546,
+      "step": 15790
+    },
+    {
+      "epoch": 0.303873911559229,
+      "grad_norm": 500.22375470777,
+      "learning_rate": 1.5957797264218145e-05,
+      "loss": 276.2161,
+      "step": 15800
+    },
+    {
+      "epoch": 0.3040662368197095,
+      "grad_norm": 549.1234763745196,
+      "learning_rate": 1.5952894428889347e-05,
+      "loss": 300.0928,
+      "step": 15810
+    },
+    {
+      "epoch": 0.30425856208019003,
+      "grad_norm": 473.8553015419329,
+      "learning_rate": 1.594798937620569e-05,
+      "loss": 278.4122,
+      "step": 15820
+    },
+    {
+      "epoch": 0.30445088734067055,
+      "grad_norm": 512.2970050190106,
+      "learning_rate": 1.594308210799422e-05,
+      "loss": 284.3674,
+      "step": 15830
+    },
+    {
+      "epoch": 0.30464321260115107,
+      "grad_norm": 618.2516934853405,
+      "learning_rate": 1.5938172626082823e-05,
+      "loss": 289.8339,
+      "step": 15840
+    },
+    {
+      "epoch": 0.3048355378616316,
+      "grad_norm": 552.9841071734476,
+      "learning_rate": 1.5933260932300192e-05,
+      "loss": 284.5295,
+      "step": 15850
+    },
+    {
+      "epoch": 0.3050278631221121,
+      "grad_norm": 575.6943627252812,
+      "learning_rate": 1.5928347028475855e-05,
+      "loss": 309.3565,
+      "step": 15860
+    },
+    {
+      "epoch": 0.3052201883825926,
+      "grad_norm": 540.8615864018034,
+      "learning_rate": 1.592343091644016e-05,
+      "loss": 282.5193,
+      "step": 15870
+    },
+    {
+      "epoch": 0.30541251364307315,
+      "grad_norm": 521.4431990604672,
+      "learning_rate": 1.5918512598024275e-05,
+      "loss": 272.7361,
+      "step": 15880
+    },
+    {
+      "epoch": 0.30560483890355367,
+      "grad_norm": 560.764237488359,
+      "learning_rate": 1.59135920750602e-05,
+      "loss": 294.7301,
+      "step": 15890
+    },
+    {
+      "epoch": 0.30579716416403424,
+      "grad_norm": 510.4617815619798,
+      "learning_rate": 1.590866934938074e-05,
+      "loss": 287.0498,
+      "step": 15900
+    },
+    {
+      "epoch": 0.30598948942451476,
+      "grad_norm": 564.1033224230508,
+      "learning_rate": 1.590374442281953e-05,
+      "loss": 278.6576,
+      "step": 15910
+    },
+    {
+      "epoch": 0.3061818146849953,
+      "grad_norm": 550.1546968774793,
+      "learning_rate": 1.5898817297211028e-05,
+      "loss": 280.9709,
+      "step": 15920
+    },
+    {
+      "epoch": 0.3063741399454758,
+      "grad_norm": 521.1080363962003,
+      "learning_rate": 1.589388797439051e-05,
+      "loss": 289.1891,
+      "step": 15930
+    },
+    {
+      "epoch": 0.3065664652059563,
+      "grad_norm": 504.7766153289126,
+      "learning_rate": 1.5888956456194056e-05,
+      "loss": 284.62,
+      "step": 15940
+    },
+    {
+      "epoch": 0.30675879046643684,
+      "grad_norm": 494.8050240522159,
+      "learning_rate": 1.588402274445858e-05,
+      "loss": 295.6602,
+      "step": 15950
+    },
+    {
+      "epoch": 0.30695111572691736,
+      "grad_norm": 496.2823060475834,
+      "learning_rate": 1.5879086841021815e-05,
+      "loss": 288.8745,
+      "step": 15960
+    },
+    {
+      "epoch": 0.3071434409873979,
+      "grad_norm": 519.9110449577126,
+      "learning_rate": 1.5874148747722294e-05,
+      "loss": 287.3097,
+      "step": 15970
+    },
+    {
+      "epoch": 0.3073357662478784,
+      "grad_norm": 487.69124540647744,
+      "learning_rate": 1.5869208466399382e-05,
+      "loss": 284.0637,
+      "step": 15980
+    },
+    {
+      "epoch": 0.3075280915083589,
+      "grad_norm": 539.631381040546,
+      "learning_rate": 1.586426599889325e-05,
+      "loss": 288.6262,
+      "step": 15990
+    },
+    {
+      "epoch": 0.3077204167688395,
+      "grad_norm": 594.2390774409421,
+      "learning_rate": 1.5859321347044882e-05,
+      "loss": 293.5237,
+      "step": 16000
+    },
+    {
+      "epoch": 0.30791274202932,
+      "grad_norm": 477.3472878789492,
+      "learning_rate": 1.5854374512696084e-05,
+      "loss": 289.5295,
+      "step": 16010
+    },
+    {
+      "epoch": 0.3081050672898005,
+      "grad_norm": 582.688043220394,
+      "learning_rate": 1.584942549768947e-05,
+      "loss": 284.3594,
+      "step": 16020
+    },
+    {
+      "epoch": 0.30829739255028105,
+      "grad_norm": 526.1582192938724,
+      "learning_rate": 1.584447430386846e-05,
+      "loss": 295.4727,
+      "step": 16030
+    },
+    {
+      "epoch": 0.30848971781076157,
+      "grad_norm": 480.97555346584437,
+      "learning_rate": 1.58395209330773e-05,
+      "loss": 273.5935,
+      "step": 16040
+    },
+    {
+      "epoch": 0.3086820430712421,
+      "grad_norm": 580.7129005239941,
+      "learning_rate": 1.5834565387161034e-05,
+      "loss": 290.0589,
+      "step": 16050
+    },
+    {
+      "epoch": 0.3088743683317226,
+      "grad_norm": 595.7538220710202,
+      "learning_rate": 1.5829607667965524e-05,
+      "loss": 282.5735,
+      "step": 16060
+    },
+    {
+      "epoch": 0.3090666935922031,
+      "grad_norm": 741.90393978779,
+      "learning_rate": 1.5824647777337433e-05,
+      "loss": 304.68,
+      "step": 16070
+    },
+    {
+      "epoch": 0.30925901885268364,
+      "grad_norm": 503.28682760979336,
+      "learning_rate": 1.5819685717124245e-05,
+      "loss": 278.2125,
+      "step": 16080
+    },
+    {
+      "epoch": 0.30945134411316416,
+      "grad_norm": 529.0526285631472,
+      "learning_rate": 1.5814721489174246e-05,
+      "loss": 281.5895,
+      "step": 16090
+    },
+    {
+      "epoch": 0.3096436693736447,
+      "grad_norm": 523.754863430061,
+      "learning_rate": 1.580975509533652e-05,
+      "loss": 287.9395,
+      "step": 16100
+    },
+    {
+      "epoch": 0.30983599463412526,
+      "grad_norm": 513.340387068993,
+      "learning_rate": 1.5804786537460972e-05,
+      "loss": 284.7428,
+      "step": 16110
+    },
+    {
+      "epoch": 0.3100283198946058,
+      "grad_norm": 498.4498258034029,
+      "learning_rate": 1.5799815817398312e-05,
+      "loss": 282.2421,
+      "step": 16120
+    },
+    {
+      "epoch": 0.3102206451550863,
+      "grad_norm": 644.6228277837,
+      "learning_rate": 1.579484293700004e-05,
+      "loss": 287.2761,
+      "step": 16130
+    },
+    {
+      "epoch": 0.3104129704155668,
+      "grad_norm": 553.0587483533304,
+      "learning_rate": 1.578986789811849e-05,
+      "loss": 293.0727,
+      "step": 16140
+    },
+    {
+      "epoch": 0.31060529567604733,
+      "grad_norm": 467.04001081599125,
+      "learning_rate": 1.5784890702606763e-05,
+      "loss": 279.3575,
+      "step": 16150
+    },
+    {
+      "epoch": 0.31079762093652785,
+      "grad_norm": 561.0758916704282,
+      "learning_rate": 1.5779911352318792e-05,
+      "loss": 283.7236,
+      "step": 16160
+    },
+    {
+      "epoch": 0.31098994619700837,
+      "grad_norm": 489.45034803369253,
+      "learning_rate": 1.5774929849109303e-05,
+      "loss": 282.6329,
+      "step": 16170
+    },
+    {
+      "epoch": 0.3111822714574889,
+      "grad_norm": 491.5180048936539,
+      "learning_rate": 1.5769946194833816e-05,
+      "loss": 296.9812,
+      "step": 16180
+    },
+    {
+      "epoch": 0.3113745967179694,
+      "grad_norm": 705.4764466767829,
+      "learning_rate": 1.5764960391348666e-05,
+      "loss": 292.4866,
+      "step": 16190
+    },
+    {
+      "epoch": 0.31156692197844993,
+      "grad_norm": 454.96297015142267,
+      "learning_rate": 1.5759972440510985e-05,
+      "loss": 272.6013,
+      "step": 16200
+    },
+    {
+      "epoch": 0.3117592472389305,
+      "grad_norm": 476.0777823907391,
+      "learning_rate": 1.5754982344178697e-05,
+      "loss": 275.2302,
+      "step": 16210
+    },
+    {
+      "epoch": 0.311951572499411,
+      "grad_norm": 460.34322903100036,
+      "learning_rate": 1.5749990104210534e-05,
+      "loss": 286.4499,
+      "step": 16220
+    },
+    {
+      "epoch": 0.31214389775989154,
+      "grad_norm": 482.8467760461587,
+      "learning_rate": 1.574499572246602e-05,
+      "loss": 280.7481,
+      "step": 16230
+    },
+    {
+      "epoch": 0.31233622302037206,
+      "grad_norm": 564.6705875897441,
+      "learning_rate": 1.5739999200805483e-05,
+      "loss": 297.4406,
+      "step": 16240
+    },
+    {
+      "epoch": 0.3125285482808526,
+      "grad_norm": 479.0560359918922,
+      "learning_rate": 1.573500054109004e-05,
+      "loss": 285.5159,
+      "step": 16250
+    },
+    {
+      "epoch": 0.3127208735413331,
+      "grad_norm": 505.97033396095134,
+      "learning_rate": 1.5729999745181617e-05,
+      "loss": 282.9737,
+      "step": 16260
+    },
+    {
+      "epoch": 0.3129131988018136,
+      "grad_norm": 504.80656370046,
+      "learning_rate": 1.572499681494292e-05,
+      "loss": 281.4261,
+      "step": 16270
+    },
+    {
+      "epoch": 0.31310552406229414,
+      "grad_norm": 517.9894277752834,
+      "learning_rate": 1.571999175223746e-05,
+      "loss": 278.8282,
+      "step": 16280
+    },
+    {
+      "epoch": 0.31329784932277466,
+      "grad_norm": 521.6867817039416,
+      "learning_rate": 1.571498455892954e-05,
+      "loss": 289.0921,
+      "step": 16290
+    },
+    {
+      "epoch": 0.3134901745832552,
+      "grad_norm": 536.1863005509615,
+      "learning_rate": 1.570997523688426e-05,
+      "loss": 289.1859,
+      "step": 16300
+    },
+    {
+      "epoch": 0.31368249984373575,
+      "grad_norm": 547.7252933988581,
+      "learning_rate": 1.570496378796751e-05,
+      "loss": 280.1739,
+      "step": 16310
+    },
+    {
+      "epoch": 0.31387482510421627,
+      "grad_norm": 520.4258705378184,
+      "learning_rate": 1.5699950214045966e-05,
+      "loss": 290.1589,
+      "step": 16320
+    },
+    {
+      "epoch": 0.3140671503646968,
+      "grad_norm": 510.2997733235102,
+      "learning_rate": 1.5694934516987102e-05,
+      "loss": 288.2046,
+      "step": 16330
+    },
+    {
+      "epoch": 0.3142594756251773,
+      "grad_norm": 522.3530989708389,
+      "learning_rate": 1.5689916698659193e-05,
+      "loss": 290.9881,
+      "step": 16340
+    },
+    {
+      "epoch": 0.31445180088565783,
+      "grad_norm": 540.4821080464747,
+      "learning_rate": 1.568489676093128e-05,
+      "loss": 273.7589,
+      "step": 16350
+    },
+    {
+      "epoch": 0.31464412614613835,
+      "grad_norm": 549.9771541360382,
+      "learning_rate": 1.5679874705673215e-05,
+      "loss": 288.1429,
+      "step": 16360
+    },
+    {
+      "epoch": 0.31483645140661887,
+      "grad_norm": 504.49624922154203,
+      "learning_rate": 1.5674850534755628e-05,
+      "loss": 287.5587,
+      "step": 16370
+    },
+    {
+      "epoch": 0.3150287766670994,
+      "grad_norm": 540.5603433636334,
+      "learning_rate": 1.566982425004994e-05,
+      "loss": 283.5713,
+      "step": 16380
+    },
+    {
+      "epoch": 0.3152211019275799,
+      "grad_norm": 564.9688463492913,
+      "learning_rate": 1.5664795853428357e-05,
+      "loss": 283.333,
+      "step": 16390
+    },
+    {
+      "epoch": 0.3154134271880604,
+      "grad_norm": 487.59097667101787,
+      "learning_rate": 1.565976534676388e-05,
+      "loss": 279.123,
+      "step": 16400
+    },
+    {
+      "epoch": 0.315605752448541,
+      "grad_norm": 514.6270182263046,
+      "learning_rate": 1.5654732731930286e-05,
+      "loss": 284.5096,
+      "step": 16410
+    },
+    {
+      "epoch": 0.3157980777090215,
+      "grad_norm": 538.9822198776212,
+      "learning_rate": 1.5649698010802138e-05,
+      "loss": 300.1915,
+      "step": 16420
+    },
+    {
+      "epoch": 0.31599040296950204,
+      "grad_norm": 481.9075128550809,
+      "learning_rate": 1.564466118525479e-05,
+      "loss": 279.3204,
+      "step": 16430
+    },
+    {
+      "epoch": 0.31618272822998256,
+      "grad_norm": 484.67382443706265,
+      "learning_rate": 1.5639622257164372e-05,
+      "loss": 283.7897,
+      "step": 16440
+    },
+    {
+      "epoch": 0.3163750534904631,
+      "grad_norm": 493.3385433360078,
+      "learning_rate": 1.5634581228407807e-05,
+      "loss": 274.0433,
+      "step": 16450
+    },
+    {
+      "epoch": 0.3165673787509436,
+      "grad_norm": 598.5308599205382,
+      "learning_rate": 1.562953810086279e-05,
+      "loss": 283.401,
+      "step": 16460
+    },
+    {
+      "epoch": 0.3167597040114241,
+      "grad_norm": 511.7552982060595,
+      "learning_rate": 1.562449287640781e-05,
+      "loss": 278.011,
+      "step": 16470
+    },
+    {
+      "epoch": 0.31695202927190463,
+      "grad_norm": 512.0514524825103,
+      "learning_rate": 1.5619445556922118e-05,
+      "loss": 289.432,
+      "step": 16480
+    },
+    {
+      "epoch": 0.31714435453238515,
+      "grad_norm": 518.3319904929779,
+      "learning_rate": 1.561439614428577e-05,
+      "loss": 287.8469,
+      "step": 16490
+    },
+    {
+      "epoch": 0.3173366797928657,
+      "grad_norm": 543.2992007596245,
+      "learning_rate": 1.5609344640379585e-05,
+      "loss": 284.8363,
+      "step": 16500
+    },
+    {
+      "epoch": 0.3175290050533462,
+      "grad_norm": 538.4865382684914,
+      "learning_rate": 1.560429104708516e-05,
+      "loss": 277.2691,
+      "step": 16510
+    },
+    {
+      "epoch": 0.31772133031382677,
+      "grad_norm": 483.3174815768532,
+      "learning_rate": 1.5599235366284874e-05,
+      "loss": 278.3753,
+      "step": 16520
+    },
+    {
+      "epoch": 0.3179136555743073,
+      "grad_norm": 493.07700587670973,
+      "learning_rate": 1.5594177599861894e-05,
+      "loss": 286.4425,
+      "step": 16530
+    },
+    {
+      "epoch": 0.3181059808347878,
+      "grad_norm": 545.0671539989104,
+      "learning_rate": 1.5589117749700147e-05,
+      "loss": 283.1981,
+      "step": 16540
+    },
+    {
+      "epoch": 0.3182983060952683,
+      "grad_norm": 485.1514583115408,
+      "learning_rate": 1.5584055817684346e-05,
+      "loss": 275.4777,
+      "step": 16550
+    },
+    {
+      "epoch": 0.31849063135574884,
+      "grad_norm": 585.9405464455358,
+      "learning_rate": 1.5578991805699975e-05,
+      "loss": 284.9682,
+      "step": 16560
+    },
+    {
+      "epoch": 0.31868295661622936,
+      "grad_norm": 510.4638462378715,
+      "learning_rate": 1.5573925715633297e-05,
+      "loss": 281.915,
+      "step": 16570
+    },
+    {
+      "epoch": 0.3188752818767099,
+      "grad_norm": 521.0007405153933,
+      "learning_rate": 1.5568857549371348e-05,
+      "loss": 286.4713,
+      "step": 16580
+    },
+    {
+      "epoch": 0.3190676071371904,
+      "grad_norm": 599.2231679338023,
+      "learning_rate": 1.5563787308801934e-05,
+      "loss": 283.5192,
+      "step": 16590
+    },
+    {
+      "epoch": 0.3192599323976709,
+      "grad_norm": 528.6500518662007,
+      "learning_rate": 1.5558714995813636e-05,
+      "loss": 279.1317,
+      "step": 16600
+    },
+    {
+      "epoch": 0.31945225765815144,
+      "grad_norm": 490.38248331872586,
+      "learning_rate": 1.5553640612295807e-05,
+      "loss": 279.3707,
+      "step": 16610
+    },
+    {
+      "epoch": 0.319644582918632,
+      "grad_norm": 481.72437032816214,
+      "learning_rate": 1.5548564160138572e-05,
+      "loss": 286.9246,
+      "step": 16620
+    },
+    {
+      "epoch": 0.31983690817911253,
+      "grad_norm": 499.9533911609183,
+      "learning_rate": 1.5543485641232825e-05,
+      "loss": 276.9663,
+      "step": 16630
+    },
+    {
+      "epoch": 0.32002923343959305,
+      "grad_norm": 499.9996559319061,
+      "learning_rate": 1.553840505747023e-05,
+      "loss": 275.5364,
+      "step": 16640
+    },
+    {
+      "epoch": 0.32022155870007357,
+      "grad_norm": 457.3820162231778,
+      "learning_rate": 1.5533322410743223e-05,
+      "loss": 281.6447,
+      "step": 16650
+    },
+    {
+      "epoch": 0.3204138839605541,
+      "grad_norm": 506.4994518370763,
+      "learning_rate": 1.5528237702945e-05,
+      "loss": 279.5296,
+      "step": 16660
+    },
+    {
+      "epoch": 0.3206062092210346,
+      "grad_norm": 471.5715076032157,
+      "learning_rate": 1.5523150935969534e-05,
+      "loss": 276.193,
+      "step": 16670
+    },
+    {
+      "epoch": 0.32079853448151513,
+      "grad_norm": 549.1394218718974,
+      "learning_rate": 1.5518062111711566e-05,
+      "loss": 277.7524,
+      "step": 16680
+    },
+    {
+      "epoch": 0.32099085974199565,
+      "grad_norm": 626.3914644799332,
+      "learning_rate": 1.5512971232066593e-05,
+      "loss": 277.2895,
+      "step": 16690
+    },
+    {
+      "epoch": 0.32118318500247617,
+      "grad_norm": 535.4837034208131,
+      "learning_rate": 1.5507878298930888e-05,
+      "loss": 284.2514,
+      "step": 16700
+    },
+    {
+      "epoch": 0.3213755102629567,
+      "grad_norm": 484.6434168970839,
+      "learning_rate": 1.5502783314201478e-05,
+      "loss": 282.1756,
+      "step": 16710
+    },
+    {
+      "epoch": 0.32156783552343726,
+      "grad_norm": 494.733905875667,
+      "learning_rate": 1.549768627977617e-05,
+      "loss": 282.7522,
+      "step": 16720
+    },
+    {
+      "epoch": 0.3217601607839178,
+      "grad_norm": 546.9176674970975,
+      "learning_rate": 1.5492587197553517e-05,
+      "loss": 288.5908,
+      "step": 16730
+    },
+    {
+      "epoch": 0.3219524860443983,
+      "grad_norm": 567.2076576521231,
+      "learning_rate": 1.5487486069432848e-05,
+      "loss": 279.9842,
+      "step": 16740
+    },
+    {
+      "epoch": 0.3221448113048788,
+      "grad_norm": 486.6796310621719,
+      "learning_rate": 1.5482382897314243e-05,
+      "loss": 277.7258,
+      "step": 16750
+    },
+    {
+      "epoch": 0.32233713656535934,
+      "grad_norm": 561.8742935723349,
+      "learning_rate": 1.5477277683098555e-05,
+      "loss": 290.0802,
+      "step": 16760
+    },
+    {
+      "epoch": 0.32252946182583986,
+      "grad_norm": 572.7351707165404,
+      "learning_rate": 1.547217042868739e-05,
+      "loss": 275.8038,
+      "step": 16770
+    },
+    {
+      "epoch": 0.3227217870863204,
+      "grad_norm": 488.6701126797395,
+      "learning_rate": 1.546706113598312e-05,
+      "loss": 274.9992,
+      "step": 16780
+    },
+    {
+      "epoch": 0.3229141123468009,
+      "grad_norm": 503.4435960152653,
+      "learning_rate": 1.5461949806888867e-05,
+      "loss": 274.8207,
+      "step": 16790
+    },
+    {
+      "epoch": 0.3231064376072814,
+      "grad_norm": 478.409455223133,
+      "learning_rate": 1.5456836443308512e-05,
+      "loss": 280.9919,
+      "step": 16800
+    },
+    {
+      "epoch": 0.32329876286776194,
+      "grad_norm": 610.258382612928,
+      "learning_rate": 1.545172104714671e-05,
+      "loss": 281.0779,
+      "step": 16810
+    },
+    {
+      "epoch": 0.3234910881282425,
+      "grad_norm": 483.17836284432815,
+      "learning_rate": 1.544660362030886e-05,
+      "loss": 279.3705,
+      "step": 16820
+    },
+    {
+      "epoch": 0.32368341338872303,
+      "grad_norm": 536.135670339746,
+      "learning_rate": 1.544148416470111e-05,
+      "loss": 284.1549,
+      "step": 16830
+    },
+    {
+      "epoch": 0.32387573864920355,
+      "grad_norm": 499.98491209411674,
+      "learning_rate": 1.5436362682230378e-05,
+      "loss": 287.784,
+      "step": 16840
+    },
+    {
+      "epoch": 0.32406806390968407,
+      "grad_norm": 532.9883687853627,
+      "learning_rate": 1.543123917480433e-05,
+      "loss": 282.5017,
+      "step": 16850
+    },
+    {
+      "epoch": 0.3242603891701646,
+      "grad_norm": 523.7473473726294,
+      "learning_rate": 1.542611364433139e-05,
+      "loss": 286.9162,
+      "step": 16860
+    },
+    {
+      "epoch": 0.3244527144306451,
+      "grad_norm": 505.14528914900814,
+      "learning_rate": 1.5420986092720735e-05,
+      "loss": 274.7977,
+      "step": 16870
+    },
+    {
+      "epoch": 0.3246450396911256,
+      "grad_norm": 539.9288293906241,
+      "learning_rate": 1.541585652188229e-05,
+      "loss": 276.965,
+      "step": 16880
+    },
+    {
+      "epoch": 0.32483736495160614,
+      "grad_norm": 493.824063240275,
+      "learning_rate": 1.5410724933726732e-05,
+      "loss": 285.4889,
+      "step": 16890
+    },
+    {
+      "epoch": 0.32502969021208666,
+      "grad_norm": 484.25876968391384,
+      "learning_rate": 1.5405591330165503e-05,
+      "loss": 271.0561,
+      "step": 16900
+    },
+    {
+      "epoch": 0.3252220154725672,
+      "grad_norm": 519.8669603964919,
+      "learning_rate": 1.5400455713110777e-05,
+      "loss": 283.1997,
+      "step": 16910
+    },
+    {
+      "epoch": 0.32541434073304776,
+      "grad_norm": 542.3394815067633,
+      "learning_rate": 1.539531808447549e-05,
+      "loss": 287.2446,
+      "step": 16920
+    },
+    {
+      "epoch": 0.3256066659935283,
+      "grad_norm": 508.3324806387887,
+      "learning_rate": 1.5390178446173325e-05,
+      "loss": 283.9394,
+      "step": 16930
+    },
+    {
+      "epoch": 0.3257989912540088,
+      "grad_norm": 482.09653086508064,
+      "learning_rate": 1.538503680011871e-05,
+      "loss": 279.455,
+      "step": 16940
+    },
+    {
+      "epoch": 0.3259913165144893,
+      "grad_norm": 503.93733588294157,
+      "learning_rate": 1.537989314822682e-05,
+      "loss": 278.5873,
+      "step": 16950
+    },
+    {
+      "epoch": 0.32618364177496983,
+      "grad_norm": 474.553669869395,
+      "learning_rate": 1.5374747492413587e-05,
+      "loss": 283.2114,
+      "step": 16960
+    },
+    {
+      "epoch": 0.32637596703545035,
+      "grad_norm": 552.9804402398042,
+      "learning_rate": 1.536959983459568e-05,
+      "loss": 283.8131,
+      "step": 16970
+    },
+    {
+      "epoch": 0.3265682922959309,
+      "grad_norm": 523.9944541708164,
+      "learning_rate": 1.536445017669052e-05,
+      "loss": 277.2351,
+      "step": 16980
+    },
+    {
+      "epoch": 0.3267606175564114,
+      "grad_norm": 535.0358808170469,
+      "learning_rate": 1.535929852061626e-05,
+      "loss": 273.6,
+      "step": 16990
+    },
+    {
+      "epoch": 0.3269529428168919,
+      "grad_norm": 492.0514777933171,
+      "learning_rate": 1.5354144868291817e-05,
+      "loss": 289.0141,
+      "step": 17000
+    },
+    {
+      "epoch": 0.32714526807737243,
+      "grad_norm": 525.554946727861,
+      "learning_rate": 1.5348989221636835e-05,
+      "loss": 276.8913,
+      "step": 17010
+    },
+    {
+      "epoch": 0.32733759333785295,
+      "grad_norm": 561.7663668886851,
+      "learning_rate": 1.5343831582571706e-05,
+      "loss": 295.5605,
+      "step": 17020
+    },
+    {
+      "epoch": 0.3275299185983335,
+      "grad_norm": 517.5711871233265,
+      "learning_rate": 1.5338671953017576e-05,
+      "loss": 275.8393,
+      "step": 17030
+    },
+    {
+      "epoch": 0.32772224385881404,
+      "grad_norm": 498.35236385335435,
+      "learning_rate": 1.5333510334896308e-05,
+      "loss": 277.6092,
+      "step": 17040
+    },
+    {
+      "epoch": 0.32791456911929456,
+      "grad_norm": 562.1171764611048,
+      "learning_rate": 1.532834673013053e-05,
+      "loss": 285.218,
+      "step": 17050
+    },
+    {
+      "epoch": 0.3281068943797751,
+      "grad_norm": 492.69519703076077,
+      "learning_rate": 1.5323181140643598e-05,
+      "loss": 276.2242,
+      "step": 17060
+    },
+    {
+      "epoch": 0.3282992196402556,
+      "grad_norm": 493.8678578390804,
+      "learning_rate": 1.5318013568359603e-05,
+      "loss": 279.1683,
+      "step": 17070
+    },
+    {
+      "epoch": 0.3284915449007361,
+      "grad_norm": 482.7530348221218,
+      "learning_rate": 1.531284401520338e-05,
+      "loss": 274.2736,
+      "step": 17080
+    },
+    {
+      "epoch": 0.32868387016121664,
+      "grad_norm": 528.8986365666045,
+      "learning_rate": 1.530767248310051e-05,
+      "loss": 278.8286,
+      "step": 17090
+    },
+    {
+      "epoch": 0.32887619542169716,
+      "grad_norm": 479.2144729451598,
+      "learning_rate": 1.53024989739773e-05,
+      "loss": 278.8063,
+      "step": 17100
+    },
+    {
+      "epoch": 0.3290685206821777,
+      "grad_norm": 481.86655683857055,
+      "learning_rate": 1.5297323489760792e-05,
+      "loss": 277.2227,
+      "step": 17110
+    },
+    {
+      "epoch": 0.3292608459426582,
+      "grad_norm": 505.88855151867824,
+      "learning_rate": 1.5292146032378778e-05,
+      "loss": 282.797,
+      "step": 17120
+    },
+    {
+      "epoch": 0.3294531712031388,
+      "grad_norm": 560.1317997466638,
+      "learning_rate": 1.5286966603759767e-05,
+      "loss": 279.9418,
+      "step": 17130
+    },
+    {
+      "epoch": 0.3296454964636193,
+      "grad_norm": 623.2484739964339,
+      "learning_rate": 1.5281785205833013e-05,
+      "loss": 276.0334,
+      "step": 17140
+    },
+    {
+      "epoch": 0.3298378217240998,
+      "grad_norm": 518.7324905589361,
+      "learning_rate": 1.52766018405285e-05,
+      "loss": 275.4337,
+      "step": 17150
+    },
+    {
+      "epoch": 0.33003014698458033,
+      "grad_norm": 499.7788592259437,
+      "learning_rate": 1.5271416509776948e-05,
+      "loss": 274.5938,
+      "step": 17160
+    },
+    {
+      "epoch": 0.33022247224506085,
+      "grad_norm": 506.7003651963242,
+      "learning_rate": 1.5266229215509806e-05,
+      "loss": 276.1904,
+      "step": 17170
+    },
+    {
+      "epoch": 0.33041479750554137,
+      "grad_norm": 591.2069086162949,
+      "learning_rate": 1.5261039959659257e-05,
+      "loss": 282.4061,
+      "step": 17180
+    },
+    {
+      "epoch": 0.3306071227660219,
+      "grad_norm": 530.2401027869066,
+      "learning_rate": 1.5255848744158214e-05,
+      "loss": 268.4294,
+      "step": 17190
+    },
+    {
+      "epoch": 0.3307994480265024,
+      "grad_norm": 503.438411688062,
+      "learning_rate": 1.5250655570940317e-05,
+      "loss": 290.5155,
+      "step": 17200
+    },
+    {
+      "epoch": 0.3309917732869829,
+      "grad_norm": 507.0000846841116,
+      "learning_rate": 1.5245460441939934e-05,
+      "loss": 282.9155,
+      "step": 17210
+    },
+    {
+      "epoch": 0.33118409854746345,
+      "grad_norm": 510.5138871865681,
+      "learning_rate": 1.5240263359092167e-05,
+      "loss": 280.4435,
+      "step": 17220
+    },
+    {
+      "epoch": 0.331376423807944,
+      "grad_norm": 467.48821668726924,
+      "learning_rate": 1.5235064324332846e-05,
+      "loss": 286.6052,
+      "step": 17230
+    },
+    {
+      "epoch": 0.33156874906842454,
+      "grad_norm": 522.1372416385899,
+      "learning_rate": 1.5229863339598528e-05,
+      "loss": 274.6612,
+      "step": 17240
+    },
+    {
+      "epoch": 0.33176107432890506,
+      "grad_norm": 501.7652116015993,
+      "learning_rate": 1.5224660406826486e-05,
+      "loss": 272.3773,
+      "step": 17250
+    },
+    {
+      "epoch": 0.3319533995893856,
+      "grad_norm": 533.9846585342436,
+      "learning_rate": 1.5219455527954732e-05,
+      "loss": 287.6437,
+      "step": 17260
+    },
+    {
+      "epoch": 0.3321457248498661,
+      "grad_norm": 468.84762379401434,
+      "learning_rate": 1.5214248704921995e-05,
+      "loss": 277.3463,
+      "step": 17270
+    },
+    {
+      "epoch": 0.3323380501103466,
+      "grad_norm": 532.0771512669993,
+      "learning_rate": 1.5209039939667731e-05,
+      "loss": 278.6008,
+      "step": 17280
+    },
+    {
+      "epoch": 0.33253037537082714,
+      "grad_norm": 521.8244517674843,
+      "learning_rate": 1.5203829234132118e-05,
+      "loss": 277.7722,
+      "step": 17290
+    },
+    {
+      "epoch": 0.33272270063130766,
+      "grad_norm": 463.5135411104897,
+      "learning_rate": 1.5198616590256064e-05,
+      "loss": 281.7509,
+      "step": 17300
+    },
+    {
+      "epoch": 0.3329150258917882,
+      "grad_norm": 500.148078934557,
+      "learning_rate": 1.5193402009981187e-05,
+      "loss": 267.7973,
+      "step": 17310
+    },
+    {
+      "epoch": 0.3331073511522687,
+      "grad_norm": 507.9846930240541,
+      "learning_rate": 1.5188185495249832e-05,
+      "loss": 283.1407,
+      "step": 17320
+    },
+    {
+      "epoch": 0.33329967641274927,
+      "grad_norm": 483.99928509173293,
+      "learning_rate": 1.5182967048005069e-05,
+      "loss": 279.5925,
+      "step": 17330
+    },
+    {
+      "epoch": 0.3334920016732298,
+      "grad_norm": 491.625892638982,
+      "learning_rate": 1.5177746670190674e-05,
+      "loss": 274.332,
+      "step": 17340
+    },
+    {
+      "epoch": 0.3336843269337103,
+      "grad_norm": 465.77803815953786,
+      "learning_rate": 1.5172524363751162e-05,
+      "loss": 273.0145,
+      "step": 17350
+    },
+    {
+      "epoch": 0.3338766521941908,
+      "grad_norm": 543.3911445883701,
+      "learning_rate": 1.5167300130631748e-05,
+      "loss": 278.0825,
+      "step": 17360
+    },
+    {
+      "epoch": 0.33406897745467135,
+      "grad_norm": 523.9792196412394,
+      "learning_rate": 1.516207397277838e-05,
+      "loss": 277.4151,
+      "step": 17370
+    },
+    {
+      "epoch": 0.33426130271515186,
+      "grad_norm": 499.3017131309414,
+      "learning_rate": 1.5156845892137711e-05,
+      "loss": 267.7405,
+      "step": 17380
+    },
+    {
+      "epoch": 0.3344536279756324,
+      "grad_norm": 518.06996096728,
+      "learning_rate": 1.5151615890657113e-05,
+      "loss": 281.213,
+      "step": 17390
+    },
+    {
+      "epoch": 0.3346459532361129,
+      "grad_norm": 491.79445451709057,
+      "learning_rate": 1.5146383970284679e-05,
+      "loss": 278.501,
+      "step": 17400
+    },
+    {
+      "epoch": 0.3348382784965934,
+      "grad_norm": 464.2798517580624,
+      "learning_rate": 1.514115013296921e-05,
+      "loss": 277.1603,
+      "step": 17410
+    },
+    {
+      "epoch": 0.33503060375707394,
+      "grad_norm": 504.1890016819973,
+      "learning_rate": 1.513591438066023e-05,
+      "loss": 279.4511,
+      "step": 17420
+    },
+    {
+      "epoch": 0.33522292901755446,
+      "grad_norm": 513.4741153847451,
+      "learning_rate": 1.5130676715307962e-05,
+      "loss": 289.9366,
+      "step": 17430
+    },
+    {
+      "epoch": 0.33541525427803504,
+      "grad_norm": 487.91926779546486,
+      "learning_rate": 1.5125437138863353e-05,
+      "loss": 288.2803,
+      "step": 17440
+    },
+    {
+      "epoch": 0.33560757953851555,
+      "grad_norm": 504.73540403510736,
+      "learning_rate": 1.512019565327806e-05,
+      "loss": 281.8395,
+      "step": 17450
+    },
+    {
+      "epoch": 0.3357999047989961,
+      "grad_norm": 528.102721630349,
+      "learning_rate": 1.5114952260504448e-05,
+      "loss": 288.5389,
+      "step": 17460
+    },
+    {
+      "epoch": 0.3359922300594766,
+      "grad_norm": 505.9273984774669,
+      "learning_rate": 1.5109706962495596e-05,
+      "loss": 280.3888,
+      "step": 17470
+    },
+    {
+      "epoch": 0.3361845553199571,
+      "grad_norm": 549.5731745610719,
+      "learning_rate": 1.510445976120529e-05,
+      "loss": 275.3772,
+      "step": 17480
+    },
+    {
+      "epoch": 0.33637688058043763,
+      "grad_norm": 513.3616780422686,
+      "learning_rate": 1.5099210658588029e-05,
+      "loss": 279.3206,
+      "step": 17490
+    },
+    {
+      "epoch": 0.33656920584091815,
+      "grad_norm": 505.5728776805078,
+      "learning_rate": 1.5093959656599008e-05,
+      "loss": 280.9716,
+      "step": 17500
+    },
+    {
+      "epoch": 0.33676153110139867,
+      "grad_norm": 511.8719650519859,
+      "learning_rate": 1.5088706757194147e-05,
+      "loss": 272.791,
+      "step": 17510
+    },
+    {
+      "epoch": 0.3369538563618792,
+      "grad_norm": 479.78900731898904,
+      "learning_rate": 1.5083451962330062e-05,
+      "loss": 275.2819,
+      "step": 17520
+    },
+    {
+      "epoch": 0.3371461816223597,
+      "grad_norm": 523.6620599199387,
+      "learning_rate": 1.5078195273964081e-05,
+      "loss": 273.6671,
+      "step": 17530
+    },
+    {
+      "epoch": 0.3373385068828403,
+      "grad_norm": 537.1971920513456,
+      "learning_rate": 1.5072936694054222e-05,
+      "loss": 274.4501,
+      "step": 17540
+    },
+    {
+      "epoch": 0.3375308321433208,
+      "grad_norm": 529.84177522821,
+      "learning_rate": 1.5067676224559231e-05,
+      "loss": 279.5901,
+      "step": 17550
+    },
+    {
+      "epoch": 0.3377231574038013,
+      "grad_norm": 508.1453158554725,
+      "learning_rate": 1.506241386743854e-05,
+      "loss": 278.676,
+      "step": 17560
+    },
+    {
+      "epoch": 0.33791548266428184,
+      "grad_norm": 523.0601712437133,
+      "learning_rate": 1.5057149624652297e-05,
+      "loss": 268.4308,
+      "step": 17570
+    },
+    {
+      "epoch": 0.33810780792476236,
+      "grad_norm": 491.62081580979446,
+      "learning_rate": 1.5051883498161334e-05,
+      "loss": 273.8124,
+      "step": 17580
+    },
+    {
+      "epoch": 0.3383001331852429,
+      "grad_norm": 501.30099692217374,
+      "learning_rate": 1.5046615489927206e-05,
+      "loss": 273.8414,
+      "step": 17590
+    },
+    {
+      "epoch": 0.3384924584457234,
+      "grad_norm": 489.6074219863892,
+      "learning_rate": 1.5041345601912154e-05,
+      "loss": 275.304,
+      "step": 17600
+    },
+    {
+      "epoch": 0.3386847837062039,
+      "grad_norm": 491.03167632258294,
+      "learning_rate": 1.5036073836079128e-05,
+      "loss": 282.0901,
+      "step": 17610
+    },
+    {
+      "epoch": 0.33887710896668444,
+      "grad_norm": 506.6295762251429,
+      "learning_rate": 1.5030800194391773e-05,
+      "loss": 269.5102,
+      "step": 17620
+    },
+    {
+      "epoch": 0.33906943422716496,
+      "grad_norm": 519.1804558435261,
+      "learning_rate": 1.5025524678814428e-05,
+      "loss": 279.2036,
+      "step": 17630
+    },
+    {
+      "epoch": 0.33926175948764553,
+      "grad_norm": 479.64658892197633,
+      "learning_rate": 1.5020247291312138e-05,
+      "loss": 280.2888,
+      "step": 17640
+    },
+    {
+      "epoch": 0.33945408474812605,
+      "grad_norm": 497.4550420103091,
+      "learning_rate": 1.5014968033850647e-05,
+      "loss": 291.1086,
+      "step": 17650
+    },
+    {
+      "epoch": 0.33964641000860657,
+      "grad_norm": 549.9160651598164,
+      "learning_rate": 1.5009686908396388e-05,
+      "loss": 279.9692,
+      "step": 17660
+    },
+    {
+      "epoch": 0.3398387352690871,
+      "grad_norm": 484.1959080508733,
+      "learning_rate": 1.5004403916916494e-05,
+      "loss": 278.8785,
+      "step": 17670
+    },
+    {
+      "epoch": 0.3400310605295676,
+      "grad_norm": 518.9782936604537,
+      "learning_rate": 1.4999119061378791e-05,
+      "loss": 279.7993,
+      "step": 17680
+    },
+    {
+      "epoch": 0.3402233857900481,
+      "grad_norm": 500.82782772726847,
+      "learning_rate": 1.4993832343751802e-05,
+      "loss": 290.7989,
+      "step": 17690
+    },
+    {
+      "epoch": 0.34041571105052865,
+      "grad_norm": 497.78970508729935,
+      "learning_rate": 1.4988543766004737e-05,
+      "loss": 272.7477,
+      "step": 17700
+    },
+    {
+      "epoch": 0.34060803631100917,
+      "grad_norm": 513.4342711301089,
+      "learning_rate": 1.498325333010751e-05,
+      "loss": 267.0763,
+      "step": 17710
+    },
+    {
+      "epoch": 0.3408003615714897,
+      "grad_norm": 499.06077693041084,
+      "learning_rate": 1.497796103803072e-05,
+      "loss": 280.9972,
+      "step": 17720
+    },
+    {
+      "epoch": 0.3409926868319702,
+      "grad_norm": 550.4500521767796,
+      "learning_rate": 1.4972666891745655e-05,
+      "loss": 279.5483,
+      "step": 17730
+    },
+    {
+      "epoch": 0.3411850120924508,
+      "grad_norm": 474.5120322786387,
+      "learning_rate": 1.49673708932243e-05,
+      "loss": 267.1358,
+      "step": 17740
+    },
+    {
+      "epoch": 0.3413773373529313,
+      "grad_norm": 511.2861397904143,
+      "learning_rate": 1.4962073044439328e-05,
+      "loss": 281.4607,
+      "step": 17750
+    },
+    {
+      "epoch": 0.3415696626134118,
+      "grad_norm": 485.9737847449894,
+      "learning_rate": 1.4956773347364095e-05,
+      "loss": 281.6694,
+      "step": 17760
+    },
+    {
+      "epoch": 0.34176198787389234,
+      "grad_norm": 543.3463788744507,
+      "learning_rate": 1.4951471803972657e-05,
+      "loss": 276.9335,
+      "step": 17770
+    },
+    {
+      "epoch": 0.34195431313437286,
+      "grad_norm": 570.2492342485151,
+      "learning_rate": 1.4946168416239746e-05,
+      "loss": 296.8713,
+      "step": 17780
+    },
+    {
+      "epoch": 0.3421466383948534,
+      "grad_norm": 542.5114847191128,
+      "learning_rate": 1.4940863186140788e-05,
+      "loss": 288.7994,
+      "step": 17790
+    },
+    {
+      "epoch": 0.3423389636553339,
+      "grad_norm": 538.8198821083244,
+      "learning_rate": 1.4935556115651898e-05,
+      "loss": 267.133,
+      "step": 17800
+    },
+    {
+      "epoch": 0.3425312889158144,
+      "grad_norm": 539.7385609767242,
+      "learning_rate": 1.4930247206749863e-05,
+      "loss": 270.0881,
+      "step": 17810
+    },
+    {
+      "epoch": 0.34272361417629493,
+      "grad_norm": 531.2502518020995,
+      "learning_rate": 1.4924936461412172e-05,
+      "loss": 279.517,
+      "step": 17820
+    },
+    {
+      "epoch": 0.34291593943677545,
+      "grad_norm": 478.10313732077884,
+      "learning_rate": 1.4919623881616989e-05,
+      "loss": 271.1312,
+      "step": 17830
+    },
+    {
+      "epoch": 0.343108264697256,
+      "grad_norm": 526.6498889840595,
+      "learning_rate": 1.4914309469343158e-05,
+      "loss": 265.5551,
+      "step": 17840
+    },
+    {
+      "epoch": 0.34330058995773655,
+      "grad_norm": 563.5118198740181,
+      "learning_rate": 1.4908993226570214e-05,
+      "loss": 290.039,
+      "step": 17850
+    },
+    {
+      "epoch": 0.34349291521821707,
+      "grad_norm": 555.612150809867,
+      "learning_rate": 1.4903675155278365e-05,
+      "loss": 271.889,
+      "step": 17860
+    },
+    {
+      "epoch": 0.3436852404786976,
+      "grad_norm": 457.22209280509924,
+      "learning_rate": 1.4898355257448508e-05,
+      "loss": 278.4439,
+      "step": 17870
+    },
+    {
+      "epoch": 0.3438775657391781,
+      "grad_norm": 450.13527306824096,
+      "learning_rate": 1.4893033535062219e-05,
+      "loss": 271.5201,
+      "step": 17880
+    },
+    {
+      "epoch": 0.3440698909996586,
+      "grad_norm": 481.2346702407072,
+      "learning_rate": 1.4887709990101748e-05,
+      "loss": 276.1723,
+      "step": 17890
+    },
+    {
+      "epoch": 0.34426221626013914,
+      "grad_norm": 482.0542976273173,
+      "learning_rate": 1.4882384624550028e-05,
+      "loss": 279.5917,
+      "step": 17900
+    },
+    {
+      "epoch": 0.34445454152061966,
+      "grad_norm": 570.5918494552345,
+      "learning_rate": 1.4877057440390672e-05,
+      "loss": 267.7453,
+      "step": 17910
+    },
+    {
+      "epoch": 0.3446468667811002,
+      "grad_norm": 489.2071285520791,
+      "learning_rate": 1.4871728439607967e-05,
+      "loss": 276.4078,
+      "step": 17920
+    },
+    {
+      "epoch": 0.3448391920415807,
+      "grad_norm": 499.4925409907639,
+      "learning_rate": 1.4866397624186877e-05,
+      "loss": 274.1754,
+      "step": 17930
+    },
+    {
+      "epoch": 0.3450315173020612,
+      "grad_norm": 489.3010735053375,
+      "learning_rate": 1.4861064996113042e-05,
+      "loss": 270.5771,
+      "step": 17940
+    },
+    {
+      "epoch": 0.3452238425625418,
+      "grad_norm": 516.4476601167114,
+      "learning_rate": 1.485573055737278e-05,
+      "loss": 281.167,
+      "step": 17950
+    },
+    {
+      "epoch": 0.3454161678230223,
+      "grad_norm": 502.5953209048784,
+      "learning_rate": 1.485039430995308e-05,
+      "loss": 276.3996,
+      "step": 17960
+    },
+    {
+      "epoch": 0.34560849308350283,
+      "grad_norm": 509.1630480153723,
+      "learning_rate": 1.4845056255841608e-05,
+      "loss": 279.1599,
+      "step": 17970
+    },
+    {
+      "epoch": 0.34580081834398335,
+      "grad_norm": 511.0730622472175,
+      "learning_rate": 1.48397163970267e-05,
+      "loss": 286.7167,
+      "step": 17980
+    },
+    {
+      "epoch": 0.34599314360446387,
+      "grad_norm": 543.6350962657646,
+      "learning_rate": 1.4834374735497362e-05,
+      "loss": 275.0893,
+      "step": 17990
+    },
+    {
+      "epoch": 0.3461854688649444,
+      "grad_norm": 549.0933276042018,
+      "learning_rate": 1.4829031273243277e-05,
+      "loss": 283.2004,
+      "step": 18000
+    },
+    {
+      "epoch": 0.3463777941254249,
+      "grad_norm": 478.9966551116924,
+      "learning_rate": 1.4823686012254798e-05,
+      "loss": 270.4572,
+      "step": 18010
+    },
+    {
+      "epoch": 0.34657011938590543,
+      "grad_norm": 451.09551493351705,
+      "learning_rate": 1.4818338954522943e-05,
+      "loss": 267.2886,
+      "step": 18020
+    },
+    {
+      "epoch": 0.34676244464638595,
+      "grad_norm": 487.70379309691907,
+      "learning_rate": 1.4812990102039411e-05,
+      "loss": 272.4028,
+      "step": 18030
+    },
+    {
+      "epoch": 0.34695476990686647,
+      "grad_norm": 461.8328123924969,
+      "learning_rate": 1.480763945679655e-05,
+      "loss": 270.2243,
+      "step": 18040
+    },
+    {
+      "epoch": 0.34714709516734704,
+      "grad_norm": 513.3241681581126,
+      "learning_rate": 1.4802287020787396e-05,
+      "loss": 273.2115,
+      "step": 18050
+    },
+    {
+      "epoch": 0.34733942042782756,
+      "grad_norm": 583.6831276889415,
+      "learning_rate": 1.4796932796005634e-05,
+      "loss": 278.092,
+      "step": 18060
+    },
+    {
+      "epoch": 0.3475317456883081,
+      "grad_norm": 500.1706108032943,
+      "learning_rate": 1.4791576784445632e-05,
+      "loss": 270.2284,
+      "step": 18070
+    },
+    {
+      "epoch": 0.3477240709487886,
+      "grad_norm": 513.7577102398449,
+      "learning_rate": 1.4786218988102414e-05,
+      "loss": 271.5205,
+      "step": 18080
+    },
+    {
+      "epoch": 0.3479163962092691,
+      "grad_norm": 517.6885947849202,
+      "learning_rate": 1.4780859408971668e-05,
+      "loss": 267.0917,
+      "step": 18090
+    },
+    {
+      "epoch": 0.34810872146974964,
+      "grad_norm": 493.7372168706724,
+      "learning_rate": 1.4775498049049754e-05,
+      "loss": 277.2727,
+      "step": 18100
+    },
+    {
+      "epoch": 0.34830104673023016,
+      "grad_norm": 502.8970072394122,
+      "learning_rate": 1.4770134910333684e-05,
+      "loss": 280.6166,
+      "step": 18110
+    },
+    {
+      "epoch": 0.3484933719907107,
+      "grad_norm": 453.5355996943572,
+      "learning_rate": 1.4764769994821145e-05,
+      "loss": 280.9081,
+      "step": 18120
+    },
+    {
+      "epoch": 0.3486856972511912,
+      "grad_norm": 584.3013429138648,
+      "learning_rate": 1.4759403304510472e-05,
+      "loss": 281.7896,
+      "step": 18130
+    },
+    {
+      "epoch": 0.3488780225116717,
+      "grad_norm": 510.58687368556485,
+      "learning_rate": 1.475403484140067e-05,
+      "loss": 281.7353,
+      "step": 18140
+    },
+    {
+      "epoch": 0.3490703477721523,
+      "grad_norm": 478.761783282778,
+      "learning_rate": 1.4748664607491408e-05,
+      "loss": 274.2493,
+      "step": 18150
+    },
+    {
+      "epoch": 0.3492626730326328,
+      "grad_norm": 478.48081702040633,
+      "learning_rate": 1.4743292604783008e-05,
+      "loss": 278.2816,
+      "step": 18160
+    },
+    {
+      "epoch": 0.34945499829311333,
+      "grad_norm": 490.50956716839545,
+      "learning_rate": 1.4737918835276451e-05,
+      "loss": 273.5343,
+      "step": 18170
+    },
+    {
+      "epoch": 0.34964732355359385,
+      "grad_norm": 499.3343957519696,
+      "learning_rate": 1.4732543300973374e-05,
+      "loss": 274.0835,
+      "step": 18180
+    },
+    {
+      "epoch": 0.34983964881407437,
+      "grad_norm": 496.47168303627546,
+      "learning_rate": 1.472716600387608e-05,
+      "loss": 274.201,
+      "step": 18190
+    },
+    {
+      "epoch": 0.3500319740745549,
+      "grad_norm": 501.4186703132157,
+      "learning_rate": 1.4721786945987519e-05,
+      "loss": 282.1016,
+      "step": 18200
+    },
+    {
+      "epoch": 0.3502242993350354,
+      "grad_norm": 482.18739096287226,
+      "learning_rate": 1.4716406129311307e-05,
+      "loss": 277.8248,
+      "step": 18210
+    },
+    {
+      "epoch": 0.3504166245955159,
+      "grad_norm": 492.65093701630104,
+      "learning_rate": 1.4711023555851702e-05,
+      "loss": 281.1395,
+      "step": 18220
+    },
+    {
+      "epoch": 0.35060894985599644,
+      "grad_norm": 510.5813213558361,
+      "learning_rate": 1.470563922761363e-05,
+      "loss": 264.6669,
+      "step": 18230
+    },
+    {
+      "epoch": 0.35080127511647696,
+      "grad_norm": 484.1044532276803,
+      "learning_rate": 1.470025314660266e-05,
+      "loss": 275.079,
+      "step": 18240
+    },
+    {
+      "epoch": 0.35099360037695754,
+      "grad_norm": 492.22338237898106,
+      "learning_rate": 1.4694865314825024e-05,
+      "loss": 279.9116,
+      "step": 18250
+    },
+    {
+      "epoch": 0.35118592563743806,
+      "grad_norm": 582.2876021990764,
+      "learning_rate": 1.4689475734287596e-05,
+      "loss": 275.7554,
+      "step": 18260
+    },
+    {
+      "epoch": 0.3513782508979186,
+      "grad_norm": 491.92303937431313,
+      "learning_rate": 1.4684084406997903e-05,
+      "loss": 278.5537,
+      "step": 18270
+    },
+    {
+      "epoch": 0.3515705761583991,
+      "grad_norm": 522.9623001197829,
+      "learning_rate": 1.467869133496413e-05,
+      "loss": 275.3861,
+      "step": 18280
+    },
+    {
+      "epoch": 0.3517629014188796,
+      "grad_norm": 509.6981462756343,
+      "learning_rate": 1.4673296520195105e-05,
+      "loss": 270.7336,
+      "step": 18290
+    },
+    {
+      "epoch": 0.35195522667936013,
+      "grad_norm": 443.24550556703184,
+      "learning_rate": 1.4667899964700309e-05,
+      "loss": 273.7989,
+      "step": 18300
+    },
+    {
+      "epoch": 0.35214755193984065,
+      "grad_norm": 489.07076765329754,
+      "learning_rate": 1.466250167048987e-05,
+      "loss": 273.8396,
+      "step": 18310
+    },
+    {
+      "epoch": 0.35233987720032117,
+      "grad_norm": 461.5638593100279,
+      "learning_rate": 1.4657101639574563e-05,
+      "loss": 281.685,
+      "step": 18320
+    },
+    {
+      "epoch": 0.3525322024608017,
+      "grad_norm": 483.6051806875028,
+      "learning_rate": 1.4651699873965808e-05,
+      "loss": 276.0564,
+      "step": 18330
+    },
+    {
+      "epoch": 0.3527245277212822,
+      "grad_norm": 475.19970902453315,
+      "learning_rate": 1.4646296375675676e-05,
+      "loss": 280.6133,
+      "step": 18340
+    },
+    {
+      "epoch": 0.35291685298176273,
+      "grad_norm": 467.261602599556,
+      "learning_rate": 1.464089114671688e-05,
+      "loss": 284.0963,
+      "step": 18350
+    },
+    {
+      "epoch": 0.3531091782422433,
+      "grad_norm": 565.2773359794652,
+      "learning_rate": 1.4635484189102776e-05,
+      "loss": 273.4974,
+      "step": 18360
+    },
+    {
+      "epoch": 0.3533015035027238,
+      "grad_norm": 610.7791523439159,
+      "learning_rate": 1.4630075504847373e-05,
+      "loss": 274.8716,
+      "step": 18370
+    },
+    {
+      "epoch": 0.35349382876320434,
+      "grad_norm": 561.8525351287174,
+      "learning_rate": 1.4624665095965311e-05,
+      "loss": 269.1808,
+      "step": 18380
+    },
+    {
+      "epoch": 0.35368615402368486,
+      "grad_norm": 534.9783175143453,
+      "learning_rate": 1.4619252964471881e-05,
+      "loss": 270.5985,
+      "step": 18390
+    },
+    {
+      "epoch": 0.3538784792841654,
+      "grad_norm": 518.9606747472129,
+      "learning_rate": 1.461383911238301e-05,
+      "loss": 284.0321,
+      "step": 18400
+    },
+    {
+      "epoch": 0.3540708045446459,
+      "grad_norm": 589.8949359179564,
+      "learning_rate": 1.4608423541715273e-05,
+      "loss": 271.4509,
+      "step": 18410
+    },
+    {
+      "epoch": 0.3542631298051264,
+      "grad_norm": 486.3948360072605,
+      "learning_rate": 1.4603006254485874e-05,
+      "loss": 274.5822,
+      "step": 18420
+    },
+    {
+      "epoch": 0.35445545506560694,
+      "grad_norm": 492.48757346366045,
+      "learning_rate": 1.4597587252712666e-05,
+      "loss": 274.8331,
+      "step": 18430
+    },
+    {
+      "epoch": 0.35464778032608746,
+      "grad_norm": 484.7721593245906,
+      "learning_rate": 1.4592166538414136e-05,
+      "loss": 266.309,
+      "step": 18440
+    },
+    {
+      "epoch": 0.354840105586568,
+      "grad_norm": 547.3019627975193,
+      "learning_rate": 1.4586744113609416e-05,
+      "loss": 276.1605,
+      "step": 18450
+    },
+    {
+      "epoch": 0.35503243084704855,
+      "grad_norm": 499.784131490363,
+      "learning_rate": 1.4581319980318266e-05,
+      "loss": 278.8771,
+      "step": 18460
+    },
+    {
+      "epoch": 0.35522475610752907,
+      "grad_norm": 547.802960934263,
+      "learning_rate": 1.4575894140561086e-05,
+      "loss": 265.872,
+      "step": 18470
+    },
+    {
+      "epoch": 0.3554170813680096,
+      "grad_norm": 584.7809345359788,
+      "learning_rate": 1.4570466596358914e-05,
+      "loss": 281.2158,
+      "step": 18480
+    },
+    {
+      "epoch": 0.3556094066284901,
+      "grad_norm": 462.78468535250977,
+      "learning_rate": 1.4565037349733415e-05,
+      "loss": 278.9555,
+      "step": 18490
+    },
+    {
+      "epoch": 0.35580173188897063,
+      "grad_norm": 611.0583743982891,
+      "learning_rate": 1.45596064027069e-05,
+      "loss": 273.9848,
+      "step": 18500
+    },
+    {
+      "epoch": 0.35599405714945115,
+      "grad_norm": 586.6556436330482,
+      "learning_rate": 1.4554173757302303e-05,
+      "loss": 266.5396,
+      "step": 18510
+    },
+    {
+      "epoch": 0.35618638240993167,
+      "grad_norm": 535.2620975374997,
+      "learning_rate": 1.4548739415543197e-05,
+      "loss": 268.6847,
+      "step": 18520
+    },
+    {
+      "epoch": 0.3563787076704122,
+      "grad_norm": 532.6024427590063,
+      "learning_rate": 1.454330337945378e-05,
+      "loss": 268.69,
+      "step": 18530
+    },
+    {
+      "epoch": 0.3565710329308927,
+      "grad_norm": 451.5225454658764,
+      "learning_rate": 1.4537865651058893e-05,
+      "loss": 266.809,
+      "step": 18540
+    },
+    {
+      "epoch": 0.3567633581913732,
+      "grad_norm": 508.726118035006,
+      "learning_rate": 1.4532426232383998e-05,
+      "loss": 271.8431,
+      "step": 18550
+    },
+    {
+      "epoch": 0.3569556834518538,
+      "grad_norm": 435.3919759200938,
+      "learning_rate": 1.4526985125455184e-05,
+      "loss": 273.8349,
+      "step": 18560
+    },
+    {
+      "epoch": 0.3571480087123343,
+      "grad_norm": 475.22283493396196,
+      "learning_rate": 1.4521542332299177e-05,
+      "loss": 277.1238,
+      "step": 18570
+    },
+    {
+      "epoch": 0.35734033397281484,
+      "grad_norm": 454.663188481646,
+      "learning_rate": 1.4516097854943325e-05,
+      "loss": 269.471,
+      "step": 18580
+    },
+    {
+      "epoch": 0.35753265923329536,
+      "grad_norm": 489.6256450352957,
+      "learning_rate": 1.4510651695415612e-05,
+      "loss": 263.6692,
+      "step": 18590
+    },
+    {
+      "epoch": 0.3577249844937759,
+      "grad_norm": 523.4703439970984,
+      "learning_rate": 1.4505203855744637e-05,
+      "loss": 287.1177,
+      "step": 18600
+    },
+    {
+      "epoch": 0.3579173097542564,
+      "grad_norm": 487.45110951567614,
+      "learning_rate": 1.4499754337959628e-05,
+      "loss": 275.8579,
+      "step": 18610
+    },
+    {
+      "epoch": 0.3581096350147369,
+      "grad_norm": 471.63288603724055,
+      "learning_rate": 1.449430314409045e-05,
+      "loss": 268.8969,
+      "step": 18620
+    },
+    {
+      "epoch": 0.35830196027521743,
+      "grad_norm": 509.35368761915225,
+      "learning_rate": 1.4488850276167572e-05,
+      "loss": 272.5918,
+      "step": 18630
+    },
+    {
+      "epoch": 0.35849428553569795,
+      "grad_norm": 496.2132943043483,
+      "learning_rate": 1.44833957362221e-05,
+      "loss": 278.5671,
+      "step": 18640
+    },
+    {
+      "epoch": 0.3586866107961785,
+      "grad_norm": 516.1593242842114,
+      "learning_rate": 1.4477939526285767e-05,
+      "loss": 281.0237,
+      "step": 18650
+    },
+    {
+      "epoch": 0.35887893605665905,
+      "grad_norm": 500.2385340599565,
+      "learning_rate": 1.4472481648390914e-05,
+      "loss": 284.4294,
+      "step": 18660
+    },
+    {
+      "epoch": 0.35907126131713957,
+      "grad_norm": 495.0134020699203,
+      "learning_rate": 1.4467022104570514e-05,
+      "loss": 276.252,
+      "step": 18670
+    },
+    {
+      "epoch": 0.3592635865776201,
+      "grad_norm": 522.2266800170015,
+      "learning_rate": 1.4461560896858156e-05,
+      "loss": 273.3659,
+      "step": 18680
+    },
+    {
+      "epoch": 0.3594559118381006,
+      "grad_norm": 493.21767026047456,
+      "learning_rate": 1.4456098027288046e-05,
+      "loss": 270.9548,
+      "step": 18690
+    },
+    {
+      "epoch": 0.3596482370985811,
+      "grad_norm": 515.6318969442698,
+      "learning_rate": 1.4450633497895017e-05,
+      "loss": 271.9803,
+      "step": 18700
+    },
+    {
+      "epoch": 0.35984056235906164,
+      "grad_norm": 492.05182539050395,
+      "learning_rate": 1.4445167310714514e-05,
+      "loss": 278.6881,
+      "step": 18710
+    },
+    {
+      "epoch": 0.36003288761954216,
+      "grad_norm": 518.9349390572469,
+      "learning_rate": 1.4439699467782602e-05,
+      "loss": 268.8568,
+      "step": 18720
+    },
+    {
+      "epoch": 0.3602252128800227,
+      "grad_norm": 472.47978878508525,
+      "learning_rate": 1.4434229971135965e-05,
+      "loss": 274.5661,
+      "step": 18730
+    },
+    {
+      "epoch": 0.3604175381405032,
+      "grad_norm": 462.5800958576359,
+      "learning_rate": 1.4428758822811894e-05,
+      "loss": 268.6313,
+      "step": 18740
+    },
+    {
+      "epoch": 0.3606098634009837,
+      "grad_norm": 556.8318001826748,
+      "learning_rate": 1.442328602484831e-05,
+      "loss": 279.1211,
+      "step": 18750
+    },
+    {
+      "epoch": 0.3608021886614643,
+      "grad_norm": 475.38132861284896,
+      "learning_rate": 1.441781157928373e-05,
+      "loss": 267.9185,
+      "step": 18760
+    },
+    {
+      "epoch": 0.3609945139219448,
+      "grad_norm": 444.2590438167381,
+      "learning_rate": 1.44123354881573e-05,
+      "loss": 264.6602,
+      "step": 18770
+    },
+    {
+      "epoch": 0.36118683918242533,
+      "grad_norm": 593.5762996181691,
+      "learning_rate": 1.4406857753508772e-05,
+      "loss": 277.1706,
+      "step": 18780
+    },
+    {
+      "epoch": 0.36137916444290585,
+      "grad_norm": 498.61812297220695,
+      "learning_rate": 1.4401378377378512e-05,
+      "loss": 274.0979,
+      "step": 18790
+    },
+    {
+      "epoch": 0.3615714897033864,
+      "grad_norm": 464.1576611937751,
+      "learning_rate": 1.43958973618075e-05,
+      "loss": 265.5725,
+      "step": 18800
+    },
+    {
+      "epoch": 0.3617638149638669,
+      "grad_norm": 521.681813003637,
+      "learning_rate": 1.4390414708837322e-05,
+      "loss": 266.8785,
+      "step": 18810
+    },
+    {
+      "epoch": 0.3619561402243474,
+      "grad_norm": 487.28500492160447,
+      "learning_rate": 1.4384930420510173e-05,
+      "loss": 268.4358,
+      "step": 18820
+    },
+    {
+      "epoch": 0.36214846548482793,
+      "grad_norm": 2610.6643410670968,
+      "learning_rate": 1.4379444498868864e-05,
+      "loss": 269.2086,
+      "step": 18830
+    },
+    {
+      "epoch": 0.36234079074530845,
+      "grad_norm": 537.8437440046431,
+      "learning_rate": 1.4373956945956807e-05,
+      "loss": 284.2879,
+      "step": 18840
+    },
+    {
+      "epoch": 0.36253311600578897,
+      "grad_norm": 499.0172686886838,
+      "learning_rate": 1.4368467763818026e-05,
+      "loss": 266.1877,
+      "step": 18850
+    },
+    {
+      "epoch": 0.3627254412662695,
+      "grad_norm": 502.55642424490685,
+      "learning_rate": 1.436297695449715e-05,
+      "loss": 277.1923,
+      "step": 18860
+    },
+    {
+      "epoch": 0.36291776652675006,
+      "grad_norm": 502.29657409854235,
+      "learning_rate": 1.4357484520039412e-05,
+      "loss": 267.7814,
+      "step": 18870
+    },
+    {
+      "epoch": 0.3631100917872306,
+      "grad_norm": 509.966259301657,
+      "learning_rate": 1.4351990462490662e-05,
+      "loss": 270.7,
+      "step": 18880
+    },
+    {
+      "epoch": 0.3633024170477111,
+      "grad_norm": 470.24477104277145,
+      "learning_rate": 1.434649478389734e-05,
+      "loss": 265.3593,
+      "step": 18890
+    },
+    {
+      "epoch": 0.3634947423081916,
+      "grad_norm": 469.71886315911667,
+      "learning_rate": 1.4340997486306491e-05,
+      "loss": 278.0755,
+      "step": 18900
+    },
+    {
+      "epoch": 0.36368706756867214,
+      "grad_norm": 453.5521401137459,
+      "learning_rate": 1.4335498571765777e-05,
+      "loss": 277.2238,
+      "step": 18910
+    },
+    {
+      "epoch": 0.36387939282915266,
+      "grad_norm": 498.7986418858856,
+      "learning_rate": 1.4329998042323447e-05,
+      "loss": 262.8456,
+      "step": 18920
+    },
+    {
+      "epoch": 0.3640717180896332,
+      "grad_norm": 492.73132779524826,
+      "learning_rate": 1.4324495900028358e-05,
+      "loss": 277.1212,
+      "step": 18930
+    },
+    {
+      "epoch": 0.3642640433501137,
+      "grad_norm": 491.47538477827686,
+      "learning_rate": 1.4318992146929967e-05,
+      "loss": 270.7526,
+      "step": 18940
+    },
+    {
+      "epoch": 0.3644563686105942,
+      "grad_norm": 453.66565326216687,
+      "learning_rate": 1.4313486785078335e-05,
+      "loss": 269.5168,
+      "step": 18950
+    },
+    {
+      "epoch": 0.36464869387107474,
+      "grad_norm": 468.63668189220294,
+      "learning_rate": 1.4307979816524111e-05,
+      "loss": 274.1952,
+      "step": 18960
+    },
+    {
+      "epoch": 0.3648410191315553,
+      "grad_norm": 475.51421977181815,
+      "learning_rate": 1.4302471243318554e-05,
+      "loss": 275.7419,
+      "step": 18970
+    },
+    {
+      "epoch": 0.36503334439203583,
+      "grad_norm": 449.11468583079665,
+      "learning_rate": 1.4296961067513519e-05,
+      "loss": 271.0766,
+      "step": 18980
+    },
+    {
+      "epoch": 0.36522566965251635,
+      "grad_norm": 496.7789064901429,
+      "learning_rate": 1.4291449291161452e-05,
+      "loss": 269.3588,
+      "step": 18990
+    },
+    {
+      "epoch": 0.36541799491299687,
+      "grad_norm": 520.561948848321,
+      "learning_rate": 1.4285935916315401e-05,
+      "loss": 268.3722,
+      "step": 19000
+    },
+    {
+      "epoch": 0.3656103201734774,
+      "grad_norm": 556.507199075098,
+      "learning_rate": 1.4280420945029004e-05,
+      "loss": 280.2751,
+      "step": 19010
+    },
+    {
+      "epoch": 0.3658026454339579,
+      "grad_norm": 459.33758252701904,
+      "learning_rate": 1.4274904379356498e-05,
+      "loss": 273.0975,
+      "step": 19020
+    },
+    {
+      "epoch": 0.3659949706944384,
+      "grad_norm": 510.58230771907785,
+      "learning_rate": 1.4269386221352714e-05,
+      "loss": 270.5178,
+      "step": 19030
+    },
+    {
+      "epoch": 0.36618729595491895,
+      "grad_norm": 467.9697920669793,
+      "learning_rate": 1.4263866473073076e-05,
+      "loss": 268.6472,
+      "step": 19040
+    },
+    {
+      "epoch": 0.36637962121539946,
+      "grad_norm": 529.010191046606,
+      "learning_rate": 1.4258345136573595e-05,
+      "loss": 264.4419,
+      "step": 19050
+    },
+    {
+      "epoch": 0.36657194647588,
+      "grad_norm": 513.1160209826901,
+      "learning_rate": 1.4252822213910878e-05,
+      "loss": 272.1163,
+      "step": 19060
+    },
+    {
+      "epoch": 0.36676427173636056,
+      "grad_norm": 519.091158590268,
+      "learning_rate": 1.4247297707142126e-05,
+      "loss": 268.8429,
+      "step": 19070
+    },
+    {
+      "epoch": 0.3669565969968411,
+      "grad_norm": 457.3375080569709,
+      "learning_rate": 1.4241771618325123e-05,
+      "loss": 268.8333,
+      "step": 19080
+    },
+    {
+      "epoch": 0.3671489222573216,
+      "grad_norm": 456.79388228896937,
+      "learning_rate": 1.4236243949518249e-05,
+      "loss": 294.9989,
+      "step": 19090
+    },
+    {
+      "epoch": 0.3673412475178021,
+      "grad_norm": 490.57042549721854,
+      "learning_rate": 1.4230714702780466e-05,
+      "loss": 269.3215,
+      "step": 19100
+    },
+    {
+      "epoch": 0.36753357277828264,
+      "grad_norm": 491.6619434120859,
+      "learning_rate": 1.422518388017133e-05,
+      "loss": 274.9493,
+      "step": 19110
+    },
+    {
+      "epoch": 0.36772589803876315,
+      "grad_norm": 458.52854246821556,
+      "learning_rate": 1.4219651483750978e-05,
+      "loss": 267.6029,
+      "step": 19120
+    },
+    {
+      "epoch": 0.3679182232992437,
+      "grad_norm": 489.66741656386336,
+      "learning_rate": 1.4214117515580139e-05,
+      "loss": 272.3837,
+      "step": 19130
+    },
+    {
+      "epoch": 0.3681105485597242,
+      "grad_norm": 528.4465308827021,
+      "learning_rate": 1.4208581977720124e-05,
+      "loss": 275.2206,
+      "step": 19140
+    },
+    {
+      "epoch": 0.3683028738202047,
+      "grad_norm": 517.4333131273371,
+      "learning_rate": 1.420304487223283e-05,
+      "loss": 265.3739,
+      "step": 19150
+    },
+    {
+      "epoch": 0.36849519908068523,
+      "grad_norm": 454.28865672013114,
+      "learning_rate": 1.4197506201180737e-05,
+      "loss": 270.5704,
+      "step": 19160
+    },
+    {
+      "epoch": 0.3686875243411658,
+      "grad_norm": 486.3927113540107,
+      "learning_rate": 1.4191965966626908e-05,
+      "loss": 261.7958,
+      "step": 19170
+    },
+    {
+      "epoch": 0.3688798496016463,
+      "grad_norm": 466.78083849002485,
+      "learning_rate": 1.418642417063499e-05,
+      "loss": 263.7457,
+      "step": 19180
+    },
+    {
+      "epoch": 0.36907217486212684,
+      "grad_norm": 451.2882059299501,
+      "learning_rate": 1.4180880815269207e-05,
+      "loss": 261.4978,
+      "step": 19190
+    },
+    {
+      "epoch": 0.36926450012260736,
+      "grad_norm": 467.6685539093852,
+      "learning_rate": 1.4175335902594372e-05,
+      "loss": 267.75,
+      "step": 19200
+    },
+    {
+      "epoch": 0.3694568253830879,
+      "grad_norm": 473.7323917470346,
+      "learning_rate": 1.416978943467587e-05,
+      "loss": 273.3609,
+      "step": 19210
+    },
+    {
+      "epoch": 0.3696491506435684,
+      "grad_norm": 509.82071496217156,
+      "learning_rate": 1.4164241413579669e-05,
+      "loss": 274.5023,
+      "step": 19220
+    },
+    {
+      "epoch": 0.3698414759040489,
+      "grad_norm": 573.7533306731347,
+      "learning_rate": 1.4158691841372318e-05,
+      "loss": 276.4014,
+      "step": 19230
+    },
+    {
+      "epoch": 0.37003380116452944,
+      "grad_norm": 484.9879111344704,
+      "learning_rate": 1.4153140720120936e-05,
+      "loss": 266.9,
+      "step": 19240
+    },
+    {
+      "epoch": 0.37022612642500996,
+      "grad_norm": 506.2506264507777,
+      "learning_rate": 1.4147588051893233e-05,
+      "loss": 266.8248,
+      "step": 19250
+    },
+    {
+      "epoch": 0.3704184516854905,
+      "grad_norm": 471.4549656744703,
+      "learning_rate": 1.4142033838757476e-05,
+      "loss": 268.2752,
+      "step": 19260
+    },
+    {
+      "epoch": 0.370610776945971,
+      "grad_norm": 463.65869752370816,
+      "learning_rate": 1.4136478082782525e-05,
+      "loss": 268.3878,
+      "step": 19270
+    },
+    {
+      "epoch": 0.3708031022064516,
+      "grad_norm": 503.20251992758267,
+      "learning_rate": 1.4130920786037798e-05,
+      "loss": 269.5339,
+      "step": 19280
+    },
+    {
+      "epoch": 0.3709954274669321,
+      "grad_norm": 508.63307278815176,
+      "learning_rate": 1.412536195059331e-05,
+      "loss": 274.3348,
+      "step": 19290
+    },
+    {
+      "epoch": 0.3711877527274126,
+      "grad_norm": 534.9970695527994,
+      "learning_rate": 1.4119801578519625e-05,
+      "loss": 275.4488,
+      "step": 19300
+    },
+    {
+      "epoch": 0.37138007798789313,
+      "grad_norm": 465.4836706905475,
+      "learning_rate": 1.4114239671887892e-05,
+      "loss": 277.5209,
+      "step": 19310
+    },
+    {
+      "epoch": 0.37157240324837365,
+      "grad_norm": 523.6302016167372,
+      "learning_rate": 1.4108676232769831e-05,
+      "loss": 269.7658,
+      "step": 19320
+    },
+    {
+      "epoch": 0.37176472850885417,
+      "grad_norm": 490.53360691166097,
+      "learning_rate": 1.410311126323773e-05,
+      "loss": 269.1929,
+      "step": 19330
+    },
+    {
+      "epoch": 0.3719570537693347,
+      "grad_norm": 478.5589465618871,
+      "learning_rate": 1.409754476536445e-05,
+      "loss": 263.6918,
+      "step": 19340
+    },
+    {
+      "epoch": 0.3721493790298152,
+      "grad_norm": 473.541200896869,
+      "learning_rate": 1.4091976741223414e-05,
+      "loss": 263.5954,
+      "step": 19350
+    },
+    {
+      "epoch": 0.3723417042902957,
+      "grad_norm": 596.4861586242048,
+      "learning_rate": 1.408640719288863e-05,
+      "loss": 282.8791,
+      "step": 19360
+    },
+    {
+      "epoch": 0.37253402955077625,
+      "grad_norm": 454.33202659460784,
+      "learning_rate": 1.408083612243465e-05,
+      "loss": 275.5541,
+      "step": 19370
+    },
+    {
+      "epoch": 0.3727263548112568,
+      "grad_norm": 487.7196103265241,
+      "learning_rate": 1.4075263531936614e-05,
+      "loss": 273.0452,
+      "step": 19380
+    },
+    {
+      "epoch": 0.37291868007173734,
+      "grad_norm": 520.2335507430199,
+      "learning_rate": 1.4069689423470219e-05,
+      "loss": 264.2038,
+      "step": 19390
+    },
+    {
+      "epoch": 0.37311100533221786,
+      "grad_norm": 493.9525057118881,
+      "learning_rate": 1.4064113799111725e-05,
+      "loss": 271.7883,
+      "step": 19400
+    },
+    {
+      "epoch": 0.3733033305926984,
+      "grad_norm": 494.2036565145267,
+      "learning_rate": 1.405853666093796e-05,
+      "loss": 266.2629,
+      "step": 19410
+    },
+    {
+      "epoch": 0.3734956558531789,
+      "grad_norm": 560.1117951292418,
+      "learning_rate": 1.405295801102632e-05,
+      "loss": 270.0681,
+      "step": 19420
+    },
+    {
+      "epoch": 0.3736879811136594,
+      "grad_norm": 490.2307029611529,
+      "learning_rate": 1.4047377851454758e-05,
+      "loss": 275.5943,
+      "step": 19430
+    },
+    {
+      "epoch": 0.37388030637413994,
+      "grad_norm": 491.3590565361957,
+      "learning_rate": 1.4041796184301788e-05,
+      "loss": 262.3873,
+      "step": 19440
+    },
+    {
+      "epoch": 0.37407263163462046,
+      "grad_norm": 492.35585428770435,
+      "learning_rate": 1.4036213011646496e-05,
+      "loss": 274.749,
+      "step": 19450
+    },
+    {
+      "epoch": 0.374264956895101,
+      "grad_norm": 568.1851303891046,
+      "learning_rate": 1.4030628335568515e-05,
+      "loss": 276.2821,
+      "step": 19460
+    },
+    {
+      "epoch": 0.3744572821555815,
+      "grad_norm": 495.200883399743,
+      "learning_rate": 1.4025042158148048e-05,
+      "loss": 268.6443,
+      "step": 19470
+    },
+    {
+      "epoch": 0.37464960741606207,
+      "grad_norm": 535.8971360833916,
+      "learning_rate": 1.4019454481465853e-05,
+      "loss": 271.2471,
+      "step": 19480
+    },
+    {
+      "epoch": 0.3748419326765426,
+      "grad_norm": 502.58996637213914,
+      "learning_rate": 1.4013865307603248e-05,
+      "loss": 264.8398,
+      "step": 19490
+    },
+    {
+      "epoch": 0.3750342579370231,
+      "grad_norm": 473.907021395529,
+      "learning_rate": 1.4008274638642103e-05,
+      "loss": 275.0792,
+      "step": 19500
+    },
+    {
+      "epoch": 0.3752265831975036,
+      "grad_norm": 447.78039303906274,
+      "learning_rate": 1.4002682476664857e-05,
+      "loss": 265.4133,
+      "step": 19510
+    },
+    {
+      "epoch": 0.37541890845798415,
+      "grad_norm": 469.60597295755775,
+      "learning_rate": 1.3997088823754494e-05,
+      "loss": 261.3435,
+      "step": 19520
+    },
+    {
+      "epoch": 0.37561123371846467,
+      "grad_norm": 548.0002170760671,
+      "learning_rate": 1.399149368199456e-05,
+      "loss": 272.6939,
+      "step": 19530
+    },
+    {
+      "epoch": 0.3758035589789452,
+      "grad_norm": 449.1889657975298,
+      "learning_rate": 1.398589705346915e-05,
+      "loss": 266.1523,
+      "step": 19540
+    },
+    {
+      "epoch": 0.3759958842394257,
+      "grad_norm": 528.7858663903933,
+      "learning_rate": 1.3980298940262918e-05,
+      "loss": 267.7942,
+      "step": 19550
+    },
+    {
+      "epoch": 0.3761882094999062,
+      "grad_norm": 490.66748329643616,
+      "learning_rate": 1.3974699344461065e-05,
+      "loss": 266.4364,
+      "step": 19560
+    },
+    {
+      "epoch": 0.37638053476038674,
+      "grad_norm": 495.24507533583477,
+      "learning_rate": 1.396909826814935e-05,
+      "loss": 271.0768,
+      "step": 19570
+    },
+    {
+      "epoch": 0.3765728600208673,
+      "grad_norm": 503.8560176492099,
+      "learning_rate": 1.3963495713414085e-05,
+      "loss": 267.4308,
+      "step": 19580
+    },
+    {
+      "epoch": 0.37676518528134784,
+      "grad_norm": 422.63297351147776,
+      "learning_rate": 1.3957891682342127e-05,
+      "loss": 270.8489,
+      "step": 19590
+    },
+    {
+      "epoch": 0.37695751054182836,
+      "grad_norm": 525.3053386685739,
+      "learning_rate": 1.3952286177020879e-05,
+      "loss": 269.879,
+      "step": 19600
+    },
+    {
+      "epoch": 0.3771498358023089,
+      "grad_norm": 464.7862153008228,
+      "learning_rate": 1.3946679199538308e-05,
+      "loss": 264.1066,
+      "step": 19610
+    },
+    {
+      "epoch": 0.3773421610627894,
+      "grad_norm": 443.0844224733384,
+      "learning_rate": 1.3941070751982917e-05,
+      "loss": 263.533,
+      "step": 19620
+    },
+    {
+      "epoch": 0.3775344863232699,
+      "grad_norm": 453.0124217089098,
+      "learning_rate": 1.3935460836443758e-05,
+      "loss": 266.8518,
+      "step": 19630
+    },
+    {
+      "epoch": 0.37772681158375043,
+      "grad_norm": 476.2180584075246,
+      "learning_rate": 1.3929849455010433e-05,
+      "loss": 273.0895,
+      "step": 19640
+    },
+    {
+      "epoch": 0.37791913684423095,
+      "grad_norm": 495.1021887537681,
+      "learning_rate": 1.3924236609773094e-05,
+      "loss": 269.4132,
+      "step": 19650
+    },
+    {
+      "epoch": 0.37811146210471147,
+      "grad_norm": 480.3981339495872,
+      "learning_rate": 1.3918622302822425e-05,
+      "loss": 267.0504,
+      "step": 19660
+    },
+    {
+      "epoch": 0.378303787365192,
+      "grad_norm": 460.9074351957358,
+      "learning_rate": 1.391300653624967e-05,
+      "loss": 275.5027,
+      "step": 19670
+    },
+    {
+      "epoch": 0.37849611262567256,
+      "grad_norm": 440.80488409349806,
+      "learning_rate": 1.39073893121466e-05,
+      "loss": 269.0255,
+      "step": 19680
+    },
+    {
+      "epoch": 0.3786884378861531,
+      "grad_norm": 496.1958422879764,
+      "learning_rate": 1.3901770632605546e-05,
+      "loss": 267.1534,
+      "step": 19690
+    },
+    {
+      "epoch": 0.3788807631466336,
+      "grad_norm": 472.26094418744105,
+      "learning_rate": 1.3896150499719372e-05,
+      "loss": 263.1359,
+      "step": 19700
+    },
+    {
+      "epoch": 0.3790730884071141,
+      "grad_norm": 485.96661106237144,
+      "learning_rate": 1.3890528915581482e-05,
+      "loss": 269.3168,
+      "step": 19710
+    },
+    {
+      "epoch": 0.37926541366759464,
+      "grad_norm": 469.9738376511845,
+      "learning_rate": 1.3884905882285829e-05,
+      "loss": 269.5454,
+      "step": 19720
+    },
+    {
+      "epoch": 0.37945773892807516,
+      "grad_norm": 531.450695464184,
+      "learning_rate": 1.3879281401926894e-05,
+      "loss": 263.9205,
+      "step": 19730
+    },
+    {
+      "epoch": 0.3796500641885557,
+      "grad_norm": 475.7810674544225,
+      "learning_rate": 1.3873655476599707e-05,
+      "loss": 264.1166,
+      "step": 19740
+    },
+    {
+      "epoch": 0.3798423894490362,
+      "grad_norm": 459.702197523339,
+      "learning_rate": 1.3868028108399829e-05,
+      "loss": 266.0681,
+      "step": 19750
+    },
+    {
+      "epoch": 0.3800347147095167,
+      "grad_norm": 485.09482582756414,
+      "learning_rate": 1.3862399299423364e-05,
+      "loss": 266.1789,
+      "step": 19760
+    },
+    {
+      "epoch": 0.38022703996999724,
+      "grad_norm": 512.4605800057825,
+      "learning_rate": 1.3856769051766947e-05,
+      "loss": 267.8942,
+      "step": 19770
+    },
+    {
+      "epoch": 0.38041936523047776,
+      "grad_norm": 455.0610010176158,
+      "learning_rate": 1.3851137367527757e-05,
+      "loss": 257.9983,
+      "step": 19780
+    },
+    {
+      "epoch": 0.38061169049095833,
+      "grad_norm": 494.80515284171105,
+      "learning_rate": 1.3845504248803501e-05,
+      "loss": 270.2606,
+      "step": 19790
+    },
+    {
+      "epoch": 0.38080401575143885,
+      "grad_norm": 463.62877696576777,
+      "learning_rate": 1.383986969769242e-05,
+      "loss": 276.2922,
+      "step": 19800
+    },
+    {
+      "epoch": 0.38099634101191937,
+      "grad_norm": 481.8753010398581,
+      "learning_rate": 1.38342337162933e-05,
+      "loss": 270.7017,
+      "step": 19810
+    },
+    {
+      "epoch": 0.3811886662723999,
+      "grad_norm": 518.2911861398973,
+      "learning_rate": 1.3828596306705442e-05,
+      "loss": 264.0347,
+      "step": 19820
+    },
+    {
+      "epoch": 0.3813809915328804,
+      "grad_norm": 476.25043948059556,
+      "learning_rate": 1.3822957471028693e-05,
+      "loss": 258.8927,
+      "step": 19830
+    },
+    {
+      "epoch": 0.38157331679336093,
+      "grad_norm": 456.0857333807183,
+      "learning_rate": 1.3817317211363422e-05,
+      "loss": 259.2476,
+      "step": 19840
+    },
+    {
+      "epoch": 0.38176564205384145,
+      "grad_norm": 591.7758307029712,
+      "learning_rate": 1.3811675529810535e-05,
+      "loss": 264.3246,
+      "step": 19850
+    },
+    {
+      "epoch": 0.38195796731432197,
+      "grad_norm": 480.1235787290731,
+      "learning_rate": 1.3806032428471463e-05,
+      "loss": 267.5823,
+      "step": 19860
+    },
+    {
+      "epoch": 0.3821502925748025,
+      "grad_norm": 490.12823640205306,
+      "learning_rate": 1.3800387909448171e-05,
+      "loss": 268.1071,
+      "step": 19870
+    },
+    {
+      "epoch": 0.382342617835283,
+      "grad_norm": 482.5637101031617,
+      "learning_rate": 1.3794741974843154e-05,
+      "loss": 272.2428,
+      "step": 19880
+    },
+    {
+      "epoch": 0.3825349430957636,
+      "grad_norm": 463.864497799353,
+      "learning_rate": 1.3789094626759419e-05,
+      "loss": 262.553,
+      "step": 19890
+    },
+    {
+      "epoch": 0.3827272683562441,
+      "grad_norm": 451.326551759344,
+      "learning_rate": 1.3783445867300515e-05,
+      "loss": 271.138,
+      "step": 19900
+    },
+    {
+      "epoch": 0.3829195936167246,
+      "grad_norm": 482.5981306829182,
+      "learning_rate": 1.3777795698570511e-05,
+      "loss": 271.6907,
+      "step": 19910
+    },
+    {
+      "epoch": 0.38311191887720514,
+      "grad_norm": 462.3405191740387,
+      "learning_rate": 1.3772144122674e-05,
+      "loss": 261.1419,
+      "step": 19920
+    },
+    {
+      "epoch": 0.38330424413768566,
+      "grad_norm": 456.3619855606467,
+      "learning_rate": 1.3766491141716103e-05,
+      "loss": 263.4425,
+      "step": 19930
+    },
+    {
+      "epoch": 0.3834965693981662,
+      "grad_norm": 470.32262807585965,
+      "learning_rate": 1.3760836757802462e-05,
+      "loss": 267.8558,
+      "step": 19940
+    },
+    {
+      "epoch": 0.3836888946586467,
+      "grad_norm": 458.30893218538426,
+      "learning_rate": 1.3755180973039241e-05,
+      "loss": 269.731,
+      "step": 19950
+    },
+    {
+      "epoch": 0.3838812199191272,
+      "grad_norm": 490.4249697361625,
+      "learning_rate": 1.3749523789533128e-05,
+      "loss": 265.6844,
+      "step": 19960
+    },
+    {
+      "epoch": 0.38407354517960773,
+      "grad_norm": 451.9255863952605,
+      "learning_rate": 1.374386520939133e-05,
+      "loss": 261.8062,
+      "step": 19970
+    },
+    {
+      "epoch": 0.38426587044008825,
+      "grad_norm": 587.6749626221176,
+      "learning_rate": 1.3738205234721568e-05,
+      "loss": 279.9919,
+      "step": 19980
+    },
+    {
+      "epoch": 0.3844581957005688,
+      "grad_norm": 516.8899606675635,
+      "learning_rate": 1.3732543867632098e-05,
+      "loss": 266.4001,
+      "step": 19990
+    },
+    {
+      "epoch": 0.38465052096104935,
+      "grad_norm": 480.5518795120685,
+      "learning_rate": 1.3726881110231682e-05,
+      "loss": 265.4916,
+      "step": 20000
+    },
+    {
+      "epoch": 0.38484284622152987,
+      "grad_norm": 555.3792537867905,
+      "learning_rate": 1.3721216964629605e-05,
+      "loss": 269.9944,
+      "step": 20010
+    },
+    {
+      "epoch": 0.3850351714820104,
+      "grad_norm": 506.87884985064585,
+      "learning_rate": 1.3715551432935664e-05,
+      "loss": 276.7794,
+      "step": 20020
+    },
+    {
+      "epoch": 0.3852274967424909,
+      "grad_norm": 478.03257009871555,
+      "learning_rate": 1.3709884517260178e-05,
+      "loss": 277.7954,
+      "step": 20030
+    },
+    {
+      "epoch": 0.3854198220029714,
+      "grad_norm": 457.637534653076,
+      "learning_rate": 1.3704216219713984e-05,
+      "loss": 260.3147,
+      "step": 20040
+    },
+    {
+      "epoch": 0.38561214726345194,
+      "grad_norm": 515.880634704802,
+      "learning_rate": 1.3698546542408424e-05,
+      "loss": 267.1376,
+      "step": 20050
+    },
+    {
+      "epoch": 0.38580447252393246,
+      "grad_norm": 479.2124703567474,
+      "learning_rate": 1.369287548745536e-05,
+      "loss": 264.1758,
+      "step": 20060
+    },
+    {
+      "epoch": 0.385996797784413,
+      "grad_norm": 683.7181567994502,
+      "learning_rate": 1.3687203056967165e-05,
+      "loss": 269.4074,
+      "step": 20070
+    },
+    {
+      "epoch": 0.3861891230448935,
+      "grad_norm": 546.1825043796105,
+      "learning_rate": 1.368152925305673e-05,
+      "loss": 272.3967,
+      "step": 20080
+    },
+    {
+      "epoch": 0.3863814483053741,
+      "grad_norm": 469.7635754984195,
+      "learning_rate": 1.3675854077837449e-05,
+      "loss": 268.2429,
+      "step": 20090
+    },
+    {
+      "epoch": 0.3865737735658546,
+      "grad_norm": 431.77778742437175,
+      "learning_rate": 1.3670177533423234e-05,
+      "loss": 258.7614,
+      "step": 20100
+    },
+    {
+      "epoch": 0.3867660988263351,
+      "grad_norm": 500.68929215513714,
+      "learning_rate": 1.3664499621928502e-05,
+      "loss": 264.5214,
+      "step": 20110
+    },
+    {
+      "epoch": 0.38695842408681563,
+      "grad_norm": 465.9139099305224,
+      "learning_rate": 1.3658820345468183e-05,
+      "loss": 282.7994,
+      "step": 20120
+    },
+    {
+      "epoch": 0.38715074934729615,
+      "grad_norm": 497.18925828839093,
+      "learning_rate": 1.365313970615771e-05,
+      "loss": 260.5789,
+      "step": 20130
+    },
+    {
+      "epoch": 0.38734307460777667,
+      "grad_norm": 522.7577508611184,
+      "learning_rate": 1.3647457706113031e-05,
+      "loss": 270.7295,
+      "step": 20140
+    },
+    {
+      "epoch": 0.3875353998682572,
+      "grad_norm": 484.58468255015976,
+      "learning_rate": 1.36417743474506e-05,
+      "loss": 271.1447,
+      "step": 20150
+    },
+    {
+      "epoch": 0.3877277251287377,
+      "grad_norm": 455.9399960984987,
+      "learning_rate": 1.3636089632287369e-05,
+      "loss": 278.3718,
+      "step": 20160
+    },
+    {
+      "epoch": 0.38792005038921823,
+      "grad_norm": 438.9810801149823,
+      "learning_rate": 1.36304035627408e-05,
+      "loss": 270.3828,
+      "step": 20170
+    },
+    {
+      "epoch": 0.38811237564969875,
+      "grad_norm": 521.92431656634,
+      "learning_rate": 1.3624716140928861e-05,
+      "loss": 276.9672,
+      "step": 20180
+    },
+    {
+      "epoch": 0.38830470091017927,
+      "grad_norm": 479.2658690926107,
+      "learning_rate": 1.3619027368970025e-05,
+      "loss": 270.019,
+      "step": 20190
+    },
+    {
+      "epoch": 0.38849702617065984,
+      "grad_norm": 539.2950852697795,
+      "learning_rate": 1.3613337248983265e-05,
+      "loss": 264.5474,
+      "step": 20200
+    },
+    {
+      "epoch": 0.38868935143114036,
+      "grad_norm": 521.2340223855009,
+      "learning_rate": 1.3607645783088055e-05,
+      "loss": 269.4898,
+      "step": 20210
+    },
+    {
+      "epoch": 0.3888816766916209,
+      "grad_norm": 478.6268067284272,
+      "learning_rate": 1.3601952973404376e-05,
+      "loss": 274.8288,
+      "step": 20220
+    },
+    {
+      "epoch": 0.3890740019521014,
+      "grad_norm": 452.593598304092,
+      "learning_rate": 1.35962588220527e-05,
+      "loss": 268.0504,
+      "step": 20230
+    },
+    {
+      "epoch": 0.3892663272125819,
+      "grad_norm": 488.63591962836944,
+      "learning_rate": 1.3590563331154008e-05,
+      "loss": 276.3167,
+      "step": 20240
+    },
+    {
+      "epoch": 0.38945865247306244,
+      "grad_norm": 527.0343169877652,
+      "learning_rate": 1.3584866502829774e-05,
+      "loss": 270.3584,
+      "step": 20250
+    },
+    {
+      "epoch": 0.38965097773354296,
+      "grad_norm": 508.6791754239123,
+      "learning_rate": 1.3579168339201975e-05,
+      "loss": 267.6406,
+      "step": 20260
+    },
+    {
+      "epoch": 0.3898433029940235,
+      "grad_norm": 500.0040765716491,
+      "learning_rate": 1.3573468842393077e-05,
+      "loss": 264.213,
+      "step": 20270
+    },
+    {
+      "epoch": 0.390035628254504,
+      "grad_norm": 480.87523411146907,
+      "learning_rate": 1.356776801452606e-05,
+      "loss": 268.1814,
+      "step": 20280
+    },
+    {
+      "epoch": 0.3902279535149845,
+      "grad_norm": 497.87665383882864,
+      "learning_rate": 1.3562065857724378e-05,
+      "loss": 268.274,
+      "step": 20290
+    },
+    {
+      "epoch": 0.3904202787754651,
+      "grad_norm": 517.1153571266282,
+      "learning_rate": 1.3556362374111993e-05,
+      "loss": 264.8947,
+      "step": 20300
+    },
+    {
+      "epoch": 0.3906126040359456,
+      "grad_norm": 516.7694257062676,
+      "learning_rate": 1.3550657565813362e-05,
+      "loss": 271.0774,
+      "step": 20310
+    },
+    {
+      "epoch": 0.39080492929642613,
+      "grad_norm": 471.1677066651554,
+      "learning_rate": 1.3544951434953423e-05,
+      "loss": 261.9783,
+      "step": 20320
+    },
+    {
+      "epoch": 0.39099725455690665,
+      "grad_norm": 475.4164801434219,
+      "learning_rate": 1.3539243983657627e-05,
+      "loss": 266.1654,
+      "step": 20330
+    },
+    {
+      "epoch": 0.39118957981738717,
+      "grad_norm": 465.25438418996094,
+      "learning_rate": 1.3533535214051896e-05,
+      "loss": 265.0667,
+      "step": 20340
+    },
+    {
+      "epoch": 0.3913819050778677,
+      "grad_norm": 636.6946393412486,
+      "learning_rate": 1.3527825128262656e-05,
+      "loss": 266.78,
+      "step": 20350
+    },
+    {
+      "epoch": 0.3915742303383482,
+      "grad_norm": 454.03043826764105,
+      "learning_rate": 1.3522113728416821e-05,
+      "loss": 273.3805,
+      "step": 20360
+    },
+    {
+      "epoch": 0.3917665555988287,
+      "grad_norm": 439.79846213408223,
+      "learning_rate": 1.3516401016641793e-05,
+      "loss": 262.7015,
+      "step": 20370
+    },
+    {
+      "epoch": 0.39195888085930924,
+      "grad_norm": 459.73154660133,
+      "learning_rate": 1.351068699506546e-05,
+      "loss": 267.3195,
+      "step": 20380
+    },
+    {
+      "epoch": 0.39215120611978976,
+      "grad_norm": 461.65266457074955,
+      "learning_rate": 1.3504971665816202e-05,
+      "loss": 263.594,
+      "step": 20390
+    },
+    {
+      "epoch": 0.39234353138027034,
+      "grad_norm": 462.50120916625644,
+      "learning_rate": 1.3499255031022887e-05,
+      "loss": 257.6826,
+      "step": 20400
+    },
+    {
+      "epoch": 0.39253585664075086,
+      "grad_norm": 448.3118307481313,
+      "learning_rate": 1.3493537092814863e-05,
+      "loss": 268.2437,
+      "step": 20410
+    },
+    {
+      "epoch": 0.3927281819012314,
+      "grad_norm": 559.7940647151852,
+      "learning_rate": 1.348781785332197e-05,
+      "loss": 265.3423,
+      "step": 20420
+    },
+    {
+      "epoch": 0.3929205071617119,
+      "grad_norm": 435.63117445349616,
+      "learning_rate": 1.3482097314674526e-05,
+      "loss": 263.3138,
+      "step": 20430
+    },
+    {
+      "epoch": 0.3931128324221924,
+      "grad_norm": 463.97021445369614,
+      "learning_rate": 1.3476375479003347e-05,
+      "loss": 263.2743,
+      "step": 20440
+    },
+    {
+      "epoch": 0.39330515768267293,
+      "grad_norm": 460.0537663900258,
+      "learning_rate": 1.3470652348439715e-05,
+      "loss": 264.744,
+      "step": 20450
+    },
+    {
+      "epoch": 0.39349748294315345,
+      "grad_norm": 479.3908671830001,
+      "learning_rate": 1.3464927925115405e-05,
+      "loss": 277.0105,
+      "step": 20460
+    },
+    {
+      "epoch": 0.393689808203634,
+      "grad_norm": 481.4851262904071,
+      "learning_rate": 1.3459202211162663e-05,
+      "loss": 263.0484,
+      "step": 20470
+    },
+    {
+      "epoch": 0.3938821334641145,
+      "grad_norm": 503.2751886846145,
+      "learning_rate": 1.345347520871423e-05,
+      "loss": 263.6013,
+      "step": 20480
+    },
+    {
+      "epoch": 0.394074458724595,
+      "grad_norm": 457.05885782521125,
+      "learning_rate": 1.3447746919903318e-05,
+      "loss": 260.8919,
+      "step": 20490
+    },
+    {
+      "epoch": 0.3942667839850756,
+      "grad_norm": 434.93552767656496,
+      "learning_rate": 1.3442017346863618e-05,
+      "loss": 266.9873,
+      "step": 20500
+    },
+    {
+      "epoch": 0.3944591092455561,
+      "grad_norm": 491.80384884722565,
+      "learning_rate": 1.3436286491729306e-05,
+      "loss": 273.6724,
+      "step": 20510
+    },
+    {
+      "epoch": 0.3946514345060366,
+      "grad_norm": 496.57294155739976,
+      "learning_rate": 1.3430554356635029e-05,
+      "loss": 268.6439,
+      "step": 20520
+    },
+    {
+      "epoch": 0.39484375976651714,
+      "grad_norm": 445.4804759916785,
+      "learning_rate": 1.342482094371591e-05,
+      "loss": 261.7515,
+      "step": 20530
+    },
+    {
+      "epoch": 0.39503608502699766,
+      "grad_norm": 449.89297898819564,
+      "learning_rate": 1.341908625510755e-05,
+      "loss": 273.4439,
+      "step": 20540
+    },
+    {
+      "epoch": 0.3952284102874782,
+      "grad_norm": 463.5409245133629,
+      "learning_rate": 1.341335029294603e-05,
+      "loss": 261.5422,
+      "step": 20550
+    },
+    {
+      "epoch": 0.3954207355479587,
+      "grad_norm": 468.96643929554153,
+      "learning_rate": 1.3407613059367898e-05,
+      "loss": 264.3,
+      "step": 20560
+    },
+    {
+      "epoch": 0.3956130608084392,
+      "grad_norm": 457.0973214500332,
+      "learning_rate": 1.3401874556510182e-05,
+      "loss": 264.3025,
+      "step": 20570
+    },
+    {
+      "epoch": 0.39580538606891974,
+      "grad_norm": 498.0662983959548,
+      "learning_rate": 1.3396134786510375e-05,
+      "loss": 269.4016,
+      "step": 20580
+    },
+    {
+      "epoch": 0.39599771132940026,
+      "grad_norm": 472.88080784831743,
+      "learning_rate": 1.3390393751506452e-05,
+      "loss": 265.9489,
+      "step": 20590
+    },
+    {
+      "epoch": 0.39619003658988083,
+      "grad_norm": 438.3456069369478,
+      "learning_rate": 1.338465145363685e-05,
+      "loss": 259.0075,
+      "step": 20600
+    },
+    {
+      "epoch": 0.39638236185036135,
+      "grad_norm": 489.050290751274,
+      "learning_rate": 1.3378907895040478e-05,
+      "loss": 267.6819,
+      "step": 20610
+    },
+    {
+      "epoch": 0.3965746871108419,
+      "grad_norm": 481.1230216655527,
+      "learning_rate": 1.3373163077856724e-05,
+      "loss": 261.676,
+      "step": 20620
+    },
+    {
+      "epoch": 0.3967670123713224,
+      "grad_norm": 477.93123040190807,
+      "learning_rate": 1.3367417004225429e-05,
+      "loss": 268.3795,
+      "step": 20630
+    },
+    {
+      "epoch": 0.3969593376318029,
+      "grad_norm": 489.0123489085017,
+      "learning_rate": 1.3361669676286919e-05,
+      "loss": 268.391,
+      "step": 20640
+    },
+    {
+      "epoch": 0.39715166289228343,
+      "grad_norm": 456.9461427745397,
+      "learning_rate": 1.335592109618197e-05,
+      "loss": 254.8374,
+      "step": 20650
+    },
+    {
+      "epoch": 0.39734398815276395,
+      "grad_norm": 468.7076295634319,
+      "learning_rate": 1.335017126605184e-05,
+      "loss": 269.8656,
+      "step": 20660
+    },
+    {
+      "epoch": 0.39753631341324447,
+      "grad_norm": 416.20876899899446,
+      "learning_rate": 1.3344420188038243e-05,
+      "loss": 259.3088,
+      "step": 20670
+    },
+    {
+      "epoch": 0.397728638673725,
+      "grad_norm": 457.12126878500476,
+      "learning_rate": 1.333866786428336e-05,
+      "loss": 267.4308,
+      "step": 20680
+    },
+    {
+      "epoch": 0.3979209639342055,
+      "grad_norm": 496.816954787181,
+      "learning_rate": 1.3332914296929838e-05,
+      "loss": 271.6264,
+      "step": 20690
+    },
+    {
+      "epoch": 0.398113289194686,
+      "grad_norm": 494.4445362407307,
+      "learning_rate": 1.3327159488120784e-05,
+      "loss": 263.8834,
+      "step": 20700
+    },
+    {
+      "epoch": 0.3983056144551666,
+      "grad_norm": 499.98248915990706,
+      "learning_rate": 1.3321403439999775e-05,
+      "loss": 265.676,
+      "step": 20710
+    },
+    {
+      "epoch": 0.3984979397156471,
+      "grad_norm": 513.2703757254474,
+      "learning_rate": 1.3315646154710835e-05,
+      "loss": 271.5798,
+      "step": 20720
+    },
+    {
+      "epoch": 0.39869026497612764,
+      "grad_norm": 487.88181581667743,
+      "learning_rate": 1.3309887634398466e-05,
+      "loss": 272.9735,
+      "step": 20730
+    },
+    {
+      "epoch": 0.39888259023660816,
+      "grad_norm": 455.8014438117221,
+      "learning_rate": 1.3304127881207614e-05,
+      "loss": 260.4117,
+      "step": 20740
+    },
+    {
+      "epoch": 0.3990749154970887,
+      "grad_norm": 508.33662467425575,
+      "learning_rate": 1.3298366897283697e-05,
+      "loss": 270.7414,
+      "step": 20750
+    },
+    {
+      "epoch": 0.3992672407575692,
+      "grad_norm": 467.02838435024915,
+      "learning_rate": 1.3292604684772585e-05,
+      "loss": 263.7842,
+      "step": 20760
+    },
+    {
+      "epoch": 0.3994595660180497,
+      "grad_norm": 476.0550107206296,
+      "learning_rate": 1.3286841245820605e-05,
+      "loss": 262.7772,
+      "step": 20770
+    },
+    {
+      "epoch": 0.39965189127853024,
+      "grad_norm": 449.47408887229426,
+      "learning_rate": 1.3281076582574548e-05,
+      "loss": 262.4583,
+      "step": 20780
+    },
+    {
+      "epoch": 0.39984421653901075,
+      "grad_norm": 474.12126929879054,
+      "learning_rate": 1.3275310697181652e-05,
+      "loss": 261.6002,
+      "step": 20790
+    },
+    {
+      "epoch": 0.4000365417994913,
+      "grad_norm": 488.14956613833726,
+      "learning_rate": 1.3269543591789616e-05,
+      "loss": 258.3711,
+      "step": 20800
+    },
+    {
+      "epoch": 0.40022886705997185,
+      "grad_norm": 481.0221593212846,
+      "learning_rate": 1.3263775268546588e-05,
+      "loss": 266.2099,
+      "step": 20810
+    },
+    {
+      "epoch": 0.40042119232045237,
+      "grad_norm": 452.76960322492914,
+      "learning_rate": 1.3258005729601178e-05,
+      "loss": 266.4222,
+      "step": 20820
+    },
+    {
+      "epoch": 0.4006135175809329,
+      "grad_norm": 501.9836865809758,
+      "learning_rate": 1.325223497710244e-05,
+      "loss": 269.5421,
+      "step": 20830
+    },
+    {
+      "epoch": 0.4008058428414134,
+      "grad_norm": 440.709245488314,
+      "learning_rate": 1.3246463013199882e-05,
+      "loss": 270.561,
+      "step": 20840
+    },
+    {
+      "epoch": 0.4009981681018939,
+      "grad_norm": 523.5778750164358,
+      "learning_rate": 1.3240689840043475e-05,
+      "loss": 265.8489,
+      "step": 20850
+    },
+    {
+      "epoch": 0.40119049336237445,
+      "grad_norm": 511.2207268835909,
+      "learning_rate": 1.323491545978362e-05,
+      "loss": 262.5227,
+      "step": 20860
+    },
+    {
+      "epoch": 0.40138281862285496,
+      "grad_norm": 484.05020111853077,
+      "learning_rate": 1.3229139874571186e-05,
+      "loss": 262.5708,
+      "step": 20870
+    },
+    {
+      "epoch": 0.4015751438833355,
+      "grad_norm": 497.8572817887082,
+      "learning_rate": 1.3223363086557477e-05,
+      "loss": 260.0307,
+      "step": 20880
+    },
+    {
+      "epoch": 0.401767469143816,
+      "grad_norm": 476.81565277174053,
+      "learning_rate": 1.3217585097894255e-05,
+      "loss": 253.8691,
+      "step": 20890
+    },
+    {
+      "epoch": 0.4019597944042965,
+      "grad_norm": 462.57048442453475,
+      "learning_rate": 1.3211805910733724e-05,
+      "loss": 269.6239,
+      "step": 20900
+    },
+    {
+      "epoch": 0.4021521196647771,
+      "grad_norm": 416.18844114101756,
+      "learning_rate": 1.3206025527228535e-05,
+      "loss": 262.9443,
+      "step": 20910
+    },
+    {
+      "epoch": 0.4023444449252576,
+      "grad_norm": 541.2801206050265,
+      "learning_rate": 1.3200243949531788e-05,
+      "loss": 268.8235,
+      "step": 20920
+    },
+    {
+      "epoch": 0.40253677018573814,
+      "grad_norm": 492.4633272315584,
+      "learning_rate": 1.3194461179797023e-05,
+      "loss": 260.9503,
+      "step": 20930
+    },
+    {
+      "epoch": 0.40272909544621865,
+      "grad_norm": 463.147895519567,
+      "learning_rate": 1.3188677220178225e-05,
+      "loss": 252.2378,
+      "step": 20940
+    },
+    {
+      "epoch": 0.4029214207066992,
+      "grad_norm": 490.0795226107119,
+      "learning_rate": 1.3182892072829828e-05,
+      "loss": 253.5228,
+      "step": 20950
+    },
+    {
+      "epoch": 0.4031137459671797,
+      "grad_norm": 807.586274480264,
+      "learning_rate": 1.3177105739906702e-05,
+      "loss": 254.9583,
+      "step": 20960
+    },
+    {
+      "epoch": 0.4033060712276602,
+      "grad_norm": 463.13335147863427,
+      "learning_rate": 1.3171318223564156e-05,
+      "loss": 261.9763,
+      "step": 20970
+    },
+    {
+      "epoch": 0.40349839648814073,
+      "grad_norm": 433.7243442154332,
+      "learning_rate": 1.3165529525957947e-05,
+      "loss": 265.0036,
+      "step": 20980
+    },
+    {
+      "epoch": 0.40369072174862125,
+      "grad_norm": 465.2932577637766,
+      "learning_rate": 1.3159739649244271e-05,
+      "loss": 265.3626,
+      "step": 20990
+    },
+    {
+      "epoch": 0.40388304700910177,
+      "grad_norm": 499.9595149596224,
+      "learning_rate": 1.3153948595579764e-05,
+      "loss": 266.7549,
+      "step": 21000
+    },
+    {
+      "epoch": 0.40407537226958234,
+      "grad_norm": 447.17264132416426,
+      "learning_rate": 1.3148156367121491e-05,
+      "loss": 268.7102,
+      "step": 21010
+    },
+    {
+      "epoch": 0.40426769753006286,
+      "grad_norm": 505.2112041051111,
+      "learning_rate": 1.3142362966026967e-05,
+      "loss": 262.4011,
+      "step": 21020
+    },
+    {
+      "epoch": 0.4044600227905434,
+      "grad_norm": 481.5070474788416,
+      "learning_rate": 1.3136568394454136e-05,
+      "loss": 263.7487,
+      "step": 21030
+    },
+    {
+      "epoch": 0.4046523480510239,
+      "grad_norm": 434.4098885621158,
+      "learning_rate": 1.313077265456138e-05,
+      "loss": 260.0427,
+      "step": 21040
+    },
+    {
+      "epoch": 0.4048446733115044,
+      "grad_norm": 504.62258472747294,
+      "learning_rate": 1.3124975748507514e-05,
+      "loss": 267.7885,
+      "step": 21050
+    },
+    {
+      "epoch": 0.40503699857198494,
+      "grad_norm": 540.5160667691246,
+      "learning_rate": 1.3119177678451793e-05,
+      "loss": 271.2426,
+      "step": 21060
+    },
+    {
+      "epoch": 0.40522932383246546,
+      "grad_norm": 462.26980614363873,
+      "learning_rate": 1.3113378446553903e-05,
+      "loss": 269.1644,
+      "step": 21070
+    },
+    {
+      "epoch": 0.405421649092946,
+      "grad_norm": 487.8128331673341,
+      "learning_rate": 1.3107578054973962e-05,
+      "loss": 265.5426,
+      "step": 21080
+    },
+    {
+      "epoch": 0.4056139743534265,
+      "grad_norm": 468.65292747516736,
+      "learning_rate": 1.3101776505872516e-05,
+      "loss": 263.9606,
+      "step": 21090
+    },
+    {
+      "epoch": 0.405806299613907,
+      "grad_norm": 538.382125547612,
+      "learning_rate": 1.309597380141055e-05,
+      "loss": 262.1304,
+      "step": 21100
+    },
+    {
+      "epoch": 0.40599862487438754,
+      "grad_norm": 452.34882209656166,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 261.6204,
+      "step": 21110
+    },
+    {
+      "epoch": 0.4061909501348681,
+      "grad_norm": 465.4738287303043,
+      "learning_rate": 1.3084364935051132e-05,
+      "loss": 261.3079,
+      "step": 21120
+    },
+    {
+      "epoch": 0.40638327539534863,
+      "grad_norm": 498.179225067857,
+      "learning_rate": 1.307855877747779e-05,
+      "loss": 267.8903,
+      "step": 21130
+    },
+    {
+      "epoch": 0.40657560065582915,
+      "grad_norm": 439.4053969868385,
+      "learning_rate": 1.307275147319215e-05,
+      "loss": 254.3707,
+      "step": 21140
+    },
+    {
+      "epoch": 0.40676792591630967,
+      "grad_norm": 472.626843666605,
+      "learning_rate": 1.3066943024357333e-05,
+      "loss": 254.7389,
+      "step": 21150
+    },
+    {
+      "epoch": 0.4069602511767902,
+      "grad_norm": 494.5128574772442,
+      "learning_rate": 1.306113343313689e-05,
+      "loss": 265.8481,
+      "step": 21160
+    },
+    {
+      "epoch": 0.4071525764372707,
+      "grad_norm": 478.0834604506982,
+      "learning_rate": 1.3055322701694801e-05,
+      "loss": 268.7855,
+      "step": 21170
+    },
+    {
+      "epoch": 0.4073449016977512,
+      "grad_norm": 514.8479937865139,
+      "learning_rate": 1.3049510832195466e-05,
+      "loss": 268.5113,
+      "step": 21180
+    },
+    {
+      "epoch": 0.40753722695823175,
+      "grad_norm": 456.1037000944643,
+      "learning_rate": 1.3043697826803707e-05,
+      "loss": 259.2839,
+      "step": 21190
+    },
+    {
+      "epoch": 0.40772955221871227,
+      "grad_norm": 485.5182817775172,
+      "learning_rate": 1.303788368768478e-05,
+      "loss": 252.2747,
+      "step": 21200
+    },
+    {
+      "epoch": 0.4079218774791928,
+      "grad_norm": 504.7148204922667,
+      "learning_rate": 1.3032068417004351e-05,
+      "loss": 266.3797,
+      "step": 21210
+    },
+    {
+      "epoch": 0.40811420273967336,
+      "grad_norm": 469.8707805820898,
+      "learning_rate": 1.302625201692851e-05,
+      "loss": 262.4649,
+      "step": 21220
+    },
+    {
+      "epoch": 0.4083065280001539,
+      "grad_norm": 441.98200994127427,
+      "learning_rate": 1.302043448962378e-05,
+      "loss": 257.2601,
+      "step": 21230
+    },
+    {
+      "epoch": 0.4084988532606344,
+      "grad_norm": 478.50193947638127,
+      "learning_rate": 1.301461583725708e-05,
+      "loss": 267.2607,
+      "step": 21240
+    },
+    {
+      "epoch": 0.4086911785211149,
+      "grad_norm": 631.6467442795822,
+      "learning_rate": 1.3008796061995772e-05,
+      "loss": 266.5208,
+      "step": 21250
+    },
+    {
+      "epoch": 0.40888350378159544,
+      "grad_norm": 482.0852494397625,
+      "learning_rate": 1.3002975166007618e-05,
+      "loss": 258.947,
+      "step": 21260
+    },
+    {
+      "epoch": 0.40907582904207596,
+      "grad_norm": 528.167999492159,
+      "learning_rate": 1.2997153151460814e-05,
+      "loss": 261.8091,
+      "step": 21270
+    },
+    {
+      "epoch": 0.4092681543025565,
+      "grad_norm": 461.33456620866116,
+      "learning_rate": 1.299133002052396e-05,
+      "loss": 259.8377,
+      "step": 21280
+    },
+    {
+      "epoch": 0.409460479563037,
+      "grad_norm": 433.35539917068337,
+      "learning_rate": 1.2985505775366079e-05,
+      "loss": 264.6542,
+      "step": 21290
+    },
+    {
+      "epoch": 0.4096528048235175,
+      "grad_norm": 482.49429400187677,
+      "learning_rate": 1.2979680418156604e-05,
+      "loss": 271.3772,
+      "step": 21300
+    },
+    {
+      "epoch": 0.40984513008399803,
+      "grad_norm": 569.2518002103108,
+      "learning_rate": 1.297385395106538e-05,
+      "loss": 265.4428,
+      "step": 21310
+    },
+    {
+      "epoch": 0.4100374553444786,
+      "grad_norm": 522.7769096559927,
+      "learning_rate": 1.2968026376262679e-05,
+      "loss": 251.5035,
+      "step": 21320
+    },
+    {
+      "epoch": 0.4102297806049591,
+      "grad_norm": 519.5724509483889,
+      "learning_rate": 1.2962197695919167e-05,
+      "loss": 268.33,
+      "step": 21330
+    },
+    {
+      "epoch": 0.41042210586543965,
+      "grad_norm": 450.5461123970926,
+      "learning_rate": 1.2956367912205939e-05,
+      "loss": 264.3903,
+      "step": 21340
+    },
+    {
+      "epoch": 0.41061443112592017,
+      "grad_norm": 428.5107754836988,
+      "learning_rate": 1.2950537027294487e-05,
+      "loss": 258.1454,
+      "step": 21350
+    },
+    {
+      "epoch": 0.4108067563864007,
+      "grad_norm": 498.8952887293403,
+      "learning_rate": 1.2944705043356722e-05,
+      "loss": 262.4074,
+      "step": 21360
+    },
+    {
+      "epoch": 0.4109990816468812,
+      "grad_norm": 451.5979215755134,
+      "learning_rate": 1.2938871962564965e-05,
+      "loss": 257.2829,
+      "step": 21370
+    },
+    {
+      "epoch": 0.4111914069073617,
+      "grad_norm": 549.7022528597881,
+      "learning_rate": 1.2933037787091935e-05,
+      "loss": 268.7281,
+      "step": 21380
+    },
+    {
+      "epoch": 0.41138373216784224,
+      "grad_norm": 540.6518831658659,
+      "learning_rate": 1.2927202519110775e-05,
+      "loss": 271.0587,
+      "step": 21390
+    },
+    {
+      "epoch": 0.41157605742832276,
+      "grad_norm": 466.1715980949634,
+      "learning_rate": 1.2921366160795017e-05,
+      "loss": 261.0138,
+      "step": 21400
+    },
+    {
+      "epoch": 0.4117683826888033,
+      "grad_norm": 417.2243875517313,
+      "learning_rate": 1.2915528714318612e-05,
+      "loss": 267.694,
+      "step": 21410
+    },
+    {
+      "epoch": 0.41196070794928386,
+      "grad_norm": 442.2236286965045,
+      "learning_rate": 1.2909690181855914e-05,
+      "loss": 259.7224,
+      "step": 21420
+    },
+    {
+      "epoch": 0.4121530332097644,
+      "grad_norm": 471.0412096039758,
+      "learning_rate": 1.290385056558168e-05,
+      "loss": 261.5262,
+      "step": 21430
+    },
+    {
+      "epoch": 0.4123453584702449,
+      "grad_norm": 440.8680488355472,
+      "learning_rate": 1.2898009867671066e-05,
+      "loss": 261.603,
+      "step": 21440
+    },
+    {
+      "epoch": 0.4125376837307254,
+      "grad_norm": 493.29723829932004,
+      "learning_rate": 1.2892168090299639e-05,
+      "loss": 256.7897,
+      "step": 21450
+    },
+    {
+      "epoch": 0.41273000899120593,
+      "grad_norm": 503.248727037316,
+      "learning_rate": 1.2886325235643367e-05,
+      "loss": 258.0366,
+      "step": 21460
+    },
+    {
+      "epoch": 0.41292233425168645,
+      "grad_norm": 441.1032360486992,
+      "learning_rate": 1.288048130587861e-05,
+      "loss": 256.1215,
+      "step": 21470
+    },
+    {
+      "epoch": 0.41311465951216697,
+      "grad_norm": 473.53679642492835,
+      "learning_rate": 1.287463630318214e-05,
+      "loss": 261.4138,
+      "step": 21480
+    },
+    {
+      "epoch": 0.4133069847726475,
+      "grad_norm": 589.302144938101,
+      "learning_rate": 1.2868790229731123e-05,
+      "loss": 262.8046,
+      "step": 21490
+    },
+    {
+      "epoch": 0.413499310033128,
+      "grad_norm": 550.2172648288438,
+      "learning_rate": 1.2862943087703127e-05,
+      "loss": 259.2098,
+      "step": 21500
+    },
+    {
+      "epoch": 0.41369163529360853,
+      "grad_norm": 478.9256732741642,
+      "learning_rate": 1.2857094879276115e-05,
+      "loss": 266.143,
+      "step": 21510
+    },
+    {
+      "epoch": 0.4138839605540891,
+      "grad_norm": 589.5836595002756,
+      "learning_rate": 1.2851245606628447e-05,
+      "loss": 262.937,
+      "step": 21520
+    },
+    {
+      "epoch": 0.4140762858145696,
+      "grad_norm": 494.61229077440214,
+      "learning_rate": 1.2845395271938876e-05,
+      "loss": 259.5427,
+      "step": 21530
+    },
+    {
+      "epoch": 0.41426861107505014,
+      "grad_norm": 547.5338493808231,
+      "learning_rate": 1.2839543877386562e-05,
+      "loss": 255.1824,
+      "step": 21540
+    },
+    {
+      "epoch": 0.41446093633553066,
+      "grad_norm": 528.3897621747665,
+      "learning_rate": 1.283369142515105e-05,
+      "loss": 264.4911,
+      "step": 21550
+    },
+    {
+      "epoch": 0.4146532615960112,
+      "grad_norm": 528.8073980332322,
+      "learning_rate": 1.282783791741228e-05,
+      "loss": 271.3275,
+      "step": 21560
+    },
+    {
+      "epoch": 0.4148455868564917,
+      "grad_norm": 1853.9268192141426,
+      "learning_rate": 1.2821983356350593e-05,
+      "loss": 280.1635,
+      "step": 21570
+    },
+    {
+      "epoch": 0.4150379121169722,
+      "grad_norm": 472.4090220473149,
+      "learning_rate": 1.2816127744146711e-05,
+      "loss": 261.4486,
+      "step": 21580
+    },
+    {
+      "epoch": 0.41523023737745274,
+      "grad_norm": 466.8237635234257,
+      "learning_rate": 1.2810271082981755e-05,
+      "loss": 261.6149,
+      "step": 21590
+    },
+    {
+      "epoch": 0.41542256263793326,
+      "grad_norm": 459.7344290567423,
+      "learning_rate": 1.2804413375037232e-05,
+      "loss": 256.9443,
+      "step": 21600
+    },
+    {
+      "epoch": 0.4156148878984138,
+      "grad_norm": 526.0563777738271,
+      "learning_rate": 1.2798554622495042e-05,
+      "loss": 260.6255,
+      "step": 21610
+    },
+    {
+      "epoch": 0.4158072131588943,
+      "grad_norm": 506.1385356690671,
+      "learning_rate": 1.2792694827537477e-05,
+      "loss": 263.7782,
+      "step": 21620
+    },
+    {
+      "epoch": 0.41599953841937487,
+      "grad_norm": 437.80211070436144,
+      "learning_rate": 1.2786833992347212e-05,
+      "loss": 258.7773,
+      "step": 21630
+    },
+    {
+      "epoch": 0.4161918636798554,
+      "grad_norm": 414.17361030926605,
+      "learning_rate": 1.2780972119107312e-05,
+      "loss": 263.2555,
+      "step": 21640
+    },
+    {
+      "epoch": 0.4163841889403359,
+      "grad_norm": 430.04944197431513,
+      "learning_rate": 1.2775109210001225e-05,
+      "loss": 254.02,
+      "step": 21650
+    },
+    {
+      "epoch": 0.41657651420081643,
+      "grad_norm": 502.979970285786,
+      "learning_rate": 1.2769245267212789e-05,
+      "loss": 260.4777,
+      "step": 21660
+    },
+    {
+      "epoch": 0.41676883946129695,
+      "grad_norm": 504.3363926046776,
+      "learning_rate": 1.2763380292926227e-05,
+      "loss": 259.9194,
+      "step": 21670
+    },
+    {
+      "epoch": 0.41696116472177747,
+      "grad_norm": 481.40444777783983,
+      "learning_rate": 1.2757514289326146e-05,
+      "loss": 257.8855,
+      "step": 21680
+    },
+    {
+      "epoch": 0.417153489982258,
+      "grad_norm": 471.2924993212668,
+      "learning_rate": 1.2751647258597531e-05,
+      "loss": 259.6434,
+      "step": 21690
+    },
+    {
+      "epoch": 0.4173458152427385,
+      "grad_norm": 444.72767899018174,
+      "learning_rate": 1.274577920292576e-05,
+      "loss": 250.2615,
+      "step": 21700
+    },
+    {
+      "epoch": 0.417538140503219,
+      "grad_norm": 469.30635449947005,
+      "learning_rate": 1.2739910124496585e-05,
+      "loss": 263.5958,
+      "step": 21710
+    },
+    {
+      "epoch": 0.41773046576369954,
+      "grad_norm": 470.64646428910925,
+      "learning_rate": 1.2734040025496137e-05,
+      "loss": 255.7097,
+      "step": 21720
+    },
+    {
+      "epoch": 0.4179227910241801,
+      "grad_norm": 456.6113086401947,
+      "learning_rate": 1.2728168908110937e-05,
+      "loss": 258.9102,
+      "step": 21730
+    },
+    {
+      "epoch": 0.41811511628466064,
+      "grad_norm": 463.02709316818124,
+      "learning_rate": 1.2722296774527871e-05,
+      "loss": 251.8326,
+      "step": 21740
+    },
+    {
+      "epoch": 0.41830744154514116,
+      "grad_norm": 425.60263211467986,
+      "learning_rate": 1.271642362693422e-05,
+      "loss": 270.1656,
+      "step": 21750
+    },
+    {
+      "epoch": 0.4184997668056217,
+      "grad_norm": 473.88188215245344,
+      "learning_rate": 1.2710549467517628e-05,
+      "loss": 257.0249,
+      "step": 21760
+    },
+    {
+      "epoch": 0.4186920920661022,
+      "grad_norm": 433.8415365562208,
+      "learning_rate": 1.270467429846613e-05,
+      "loss": 246.6032,
+      "step": 21770
+    },
+    {
+      "epoch": 0.4188844173265827,
+      "grad_norm": 453.1147717742959,
+      "learning_rate": 1.2698798121968122e-05,
+      "loss": 256.7895,
+      "step": 21780
+    },
+    {
+      "epoch": 0.41907674258706323,
+      "grad_norm": 453.940042470423,
+      "learning_rate": 1.2692920940212387e-05,
+      "loss": 267.6017,
+      "step": 21790
+    },
+    {
+      "epoch": 0.41926906784754375,
+      "grad_norm": 422.5455373258571,
+      "learning_rate": 1.2687042755388077e-05,
+      "loss": 253.5791,
+      "step": 21800
+    },
+    {
+      "epoch": 0.41946139310802427,
+      "grad_norm": 450.2053775453577,
+      "learning_rate": 1.2681163569684718e-05,
+      "loss": 262.1306,
+      "step": 21810
+    },
+    {
+      "epoch": 0.4196537183685048,
+      "grad_norm": 451.0061372027502,
+      "learning_rate": 1.2675283385292212e-05,
+      "loss": 261.4113,
+      "step": 21820
+    },
+    {
+      "epoch": 0.41984604362898537,
+      "grad_norm": 481.0762192391593,
+      "learning_rate": 1.2669402204400825e-05,
+      "loss": 254.4698,
+      "step": 21830
+    },
+    {
+      "epoch": 0.4200383688894659,
+      "grad_norm": 469.2880736612743,
+      "learning_rate": 1.2663520029201206e-05,
+      "loss": 251.548,
+      "step": 21840
+    },
+    {
+      "epoch": 0.4202306941499464,
+      "grad_norm": 480.6530382268694,
+      "learning_rate": 1.2657636861884363e-05,
+      "loss": 256.3476,
+      "step": 21850
+    },
+    {
+      "epoch": 0.4204230194104269,
+      "grad_norm": 503.05366976347256,
+      "learning_rate": 1.2651752704641686e-05,
+      "loss": 259.4078,
+      "step": 21860
+    },
+    {
+      "epoch": 0.42061534467090744,
+      "grad_norm": 438.03418759023884,
+      "learning_rate": 1.2645867559664918e-05,
+      "loss": 261.3653,
+      "step": 21870
+    },
+    {
+      "epoch": 0.42080766993138796,
+      "grad_norm": 415.9489057170127,
+      "learning_rate": 1.2639981429146184e-05,
+      "loss": 250.693,
+      "step": 21880
+    },
+    {
+      "epoch": 0.4209999951918685,
+      "grad_norm": 475.3108477134088,
+      "learning_rate": 1.2634094315277967e-05,
+      "loss": 256.4124,
+      "step": 21890
+    },
+    {
+      "epoch": 0.421192320452349,
+      "grad_norm": 500.94492016642886,
+      "learning_rate": 1.262820622025312e-05,
+      "loss": 265.2192,
+      "step": 21900
+    },
+    {
+      "epoch": 0.4213846457128295,
+      "grad_norm": 531.1664146710855,
+      "learning_rate": 1.2622317146264864e-05,
+      "loss": 262.586,
+      "step": 21910
+    },
+    {
+      "epoch": 0.42157697097331004,
+      "grad_norm": 461.37948060079873,
+      "learning_rate": 1.2616427095506779e-05,
+      "loss": 261.5061,
+      "step": 21920
+    },
+    {
+      "epoch": 0.4217692962337906,
+      "grad_norm": 427.53926808009055,
+      "learning_rate": 1.2610536070172815e-05,
+      "loss": 261.1253,
+      "step": 21930
+    },
+    {
+      "epoch": 0.42196162149427113,
+      "grad_norm": 476.7539894271758,
+      "learning_rate": 1.2604644072457277e-05,
+      "loss": 256.7359,
+      "step": 21940
+    },
+    {
+      "epoch": 0.42215394675475165,
+      "grad_norm": 464.2967261014823,
+      "learning_rate": 1.259875110455484e-05,
+      "loss": 260.7578,
+      "step": 21950
+    },
+    {
+      "epoch": 0.42234627201523217,
+      "grad_norm": 449.9215572169176,
+      "learning_rate": 1.2592857168660535e-05,
+      "loss": 259.3151,
+      "step": 21960
+    },
+    {
+      "epoch": 0.4225385972757127,
+      "grad_norm": 469.06798265730447,
+      "learning_rate": 1.2586962266969758e-05,
+      "loss": 261.4145,
+      "step": 21970
+    },
+    {
+      "epoch": 0.4227309225361932,
+      "grad_norm": 521.4087708846176,
+      "learning_rate": 1.2581066401678261e-05,
+      "loss": 271.0152,
+      "step": 21980
+    },
+    {
+      "epoch": 0.42292324779667373,
+      "grad_norm": 445.9083651602901,
+      "learning_rate": 1.2575169574982158e-05,
+      "loss": 256.6363,
+      "step": 21990
+    },
+    {
+      "epoch": 0.42311557305715425,
+      "grad_norm": 440.7079893689907,
+      "learning_rate": 1.2569271789077919e-05,
+      "loss": 261.3805,
+      "step": 22000
+    },
+    {
+      "epoch": 0.42330789831763477,
+      "grad_norm": 485.5922587252784,
+      "learning_rate": 1.256337304616237e-05,
+      "loss": 254.0244,
+      "step": 22010
+    },
+    {
+      "epoch": 0.4235002235781153,
+      "grad_norm": 435.75623709112523,
+      "learning_rate": 1.25574733484327e-05,
+      "loss": 254.9253,
+      "step": 22020
+    },
+    {
+      "epoch": 0.4236925488385958,
+      "grad_norm": 499.78357667385853,
+      "learning_rate": 1.2551572698086446e-05,
+      "loss": 262.204,
+      "step": 22030
+    },
+    {
+      "epoch": 0.4238848740990764,
+      "grad_norm": 461.2841644547413,
+      "learning_rate": 1.2545671097321497e-05,
+      "loss": 258.2505,
+      "step": 22040
+    },
+    {
+      "epoch": 0.4240771993595569,
+      "grad_norm": 426.4627395338991,
+      "learning_rate": 1.2539768548336112e-05,
+      "loss": 259.5051,
+      "step": 22050
+    },
+    {
+      "epoch": 0.4242695246200374,
+      "grad_norm": 483.58233623795405,
+      "learning_rate": 1.2533865053328886e-05,
+      "loss": 259.016,
+      "step": 22060
+    },
+    {
+      "epoch": 0.42446184988051794,
+      "grad_norm": 453.30443322323816,
+      "learning_rate": 1.2527960614498778e-05,
+      "loss": 263.4157,
+      "step": 22070
+    },
+    {
+      "epoch": 0.42465417514099846,
+      "grad_norm": 510.33885311252754,
+      "learning_rate": 1.252205523404509e-05,
+      "loss": 264.2783,
+      "step": 22080
+    },
+    {
+      "epoch": 0.424846500401479,
+      "grad_norm": 468.45344532539286,
+      "learning_rate": 1.2516148914167481e-05,
+      "loss": 262.4432,
+      "step": 22090
+    },
+    {
+      "epoch": 0.4250388256619595,
+      "grad_norm": 425.64539208833423,
+      "learning_rate": 1.2510241657065958e-05,
+      "loss": 250.1267,
+      "step": 22100
+    },
+    {
+      "epoch": 0.42523115092244,
+      "grad_norm": 444.8032661433248,
+      "learning_rate": 1.2504333464940874e-05,
+      "loss": 251.268,
+      "step": 22110
+    },
+    {
+      "epoch": 0.42542347618292053,
+      "grad_norm": 2047.4576304833924,
+      "learning_rate": 1.2498424339992934e-05,
+      "loss": 258.2935,
+      "step": 22120
+    },
+    {
+      "epoch": 0.42561580144340105,
+      "grad_norm": 496.2123843311693,
+      "learning_rate": 1.2492514284423194e-05,
+      "loss": 259.7029,
+      "step": 22130
+    },
+    {
+      "epoch": 0.42580812670388163,
+      "grad_norm": 451.31672227430596,
+      "learning_rate": 1.2486603300433045e-05,
+      "loss": 256.3716,
+      "step": 22140
+    },
+    {
+      "epoch": 0.42600045196436215,
+      "grad_norm": 476.5778994566375,
+      "learning_rate": 1.2480691390224238e-05,
+      "loss": 262.5988,
+      "step": 22150
+    },
+    {
+      "epoch": 0.42619277722484267,
+      "grad_norm": 436.23722902174603,
+      "learning_rate": 1.2474778555998855e-05,
+      "loss": 246.5431,
+      "step": 22160
+    },
+    {
+      "epoch": 0.4263851024853232,
+      "grad_norm": 459.69425102316336,
+      "learning_rate": 1.246886479995933e-05,
+      "loss": 262.2735,
+      "step": 22170
+    },
+    {
+      "epoch": 0.4265774277458037,
+      "grad_norm": 455.6661641040029,
+      "learning_rate": 1.2462950124308444e-05,
+      "loss": 261.5882,
+      "step": 22180
+    },
+    {
+      "epoch": 0.4267697530062842,
+      "grad_norm": 503.4685081267895,
+      "learning_rate": 1.2457034531249313e-05,
+      "loss": 257.1507,
+      "step": 22190
+    },
+    {
+      "epoch": 0.42696207826676474,
+      "grad_norm": 481.41008037318215,
+      "learning_rate": 1.2451118022985402e-05,
+      "loss": 267.3524,
+      "step": 22200
+    },
+    {
+      "epoch": 0.42715440352724526,
+      "grad_norm": 689.4060625509051,
+      "learning_rate": 1.2445200601720504e-05,
+      "loss": 250.75,
+      "step": 22210
+    },
+    {
+      "epoch": 0.4273467287877258,
+      "grad_norm": 476.36390421373187,
+      "learning_rate": 1.243928226965877e-05,
+      "loss": 256.4727,
+      "step": 22220
+    },
+    {
+      "epoch": 0.4275390540482063,
+      "grad_norm": 440.12614855325927,
+      "learning_rate": 1.2433363029004675e-05,
+      "loss": 254.9294,
+      "step": 22230
+    },
+    {
+      "epoch": 0.4277313793086869,
+      "grad_norm": 442.62699204163243,
+      "learning_rate": 1.2427442881963042e-05,
+      "loss": 257.3285,
+      "step": 22240
+    },
+    {
+      "epoch": 0.4279237045691674,
+      "grad_norm": 466.74214551272325,
+      "learning_rate": 1.242152183073902e-05,
+      "loss": 260.5592,
+      "step": 22250
+    },
+    {
+      "epoch": 0.4281160298296479,
+      "grad_norm": 471.2808107279145,
+      "learning_rate": 1.2415599877538111e-05,
+      "loss": 250.9756,
+      "step": 22260
+    },
+    {
+      "epoch": 0.42830835509012843,
+      "grad_norm": 441.43787592468067,
+      "learning_rate": 1.2409677024566145e-05,
+      "loss": 265.465,
+      "step": 22270
+    },
+    {
+      "epoch": 0.42850068035060895,
+      "grad_norm": 483.1493834726963,
+      "learning_rate": 1.2403753274029281e-05,
+      "loss": 261.402,
+      "step": 22280
+    },
+    {
+      "epoch": 0.4286930056110895,
+      "grad_norm": 457.95176673999674,
+      "learning_rate": 1.2397828628134028e-05,
+      "loss": 250.1696,
+      "step": 22290
+    },
+    {
+      "epoch": 0.42888533087157,
+      "grad_norm": 476.74012808717566,
+      "learning_rate": 1.2391903089087208e-05,
+      "loss": 257.1226,
+      "step": 22300
+    },
+    {
+      "epoch": 0.4290776561320505,
+      "grad_norm": 485.91918552417684,
+      "learning_rate": 1.2385976659095993e-05,
+      "loss": 255.824,
+      "step": 22310
+    },
+    {
+      "epoch": 0.42926998139253103,
+      "grad_norm": 486.2026801353059,
+      "learning_rate": 1.2380049340367876e-05,
+      "loss": 251.959,
+      "step": 22320
+    },
+    {
+      "epoch": 0.42946230665301155,
+      "grad_norm": 530.8842861874508,
+      "learning_rate": 1.2374121135110688e-05,
+      "loss": 267.2906,
+      "step": 22330
+    },
+    {
+      "epoch": 0.4296546319134921,
+      "grad_norm": 449.3700140095058,
+      "learning_rate": 1.2368192045532586e-05,
+      "loss": 257.396,
+      "step": 22340
+    },
+    {
+      "epoch": 0.42984695717397264,
+      "grad_norm": 499.66813060099645,
+      "learning_rate": 1.236226207384206e-05,
+      "loss": 253.6241,
+      "step": 22350
+    },
+    {
+      "epoch": 0.43003928243445316,
+      "grad_norm": 459.71767235372903,
+      "learning_rate": 1.2356331222247929e-05,
+      "loss": 260.5438,
+      "step": 22360
+    },
+    {
+      "epoch": 0.4302316076949337,
+      "grad_norm": 455.8846773640258,
+      "learning_rate": 1.2350399492959328e-05,
+      "loss": 262.5015,
+      "step": 22370
+    },
+    {
+      "epoch": 0.4304239329554142,
+      "grad_norm": 493.59394894441596,
+      "learning_rate": 1.234446688818574e-05,
+      "loss": 258.3056,
+      "step": 22380
+    },
+    {
+      "epoch": 0.4306162582158947,
+      "grad_norm": 476.04763204624425,
+      "learning_rate": 1.233853341013695e-05,
+      "loss": 255.3356,
+      "step": 22390
+    },
+    {
+      "epoch": 0.43080858347637524,
+      "grad_norm": 449.13623647617436,
+      "learning_rate": 1.233259906102309e-05,
+      "loss": 256.283,
+      "step": 22400
+    },
+    {
+      "epoch": 0.43100090873685576,
+      "grad_norm": 418.47346689864503,
+      "learning_rate": 1.2326663843054603e-05,
+      "loss": 253.7366,
+      "step": 22410
+    },
+    {
+      "epoch": 0.4311932339973363,
+      "grad_norm": 467.19840956139757,
+      "learning_rate": 1.2320727758442264e-05,
+      "loss": 260.7724,
+      "step": 22420
+    },
+    {
+      "epoch": 0.4313855592578168,
+      "grad_norm": 476.0566488083017,
+      "learning_rate": 1.231479080939716e-05,
+      "loss": 265.8249,
+      "step": 22430
+    },
+    {
+      "epoch": 0.43157788451829737,
+      "grad_norm": 426.5182115420512,
+      "learning_rate": 1.230885299813071e-05,
+      "loss": 265.0076,
+      "step": 22440
+    },
+    {
+      "epoch": 0.4317702097787779,
+      "grad_norm": 502.60556109571854,
+      "learning_rate": 1.2302914326854651e-05,
+      "loss": 258.2654,
+      "step": 22450
+    },
+    {
+      "epoch": 0.4319625350392584,
+      "grad_norm": 476.6196111642156,
+      "learning_rate": 1.2296974797781036e-05,
+      "loss": 265.3442,
+      "step": 22460
+    },
+    {
+      "epoch": 0.43215486029973893,
+      "grad_norm": 491.5977996643741,
+      "learning_rate": 1.2291034413122247e-05,
+      "loss": 268.0509,
+      "step": 22470
+    },
+    {
+      "epoch": 0.43234718556021945,
+      "grad_norm": 485.45530561680806,
+      "learning_rate": 1.2285093175090976e-05,
+      "loss": 256.3687,
+      "step": 22480
+    },
+    {
+      "epoch": 0.43253951082069997,
+      "grad_norm": 477.6220432991964,
+      "learning_rate": 1.227915108590024e-05,
+      "loss": 251.5692,
+      "step": 22490
+    },
+    {
+      "epoch": 0.4327318360811805,
+      "grad_norm": 426.1059902665939,
+      "learning_rate": 1.2273208147763363e-05,
+      "loss": 246.4806,
+      "step": 22500
+    },
+    {
+      "epoch": 0.432924161341661,
+      "grad_norm": 474.1517991404114,
+      "learning_rate": 1.2267264362893995e-05,
+      "loss": 258.7537,
+      "step": 22510
+    },
+    {
+      "epoch": 0.4331164866021415,
+      "grad_norm": 470.00063181770497,
+      "learning_rate": 1.2261319733506096e-05,
+      "loss": 257.5305,
+      "step": 22520
+    },
+    {
+      "epoch": 0.43330881186262205,
+      "grad_norm": 493.36629571964477,
+      "learning_rate": 1.2255374261813944e-05,
+      "loss": 264.3011,
+      "step": 22530
+    },
+    {
+      "epoch": 0.43350113712310256,
+      "grad_norm": 456.7249590892431,
+      "learning_rate": 1.2249427950032127e-05,
+      "loss": 261.1499,
+      "step": 22540
+    },
+    {
+      "epoch": 0.43369346238358314,
+      "grad_norm": 506.462238555744,
+      "learning_rate": 1.224348080037555e-05,
+      "loss": 267.5491,
+      "step": 22550
+    },
+    {
+      "epoch": 0.43388578764406366,
+      "grad_norm": 478.62500351555593,
+      "learning_rate": 1.2237532815059427e-05,
+      "loss": 254.4973,
+      "step": 22560
+    },
+    {
+      "epoch": 0.4340781129045442,
+      "grad_norm": 435.6395897127192,
+      "learning_rate": 1.2231583996299285e-05,
+      "loss": 255.2659,
+      "step": 22570
+    },
+    {
+      "epoch": 0.4342704381650247,
+      "grad_norm": 466.68234096018296,
+      "learning_rate": 1.2225634346310962e-05,
+      "loss": 252.281,
+      "step": 22580
+    },
+    {
+      "epoch": 0.4344627634255052,
+      "grad_norm": 444.44283779714925,
+      "learning_rate": 1.22196838673106e-05,
+      "loss": 260.5664,
+      "step": 22590
+    },
+    {
+      "epoch": 0.43465508868598574,
+      "grad_norm": 482.24347012851155,
+      "learning_rate": 1.2213732561514657e-05,
+      "loss": 262.3756,
+      "step": 22600
+    },
+    {
+      "epoch": 0.43484741394646625,
+      "grad_norm": 437.2513456863256,
+      "learning_rate": 1.2207780431139894e-05,
+      "loss": 260.8468,
+      "step": 22610
+    },
+    {
+      "epoch": 0.4350397392069468,
+      "grad_norm": 478.72624998312307,
+      "learning_rate": 1.2201827478403385e-05,
+      "loss": 255.498,
+      "step": 22620
+    },
+    {
+      "epoch": 0.4352320644674273,
+      "grad_norm": 429.7146375248195,
+      "learning_rate": 1.2195873705522508e-05,
+      "loss": 244.75,
+      "step": 22630
+    },
+    {
+      "epoch": 0.4354243897279078,
+      "grad_norm": 530.5027567651761,
+      "learning_rate": 1.2189919114714936e-05,
+      "loss": 247.1865,
+      "step": 22640
+    },
+    {
+      "epoch": 0.4356167149883884,
+      "grad_norm": 478.31770513426164,
+      "learning_rate": 1.2183963708198668e-05,
+      "loss": 265.0552,
+      "step": 22650
+    },
+    {
+      "epoch": 0.4358090402488689,
+      "grad_norm": 457.59647146706635,
+      "learning_rate": 1.2178007488191983e-05,
+      "loss": 253.0527,
+      "step": 22660
+    },
+    {
+      "epoch": 0.4360013655093494,
+      "grad_norm": 443.13810983288755,
+      "learning_rate": 1.2172050456913482e-05,
+      "loss": 259.3701,
+      "step": 22670
+    },
+    {
+      "epoch": 0.43619369076982994,
+      "grad_norm": 459.5323209224299,
+      "learning_rate": 1.2166092616582055e-05,
+      "loss": 247.5747,
+      "step": 22680
+    },
+    {
+      "epoch": 0.43638601603031046,
+      "grad_norm": 492.00659687861616,
+      "learning_rate": 1.2160133969416903e-05,
+      "loss": 264.748,
+      "step": 22690
+    },
+    {
+      "epoch": 0.436578341290791,
+      "grad_norm": 450.67653171967225,
+      "learning_rate": 1.2154174517637526e-05,
+      "loss": 255.2781,
+      "step": 22700
+    },
+    {
+      "epoch": 0.4367706665512715,
+      "grad_norm": 451.8655364691753,
+      "learning_rate": 1.2148214263463718e-05,
+      "loss": 248.3648,
+      "step": 22710
+    },
+    {
+      "epoch": 0.436962991811752,
+      "grad_norm": 468.4579968499324,
+      "learning_rate": 1.2142253209115577e-05,
+      "loss": 256.2649,
+      "step": 22720
+    },
+    {
+      "epoch": 0.43715531707223254,
+      "grad_norm": 450.5044203985601,
+      "learning_rate": 1.2136291356813494e-05,
+      "loss": 265.551,
+      "step": 22730
+    },
+    {
+      "epoch": 0.43734764233271306,
+      "grad_norm": 466.91794975899256,
+      "learning_rate": 1.2130328708778162e-05,
+      "loss": 255.8264,
+      "step": 22740
+    },
+    {
+      "epoch": 0.43753996759319363,
+      "grad_norm": 441.5102350206535,
+      "learning_rate": 1.2124365267230571e-05,
+      "loss": 259.3831,
+      "step": 22750
+    },
+    {
+      "epoch": 0.43773229285367415,
+      "grad_norm": 487.24572690603037,
+      "learning_rate": 1.2118401034392003e-05,
+      "loss": 258.8978,
+      "step": 22760
+    },
+    {
+      "epoch": 0.4379246181141547,
+      "grad_norm": 460.67781077753483,
+      "learning_rate": 1.2112436012484035e-05,
+      "loss": 251.7436,
+      "step": 22770
+    },
+    {
+      "epoch": 0.4381169433746352,
+      "grad_norm": 446.12348528741194,
+      "learning_rate": 1.210647020372854e-05,
+      "loss": 258.8787,
+      "step": 22780
+    },
+    {
+      "epoch": 0.4383092686351157,
+      "grad_norm": 447.42075578421685,
+      "learning_rate": 1.2100503610347686e-05,
+      "loss": 253.0285,
+      "step": 22790
+    },
+    {
+      "epoch": 0.43850159389559623,
+      "grad_norm": 454.04740862017104,
+      "learning_rate": 1.2094536234563927e-05,
+      "loss": 259.1296,
+      "step": 22800
+    },
+    {
+      "epoch": 0.43869391915607675,
+      "grad_norm": 446.70185991473926,
+      "learning_rate": 1.2088568078600013e-05,
+      "loss": 254.9898,
+      "step": 22810
+    },
+    {
+      "epoch": 0.43888624441655727,
+      "grad_norm": 485.8524665221522,
+      "learning_rate": 1.2082599144678983e-05,
+      "loss": 263.8659,
+      "step": 22820
+    },
+    {
+      "epoch": 0.4390785696770378,
+      "grad_norm": 465.4629602099813,
+      "learning_rate": 1.2076629435024168e-05,
+      "loss": 258.2121,
+      "step": 22830
+    },
+    {
+      "epoch": 0.4392708949375183,
+      "grad_norm": 425.6049277705674,
+      "learning_rate": 1.2070658951859183e-05,
+      "loss": 269.3211,
+      "step": 22840
+    },
+    {
+      "epoch": 0.4394632201979989,
+      "grad_norm": 437.8373454637032,
+      "learning_rate": 1.2064687697407939e-05,
+      "loss": 266.0314,
+      "step": 22850
+    },
+    {
+      "epoch": 0.4396555454584794,
+      "grad_norm": 464.0106575435338,
+      "learning_rate": 1.2058715673894625e-05,
+      "loss": 250.2758,
+      "step": 22860
+    },
+    {
+      "epoch": 0.4398478707189599,
+      "grad_norm": 586.2208323692147,
+      "learning_rate": 1.2052742883543724e-05,
+      "loss": 268.0162,
+      "step": 22870
+    },
+    {
+      "epoch": 0.44004019597944044,
+      "grad_norm": 484.90330169836903,
+      "learning_rate": 1.2046769328580004e-05,
+      "loss": 251.0423,
+      "step": 22880
+    },
+    {
+      "epoch": 0.44023252123992096,
+      "grad_norm": 456.29633878520326,
+      "learning_rate": 1.204079501122851e-05,
+      "loss": 257.7815,
+      "step": 22890
+    },
+    {
+      "epoch": 0.4404248465004015,
+      "grad_norm": 479.2590284121029,
+      "learning_rate": 1.2034819933714576e-05,
+      "loss": 253.3188,
+      "step": 22900
+    },
+    {
+      "epoch": 0.440617171760882,
+      "grad_norm": 455.2239237704403,
+      "learning_rate": 1.2028844098263827e-05,
+      "loss": 257.8539,
+      "step": 22910
+    },
+    {
+      "epoch": 0.4408094970213625,
+      "grad_norm": 541.7350127978118,
+      "learning_rate": 1.2022867507102159e-05,
+      "loss": 261.571,
+      "step": 22920
+    },
+    {
+      "epoch": 0.44100182228184304,
+      "grad_norm": 480.2745326515579,
+      "learning_rate": 1.2016890162455752e-05,
+      "loss": 255.3,
+      "step": 22930
+    },
+    {
+      "epoch": 0.44119414754232356,
+      "grad_norm": 443.47417519914757,
+      "learning_rate": 1.2010912066551072e-05,
+      "loss": 250.8066,
+      "step": 22940
+    },
+    {
+      "epoch": 0.4413864728028041,
+      "grad_norm": 478.64850379222105,
+      "learning_rate": 1.2004933221614854e-05,
+      "loss": 250.1601,
+      "step": 22950
+    },
+    {
+      "epoch": 0.44157879806328465,
+      "grad_norm": 544.0752856101535,
+      "learning_rate": 1.1998953629874126e-05,
+      "loss": 251.2745,
+      "step": 22960
+    },
+    {
+      "epoch": 0.44177112332376517,
+      "grad_norm": 452.350869860681,
+      "learning_rate": 1.1992973293556184e-05,
+      "loss": 262.7483,
+      "step": 22970
+    },
+    {
+      "epoch": 0.4419634485842457,
+      "grad_norm": 482.2065768136144,
+      "learning_rate": 1.1986992214888607e-05,
+      "loss": 263.3633,
+      "step": 22980
+    },
+    {
+      "epoch": 0.4421557738447262,
+      "grad_norm": 463.3403672791953,
+      "learning_rate": 1.1981010396099244e-05,
+      "loss": 255.3194,
+      "step": 22990
+    },
+    {
+      "epoch": 0.4423480991052067,
+      "grad_norm": 466.78729720574364,
+      "learning_rate": 1.1975027839416227e-05,
+      "loss": 255.3379,
+      "step": 23000
+    },
+    {
+      "epoch": 0.44254042436568725,
+      "grad_norm": 499.88773139956356,
+      "learning_rate": 1.1969044547067961e-05,
+      "loss": 254.3,
+      "step": 23010
+    },
+    {
+      "epoch": 0.44273274962616777,
+      "grad_norm": 432.9218738967188,
+      "learning_rate": 1.1963060521283117e-05,
+      "loss": 250.6967,
+      "step": 23020
+    },
+    {
+      "epoch": 0.4429250748866483,
+      "grad_norm": 441.4274391268516,
+      "learning_rate": 1.1957075764290652e-05,
+      "loss": 253.8183,
+      "step": 23030
+    },
+    {
+      "epoch": 0.4431174001471288,
+      "grad_norm": 420.4371347752832,
+      "learning_rate": 1.1951090278319784e-05,
+      "loss": 263.0922,
+      "step": 23040
+    },
+    {
+      "epoch": 0.4433097254076093,
+      "grad_norm": 438.9999602213165,
+      "learning_rate": 1.1945104065600013e-05,
+      "loss": 248.7696,
+      "step": 23050
+    },
+    {
+      "epoch": 0.4435020506680899,
+      "grad_norm": 450.09212765840107,
+      "learning_rate": 1.1939117128361101e-05,
+      "loss": 254.6416,
+      "step": 23060
+    },
+    {
+      "epoch": 0.4436943759285704,
+      "grad_norm": 595.5102962288122,
+      "learning_rate": 1.1933129468833087e-05,
+      "loss": 249.2376,
+      "step": 23070
+    },
+    {
+      "epoch": 0.44388670118905094,
+      "grad_norm": 440.65433686268125,
+      "learning_rate": 1.192714108924627e-05,
+      "loss": 251.1425,
+      "step": 23080
+    },
+    {
+      "epoch": 0.44407902644953146,
+      "grad_norm": 517.268187199839,
+      "learning_rate": 1.1921151991831225e-05,
+      "loss": 258.6075,
+      "step": 23090
+    },
+    {
+      "epoch": 0.444271351710012,
+      "grad_norm": 432.79297789317343,
+      "learning_rate": 1.1915162178818793e-05,
+      "loss": 261.2917,
+      "step": 23100
+    },
+    {
+      "epoch": 0.4444636769704925,
+      "grad_norm": 515.8398529898058,
+      "learning_rate": 1.1909171652440079e-05,
+      "loss": 258.0972,
+      "step": 23110
+    },
+    {
+      "epoch": 0.444656002230973,
+      "grad_norm": 422.71187944443324,
+      "learning_rate": 1.1903180414926457e-05,
+      "loss": 253.5417,
+      "step": 23120
+    },
+    {
+      "epoch": 0.44484832749145353,
+      "grad_norm": 481.9994465707499,
+      "learning_rate": 1.189718846850956e-05,
+      "loss": 249.982,
+      "step": 23130
+    },
+    {
+      "epoch": 0.44504065275193405,
+      "grad_norm": 451.9385686125299,
+      "learning_rate": 1.1891195815421293e-05,
+      "loss": 242.0835,
+      "step": 23140
+    },
+    {
+      "epoch": 0.44523297801241457,
+      "grad_norm": 458.08573929983805,
+      "learning_rate": 1.1885202457893819e-05,
+      "loss": 253.6223,
+      "step": 23150
+    },
+    {
+      "epoch": 0.44542530327289515,
+      "grad_norm": 433.60250292147254,
+      "learning_rate": 1.1879208398159563e-05,
+      "loss": 256.1767,
+      "step": 23160
+    },
+    {
+      "epoch": 0.44561762853337566,
+      "grad_norm": 417.2649607614848,
+      "learning_rate": 1.1873213638451215e-05,
+      "loss": 247.6789,
+      "step": 23170
+    },
+    {
+      "epoch": 0.4458099537938562,
+      "grad_norm": 454.1867546595952,
+      "learning_rate": 1.1867218181001725e-05,
+      "loss": 256.11,
+      "step": 23180
+    },
+    {
+      "epoch": 0.4460022790543367,
+      "grad_norm": 488.40089969381876,
+      "learning_rate": 1.1861222028044301e-05,
+      "loss": 263.1078,
+      "step": 23190
+    },
+    {
+      "epoch": 0.4461946043148172,
+      "grad_norm": 462.6002840704471,
+      "learning_rate": 1.1855225181812408e-05,
+      "loss": 248.574,
+      "step": 23200
+    },
+    {
+      "epoch": 0.44638692957529774,
+      "grad_norm": 454.15626057905314,
+      "learning_rate": 1.1849227644539781e-05,
+      "loss": 249.628,
+      "step": 23210
+    },
+    {
+      "epoch": 0.44657925483577826,
+      "grad_norm": 435.102089963084,
+      "learning_rate": 1.1843229418460391e-05,
+      "loss": 247.6553,
+      "step": 23220
+    },
+    {
+      "epoch": 0.4467715800962588,
+      "grad_norm": 447.0186794227221,
+      "learning_rate": 1.1837230505808485e-05,
+      "loss": 247.3917,
+      "step": 23230
+    },
+    {
+      "epoch": 0.4469639053567393,
+      "grad_norm": 482.5627136822678,
+      "learning_rate": 1.1831230908818563e-05,
+      "loss": 249.0831,
+      "step": 23240
+    },
+    {
+      "epoch": 0.4471562306172198,
+      "grad_norm": 568.4407633078066,
+      "learning_rate": 1.1825230629725366e-05,
+      "loss": 259.6739,
+      "step": 23250
+    },
+    {
+      "epoch": 0.4473485558777004,
+      "grad_norm": 441.3757223921175,
+      "learning_rate": 1.1819229670763908e-05,
+      "loss": 257.217,
+      "step": 23260
+    },
+    {
+      "epoch": 0.4475408811381809,
+      "grad_norm": 515.8922182715229,
+      "learning_rate": 1.1813228034169442e-05,
+      "loss": 262.5166,
+      "step": 23270
+    },
+    {
+      "epoch": 0.44773320639866143,
+      "grad_norm": 446.86607750340266,
+      "learning_rate": 1.180722572217748e-05,
+      "loss": 262.1952,
+      "step": 23280
+    },
+    {
+      "epoch": 0.44792553165914195,
+      "grad_norm": 503.1736591134505,
+      "learning_rate": 1.1801222737023783e-05,
+      "loss": 255.8003,
+      "step": 23290
+    },
+    {
+      "epoch": 0.44811785691962247,
+      "grad_norm": 425.88902325715725,
+      "learning_rate": 1.1795219080944366e-05,
+      "loss": 253.7978,
+      "step": 23300
+    },
+    {
+      "epoch": 0.448310182180103,
+      "grad_norm": 434.44983563432226,
+      "learning_rate": 1.178921475617549e-05,
+      "loss": 253.9643,
+      "step": 23310
+    },
+    {
+      "epoch": 0.4485025074405835,
+      "grad_norm": 459.8931623631675,
+      "learning_rate": 1.1783209764953666e-05,
+      "loss": 258.2973,
+      "step": 23320
+    },
+    {
+      "epoch": 0.44869483270106403,
+      "grad_norm": 465.89200127424294,
+      "learning_rate": 1.1777204109515653e-05,
+      "loss": 261.593,
+      "step": 23330
+    },
+    {
+      "epoch": 0.44888715796154455,
+      "grad_norm": 470.4229758549265,
+      "learning_rate": 1.1771197792098465e-05,
+      "loss": 252.345,
+      "step": 23340
+    },
+    {
+      "epoch": 0.44907948322202507,
+      "grad_norm": 439.1415837531606,
+      "learning_rate": 1.1765190814939351e-05,
+      "loss": 255.1918,
+      "step": 23350
+    },
+    {
+      "epoch": 0.4492718084825056,
+      "grad_norm": 473.3546636933446,
+      "learning_rate": 1.175918318027581e-05,
+      "loss": 248.6485,
+      "step": 23360
+    },
+    {
+      "epoch": 0.44946413374298616,
+      "grad_norm": 473.26223567284796,
+      "learning_rate": 1.1753174890345591e-05,
+      "loss": 255.5117,
+      "step": 23370
+    },
+    {
+      "epoch": 0.4496564590034667,
+      "grad_norm": 474.064042377261,
+      "learning_rate": 1.174716594738668e-05,
+      "loss": 256.0353,
+      "step": 23380
+    },
+    {
+      "epoch": 0.4498487842639472,
+      "grad_norm": 530.1382579394015,
+      "learning_rate": 1.1741156353637304e-05,
+      "loss": 247.7151,
+      "step": 23390
+    },
+    {
+      "epoch": 0.4500411095244277,
+      "grad_norm": 427.55116437237126,
+      "learning_rate": 1.1735146111335945e-05,
+      "loss": 251.0648,
+      "step": 23400
+    },
+    {
+      "epoch": 0.45023343478490824,
+      "grad_norm": 506.92651809226555,
+      "learning_rate": 1.1729135222721315e-05,
+      "loss": 250.9444,
+      "step": 23410
+    },
+    {
+      "epoch": 0.45042576004538876,
+      "grad_norm": 473.1743739295858,
+      "learning_rate": 1.1723123690032376e-05,
+      "loss": 242.5035,
+      "step": 23420
+    },
+    {
+      "epoch": 0.4506180853058693,
+      "grad_norm": 434.1947656086353,
+      "learning_rate": 1.171711151550832e-05,
+      "loss": 264.2066,
+      "step": 23430
+    },
+    {
+      "epoch": 0.4508104105663498,
+      "grad_norm": 477.5351230643194,
+      "learning_rate": 1.1711098701388581e-05,
+      "loss": 253.424,
+      "step": 23440
+    },
+    {
+      "epoch": 0.4510027358268303,
+      "grad_norm": 447.8328377689963,
+      "learning_rate": 1.1705085249912837e-05,
+      "loss": 250.5625,
+      "step": 23450
+    },
+    {
+      "epoch": 0.45119506108731083,
+      "grad_norm": 420.65551830062424,
+      "learning_rate": 1.1699071163320997e-05,
+      "loss": 255.4377,
+      "step": 23460
+    },
+    {
+      "epoch": 0.4513873863477914,
+      "grad_norm": 446.5525879917818,
+      "learning_rate": 1.169305644385321e-05,
+      "loss": 249.6609,
+      "step": 23470
+    },
+    {
+      "epoch": 0.4515797116082719,
+      "grad_norm": 520.951696992055,
+      "learning_rate": 1.168704109374986e-05,
+      "loss": 252.9367,
+      "step": 23480
+    },
+    {
+      "epoch": 0.45177203686875245,
+      "grad_norm": 441.8970751124328,
+      "learning_rate": 1.1681025115251566e-05,
+      "loss": 256.3322,
+      "step": 23490
+    },
+    {
+      "epoch": 0.45196436212923297,
+      "grad_norm": 470.8406876867624,
+      "learning_rate": 1.1675008510599176e-05,
+      "loss": 246.6226,
+      "step": 23500
+    },
+    {
+      "epoch": 0.4521566873897135,
+      "grad_norm": 478.44083732437895,
+      "learning_rate": 1.1668991282033784e-05,
+      "loss": 251.7677,
+      "step": 23510
+    },
+    {
+      "epoch": 0.452349012650194,
+      "grad_norm": 435.3744683996766,
+      "learning_rate": 1.16629734317967e-05,
+      "loss": 258.2138,
+      "step": 23520
+    },
+    {
+      "epoch": 0.4525413379106745,
+      "grad_norm": 411.29350727643117,
+      "learning_rate": 1.165695496212948e-05,
+      "loss": 263.2326,
+      "step": 23530
+    },
+    {
+      "epoch": 0.45273366317115504,
+      "grad_norm": 482.24576860111307,
+      "learning_rate": 1.1650935875273901e-05,
+      "loss": 256.6767,
+      "step": 23540
+    },
+    {
+      "epoch": 0.45292598843163556,
+      "grad_norm": 418.03357553708787,
+      "learning_rate": 1.1644916173471976e-05,
+      "loss": 251.8268,
+      "step": 23550
+    },
+    {
+      "epoch": 0.4531183136921161,
+      "grad_norm": 427.5402308809369,
+      "learning_rate": 1.1638895858965942e-05,
+      "loss": 251.8567,
+      "step": 23560
+    },
+    {
+      "epoch": 0.45331063895259666,
+      "grad_norm": 411.4118899248775,
+      "learning_rate": 1.1632874933998268e-05,
+      "loss": 250.1026,
+      "step": 23570
+    },
+    {
+      "epoch": 0.4535029642130772,
+      "grad_norm": 422.26116455188844,
+      "learning_rate": 1.1626853400811649e-05,
+      "loss": 254.3002,
+      "step": 23580
+    },
+    {
+      "epoch": 0.4536952894735577,
+      "grad_norm": 409.3825065374503,
+      "learning_rate": 1.1620831261649003e-05,
+      "loss": 248.4919,
+      "step": 23590
+    },
+    {
+      "epoch": 0.4538876147340382,
+      "grad_norm": 502.34269113830806,
+      "learning_rate": 1.1614808518753485e-05,
+      "loss": 253.1792,
+      "step": 23600
+    },
+    {
+      "epoch": 0.45407993999451873,
+      "grad_norm": 427.2176160351355,
+      "learning_rate": 1.1608785174368461e-05,
+      "loss": 255.2146,
+      "step": 23610
+    },
+    {
+      "epoch": 0.45427226525499925,
+      "grad_norm": 456.8769255260778,
+      "learning_rate": 1.1602761230737531e-05,
+      "loss": 264.9469,
+      "step": 23620
+    },
+    {
+      "epoch": 0.45446459051547977,
+      "grad_norm": 470.04123453353424,
+      "learning_rate": 1.1596736690104514e-05,
+      "loss": 267.1819,
+      "step": 23630
+    },
+    {
+      "epoch": 0.4546569157759603,
+      "grad_norm": 465.0725243037591,
+      "learning_rate": 1.1590711554713452e-05,
+      "loss": 249.3309,
+      "step": 23640
+    },
+    {
+      "epoch": 0.4548492410364408,
+      "grad_norm": 448.69630738371256,
+      "learning_rate": 1.1584685826808604e-05,
+      "loss": 261.6062,
+      "step": 23650
+    },
+    {
+      "epoch": 0.45504156629692133,
+      "grad_norm": 546.2729015603361,
+      "learning_rate": 1.157865950863446e-05,
+      "loss": 254.6912,
+      "step": 23660
+    },
+    {
+      "epoch": 0.4552338915574019,
+      "grad_norm": 496.91664114275807,
+      "learning_rate": 1.1572632602435717e-05,
+      "loss": 261.0411,
+      "step": 23670
+    },
+    {
+      "epoch": 0.4554262168178824,
+      "grad_norm": 434.51523632768345,
+      "learning_rate": 1.1566605110457305e-05,
+      "loss": 249.2502,
+      "step": 23680
+    },
+    {
+      "epoch": 0.45561854207836294,
+      "grad_norm": 463.51171216675505,
+      "learning_rate": 1.1560577034944364e-05,
+      "loss": 255.7892,
+      "step": 23690
+    },
+    {
+      "epoch": 0.45581086733884346,
+      "grad_norm": 458.64992963281185,
+      "learning_rate": 1.1554548378142249e-05,
+      "loss": 256.925,
+      "step": 23700
+    },
+    {
+      "epoch": 0.456003192599324,
+      "grad_norm": 515.4003121675826,
+      "learning_rate": 1.1548519142296541e-05,
+      "loss": 255.3865,
+      "step": 23710
+    },
+    {
+      "epoch": 0.4561955178598045,
+      "grad_norm": 451.14221472506534,
+      "learning_rate": 1.1542489329653024e-05,
+      "loss": 260.194,
+      "step": 23720
+    },
+    {
+      "epoch": 0.456387843120285,
+      "grad_norm": 454.22786161886063,
+      "learning_rate": 1.153645894245771e-05,
+      "loss": 245.0625,
+      "step": 23730
+    },
+    {
+      "epoch": 0.45658016838076554,
+      "grad_norm": 502.9341806779307,
+      "learning_rate": 1.1530427982956813e-05,
+      "loss": 259.9284,
+      "step": 23740
+    },
+    {
+      "epoch": 0.45677249364124606,
+      "grad_norm": 459.0166320562402,
+      "learning_rate": 1.1524396453396767e-05,
+      "loss": 252.9541,
+      "step": 23750
+    },
+    {
+      "epoch": 0.4569648189017266,
+      "grad_norm": 491.7858496182521,
+      "learning_rate": 1.1518364356024219e-05,
+      "loss": 257.7152,
+      "step": 23760
+    },
+    {
+      "epoch": 0.45715714416220715,
+      "grad_norm": 463.23611240589327,
+      "learning_rate": 1.1512331693086025e-05,
+      "loss": 246.3303,
+      "step": 23770
+    },
+    {
+      "epoch": 0.45734946942268767,
+      "grad_norm": 486.631687372408,
+      "learning_rate": 1.1506298466829256e-05,
+      "loss": 259.1904,
+      "step": 23780
+    },
+    {
+      "epoch": 0.4575417946831682,
+      "grad_norm": 469.2758785973338,
+      "learning_rate": 1.1500264679501181e-05,
+      "loss": 250.8088,
+      "step": 23790
+    },
+    {
+      "epoch": 0.4577341199436487,
+      "grad_norm": 453.26075096354646,
+      "learning_rate": 1.1494230333349292e-05,
+      "loss": 256.325,
+      "step": 23800
+    },
+    {
+      "epoch": 0.45792644520412923,
+      "grad_norm": 460.5524748738918,
+      "learning_rate": 1.1488195430621284e-05,
+      "loss": 252.0218,
+      "step": 23810
+    },
+    {
+      "epoch": 0.45811877046460975,
+      "grad_norm": 439.6939650960091,
+      "learning_rate": 1.1482159973565051e-05,
+      "loss": 259.9782,
+      "step": 23820
+    },
+    {
+      "epoch": 0.45831109572509027,
+      "grad_norm": 508.0750320206364,
+      "learning_rate": 1.147612396442871e-05,
+      "loss": 252.5928,
+      "step": 23830
+    },
+    {
+      "epoch": 0.4585034209855708,
+      "grad_norm": 424.31436388147495,
+      "learning_rate": 1.1470087405460572e-05,
+      "loss": 246.4426,
+      "step": 23840
+    },
+    {
+      "epoch": 0.4586957462460513,
+      "grad_norm": 450.3182060568921,
+      "learning_rate": 1.1464050298909153e-05,
+      "loss": 245.9021,
+      "step": 23850
+    },
+    {
+      "epoch": 0.4588880715065318,
+      "grad_norm": 433.24856456922555,
+      "learning_rate": 1.1458012647023178e-05,
+      "loss": 238.0203,
+      "step": 23860
+    },
+    {
+      "epoch": 0.45908039676701234,
+      "grad_norm": 482.7448499182103,
+      "learning_rate": 1.1451974452051572e-05,
+      "loss": 255.4715,
+      "step": 23870
+    },
+    {
+      "epoch": 0.4592727220274929,
+      "grad_norm": 475.0546091941969,
+      "learning_rate": 1.1445935716243463e-05,
+      "loss": 251.2,
+      "step": 23880
+    },
+    {
+      "epoch": 0.45946504728797344,
+      "grad_norm": 484.97060105554647,
+      "learning_rate": 1.143989644184818e-05,
+      "loss": 253.436,
+      "step": 23890
+    },
+    {
+      "epoch": 0.45965737254845396,
+      "grad_norm": 468.34297447904,
+      "learning_rate": 1.1433856631115252e-05,
+      "loss": 261.0416,
+      "step": 23900
+    },
+    {
+      "epoch": 0.4598496978089345,
+      "grad_norm": 405.1608644677801,
+      "learning_rate": 1.142781628629441e-05,
+      "loss": 248.2056,
+      "step": 23910
+    },
+    {
+      "epoch": 0.460042023069415,
+      "grad_norm": 433.94796888196214,
+      "learning_rate": 1.1421775409635585e-05,
+      "loss": 253.7709,
+      "step": 23920
+    },
+    {
+      "epoch": 0.4602343483298955,
+      "grad_norm": 599.1512606561172,
+      "learning_rate": 1.1415734003388899e-05,
+      "loss": 257.7739,
+      "step": 23930
+    },
+    {
+      "epoch": 0.46042667359037603,
+      "grad_norm": 467.75025045078866,
+      "learning_rate": 1.1409692069804678e-05,
+      "loss": 256.9482,
+      "step": 23940
+    },
+    {
+      "epoch": 0.46061899885085655,
+      "grad_norm": 457.6043584147629,
+      "learning_rate": 1.1403649611133444e-05,
+      "loss": 244.3143,
+      "step": 23950
+    },
+    {
+      "epoch": 0.4608113241113371,
+      "grad_norm": 463.9535672458291,
+      "learning_rate": 1.1397606629625913e-05,
+      "loss": 258.4995,
+      "step": 23960
+    },
+    {
+      "epoch": 0.4610036493718176,
+      "grad_norm": 492.8718149401336,
+      "learning_rate": 1.1391563127532992e-05,
+      "loss": 250.6034,
+      "step": 23970
+    },
+    {
+      "epoch": 0.46119597463229817,
+      "grad_norm": 437.8979606711717,
+      "learning_rate": 1.1385519107105791e-05,
+      "loss": 252.3589,
+      "step": 23980
+    },
+    {
+      "epoch": 0.4613882998927787,
+      "grad_norm": 521.9274380503093,
+      "learning_rate": 1.1379474570595604e-05,
+      "loss": 255.4463,
+      "step": 23990
+    },
+    {
+      "epoch": 0.4615806251532592,
+      "grad_norm": 428.0969800586121,
+      "learning_rate": 1.1373429520253922e-05,
+      "loss": 244.703,
+      "step": 24000
+    },
+    {
+      "epoch": 0.4617729504137397,
+      "grad_norm": 485.61974563074085,
+      "learning_rate": 1.1367383958332427e-05,
+      "loss": 247.4231,
+      "step": 24010
+    },
+    {
+      "epoch": 0.46196527567422024,
+      "grad_norm": 522.8341813022549,
+      "learning_rate": 1.1361337887082991e-05,
+      "loss": 251.8473,
+      "step": 24020
+    },
+    {
+      "epoch": 0.46215760093470076,
+      "grad_norm": 455.9985505121523,
+      "learning_rate": 1.1355291308757672e-05,
+      "loss": 255.5515,
+      "step": 24030
+    },
+    {
+      "epoch": 0.4623499261951813,
+      "grad_norm": 445.6984750674221,
+      "learning_rate": 1.1349244225608727e-05,
+      "loss": 252.9669,
+      "step": 24040
+    },
+    {
+      "epoch": 0.4625422514556618,
+      "grad_norm": 428.94798433415053,
+      "learning_rate": 1.1343196639888591e-05,
+      "loss": 256.4869,
+      "step": 24050
+    },
+    {
+      "epoch": 0.4627345767161423,
+      "grad_norm": 510.4003073113367,
+      "learning_rate": 1.133714855384989e-05,
+      "loss": 257.6965,
+      "step": 24060
+    },
+    {
+      "epoch": 0.46292690197662284,
+      "grad_norm": 460.91183608948995,
+      "learning_rate": 1.1331099969745439e-05,
+      "loss": 250.8719,
+      "step": 24070
+    },
+    {
+      "epoch": 0.4631192272371034,
+      "grad_norm": 495.272702207888,
+      "learning_rate": 1.132505088982823e-05,
+      "loss": 264.4658,
+      "step": 24080
+    },
+    {
+      "epoch": 0.46331155249758393,
+      "grad_norm": 452.1273384529998,
+      "learning_rate": 1.131900131635145e-05,
+      "loss": 247.3377,
+      "step": 24090
+    },
+    {
+      "epoch": 0.46350387775806445,
+      "grad_norm": 478.12129665054374,
+      "learning_rate": 1.1312951251568461e-05,
+      "loss": 254.158,
+      "step": 24100
+    },
+    {
+      "epoch": 0.463696203018545,
+      "grad_norm": 421.7954517930514,
+      "learning_rate": 1.1306900697732816e-05,
+      "loss": 246.6665,
+      "step": 24110
+    },
+    {
+      "epoch": 0.4638885282790255,
+      "grad_norm": 438.8070098208251,
+      "learning_rate": 1.1300849657098248e-05,
+      "loss": 252.4575,
+      "step": 24120
+    },
+    {
+      "epoch": 0.464080853539506,
+      "grad_norm": 431.1325105841485,
+      "learning_rate": 1.1294798131918665e-05,
+      "loss": 246.7184,
+      "step": 24130
+    },
+    {
+      "epoch": 0.46427317879998653,
+      "grad_norm": 436.79399854155474,
+      "learning_rate": 1.1288746124448164e-05,
+      "loss": 248.6332,
+      "step": 24140
+    },
+    {
+      "epoch": 0.46446550406046705,
+      "grad_norm": 407.84909784600154,
+      "learning_rate": 1.1282693636941013e-05,
+      "loss": 243.9928,
+      "step": 24150
+    },
+    {
+      "epoch": 0.46465782932094757,
+      "grad_norm": 425.53119656555805,
+      "learning_rate": 1.1276640671651671e-05,
+      "loss": 259.5484,
+      "step": 24160
+    },
+    {
+      "epoch": 0.4648501545814281,
+      "grad_norm": 467.32434582852153,
+      "learning_rate": 1.1270587230834757e-05,
+      "loss": 256.35,
+      "step": 24170
+    },
+    {
+      "epoch": 0.46504247984190866,
+      "grad_norm": 446.9943765525869,
+      "learning_rate": 1.1264533316745088e-05,
+      "loss": 249.4387,
+      "step": 24180
+    },
+    {
+      "epoch": 0.4652348051023892,
+      "grad_norm": 450.3452801542163,
+      "learning_rate": 1.1258478931637641e-05,
+      "loss": 250.4805,
+      "step": 24190
+    },
+    {
+      "epoch": 0.4654271303628697,
+      "grad_norm": 476.0199473912133,
+      "learning_rate": 1.1252424077767577e-05,
+      "loss": 237.7216,
+      "step": 24200
+    },
+    {
+      "epoch": 0.4656194556233502,
+      "grad_norm": 534.2949042190356,
+      "learning_rate": 1.1246368757390231e-05,
+      "loss": 246.3689,
+      "step": 24210
+    },
+    {
+      "epoch": 0.46581178088383074,
+      "grad_norm": 431.38590706435883,
+      "learning_rate": 1.1240312972761105e-05,
+      "loss": 246.3344,
+      "step": 24220
+    },
+    {
+      "epoch": 0.46600410614431126,
+      "grad_norm": 472.0800795568094,
+      "learning_rate": 1.1234256726135882e-05,
+      "loss": 250.9139,
+      "step": 24230
+    },
+    {
+      "epoch": 0.4661964314047918,
+      "grad_norm": 475.8731776955852,
+      "learning_rate": 1.1228200019770412e-05,
+      "loss": 238.8638,
+      "step": 24240
+    },
+    {
+      "epoch": 0.4663887566652723,
+      "grad_norm": 591.6405623294445,
+      "learning_rate": 1.1222142855920719e-05,
+      "loss": 254.9071,
+      "step": 24250
+    },
+    {
+      "epoch": 0.4665810819257528,
+      "grad_norm": 464.7131966163036,
+      "learning_rate": 1.1216085236842997e-05,
+      "loss": 248.4065,
+      "step": 24260
+    },
+    {
+      "epoch": 0.46677340718623334,
+      "grad_norm": 454.0690584045458,
+      "learning_rate": 1.1210027164793609e-05,
+      "loss": 257.0167,
+      "step": 24270
+    },
+    {
+      "epoch": 0.46696573244671385,
+      "grad_norm": 511.8551061919269,
+      "learning_rate": 1.1203968642029086e-05,
+      "loss": 254.8844,
+      "step": 24280
+    },
+    {
+      "epoch": 0.46715805770719443,
+      "grad_norm": 476.62536954453935,
+      "learning_rate": 1.1197909670806126e-05,
+      "loss": 250.0282,
+      "step": 24290
+    },
+    {
+      "epoch": 0.46735038296767495,
+      "grad_norm": 506.67464828242356,
+      "learning_rate": 1.1191850253381602e-05,
+      "loss": 260.3468,
+      "step": 24300
+    },
+    {
+      "epoch": 0.46754270822815547,
+      "grad_norm": 413.12764072167204,
+      "learning_rate": 1.1185790392012538e-05,
+      "loss": 249.6076,
+      "step": 24310
+    },
+    {
+      "epoch": 0.467735033488636,
+      "grad_norm": 490.1677673641274,
+      "learning_rate": 1.117973008895614e-05,
+      "loss": 250.4889,
+      "step": 24320
+    },
+    {
+      "epoch": 0.4679273587491165,
+      "grad_norm": 442.1200404680506,
+      "learning_rate": 1.1173669346469767e-05,
+      "loss": 256.6804,
+      "step": 24330
+    },
+    {
+      "epoch": 0.468119684009597,
+      "grad_norm": 420.3265496395681,
+      "learning_rate": 1.1167608166810948e-05,
+      "loss": 248.3983,
+      "step": 24340
+    },
+    {
+      "epoch": 0.46831200927007754,
+      "grad_norm": 481.82760767730286,
+      "learning_rate": 1.1161546552237368e-05,
+      "loss": 262.9398,
+      "step": 24350
+    },
+    {
+      "epoch": 0.46850433453055806,
+      "grad_norm": 502.98826463841664,
+      "learning_rate": 1.1155484505006884e-05,
+      "loss": 252.0988,
+      "step": 24360
+    },
+    {
+      "epoch": 0.4686966597910386,
+      "grad_norm": 453.78842173853565,
+      "learning_rate": 1.1149422027377501e-05,
+      "loss": 252.8134,
+      "step": 24370
+    },
+    {
+      "epoch": 0.4688889850515191,
+      "grad_norm": 421.40902934543436,
+      "learning_rate": 1.1143359121607397e-05,
+      "loss": 248.2464,
+      "step": 24380
+    },
+    {
+      "epoch": 0.4690813103119997,
+      "grad_norm": 457.17537546098157,
+      "learning_rate": 1.1137295789954904e-05,
+      "loss": 245.1052,
+      "step": 24390
+    },
+    {
+      "epoch": 0.4692736355724802,
+      "grad_norm": 443.40464034342074,
+      "learning_rate": 1.1131232034678513e-05,
+      "loss": 248.9347,
+      "step": 24400
+    },
+    {
+      "epoch": 0.4694659608329607,
+      "grad_norm": 454.3725000763719,
+      "learning_rate": 1.1125167858036874e-05,
+      "loss": 249.8428,
+      "step": 24410
+    },
+    {
+      "epoch": 0.46965828609344124,
+      "grad_norm": 658.893834386438,
+      "learning_rate": 1.1119103262288788e-05,
+      "loss": 249.1869,
+      "step": 24420
+    },
+    {
+      "epoch": 0.46985061135392175,
+      "grad_norm": 403.0624609424853,
+      "learning_rate": 1.1113038249693221e-05,
+      "loss": 247.1238,
+      "step": 24430
+    },
+    {
+      "epoch": 0.4700429366144023,
+      "grad_norm": 434.98247786393125,
+      "learning_rate": 1.1106972822509287e-05,
+      "loss": 248.6417,
+      "step": 24440
+    },
+    {
+      "epoch": 0.4702352618748828,
+      "grad_norm": 588.2224711232709,
+      "learning_rate": 1.1100906982996257e-05,
+      "loss": 257.6896,
+      "step": 24450
+    },
+    {
+      "epoch": 0.4704275871353633,
+      "grad_norm": 409.4694456668267,
+      "learning_rate": 1.109484073341356e-05,
+      "loss": 250.3586,
+      "step": 24460
+    },
+    {
+      "epoch": 0.47061991239584383,
+      "grad_norm": 447.49656036589306,
+      "learning_rate": 1.1088774076020772e-05,
+      "loss": 240.8952,
+      "step": 24470
+    },
+    {
+      "epoch": 0.47081223765632435,
+      "grad_norm": 493.25521247994993,
+      "learning_rate": 1.1082707013077625e-05,
+      "loss": 246.0719,
+      "step": 24480
+    },
+    {
+      "epoch": 0.4710045629168049,
+      "grad_norm": 464.28867493228563,
+      "learning_rate": 1.1076639546843993e-05,
+      "loss": 251.0546,
+      "step": 24490
+    },
+    {
+      "epoch": 0.47119688817728544,
+      "grad_norm": 436.4159537450911,
+      "learning_rate": 1.1070571679579913e-05,
+      "loss": 252.1561,
+      "step": 24500
+    },
+    {
+      "epoch": 0.47138921343776596,
+      "grad_norm": 421.21392822926066,
+      "learning_rate": 1.1064503413545565e-05,
+      "loss": 254.469,
+      "step": 24510
+    },
+    {
+      "epoch": 0.4715815386982465,
+      "grad_norm": 460.5343484928838,
+      "learning_rate": 1.1058434751001272e-05,
+      "loss": 251.8184,
+      "step": 24520
+    },
+    {
+      "epoch": 0.471773863958727,
+      "grad_norm": 411.91893550393957,
+      "learning_rate": 1.1052365694207516e-05,
+      "loss": 247.0095,
+      "step": 24530
+    },
+    {
+      "epoch": 0.4719661892192075,
+      "grad_norm": 449.8590650120734,
+      "learning_rate": 1.1046296245424922e-05,
+      "loss": 244.535,
+      "step": 24540
+    },
+    {
+      "epoch": 0.47215851447968804,
+      "grad_norm": 434.80418999565836,
+      "learning_rate": 1.1040226406914254e-05,
+      "loss": 245.4892,
+      "step": 24550
+    },
+    {
+      "epoch": 0.47235083974016856,
+      "grad_norm": 455.9794907514911,
+      "learning_rate": 1.103415618093643e-05,
+      "loss": 257.8027,
+      "step": 24560
+    },
+    {
+      "epoch": 0.4725431650006491,
+      "grad_norm": 477.37782837369866,
+      "learning_rate": 1.1028085569752512e-05,
+      "loss": 250.1352,
+      "step": 24570
+    },
+    {
+      "epoch": 0.4727354902611296,
+      "grad_norm": 421.5207501313153,
+      "learning_rate": 1.1022014575623695e-05,
+      "loss": 253.0865,
+      "step": 24580
+    },
+    {
+      "epoch": 0.4729278155216102,
+      "grad_norm": 481.2263235604556,
+      "learning_rate": 1.1015943200811325e-05,
+      "loss": 247.887,
+      "step": 24590
+    },
+    {
+      "epoch": 0.4731201407820907,
+      "grad_norm": 429.53295926869913,
+      "learning_rate": 1.1009871447576894e-05,
+      "loss": 243.6173,
+      "step": 24600
+    },
+    {
+      "epoch": 0.4733124660425712,
+      "grad_norm": 464.9981357871874,
+      "learning_rate": 1.100379931818203e-05,
+      "loss": 252.1286,
+      "step": 24610
+    },
+    {
+      "epoch": 0.47350479130305173,
+      "grad_norm": 517.5348927538537,
+      "learning_rate": 1.0997726814888497e-05,
+      "loss": 261.3658,
+      "step": 24620
+    },
+    {
+      "epoch": 0.47369711656353225,
+      "grad_norm": 442.3162599786568,
+      "learning_rate": 1.0991653939958203e-05,
+      "loss": 241.7305,
+      "step": 24630
+    },
+    {
+      "epoch": 0.47388944182401277,
+      "grad_norm": 424.1894663482793,
+      "learning_rate": 1.0985580695653193e-05,
+      "loss": 253.9543,
+      "step": 24640
+    },
+    {
+      "epoch": 0.4740817670844933,
+      "grad_norm": 448.4381082664568,
+      "learning_rate": 1.0979507084235653e-05,
+      "loss": 251.0728,
+      "step": 24650
+    },
+    {
+      "epoch": 0.4742740923449738,
+      "grad_norm": 458.87743375958837,
+      "learning_rate": 1.0973433107967901e-05,
+      "loss": 246.6305,
+      "step": 24660
+    },
+    {
+      "epoch": 0.4744664176054543,
+      "grad_norm": 517.6346244095042,
+      "learning_rate": 1.0967358769112389e-05,
+      "loss": 251.4199,
+      "step": 24670
+    },
+    {
+      "epoch": 0.47465874286593485,
+      "grad_norm": 451.07296821258916,
+      "learning_rate": 1.0961284069931717e-05,
+      "loss": 245.2185,
+      "step": 24680
+    },
+    {
+      "epoch": 0.4748510681264154,
+      "grad_norm": 447.18620961813315,
+      "learning_rate": 1.0955209012688602e-05,
+      "loss": 249.5758,
+      "step": 24690
+    },
+    {
+      "epoch": 0.47504339338689594,
+      "grad_norm": 409.3526850856575,
+      "learning_rate": 1.094913359964591e-05,
+      "loss": 244.5644,
+      "step": 24700
+    },
+    {
+      "epoch": 0.47523571864737646,
+      "grad_norm": 426.6026512686952,
+      "learning_rate": 1.0943057833066622e-05,
+      "loss": 252.9506,
+      "step": 24710
+    },
+    {
+      "epoch": 0.475428043907857,
+      "grad_norm": 416.51572810563215,
+      "learning_rate": 1.093698171521387e-05,
+      "loss": 249.4896,
+      "step": 24720
+    },
+    {
+      "epoch": 0.4756203691683375,
+      "grad_norm": 462.1845830799221,
+      "learning_rate": 1.0930905248350903e-05,
+      "loss": 240.598,
+      "step": 24730
+    },
+    {
+      "epoch": 0.475812694428818,
+      "grad_norm": 472.54977842161674,
+      "learning_rate": 1.0924828434741101e-05,
+      "loss": 251.475,
+      "step": 24740
+    },
+    {
+      "epoch": 0.47600501968929854,
+      "grad_norm": 436.0025743451127,
+      "learning_rate": 1.0918751276647988e-05,
+      "loss": 251.9488,
+      "step": 24750
+    },
+    {
+      "epoch": 0.47619734494977906,
+      "grad_norm": 458.33776527955024,
+      "learning_rate": 1.0912673776335194e-05,
+      "loss": 245.6878,
+      "step": 24760
+    },
+    {
+      "epoch": 0.4763896702102596,
+      "grad_norm": 455.90399573802995,
+      "learning_rate": 1.0906595936066496e-05,
+      "loss": 248.6308,
+      "step": 24770
+    },
+    {
+      "epoch": 0.4765819954707401,
+      "grad_norm": 415.12883392843173,
+      "learning_rate": 1.090051775810578e-05,
+      "loss": 246.4492,
+      "step": 24780
+    },
+    {
+      "epoch": 0.4767743207312206,
+      "grad_norm": 476.5103535122165,
+      "learning_rate": 1.0894439244717075e-05,
+      "loss": 254.9108,
+      "step": 24790
+    },
+    {
+      "epoch": 0.4769666459917012,
+      "grad_norm": 466.93904612563324,
+      "learning_rate": 1.0888360398164521e-05,
+      "loss": 260.6951,
+      "step": 24800
+    },
+    {
+      "epoch": 0.4771589712521817,
+      "grad_norm": 425.51825024616903,
+      "learning_rate": 1.088228122071239e-05,
+      "loss": 252.0665,
+      "step": 24810
+    },
+    {
+      "epoch": 0.4773512965126622,
+      "grad_norm": 458.7198288357738,
+      "learning_rate": 1.0876201714625076e-05,
+      "loss": 250.3464,
+      "step": 24820
+    },
+    {
+      "epoch": 0.47754362177314275,
+      "grad_norm": 453.30732082967415,
+      "learning_rate": 1.0870121882167095e-05,
+      "loss": 247.5671,
+      "step": 24830
+    },
+    {
+      "epoch": 0.47773594703362326,
+      "grad_norm": 436.2087674068979,
+      "learning_rate": 1.0864041725603085e-05,
+      "loss": 247.0716,
+      "step": 24840
+    },
+    {
+      "epoch": 0.4779282722941038,
+      "grad_norm": 454.10356148142944,
+      "learning_rate": 1.08579612471978e-05,
+      "loss": 243.4011,
+      "step": 24850
+    },
+    {
+      "epoch": 0.4781205975545843,
+      "grad_norm": 513.6293858830724,
+      "learning_rate": 1.0851880449216123e-05,
+      "loss": 243.0631,
+      "step": 24860
+    },
+    {
+      "epoch": 0.4783129228150648,
+      "grad_norm": 443.6660767597183,
+      "learning_rate": 1.0845799333923045e-05,
+      "loss": 251.9735,
+      "step": 24870
+    },
+    {
+      "epoch": 0.47850524807554534,
+      "grad_norm": 439.97080458978786,
+      "learning_rate": 1.0839717903583684e-05,
+      "loss": 246.0687,
+      "step": 24880
+    },
+    {
+      "epoch": 0.47869757333602586,
+      "grad_norm": 455.412674235121,
+      "learning_rate": 1.0833636160463273e-05,
+      "loss": 247.1598,
+      "step": 24890
+    },
+    {
+      "epoch": 0.47888989859650644,
+      "grad_norm": 446.6137596327817,
+      "learning_rate": 1.0827554106827162e-05,
+      "loss": 250.6849,
+      "step": 24900
+    },
+    {
+      "epoch": 0.47908222385698696,
+      "grad_norm": 433.7040588188312,
+      "learning_rate": 1.0821471744940812e-05,
+      "loss": 245.5958,
+      "step": 24910
+    },
+    {
+      "epoch": 0.4792745491174675,
+      "grad_norm": 417.6454876241638,
+      "learning_rate": 1.0815389077069805e-05,
+      "loss": 241.6265,
+      "step": 24920
+    },
+    {
+      "epoch": 0.479466874377948,
+      "grad_norm": 489.9466623664129,
+      "learning_rate": 1.0809306105479834e-05,
+      "loss": 251.8081,
+      "step": 24930
+    },
+    {
+      "epoch": 0.4796591996384285,
+      "grad_norm": 441.5493420550708,
+      "learning_rate": 1.0803222832436701e-05,
+      "loss": 242.8504,
+      "step": 24940
+    },
+    {
+      "epoch": 0.47985152489890903,
+      "grad_norm": 493.3406496849026,
+      "learning_rate": 1.0797139260206331e-05,
+      "loss": 248.1648,
+      "step": 24950
+    },
+    {
+      "epoch": 0.48004385015938955,
+      "grad_norm": 438.3408598809831,
+      "learning_rate": 1.079105539105475e-05,
+      "loss": 251.575,
+      "step": 24960
+    },
+    {
+      "epoch": 0.48023617541987007,
+      "grad_norm": 552.3292854874302,
+      "learning_rate": 1.0784971227248104e-05,
+      "loss": 254.1856,
+      "step": 24970
+    },
+    {
+      "epoch": 0.4804285006803506,
+      "grad_norm": 438.2062557228392,
+      "learning_rate": 1.077888677105264e-05,
+      "loss": 254.4444,
+      "step": 24980
+    },
+    {
+      "epoch": 0.4806208259408311,
+      "grad_norm": 422.1076010878798,
+      "learning_rate": 1.0772802024734716e-05,
+      "loss": 244.7023,
+      "step": 24990
+    },
+    {
+      "epoch": 0.4808131512013117,
+      "grad_norm": 450.69549656803775,
+      "learning_rate": 1.07667169905608e-05,
+      "loss": 253.7605,
+      "step": 25000
+    },
+    {
+      "epoch": 0.4810054764617922,
+      "grad_norm": 466.94073546656665,
+      "learning_rate": 1.0760631670797468e-05,
+      "loss": 245.616,
+      "step": 25010
+    },
+    {
+      "epoch": 0.4811978017222727,
+      "grad_norm": 467.1817616308966,
+      "learning_rate": 1.07545460677114e-05,
+      "loss": 241.4176,
+      "step": 25020
+    },
+    {
+      "epoch": 0.48139012698275324,
+      "grad_norm": 511.62849953671076,
+      "learning_rate": 1.0748460183569385e-05,
+      "loss": 250.8242,
+      "step": 25030
+    },
+    {
+      "epoch": 0.48158245224323376,
+      "grad_norm": 447.31858981256545,
+      "learning_rate": 1.0742374020638315e-05,
+      "loss": 248.7033,
+      "step": 25040
+    },
+    {
+      "epoch": 0.4817747775037143,
+      "grad_norm": 421.99882091451155,
+      "learning_rate": 1.073628758118518e-05,
+      "loss": 255.6246,
+      "step": 25050
+    },
+    {
+      "epoch": 0.4819671027641948,
+      "grad_norm": 463.3980795704606,
+      "learning_rate": 1.0730200867477083e-05,
+      "loss": 245.073,
+      "step": 25060
+    },
+    {
+      "epoch": 0.4821594280246753,
+      "grad_norm": 467.94884056645355,
+      "learning_rate": 1.0724113881781222e-05,
+      "loss": 247.966,
+      "step": 25070
+    },
+    {
+      "epoch": 0.48235175328515584,
+      "grad_norm": 511.85541483845395,
+      "learning_rate": 1.0718026626364902e-05,
+      "loss": 257.6318,
+      "step": 25080
+    },
+    {
+      "epoch": 0.48254407854563636,
+      "grad_norm": 429.096382339394,
+      "learning_rate": 1.0711939103495515e-05,
+      "loss": 245.9294,
+      "step": 25090
+    },
+    {
+      "epoch": 0.48273640380611693,
+      "grad_norm": 427.3352790828323,
+      "learning_rate": 1.0705851315440575e-05,
+      "loss": 248.8446,
+      "step": 25100
+    },
+    {
+      "epoch": 0.48292872906659745,
+      "grad_norm": 416.78510611831143,
+      "learning_rate": 1.0699763264467675e-05,
+      "loss": 248.8376,
+      "step": 25110
+    },
+    {
+      "epoch": 0.48312105432707797,
+      "grad_norm": 512.1657711730046,
+      "learning_rate": 1.069367495284452e-05,
+      "loss": 246.5034,
+      "step": 25120
+    },
+    {
+      "epoch": 0.4833133795875585,
+      "grad_norm": 442.83674156966015,
+      "learning_rate": 1.06875863828389e-05,
+      "loss": 238.0761,
+      "step": 25130
+    },
+    {
+      "epoch": 0.483505704848039,
+      "grad_norm": 488.48749967161496,
+      "learning_rate": 1.0681497556718706e-05,
+      "loss": 236.0335,
+      "step": 25140
+    },
+    {
+      "epoch": 0.4836980301085195,
+      "grad_norm": 456.7560492291506,
+      "learning_rate": 1.067540847675193e-05,
+      "loss": 272.2428,
+      "step": 25150
+    },
+    {
+      "epoch": 0.48389035536900005,
+      "grad_norm": 442.06040364008766,
+      "learning_rate": 1.0669319145206646e-05,
+      "loss": 250.605,
+      "step": 25160
+    },
+    {
+      "epoch": 0.48408268062948057,
+      "grad_norm": 566.2449344280966,
+      "learning_rate": 1.066322956435104e-05,
+      "loss": 253.4595,
+      "step": 25170
+    },
+    {
+      "epoch": 0.4842750058899611,
+      "grad_norm": 425.96797194049253,
+      "learning_rate": 1.0657139736453375e-05,
+      "loss": 256.1166,
+      "step": 25180
+    },
+    {
+      "epoch": 0.4844673311504416,
+      "grad_norm": 438.731119986068,
+      "learning_rate": 1.0651049663782007e-05,
+      "loss": 255.1635,
+      "step": 25190
+    },
+    {
+      "epoch": 0.4846596564109221,
+      "grad_norm": 451.60955684592363,
+      "learning_rate": 1.0644959348605397e-05,
+      "loss": 248.3435,
+      "step": 25200
+    },
+    {
+      "epoch": 0.4848519816714027,
+      "grad_norm": 429.3271931261198,
+      "learning_rate": 1.0638868793192079e-05,
+      "loss": 244.5909,
+      "step": 25210
+    },
+    {
+      "epoch": 0.4850443069318832,
+      "grad_norm": 481.2447928300619,
+      "learning_rate": 1.0632777999810685e-05,
+      "loss": 241.698,
+      "step": 25220
+    },
+    {
+      "epoch": 0.48523663219236374,
+      "grad_norm": 470.2915240312469,
+      "learning_rate": 1.062668697072994e-05,
+      "loss": 246.9471,
+      "step": 25230
+    },
+    {
+      "epoch": 0.48542895745284426,
+      "grad_norm": 443.08875142900945,
+      "learning_rate": 1.0620595708218646e-05,
+      "loss": 244.9708,
+      "step": 25240
+    },
+    {
+      "epoch": 0.4856212827133248,
+      "grad_norm": 443.63371264850485,
+      "learning_rate": 1.06145042145457e-05,
+      "loss": 251.5037,
+      "step": 25250
+    },
+    {
+      "epoch": 0.4858136079738053,
+      "grad_norm": 402.69981315065496,
+      "learning_rate": 1.0608412491980084e-05,
+      "loss": 242.0354,
+      "step": 25260
+    },
+    {
+      "epoch": 0.4860059332342858,
+      "grad_norm": 426.4094260441273,
+      "learning_rate": 1.0602320542790866e-05,
+      "loss": 245.9982,
+      "step": 25270
+    },
+    {
+      "epoch": 0.48619825849476633,
+      "grad_norm": 471.6454580413846,
+      "learning_rate": 1.0596228369247188e-05,
+      "loss": 249.9412,
+      "step": 25280
+    },
+    {
+      "epoch": 0.48639058375524685,
+      "grad_norm": 458.64366829222826,
+      "learning_rate": 1.0590135973618294e-05,
+      "loss": 243.9949,
+      "step": 25290
+    },
+    {
+      "epoch": 0.48658290901572737,
+      "grad_norm": 472.520503573571,
+      "learning_rate": 1.0584043358173492e-05,
+      "loss": 245.1111,
+      "step": 25300
+    },
+    {
+      "epoch": 0.48677523427620795,
+      "grad_norm": 439.99698347493904,
+      "learning_rate": 1.0577950525182189e-05,
+      "loss": 247.9348,
+      "step": 25310
+    },
+    {
+      "epoch": 0.48696755953668847,
+      "grad_norm": 453.09732479627297,
+      "learning_rate": 1.0571857476913856e-05,
+      "loss": 238.9893,
+      "step": 25320
+    },
+    {
+      "epoch": 0.487159884797169,
+      "grad_norm": 455.28499634673665,
+      "learning_rate": 1.056576421563806e-05,
+      "loss": 250.3431,
+      "step": 25330
+    },
+    {
+      "epoch": 0.4873522100576495,
+      "grad_norm": 436.19548366840934,
+      "learning_rate": 1.0559670743624439e-05,
+      "loss": 237.5524,
+      "step": 25340
+    },
+    {
+      "epoch": 0.48754453531813,
+      "grad_norm": 473.48646363292704,
+      "learning_rate": 1.0553577063142705e-05,
+      "loss": 254.6918,
+      "step": 25350
+    },
+    {
+      "epoch": 0.48773686057861054,
+      "grad_norm": 484.1705295658253,
+      "learning_rate": 1.0547483176462662e-05,
+      "loss": 251.9564,
+      "step": 25360
+    },
+    {
+      "epoch": 0.48792918583909106,
+      "grad_norm": 2366.65366049417,
+      "learning_rate": 1.0541389085854177e-05,
+      "loss": 249.7325,
+      "step": 25370
+    },
+    {
+      "epoch": 0.4881215110995716,
+      "grad_norm": 492.1477589057321,
+      "learning_rate": 1.0535294793587197e-05,
+      "loss": 250.2501,
+      "step": 25380
+    },
+    {
+      "epoch": 0.4883138363600521,
+      "grad_norm": 424.883471031728,
+      "learning_rate": 1.0529200301931747e-05,
+      "loss": 238.7042,
+      "step": 25390
+    },
+    {
+      "epoch": 0.4885061616205326,
+      "grad_norm": 423.5637215784562,
+      "learning_rate": 1.052310561315793e-05,
+      "loss": 246.725,
+      "step": 25400
+    },
+    {
+      "epoch": 0.4886984868810132,
+      "grad_norm": 423.773731351179,
+      "learning_rate": 1.0517010729535903e-05,
+      "loss": 251.5593,
+      "step": 25410
+    },
+    {
+      "epoch": 0.4888908121414937,
+      "grad_norm": 474.33298819325177,
+      "learning_rate": 1.0510915653335925e-05,
+      "loss": 243.2532,
+      "step": 25420
+    },
+    {
+      "epoch": 0.48908313740197423,
+      "grad_norm": 477.6213952086895,
+      "learning_rate": 1.05048203868283e-05,
+      "loss": 254.729,
+      "step": 25430
+    },
+    {
+      "epoch": 0.48927546266245475,
+      "grad_norm": 472.98644804818156,
+      "learning_rate": 1.0498724932283419e-05,
+      "loss": 246.5291,
+      "step": 25440
+    },
+    {
+      "epoch": 0.48946778792293527,
+      "grad_norm": 483.10040934297575,
+      "learning_rate": 1.0492629291971738e-05,
+      "loss": 250.702,
+      "step": 25450
+    },
+    {
+      "epoch": 0.4896601131834158,
+      "grad_norm": 440.58067932925127,
+      "learning_rate": 1.0486533468163782e-05,
+      "loss": 244.2888,
+      "step": 25460
+    },
+    {
+      "epoch": 0.4898524384438963,
+      "grad_norm": 394.8266921690608,
+      "learning_rate": 1.0480437463130145e-05,
+      "loss": 241.5677,
+      "step": 25470
+    },
+    {
+      "epoch": 0.49004476370437683,
+      "grad_norm": 471.3324989295299,
+      "learning_rate": 1.0474341279141486e-05,
+      "loss": 252.2472,
+      "step": 25480
+    },
+    {
+      "epoch": 0.49023708896485735,
+      "grad_norm": 425.21399499721247,
+      "learning_rate": 1.0468244918468538e-05,
+      "loss": 249.1725,
+      "step": 25490
+    },
+    {
+      "epoch": 0.49042941422533787,
+      "grad_norm": 447.2930020504913,
+      "learning_rate": 1.0462148383382086e-05,
+      "loss": 244.4379,
+      "step": 25500
+    },
+    {
+      "epoch": 0.49062173948581844,
+      "grad_norm": 485.2339050530055,
+      "learning_rate": 1.0456051676152996e-05,
+      "loss": 242.2057,
+      "step": 25510
+    },
+    {
+      "epoch": 0.49081406474629896,
+      "grad_norm": 646.7097083310359,
+      "learning_rate": 1.0449954799052189e-05,
+      "loss": 251.574,
+      "step": 25520
+    },
+    {
+      "epoch": 0.4910063900067795,
+      "grad_norm": 547.2044713815735,
+      "learning_rate": 1.044385775435065e-05,
+      "loss": 253.0338,
+      "step": 25530
+    },
+    {
+      "epoch": 0.49119871526726,
+      "grad_norm": 440.4239265355097,
+      "learning_rate": 1.043776054431943e-05,
+      "loss": 245.8395,
+      "step": 25540
+    },
+    {
+      "epoch": 0.4913910405277405,
+      "grad_norm": 416.9467677099068,
+      "learning_rate": 1.0431663171229636e-05,
+      "loss": 245.2929,
+      "step": 25550
+    },
+    {
+      "epoch": 0.49158336578822104,
+      "grad_norm": 431.6267169132507,
+      "learning_rate": 1.0425565637352441e-05,
+      "loss": 249.7551,
+      "step": 25560
+    },
+    {
+      "epoch": 0.49177569104870156,
+      "grad_norm": 465.0519131476555,
+      "learning_rate": 1.0419467944959072e-05,
+      "loss": 249.6737,
+      "step": 25570
+    },
+    {
+      "epoch": 0.4919680163091821,
+      "grad_norm": 452.2856803244398,
+      "learning_rate": 1.0413370096320823e-05,
+      "loss": 238.5953,
+      "step": 25580
+    },
+    {
+      "epoch": 0.4921603415696626,
+      "grad_norm": 455.0734526442802,
+      "learning_rate": 1.0407272093709038e-05,
+      "loss": 245.2968,
+      "step": 25590
+    },
+    {
+      "epoch": 0.4923526668301431,
+      "grad_norm": 463.6321192841197,
+      "learning_rate": 1.0401173939395128e-05,
+      "loss": 244.2217,
+      "step": 25600
+    },
+    {
+      "epoch": 0.4925449920906237,
+      "grad_norm": 485.84517009301635,
+      "learning_rate": 1.0395075635650549e-05,
+      "loss": 241.1392,
+      "step": 25610
+    },
+    {
+      "epoch": 0.4927373173511042,
+      "grad_norm": 419.260555430691,
+      "learning_rate": 1.038897718474682e-05,
+      "loss": 246.546,
+      "step": 25620
+    },
+    {
+      "epoch": 0.49292964261158473,
+      "grad_norm": 433.4192507139854,
+      "learning_rate": 1.0382878588955517e-05,
+      "loss": 251.3673,
+      "step": 25630
+    },
+    {
+      "epoch": 0.49312196787206525,
+      "grad_norm": 452.670606842314,
+      "learning_rate": 1.0376779850548257e-05,
+      "loss": 246.1752,
+      "step": 25640
+    },
+    {
+      "epoch": 0.49331429313254577,
+      "grad_norm": 429.65471329485945,
+      "learning_rate": 1.0370680971796732e-05,
+      "loss": 241.0601,
+      "step": 25650
+    },
+    {
+      "epoch": 0.4935066183930263,
+      "grad_norm": 467.61475859467316,
+      "learning_rate": 1.0364581954972662e-05,
+      "loss": 240.7547,
+      "step": 25660
+    },
+    {
+      "epoch": 0.4936989436535068,
+      "grad_norm": 442.5152961101893,
+      "learning_rate": 1.0358482802347838e-05,
+      "loss": 248.5261,
+      "step": 25670
+    },
+    {
+      "epoch": 0.4938912689139873,
+      "grad_norm": 552.073007450886,
+      "learning_rate": 1.0352383516194088e-05,
+      "loss": 246.4259,
+      "step": 25680
+    },
+    {
+      "epoch": 0.49408359417446784,
+      "grad_norm": 446.08519703837686,
+      "learning_rate": 1.0346284098783304e-05,
+      "loss": 247.5358,
+      "step": 25690
+    },
+    {
+      "epoch": 0.49427591943494836,
+      "grad_norm": 442.76307004936734,
+      "learning_rate": 1.0340184552387406e-05,
+      "loss": 246.4056,
+      "step": 25700
+    },
+    {
+      "epoch": 0.4944682446954289,
+      "grad_norm": 497.5888563331776,
+      "learning_rate": 1.0334084879278381e-05,
+      "loss": 246.5515,
+      "step": 25710
+    },
+    {
+      "epoch": 0.49466056995590946,
+      "grad_norm": 425.2440346101982,
+      "learning_rate": 1.032798508172826e-05,
+      "loss": 246.9058,
+      "step": 25720
+    },
+    {
+      "epoch": 0.49485289521639,
+      "grad_norm": 402.90499175113706,
+      "learning_rate": 1.0321885162009111e-05,
+      "loss": 244.8918,
+      "step": 25730
+    },
+    {
+      "epoch": 0.4950452204768705,
+      "grad_norm": 402.03117143256645,
+      "learning_rate": 1.0315785122393053e-05,
+      "loss": 242.0849,
+      "step": 25740
+    },
+    {
+      "epoch": 0.495237545737351,
+      "grad_norm": 476.15911193233376,
+      "learning_rate": 1.0309684965152254e-05,
+      "loss": 248.6827,
+      "step": 25750
+    },
+    {
+      "epoch": 0.49542987099783153,
+      "grad_norm": 475.220916630608,
+      "learning_rate": 1.030358469255892e-05,
+      "loss": 250.3099,
+      "step": 25760
+    },
+    {
+      "epoch": 0.49562219625831205,
+      "grad_norm": 501.3222759334091,
+      "learning_rate": 1.0297484306885304e-05,
+      "loss": 244.1939,
+      "step": 25770
+    },
+    {
+      "epoch": 0.4958145215187926,
+      "grad_norm": 397.63599090176484,
+      "learning_rate": 1.0291383810403697e-05,
+      "loss": 245.2867,
+      "step": 25780
+    },
+    {
+      "epoch": 0.4960068467792731,
+      "grad_norm": 450.37578679972626,
+      "learning_rate": 1.028528320538643e-05,
+      "loss": 246.428,
+      "step": 25790
+    },
+    {
+      "epoch": 0.4961991720397536,
+      "grad_norm": 424.172891105323,
+      "learning_rate": 1.0279182494105879e-05,
+      "loss": 240.4534,
+      "step": 25800
+    },
+    {
+      "epoch": 0.49639149730023413,
+      "grad_norm": 442.1861728522265,
+      "learning_rate": 1.0273081678834462e-05,
+      "loss": 239.6894,
+      "step": 25810
+    },
+    {
+      "epoch": 0.4965838225607147,
+      "grad_norm": 420.1162594312604,
+      "learning_rate": 1.026698076184463e-05,
+      "loss": 238.7768,
+      "step": 25820
+    },
+    {
+      "epoch": 0.4967761478211952,
+      "grad_norm": 512.6485580085946,
+      "learning_rate": 1.0260879745408876e-05,
+      "loss": 246.5191,
+      "step": 25830
+    },
+    {
+      "epoch": 0.49696847308167574,
+      "grad_norm": 503.2180245457224,
+      "learning_rate": 1.0254778631799722e-05,
+      "loss": 237.8291,
+      "step": 25840
+    },
+    {
+      "epoch": 0.49716079834215626,
+      "grad_norm": 460.8060087247976,
+      "learning_rate": 1.024867742328974e-05,
+      "loss": 251.3185,
+      "step": 25850
+    },
+    {
+      "epoch": 0.4973531236026368,
+      "grad_norm": 502.38381757118526,
+      "learning_rate": 1.024257612215152e-05,
+      "loss": 247.1877,
+      "step": 25860
+    },
+    {
+      "epoch": 0.4975454488631173,
+      "grad_norm": 466.06905872812473,
+      "learning_rate": 1.02364747306577e-05,
+      "loss": 247.8401,
+      "step": 25870
+    },
+    {
+      "epoch": 0.4977377741235978,
+      "grad_norm": 464.63045897343216,
+      "learning_rate": 1.023037325108095e-05,
+      "loss": 244.7431,
+      "step": 25880
+    },
+    {
+      "epoch": 0.49793009938407834,
+      "grad_norm": 434.22745832265616,
+      "learning_rate": 1.022427168569397e-05,
+      "loss": 240.8298,
+      "step": 25890
+    },
+    {
+      "epoch": 0.49812242464455886,
+      "grad_norm": 412.5713094350459,
+      "learning_rate": 1.021817003676949e-05,
+      "loss": 247.3949,
+      "step": 25900
+    },
+    {
+      "epoch": 0.4983147499050394,
+      "grad_norm": 523.4399446351441,
+      "learning_rate": 1.0212068306580274e-05,
+      "loss": 250.9017,
+      "step": 25910
+    },
+    {
+      "epoch": 0.49850707516551995,
+      "grad_norm": 429.88017138861875,
+      "learning_rate": 1.0205966497399118e-05,
+      "loss": 236.3815,
+      "step": 25920
+    },
+    {
+      "epoch": 0.49869940042600047,
+      "grad_norm": 466.77705788075514,
+      "learning_rate": 1.0199864611498841e-05,
+      "loss": 250.0941,
+      "step": 25930
+    },
+    {
+      "epoch": 0.498891725686481,
+      "grad_norm": 418.65971555969014,
+      "learning_rate": 1.0193762651152299e-05,
+      "loss": 250.9745,
+      "step": 25940
+    },
+    {
+      "epoch": 0.4990840509469615,
+      "grad_norm": 449.96314508050347,
+      "learning_rate": 1.018766061863237e-05,
+      "loss": 243.4652,
+      "step": 25950
+    },
+    {
+      "epoch": 0.49927637620744203,
+      "grad_norm": 489.6657370422981,
+      "learning_rate": 1.018155851621196e-05,
+      "loss": 245.073,
+      "step": 25960
+    },
+    {
+      "epoch": 0.49946870146792255,
+      "grad_norm": 420.4610545344127,
+      "learning_rate": 1.0175456346164e-05,
+      "loss": 242.4358,
+      "step": 25970
+    },
+    {
+      "epoch": 0.49966102672840307,
+      "grad_norm": 438.7971060637843,
+      "learning_rate": 1.0169354110761447e-05,
+      "loss": 243.3489,
+      "step": 25980
+    },
+    {
+      "epoch": 0.4998533519888836,
+      "grad_norm": 436.5208933561538,
+      "learning_rate": 1.0163251812277289e-05,
+      "loss": 241.6723,
+      "step": 25990
+    },
+    {
+      "epoch": 0.5000456772493641,
+      "grad_norm": 381.4129983150638,
+      "learning_rate": 1.0157149452984523e-05,
+      "loss": 245.7078,
+      "step": 26000
+    },
+    {
+      "epoch": 0.5002380025098446,
+      "grad_norm": 433.79289200845756,
+      "learning_rate": 1.0151047035156182e-05,
+      "loss": 244.9868,
+      "step": 26010
+    },
+    {
+      "epoch": 0.5004303277703251,
+      "grad_norm": 460.1908207982285,
+      "learning_rate": 1.0144944561065316e-05,
+      "loss": 241.7464,
+      "step": 26020
+    },
+    {
+      "epoch": 0.5006226530308057,
+      "grad_norm": 433.40965322735116,
+      "learning_rate": 1.0138842032984996e-05,
+      "loss": 257.1567,
+      "step": 26030
+    },
+    {
+      "epoch": 0.5008149782912862,
+      "grad_norm": 426.9756940354048,
+      "learning_rate": 1.013273945318831e-05,
+      "loss": 249.8501,
+      "step": 26040
+    },
+    {
+      "epoch": 0.5010073035517667,
+      "grad_norm": 411.1509040454855,
+      "learning_rate": 1.0126636823948373e-05,
+      "loss": 239.5032,
+      "step": 26050
+    },
+    {
+      "epoch": 0.5011996288122472,
+      "grad_norm": 478.99791632474876,
+      "learning_rate": 1.0120534147538305e-05,
+      "loss": 247.6124,
+      "step": 26060
+    },
+    {
+      "epoch": 0.5013919540727279,
+      "grad_norm": 409.0573405470129,
+      "learning_rate": 1.011443142623126e-05,
+      "loss": 251.2825,
+      "step": 26070
+    },
+    {
+      "epoch": 0.5015842793332084,
+      "grad_norm": 423.1886250758101,
+      "learning_rate": 1.0108328662300399e-05,
+      "loss": 243.0762,
+      "step": 26080
+    },
+    {
+      "epoch": 0.5017766045936889,
+      "grad_norm": 449.89564875024155,
+      "learning_rate": 1.0102225858018902e-05,
+      "loss": 239.3468,
+      "step": 26090
+    },
+    {
+      "epoch": 0.5019689298541694,
+      "grad_norm": 424.42439216757805,
+      "learning_rate": 1.009612301565996e-05,
+      "loss": 244.7099,
+      "step": 26100
+    },
+    {
+      "epoch": 0.5021612551146499,
+      "grad_norm": 425.75243466856773,
+      "learning_rate": 1.0090020137496783e-05,
+      "loss": 242.121,
+      "step": 26110
+    },
+    {
+      "epoch": 0.5023535803751304,
+      "grad_norm": 516.8689704463773,
+      "learning_rate": 1.008391722580259e-05,
+      "loss": 253.884,
+      "step": 26120
+    },
+    {
+      "epoch": 0.502545905635611,
+      "grad_norm": 462.37143231682705,
+      "learning_rate": 1.0077814282850617e-05,
+      "loss": 258.1923,
+      "step": 26130
+    },
+    {
+      "epoch": 0.5027382308960915,
+      "grad_norm": 424.00888323067863,
+      "learning_rate": 1.0071711310914111e-05,
+      "loss": 243.5125,
+      "step": 26140
+    },
+    {
+      "epoch": 0.502930556156572,
+      "grad_norm": 420.92516082000077,
+      "learning_rate": 1.0065608312266324e-05,
+      "loss": 246.5957,
+      "step": 26150
+    },
+    {
+      "epoch": 0.5031228814170525,
+      "grad_norm": 482.59303401324553,
+      "learning_rate": 1.005950528918052e-05,
+      "loss": 246.1805,
+      "step": 26160
+    },
+    {
+      "epoch": 0.503315206677533,
+      "grad_norm": 487.29229809229616,
+      "learning_rate": 1.0053402243929986e-05,
+      "loss": 251.2692,
+      "step": 26170
+    },
+    {
+      "epoch": 0.5035075319380136,
+      "grad_norm": 457.9805797458841,
+      "learning_rate": 1.0047299178787993e-05,
+      "loss": 245.5148,
+      "step": 26180
+    },
+    {
+      "epoch": 0.5036998571984941,
+      "grad_norm": 458.17136665917167,
+      "learning_rate": 1.0041196096027841e-05,
+      "loss": 241.697,
+      "step": 26190
+    },
+    {
+      "epoch": 0.5038921824589746,
+      "grad_norm": 435.9264394427311,
+      "learning_rate": 1.003509299792282e-05,
+      "loss": 242.2176,
+      "step": 26200
+    },
+    {
+      "epoch": 0.5040845077194551,
+      "grad_norm": 521.6332069641547,
+      "learning_rate": 1.0028989886746241e-05,
+      "loss": 234.7725,
+      "step": 26210
+    },
+    {
+      "epoch": 0.5042768329799356,
+      "grad_norm": 484.582780022787,
+      "learning_rate": 1.0022886764771405e-05,
+      "loss": 242.2138,
+      "step": 26220
+    },
+    {
+      "epoch": 0.5044691582404162,
+      "grad_norm": 444.34829278353914,
+      "learning_rate": 1.0016783634271626e-05,
+      "loss": 239.8629,
+      "step": 26230
+    },
+    {
+      "epoch": 0.5046614835008967,
+      "grad_norm": 457.58805933722306,
+      "learning_rate": 1.001068049752022e-05,
+      "loss": 243.8282,
+      "step": 26240
+    },
+    {
+      "epoch": 0.5048538087613772,
+      "grad_norm": 412.9655789715028,
+      "learning_rate": 1.0004577356790506e-05,
+      "loss": 237.5452,
+      "step": 26250
+    },
+    {
+      "epoch": 0.5050461340218577,
+      "grad_norm": 435.98213475968964,
+      "learning_rate": 9.998474214355805e-06,
+      "loss": 243.1981,
+      "step": 26260
+    },
+    {
+      "epoch": 0.5052384592823382,
+      "grad_norm": 489.4462227900642,
+      "learning_rate": 9.992371072489434e-06,
+      "loss": 249.8664,
+      "step": 26270
+    },
+    {
+      "epoch": 0.5054307845428189,
+      "grad_norm": 424.6057841930922,
+      "learning_rate": 9.986267933464707e-06,
+      "loss": 251.808,
+      "step": 26280
+    },
+    {
+      "epoch": 0.5056231098032994,
+      "grad_norm": 424.2281977649041,
+      "learning_rate": 9.980164799554953e-06,
+      "loss": 247.8646,
+      "step": 26290
+    },
+    {
+      "epoch": 0.5058154350637799,
+      "grad_norm": 487.14898516916384,
+      "learning_rate": 9.97406167303348e-06,
+      "loss": 244.8658,
+      "step": 26300
+    },
+    {
+      "epoch": 0.5060077603242604,
+      "grad_norm": 494.80154517631826,
+      "learning_rate": 9.967958556173612e-06,
+      "loss": 246.6637,
+      "step": 26310
+    },
+    {
+      "epoch": 0.5062000855847409,
+      "grad_norm": 513.2465036657109,
+      "learning_rate": 9.961855451248645e-06,
+      "loss": 239.1971,
+      "step": 26320
+    },
+    {
+      "epoch": 0.5063924108452215,
+      "grad_norm": 446.21360839134826,
+      "learning_rate": 9.955752360531896e-06,
+      "loss": 250.5916,
+      "step": 26330
+    },
+    {
+      "epoch": 0.506584736105702,
+      "grad_norm": 501.0889678350533,
+      "learning_rate": 9.949649286296663e-06,
+      "loss": 244.9018,
+      "step": 26340
+    },
+    {
+      "epoch": 0.5067770613661825,
+      "grad_norm": 465.91921016468143,
+      "learning_rate": 9.943546230816236e-06,
+      "loss": 244.6792,
+      "step": 26350
+    },
+    {
+      "epoch": 0.506969386626663,
+      "grad_norm": 457.0191770380282,
+      "learning_rate": 9.937443196363908e-06,
+      "loss": 249.4264,
+      "step": 26360
+    },
+    {
+      "epoch": 0.5071617118871435,
+      "grad_norm": 416.0070003608556,
+      "learning_rate": 9.931340185212955e-06,
+      "loss": 238.2517,
+      "step": 26370
+    },
+    {
+      "epoch": 0.5073540371476241,
+      "grad_norm": 476.43934334898836,
+      "learning_rate": 9.925237199636652e-06,
+      "loss": 236.8563,
+      "step": 26380
+    },
+    {
+      "epoch": 0.5075463624081046,
+      "grad_norm": 472.0325579638252,
+      "learning_rate": 9.919134241908252e-06,
+      "loss": 242.1155,
+      "step": 26390
+    },
+    {
+      "epoch": 0.5077386876685851,
+      "grad_norm": 464.23313098685463,
+      "learning_rate": 9.913031314301016e-06,
+      "loss": 249.2232,
+      "step": 26400
+    },
+    {
+      "epoch": 0.5079310129290656,
+      "grad_norm": 439.9356440277744,
+      "learning_rate": 9.906928419088178e-06,
+      "loss": 239.3025,
+      "step": 26410
+    },
+    {
+      "epoch": 0.5081233381895461,
+      "grad_norm": 452.02894711317396,
+      "learning_rate": 9.900825558542965e-06,
+      "loss": 248.1302,
+      "step": 26420
+    },
+    {
+      "epoch": 0.5083156634500267,
+      "grad_norm": 471.0250597842824,
+      "learning_rate": 9.894722734938595e-06,
+      "loss": 253.5517,
+      "step": 26430
+    },
+    {
+      "epoch": 0.5085079887105072,
+      "grad_norm": 942.0966381147898,
+      "learning_rate": 9.888619950548267e-06,
+      "loss": 241.6604,
+      "step": 26440
+    },
+    {
+      "epoch": 0.5087003139709877,
+      "grad_norm": 408.9074201010136,
+      "learning_rate": 9.88251720764517e-06,
+      "loss": 236.8139,
+      "step": 26450
+    },
+    {
+      "epoch": 0.5088926392314682,
+      "grad_norm": 436.98994804399365,
+      "learning_rate": 9.876414508502468e-06,
+      "loss": 242.7639,
+      "step": 26460
+    },
+    {
+      "epoch": 0.5090849644919487,
+      "grad_norm": 542.8409474696687,
+      "learning_rate": 9.870311855393324e-06,
+      "loss": 247.9546,
+      "step": 26470
+    },
+    {
+      "epoch": 0.5092772897524294,
+      "grad_norm": 492.2766221913677,
+      "learning_rate": 9.864209250590875e-06,
+      "loss": 236.1556,
+      "step": 26480
+    },
+    {
+      "epoch": 0.5094696150129099,
+      "grad_norm": 422.21524007550494,
+      "learning_rate": 9.858106696368235e-06,
+      "loss": 246.9354,
+      "step": 26490
+    },
+    {
+      "epoch": 0.5096619402733904,
+      "grad_norm": 413.45117700231884,
+      "learning_rate": 9.852004194998503e-06,
+      "loss": 240.2883,
+      "step": 26500
+    },
+    {
+      "epoch": 0.5098542655338709,
+      "grad_norm": 436.66097150765677,
+      "learning_rate": 9.845901748754767e-06,
+      "loss": 241.1681,
+      "step": 26510
+    },
+    {
+      "epoch": 0.5100465907943514,
+      "grad_norm": 401.0233561621916,
+      "learning_rate": 9.839799359910087e-06,
+      "loss": 245.1936,
+      "step": 26520
+    },
+    {
+      "epoch": 0.510238916054832,
+      "grad_norm": 419.60610598498783,
+      "learning_rate": 9.833697030737495e-06,
+      "loss": 255.0598,
+      "step": 26530
+    },
+    {
+      "epoch": 0.5104312413153125,
+      "grad_norm": 438.01226464221594,
+      "learning_rate": 9.827594763510016e-06,
+      "loss": 247.8375,
+      "step": 26540
+    },
+    {
+      "epoch": 0.510623566575793,
+      "grad_norm": 424.9794514517412,
+      "learning_rate": 9.821492560500641e-06,
+      "loss": 240.593,
+      "step": 26550
+    },
+    {
+      "epoch": 0.5108158918362735,
+      "grad_norm": 429.37616141942925,
+      "learning_rate": 9.815390423982339e-06,
+      "loss": 242.2971,
+      "step": 26560
+    },
+    {
+      "epoch": 0.511008217096754,
+      "grad_norm": 467.3883836438913,
+      "learning_rate": 9.809288356228051e-06,
+      "loss": 241.1749,
+      "step": 26570
+    },
+    {
+      "epoch": 0.5112005423572346,
+      "grad_norm": 455.14106394254037,
+      "learning_rate": 9.803186359510701e-06,
+      "loss": 240.6573,
+      "step": 26580
+    },
+    {
+      "epoch": 0.5113928676177151,
+      "grad_norm": 476.20805163386433,
+      "learning_rate": 9.797084436103186e-06,
+      "loss": 243.7622,
+      "step": 26590
+    },
+    {
+      "epoch": 0.5115851928781956,
+      "grad_norm": 431.13745569498144,
+      "learning_rate": 9.790982588278362e-06,
+      "loss": 245.6231,
+      "step": 26600
+    },
+    {
+      "epoch": 0.5117775181386761,
+      "grad_norm": 441.6160319727721,
+      "learning_rate": 9.784880818309076e-06,
+      "loss": 240.248,
+      "step": 26610
+    },
+    {
+      "epoch": 0.5119698433991566,
+      "grad_norm": 436.37144776940096,
+      "learning_rate": 9.778779128468133e-06,
+      "loss": 258.2943,
+      "step": 26620
+    },
+    {
+      "epoch": 0.5121621686596372,
+      "grad_norm": 438.78771472645104,
+      "learning_rate": 9.77267752102831e-06,
+      "loss": 243.6253,
+      "step": 26630
+    },
+    {
+      "epoch": 0.5123544939201177,
+      "grad_norm": 449.6758644467159,
+      "learning_rate": 9.766575998262353e-06,
+      "loss": 239.4953,
+      "step": 26640
+    },
+    {
+      "epoch": 0.5125468191805982,
+      "grad_norm": 457.8962878078774,
+      "learning_rate": 9.760474562442984e-06,
+      "loss": 256.9004,
+      "step": 26650
+    },
+    {
+      "epoch": 0.5127391444410787,
+      "grad_norm": 439.947448184104,
+      "learning_rate": 9.754373215842884e-06,
+      "loss": 245.4871,
+      "step": 26660
+    },
+    {
+      "epoch": 0.5129314697015592,
+      "grad_norm": 497.6137428470242,
+      "learning_rate": 9.748271960734708e-06,
+      "loss": 250.7543,
+      "step": 26670
+    },
+    {
+      "epoch": 0.5131237949620397,
+      "grad_norm": 429.4401569117728,
+      "learning_rate": 9.742170799391063e-06,
+      "loss": 234.9992,
+      "step": 26680
+    },
+    {
+      "epoch": 0.5133161202225204,
+      "grad_norm": 430.57617568758025,
+      "learning_rate": 9.736069734084541e-06,
+      "loss": 234.2473,
+      "step": 26690
+    },
+    {
+      "epoch": 0.5135084454830009,
+      "grad_norm": 451.1655195728304,
+      "learning_rate": 9.729968767087685e-06,
+      "loss": 239.7917,
+      "step": 26700
+    },
+    {
+      "epoch": 0.5137007707434814,
+      "grad_norm": 447.969824881567,
+      "learning_rate": 9.723867900673e-06,
+      "loss": 235.8606,
+      "step": 26710
+    },
+    {
+      "epoch": 0.5138930960039619,
+      "grad_norm": 430.6511466725646,
+      "learning_rate": 9.717767137112964e-06,
+      "loss": 242.5631,
+      "step": 26720
+    },
+    {
+      "epoch": 0.5140854212644425,
+      "grad_norm": 479.60565448422625,
+      "learning_rate": 9.711666478680007e-06,
+      "loss": 246.9556,
+      "step": 26730
+    },
+    {
+      "epoch": 0.514277746524923,
+      "grad_norm": 433.5290199664727,
+      "learning_rate": 9.705565927646526e-06,
+      "loss": 240.6249,
+      "step": 26740
+    },
+    {
+      "epoch": 0.5144700717854035,
+      "grad_norm": 493.95997173958085,
+      "learning_rate": 9.699465486284871e-06,
+      "loss": 240.8255,
+      "step": 26750
+    },
+    {
+      "epoch": 0.514662397045884,
+      "grad_norm": 458.84923128222306,
+      "learning_rate": 9.693365156867363e-06,
+      "loss": 237.2341,
+      "step": 26760
+    },
+    {
+      "epoch": 0.5148547223063645,
+      "grad_norm": 489.1075081727761,
+      "learning_rate": 9.68726494166627e-06,
+      "loss": 238.0743,
+      "step": 26770
+    },
+    {
+      "epoch": 0.515047047566845,
+      "grad_norm": 456.55188867455564,
+      "learning_rate": 9.681164842953816e-06,
+      "loss": 241.2977,
+      "step": 26780
+    },
+    {
+      "epoch": 0.5152393728273256,
+      "grad_norm": 435.1984211593493,
+      "learning_rate": 9.675064863002196e-06,
+      "loss": 242.6891,
+      "step": 26790
+    },
+    {
+      "epoch": 0.5154316980878061,
+      "grad_norm": 486.9208367025354,
+      "learning_rate": 9.668965004083549e-06,
+      "loss": 246.2692,
+      "step": 26800
+    },
+    {
+      "epoch": 0.5156240233482866,
+      "grad_norm": 488.1283763686715,
+      "learning_rate": 9.66286526846997e-06,
+      "loss": 244.1956,
+      "step": 26810
+    },
+    {
+      "epoch": 0.5158163486087671,
+      "grad_norm": 430.6743548810669,
+      "learning_rate": 9.656765658433507e-06,
+      "loss": 238.2999,
+      "step": 26820
+    },
+    {
+      "epoch": 0.5160086738692476,
+      "grad_norm": 480.70242654871583,
+      "learning_rate": 9.650666176246171e-06,
+      "loss": 240.7867,
+      "step": 26830
+    },
+    {
+      "epoch": 0.5162009991297282,
+      "grad_norm": 475.7212384357653,
+      "learning_rate": 9.644566824179916e-06,
+      "loss": 235.7113,
+      "step": 26840
+    },
+    {
+      "epoch": 0.5163933243902087,
+      "grad_norm": 424.0596093705765,
+      "learning_rate": 9.638467604506648e-06,
+      "loss": 243.6052,
+      "step": 26850
+    },
+    {
+      "epoch": 0.5165856496506892,
+      "grad_norm": 412.0695301887001,
+      "learning_rate": 9.632368519498224e-06,
+      "loss": 239.4665,
+      "step": 26860
+    },
+    {
+      "epoch": 0.5167779749111697,
+      "grad_norm": 438.08596085568695,
+      "learning_rate": 9.626269571426456e-06,
+      "loss": 249.0358,
+      "step": 26870
+    },
+    {
+      "epoch": 0.5169703001716502,
+      "grad_norm": 424.6792926233746,
+      "learning_rate": 9.620170762563103e-06,
+      "loss": 235.2679,
+      "step": 26880
+    },
+    {
+      "epoch": 0.5171626254321309,
+      "grad_norm": 455.23997407909974,
+      "learning_rate": 9.614072095179862e-06,
+      "loss": 240.3182,
+      "step": 26890
+    },
+    {
+      "epoch": 0.5173549506926114,
+      "grad_norm": 460.3046566571023,
+      "learning_rate": 9.607973571548396e-06,
+      "loss": 248.8181,
+      "step": 26900
+    },
+    {
+      "epoch": 0.5175472759530919,
+      "grad_norm": 431.60376319277043,
+      "learning_rate": 9.601875193940301e-06,
+      "loss": 235.6848,
+      "step": 26910
+    },
+    {
+      "epoch": 0.5177396012135724,
+      "grad_norm": 427.4764491510471,
+      "learning_rate": 9.595776964627119e-06,
+      "loss": 238.2073,
+      "step": 26920
+    },
+    {
+      "epoch": 0.517931926474053,
+      "grad_norm": 405.6455663795896,
+      "learning_rate": 9.589678885880342e-06,
+      "loss": 243.4056,
+      "step": 26930
+    },
+    {
+      "epoch": 0.5181242517345335,
+      "grad_norm": 450.5246242432972,
+      "learning_rate": 9.583580959971402e-06,
+      "loss": 240.6424,
+      "step": 26940
+    },
+    {
+      "epoch": 0.518316576995014,
+      "grad_norm": 434.50219769503724,
+      "learning_rate": 9.577483189171681e-06,
+      "loss": 247.0905,
+      "step": 26950
+    },
+    {
+      "epoch": 0.5185089022554945,
+      "grad_norm": 405.72505005902167,
+      "learning_rate": 9.571385575752487e-06,
+      "loss": 237.8846,
+      "step": 26960
+    },
+    {
+      "epoch": 0.518701227515975,
+      "grad_norm": 499.28480534345977,
+      "learning_rate": 9.565288121985093e-06,
+      "loss": 249.2304,
+      "step": 26970
+    },
+    {
+      "epoch": 0.5188935527764555,
+      "grad_norm": 430.11596495560156,
+      "learning_rate": 9.559190830140695e-06,
+      "loss": 244.7042,
+      "step": 26980
+    },
+    {
+      "epoch": 0.5190858780369361,
+      "grad_norm": 440.6001477530912,
+      "learning_rate": 9.553093702490433e-06,
+      "loss": 237.3319,
+      "step": 26990
+    },
+    {
+      "epoch": 0.5192782032974166,
+      "grad_norm": 439.4747838485815,
+      "learning_rate": 9.54699674130538e-06,
+      "loss": 240.8133,
+      "step": 27000
+    },
+    {
+      "epoch": 0.5194705285578971,
+      "grad_norm": 386.079623940713,
+      "learning_rate": 9.540899948856561e-06,
+      "loss": 233.2224,
+      "step": 27010
+    },
+    {
+      "epoch": 0.5196628538183776,
+      "grad_norm": 410.97870398563015,
+      "learning_rate": 9.534803327414931e-06,
+      "loss": 246.0451,
+      "step": 27020
+    },
+    {
+      "epoch": 0.5198551790788581,
+      "grad_norm": 465.3544311644236,
+      "learning_rate": 9.52870687925138e-06,
+      "loss": 243.134,
+      "step": 27030
+    },
+    {
+      "epoch": 0.5200475043393387,
+      "grad_norm": 475.8821377875846,
+      "learning_rate": 9.522610606636728e-06,
+      "loss": 243.3544,
+      "step": 27040
+    },
+    {
+      "epoch": 0.5202398295998192,
+      "grad_norm": 448.41143347877056,
+      "learning_rate": 9.516514511841745e-06,
+      "loss": 243.6413,
+      "step": 27050
+    },
+    {
+      "epoch": 0.5204321548602997,
+      "grad_norm": 432.3181110874596,
+      "learning_rate": 9.51041859713712e-06,
+      "loss": 233.4422,
+      "step": 27060
+    },
+    {
+      "epoch": 0.5206244801207802,
+      "grad_norm": 426.23762058439485,
+      "learning_rate": 9.50432286479348e-06,
+      "loss": 239.3122,
+      "step": 27070
+    },
+    {
+      "epoch": 0.5208168053812607,
+      "grad_norm": 445.9134298836942,
+      "learning_rate": 9.498227317081387e-06,
+      "loss": 241.3759,
+      "step": 27080
+    },
+    {
+      "epoch": 0.5210091306417413,
+      "grad_norm": 475.92520609272293,
+      "learning_rate": 9.492131956271334e-06,
+      "loss": 240.0812,
+      "step": 27090
+    },
+    {
+      "epoch": 0.5212014559022219,
+      "grad_norm": 473.55989336650714,
+      "learning_rate": 9.486036784633738e-06,
+      "loss": 241.4949,
+      "step": 27100
+    },
+    {
+      "epoch": 0.5213937811627024,
+      "grad_norm": 467.69774689538417,
+      "learning_rate": 9.47994180443895e-06,
+      "loss": 235.2704,
+      "step": 27110
+    },
+    {
+      "epoch": 0.5215861064231829,
+      "grad_norm": 453.1111708888243,
+      "learning_rate": 9.473847017957254e-06,
+      "loss": 240.8744,
+      "step": 27120
+    },
+    {
+      "epoch": 0.5217784316836634,
+      "grad_norm": 446.0245501939967,
+      "learning_rate": 9.467752427458851e-06,
+      "loss": 243.1803,
+      "step": 27130
+    },
+    {
+      "epoch": 0.521970756944144,
+      "grad_norm": 387.91540224222604,
+      "learning_rate": 9.461658035213878e-06,
+      "loss": 239.4184,
+      "step": 27140
+    },
+    {
+      "epoch": 0.5221630822046245,
+      "grad_norm": 427.88291545084644,
+      "learning_rate": 9.455563843492397e-06,
+      "loss": 242.6393,
+      "step": 27150
+    },
+    {
+      "epoch": 0.522355407465105,
+      "grad_norm": 482.08854329715166,
+      "learning_rate": 9.449469854564393e-06,
+      "loss": 236.5583,
+      "step": 27160
+    },
+    {
+      "epoch": 0.5225477327255855,
+      "grad_norm": 418.32751303485225,
+      "learning_rate": 9.443376070699778e-06,
+      "loss": 242.7046,
+      "step": 27170
+    },
+    {
+      "epoch": 0.522740057986066,
+      "grad_norm": 476.7775685830085,
+      "learning_rate": 9.437282494168379e-06,
+      "loss": 237.5189,
+      "step": 27180
+    },
+    {
+      "epoch": 0.5229323832465466,
+      "grad_norm": 437.47874987469135,
+      "learning_rate": 9.431189127239962e-06,
+      "loss": 248.3712,
+      "step": 27190
+    },
+    {
+      "epoch": 0.5231247085070271,
+      "grad_norm": 461.8676345622946,
+      "learning_rate": 9.4250959721842e-06,
+      "loss": 236.9921,
+      "step": 27200
+    },
+    {
+      "epoch": 0.5233170337675076,
+      "grad_norm": 459.48575264396266,
+      "learning_rate": 9.419003031270692e-06,
+      "loss": 241.2283,
+      "step": 27210
+    },
+    {
+      "epoch": 0.5235093590279881,
+      "grad_norm": 411.71970850264603,
+      "learning_rate": 9.412910306768959e-06,
+      "loss": 235.0745,
+      "step": 27220
+    },
+    {
+      "epoch": 0.5237016842884686,
+      "grad_norm": 429.01473497405584,
+      "learning_rate": 9.40681780094844e-06,
+      "loss": 233.6244,
+      "step": 27230
+    },
+    {
+      "epoch": 0.5238940095489492,
+      "grad_norm": 456.5527863494388,
+      "learning_rate": 9.400725516078496e-06,
+      "loss": 238.7592,
+      "step": 27240
+    },
+    {
+      "epoch": 0.5240863348094297,
+      "grad_norm": 457.7871086447985,
+      "learning_rate": 9.394633454428396e-06,
+      "loss": 245.8387,
+      "step": 27250
+    },
+    {
+      "epoch": 0.5242786600699102,
+      "grad_norm": 413.2930024819135,
+      "learning_rate": 9.388541618267341e-06,
+      "loss": 243.7098,
+      "step": 27260
+    },
+    {
+      "epoch": 0.5244709853303907,
+      "grad_norm": 431.8288670302755,
+      "learning_rate": 9.382450009864434e-06,
+      "loss": 245.2352,
+      "step": 27270
+    },
+    {
+      "epoch": 0.5246633105908712,
+      "grad_norm": 410.1399914861093,
+      "learning_rate": 9.376358631488697e-06,
+      "loss": 230.0802,
+      "step": 27280
+    },
+    {
+      "epoch": 0.5248556358513518,
+      "grad_norm": 407.2477501744951,
+      "learning_rate": 9.37026748540907e-06,
+      "loss": 242.6817,
+      "step": 27290
+    },
+    {
+      "epoch": 0.5250479611118324,
+      "grad_norm": 466.8828377221575,
+      "learning_rate": 9.364176573894404e-06,
+      "loss": 234.4508,
+      "step": 27300
+    },
+    {
+      "epoch": 0.5252402863723129,
+      "grad_norm": 502.122779679048,
+      "learning_rate": 9.358085899213467e-06,
+      "loss": 241.0622,
+      "step": 27310
+    },
+    {
+      "epoch": 0.5254326116327934,
+      "grad_norm": 413.17550450905907,
+      "learning_rate": 9.351995463634925e-06,
+      "loss": 237.4504,
+      "step": 27320
+    },
+    {
+      "epoch": 0.5256249368932739,
+      "grad_norm": 429.2760879714301,
+      "learning_rate": 9.345905269427374e-06,
+      "loss": 236.3551,
+      "step": 27330
+    },
+    {
+      "epoch": 0.5258172621537545,
+      "grad_norm": 389.53842007431916,
+      "learning_rate": 9.33981531885931e-06,
+      "loss": 243.789,
+      "step": 27340
+    },
+    {
+      "epoch": 0.526009587414235,
+      "grad_norm": 459.72382965221465,
+      "learning_rate": 9.333725614199132e-06,
+      "loss": 241.6083,
+      "step": 27350
+    },
+    {
+      "epoch": 0.5262019126747155,
+      "grad_norm": 494.718895807746,
+      "learning_rate": 9.32763615771516e-06,
+      "loss": 243.3535,
+      "step": 27360
+    },
+    {
+      "epoch": 0.526394237935196,
+      "grad_norm": 563.2931188281974,
+      "learning_rate": 9.321546951675616e-06,
+      "loss": 242.2146,
+      "step": 27370
+    },
+    {
+      "epoch": 0.5265865631956765,
+      "grad_norm": 404.87914951646826,
+      "learning_rate": 9.315457998348627e-06,
+      "loss": 236.3093,
+      "step": 27380
+    },
+    {
+      "epoch": 0.5267788884561571,
+      "grad_norm": 421.0440915530679,
+      "learning_rate": 9.309369300002224e-06,
+      "loss": 249.0604,
+      "step": 27390
+    },
+    {
+      "epoch": 0.5269712137166376,
+      "grad_norm": 455.40221618630517,
+      "learning_rate": 9.303280858904356e-06,
+      "loss": 234.8021,
+      "step": 27400
+    },
+    {
+      "epoch": 0.5271635389771181,
+      "grad_norm": 439.8625948302137,
+      "learning_rate": 9.297192677322862e-06,
+      "loss": 245.3899,
+      "step": 27410
+    },
+    {
+      "epoch": 0.5273558642375986,
+      "grad_norm": 419.8388523998324,
+      "learning_rate": 9.291104757525486e-06,
+      "loss": 245.9283,
+      "step": 27420
+    },
+    {
+      "epoch": 0.5275481894980791,
+      "grad_norm": 472.0594473081434,
+      "learning_rate": 9.28501710177988e-06,
+      "loss": 232.1825,
+      "step": 27430
+    },
+    {
+      "epoch": 0.5277405147585597,
+      "grad_norm": 452.6552138263836,
+      "learning_rate": 9.278929712353595e-06,
+      "loss": 245.7812,
+      "step": 27440
+    },
+    {
+      "epoch": 0.5279328400190402,
+      "grad_norm": 415.5844956131069,
+      "learning_rate": 9.272842591514083e-06,
+      "loss": 234.7244,
+      "step": 27450
+    },
+    {
+      "epoch": 0.5281251652795207,
+      "grad_norm": 458.11289825003047,
+      "learning_rate": 9.2667557415287e-06,
+      "loss": 242.4748,
+      "step": 27460
+    },
+    {
+      "epoch": 0.5283174905400012,
+      "grad_norm": 496.39703944737977,
+      "learning_rate": 9.260669164664687e-06,
+      "loss": 245.5343,
+      "step": 27470
+    },
+    {
+      "epoch": 0.5285098158004817,
+      "grad_norm": 427.5664185579466,
+      "learning_rate": 9.254582863189205e-06,
+      "loss": 246.3929,
+      "step": 27480
+    },
+    {
+      "epoch": 0.5287021410609622,
+      "grad_norm": 511.72804177530463,
+      "learning_rate": 9.248496839369293e-06,
+      "loss": 234.7416,
+      "step": 27490
+    },
+    {
+      "epoch": 0.5288944663214428,
+      "grad_norm": 431.86408011885624,
+      "learning_rate": 9.242411095471897e-06,
+      "loss": 258.5262,
+      "step": 27500
+    },
+    {
+      "epoch": 0.5290867915819234,
+      "grad_norm": 399.58378239258275,
+      "learning_rate": 9.236325633763856e-06,
+      "loss": 238.2345,
+      "step": 27510
+    },
+    {
+      "epoch": 0.5292791168424039,
+      "grad_norm": 390.55367240708824,
+      "learning_rate": 9.230240456511905e-06,
+      "loss": 250.9607,
+      "step": 27520
+    },
+    {
+      "epoch": 0.5294714421028844,
+      "grad_norm": 425.48540248480083,
+      "learning_rate": 9.224155565982673e-06,
+      "loss": 235.8497,
+      "step": 27530
+    },
+    {
+      "epoch": 0.529663767363365,
+      "grad_norm": 432.15339751457645,
+      "learning_rate": 9.218070964442673e-06,
+      "loss": 241.456,
+      "step": 27540
+    },
+    {
+      "epoch": 0.5298560926238455,
+      "grad_norm": 435.0571206833872,
+      "learning_rate": 9.21198665415833e-06,
+      "loss": 240.1762,
+      "step": 27550
+    },
+    {
+      "epoch": 0.530048417884326,
+      "grad_norm": 425.7631194357518,
+      "learning_rate": 9.205902637395943e-06,
+      "loss": 242.1126,
+      "step": 27560
+    },
+    {
+      "epoch": 0.5302407431448065,
+      "grad_norm": 464.8717870887204,
+      "learning_rate": 9.199818916421706e-06,
+      "loss": 234.666,
+      "step": 27570
+    },
+    {
+      "epoch": 0.530433068405287,
+      "grad_norm": 415.10545006700994,
+      "learning_rate": 9.193735493501707e-06,
+      "loss": 239.2713,
+      "step": 27580
+    },
+    {
+      "epoch": 0.5306253936657676,
+      "grad_norm": 453.1645297087392,
+      "learning_rate": 9.187652370901925e-06,
+      "loss": 245.0213,
+      "step": 27590
+    },
+    {
+      "epoch": 0.5308177189262481,
+      "grad_norm": 425.0915885104322,
+      "learning_rate": 9.181569550888217e-06,
+      "loss": 246.4904,
+      "step": 27600
+    },
+    {
+      "epoch": 0.5310100441867286,
+      "grad_norm": 449.4949800991947,
+      "learning_rate": 9.175487035726332e-06,
+      "loss": 236.694,
+      "step": 27610
+    },
+    {
+      "epoch": 0.5312023694472091,
+      "grad_norm": 410.62219552058235,
+      "learning_rate": 9.169404827681912e-06,
+      "loss": 234.5753,
+      "step": 27620
+    },
+    {
+      "epoch": 0.5313946947076896,
+      "grad_norm": 430.7558462469215,
+      "learning_rate": 9.163322929020476e-06,
+      "loss": 234.0876,
+      "step": 27630
+    },
+    {
+      "epoch": 0.5315870199681701,
+      "grad_norm": 420.93098003193916,
+      "learning_rate": 9.157241342007428e-06,
+      "loss": 240.2349,
+      "step": 27640
+    },
+    {
+      "epoch": 0.5317793452286507,
+      "grad_norm": 458.06884617397304,
+      "learning_rate": 9.15116006890806e-06,
+      "loss": 242.8684,
+      "step": 27650
+    },
+    {
+      "epoch": 0.5319716704891312,
+      "grad_norm": 442.557301828324,
+      "learning_rate": 9.145079111987552e-06,
+      "loss": 243.5506,
+      "step": 27660
+    },
+    {
+      "epoch": 0.5321639957496117,
+      "grad_norm": 396.668803221925,
+      "learning_rate": 9.138998473510953e-06,
+      "loss": 237.6186,
+      "step": 27670
+    },
+    {
+      "epoch": 0.5323563210100922,
+      "grad_norm": 412.50622043995537,
+      "learning_rate": 9.1329181557432e-06,
+      "loss": 240.7753,
+      "step": 27680
+    },
+    {
+      "epoch": 0.5325486462705727,
+      "grad_norm": 444.6259890702195,
+      "learning_rate": 9.126838160949119e-06,
+      "loss": 239.7045,
+      "step": 27690
+    },
+    {
+      "epoch": 0.5327409715310533,
+      "grad_norm": 419.37101807850524,
+      "learning_rate": 9.120758491393402e-06,
+      "loss": 246.308,
+      "step": 27700
+    },
+    {
+      "epoch": 0.5329332967915339,
+      "grad_norm": 410.37578221719366,
+      "learning_rate": 9.114679149340623e-06,
+      "loss": 244.9286,
+      "step": 27710
+    },
+    {
+      "epoch": 0.5331256220520144,
+      "grad_norm": 389.95589899489056,
+      "learning_rate": 9.10860013705524e-06,
+      "loss": 238.3731,
+      "step": 27720
+    },
+    {
+      "epoch": 0.5333179473124949,
+      "grad_norm": 400.30015713332745,
+      "learning_rate": 9.102521456801582e-06,
+      "loss": 233.7713,
+      "step": 27730
+    },
+    {
+      "epoch": 0.5335102725729755,
+      "grad_norm": 426.4517554236645,
+      "learning_rate": 9.096443110843864e-06,
+      "loss": 237.3978,
+      "step": 27740
+    },
+    {
+      "epoch": 0.533702597833456,
+      "grad_norm": 431.4516304307552,
+      "learning_rate": 9.090365101446157e-06,
+      "loss": 239.0948,
+      "step": 27750
+    },
+    {
+      "epoch": 0.5338949230939365,
+      "grad_norm": 460.3296232740923,
+      "learning_rate": 9.084287430872435e-06,
+      "loss": 239.439,
+      "step": 27760
+    },
+    {
+      "epoch": 0.534087248354417,
+      "grad_norm": 459.6042345801548,
+      "learning_rate": 9.078210101386518e-06,
+      "loss": 236.2329,
+      "step": 27770
+    },
+    {
+      "epoch": 0.5342795736148975,
+      "grad_norm": 464.4047365868709,
+      "learning_rate": 9.072133115252113e-06,
+      "loss": 238.4873,
+      "step": 27780
+    },
+    {
+      "epoch": 0.534471898875378,
+      "grad_norm": 454.55527372899735,
+      "learning_rate": 9.066056474732798e-06,
+      "loss": 238.1163,
+      "step": 27790
+    },
+    {
+      "epoch": 0.5346642241358586,
+      "grad_norm": 452.77022611762703,
+      "learning_rate": 9.059980182092022e-06,
+      "loss": 233.5605,
+      "step": 27800
+    },
+    {
+      "epoch": 0.5348565493963391,
+      "grad_norm": 434.5649516138401,
+      "learning_rate": 9.053904239593106e-06,
+      "loss": 235.687,
+      "step": 27810
+    },
+    {
+      "epoch": 0.5350488746568196,
+      "grad_norm": 470.9282306823315,
+      "learning_rate": 9.047828649499236e-06,
+      "loss": 241.3732,
+      "step": 27820
+    },
+    {
+      "epoch": 0.5352411999173001,
+      "grad_norm": 441.4699336661971,
+      "learning_rate": 9.041753414073463e-06,
+      "loss": 236.423,
+      "step": 27830
+    },
+    {
+      "epoch": 0.5354335251777806,
+      "grad_norm": 475.386688320379,
+      "learning_rate": 9.035678535578723e-06,
+      "loss": 236.9872,
+      "step": 27840
+    },
+    {
+      "epoch": 0.5356258504382612,
+      "grad_norm": 446.43459437258434,
+      "learning_rate": 9.029604016277798e-06,
+      "loss": 238.6817,
+      "step": 27850
+    },
+    {
+      "epoch": 0.5358181756987417,
+      "grad_norm": 486.423545734665,
+      "learning_rate": 9.02352985843335e-06,
+      "loss": 243.9364,
+      "step": 27860
+    },
+    {
+      "epoch": 0.5360105009592222,
+      "grad_norm": 414.58581917593847,
+      "learning_rate": 9.017456064307904e-06,
+      "loss": 239.5746,
+      "step": 27870
+    },
+    {
+      "epoch": 0.5362028262197027,
+      "grad_norm": 432.0216440586158,
+      "learning_rate": 9.01138263616385e-06,
+      "loss": 240.0917,
+      "step": 27880
+    },
+    {
+      "epoch": 0.5363951514801832,
+      "grad_norm": 428.7506577189589,
+      "learning_rate": 9.005309576263436e-06,
+      "loss": 237.1916,
+      "step": 27890
+    },
+    {
+      "epoch": 0.5365874767406638,
+      "grad_norm": 453.0900898673735,
+      "learning_rate": 8.999236886868772e-06,
+      "loss": 233.2634,
+      "step": 27900
+    },
+    {
+      "epoch": 0.5367798020011444,
+      "grad_norm": 446.1302851353409,
+      "learning_rate": 8.993164570241844e-06,
+      "loss": 235.4085,
+      "step": 27910
+    },
+    {
+      "epoch": 0.5369721272616249,
+      "grad_norm": 424.69855258932245,
+      "learning_rate": 8.987092628644483e-06,
+      "loss": 238.4318,
+      "step": 27920
+    },
+    {
+      "epoch": 0.5371644525221054,
+      "grad_norm": 435.9865284409231,
+      "learning_rate": 8.981021064338388e-06,
+      "loss": 237.3137,
+      "step": 27930
+    },
+    {
+      "epoch": 0.537356777782586,
+      "grad_norm": 450.13633070604385,
+      "learning_rate": 8.974949879585118e-06,
+      "loss": 239.8066,
+      "step": 27940
+    },
+    {
+      "epoch": 0.5375491030430665,
+      "grad_norm": 452.0404500781918,
+      "learning_rate": 8.968879076646093e-06,
+      "loss": 237.128,
+      "step": 27950
+    },
+    {
+      "epoch": 0.537741428303547,
+      "grad_norm": 406.93366557763164,
+      "learning_rate": 8.96280865778258e-06,
+      "loss": 236.3275,
+      "step": 27960
+    },
+    {
+      "epoch": 0.5379337535640275,
+      "grad_norm": 424.73323086920385,
+      "learning_rate": 8.956738625255709e-06,
+      "loss": 241.5757,
+      "step": 27970
+    },
+    {
+      "epoch": 0.538126078824508,
+      "grad_norm": 446.79774320592577,
+      "learning_rate": 8.950668981326473e-06,
+      "loss": 234.8669,
+      "step": 27980
+    },
+    {
+      "epoch": 0.5383184040849885,
+      "grad_norm": 445.36588525474866,
+      "learning_rate": 8.94459972825571e-06,
+      "loss": 236.9913,
+      "step": 27990
+    },
+    {
+      "epoch": 0.5385107293454691,
+      "grad_norm": 415.8629825529101,
+      "learning_rate": 8.938530868304121e-06,
+      "loss": 235.3133,
+      "step": 28000
+    },
+    {
+      "epoch": 0.5387030546059496,
+      "grad_norm": 438.071277790469,
+      "learning_rate": 8.932462403732248e-06,
+      "loss": 237.3404,
+      "step": 28010
+    },
+    {
+      "epoch": 0.5388953798664301,
+      "grad_norm": 410.3930296964305,
+      "learning_rate": 8.926394336800502e-06,
+      "loss": 245.5063,
+      "step": 28020
+    },
+    {
+      "epoch": 0.5390877051269106,
+      "grad_norm": 445.83349527723215,
+      "learning_rate": 8.920326669769134e-06,
+      "loss": 243.1,
+      "step": 28030
+    },
+    {
+      "epoch": 0.5392800303873911,
+      "grad_norm": 574.7024178289419,
+      "learning_rate": 8.914259404898247e-06,
+      "loss": 239.4048,
+      "step": 28040
+    },
+    {
+      "epoch": 0.5394723556478717,
+      "grad_norm": 429.4323270904091,
+      "learning_rate": 8.908192544447803e-06,
+      "loss": 238.0667,
+      "step": 28050
+    },
+    {
+      "epoch": 0.5396646809083522,
+      "grad_norm": 403.7219766955144,
+      "learning_rate": 8.902126090677605e-06,
+      "loss": 227.862,
+      "step": 28060
+    },
+    {
+      "epoch": 0.5398570061688327,
+      "grad_norm": 402.35099634705676,
+      "learning_rate": 8.896060045847305e-06,
+      "loss": 238.9521,
+      "step": 28070
+    },
+    {
+      "epoch": 0.5400493314293132,
+      "grad_norm": 439.902198155717,
+      "learning_rate": 8.889994412216403e-06,
+      "loss": 235.4164,
+      "step": 28080
+    },
+    {
+      "epoch": 0.5402416566897937,
+      "grad_norm": 436.839214987937,
+      "learning_rate": 8.883929192044254e-06,
+      "loss": 233.4839,
+      "step": 28090
+    },
+    {
+      "epoch": 0.5404339819502743,
+      "grad_norm": 414.3816461233584,
+      "learning_rate": 8.877864387590049e-06,
+      "loss": 241.2454,
+      "step": 28100
+    },
+    {
+      "epoch": 0.5406263072107548,
+      "grad_norm": 431.4843908096932,
+      "learning_rate": 8.871800001112822e-06,
+      "loss": 239.5498,
+      "step": 28110
+    },
+    {
+      "epoch": 0.5408186324712354,
+      "grad_norm": 461.2126036583687,
+      "learning_rate": 8.865736034871468e-06,
+      "loss": 243.9344,
+      "step": 28120
+    },
+    {
+      "epoch": 0.5410109577317159,
+      "grad_norm": 476.51559008933083,
+      "learning_rate": 8.859672491124706e-06,
+      "loss": 239.6141,
+      "step": 28130
+    },
+    {
+      "epoch": 0.5412032829921964,
+      "grad_norm": 421.76539767904706,
+      "learning_rate": 8.853609372131105e-06,
+      "loss": 240.9978,
+      "step": 28140
+    },
+    {
+      "epoch": 0.541395608252677,
+      "grad_norm": 437.24856175150074,
+      "learning_rate": 8.84754668014908e-06,
+      "loss": 237.6165,
+      "step": 28150
+    },
+    {
+      "epoch": 0.5415879335131575,
+      "grad_norm": 403.52709788469804,
+      "learning_rate": 8.841484417436886e-06,
+      "loss": 235.1001,
+      "step": 28160
+    },
+    {
+      "epoch": 0.541780258773638,
+      "grad_norm": 414.3479546667509,
+      "learning_rate": 8.835422586252613e-06,
+      "loss": 244.4629,
+      "step": 28170
+    },
+    {
+      "epoch": 0.5419725840341185,
+      "grad_norm": 417.15151179403546,
+      "learning_rate": 8.829361188854194e-06,
+      "loss": 227.7996,
+      "step": 28180
+    },
+    {
+      "epoch": 0.542164909294599,
+      "grad_norm": 414.44938855553704,
+      "learning_rate": 8.823300227499393e-06,
+      "loss": 234.9486,
+      "step": 28190
+    },
+    {
+      "epoch": 0.5423572345550796,
+      "grad_norm": 521.6171209970335,
+      "learning_rate": 8.817239704445827e-06,
+      "loss": 240.6794,
+      "step": 28200
+    },
+    {
+      "epoch": 0.5425495598155601,
+      "grad_norm": 442.6538531980378,
+      "learning_rate": 8.811179621950937e-06,
+      "loss": 233.0766,
+      "step": 28210
+    },
+    {
+      "epoch": 0.5427418850760406,
+      "grad_norm": 467.63584333633014,
+      "learning_rate": 8.805119982272001e-06,
+      "loss": 238.1453,
+      "step": 28220
+    },
+    {
+      "epoch": 0.5429342103365211,
+      "grad_norm": 409.399843981876,
+      "learning_rate": 8.799060787666142e-06,
+      "loss": 240.8451,
+      "step": 28230
+    },
+    {
+      "epoch": 0.5431265355970016,
+      "grad_norm": 393.13164901766714,
+      "learning_rate": 8.793002040390304e-06,
+      "loss": 240.9221,
+      "step": 28240
+    },
+    {
+      "epoch": 0.5433188608574822,
+      "grad_norm": 423.8686013710147,
+      "learning_rate": 8.786943742701273e-06,
+      "loss": 235.5499,
+      "step": 28250
+    },
+    {
+      "epoch": 0.5435111861179627,
+      "grad_norm": 441.5146110006175,
+      "learning_rate": 8.780885896855659e-06,
+      "loss": 232.8464,
+      "step": 28260
+    },
+    {
+      "epoch": 0.5437035113784432,
+      "grad_norm": 457.2506720772257,
+      "learning_rate": 8.774828505109918e-06,
+      "loss": 243.7003,
+      "step": 28270
+    },
+    {
+      "epoch": 0.5438958366389237,
+      "grad_norm": 462.57648847522546,
+      "learning_rate": 8.768771569720324e-06,
+      "loss": 237.9007,
+      "step": 28280
+    },
+    {
+      "epoch": 0.5440881618994042,
+      "grad_norm": 453.40683408313174,
+      "learning_rate": 8.762715092942983e-06,
+      "loss": 246.5548,
+      "step": 28290
+    },
+    {
+      "epoch": 0.5442804871598848,
+      "grad_norm": 420.21343945072533,
+      "learning_rate": 8.756659077033838e-06,
+      "loss": 238.1265,
+      "step": 28300
+    },
+    {
+      "epoch": 0.5444728124203653,
+      "grad_norm": 421.3046042008414,
+      "learning_rate": 8.750603524248653e-06,
+      "loss": 244.3051,
+      "step": 28310
+    },
+    {
+      "epoch": 0.5446651376808459,
+      "grad_norm": 407.10993830391544,
+      "learning_rate": 8.744548436843021e-06,
+      "loss": 240.005,
+      "step": 28320
+    },
+    {
+      "epoch": 0.5448574629413264,
+      "grad_norm": 489.2844336922994,
+      "learning_rate": 8.738493817072359e-06,
+      "loss": 238.9589,
+      "step": 28330
+    },
+    {
+      "epoch": 0.5450497882018069,
+      "grad_norm": 454.2845303295166,
+      "learning_rate": 8.73243966719192e-06,
+      "loss": 240.3854,
+      "step": 28340
+    },
+    {
+      "epoch": 0.5452421134622875,
+      "grad_norm": 450.523359024384,
+      "learning_rate": 8.726385989456764e-06,
+      "loss": 235.6842,
+      "step": 28350
+    },
+    {
+      "epoch": 0.545434438722768,
+      "grad_norm": 421.99590031991517,
+      "learning_rate": 8.7203327861218e-06,
+      "loss": 238.8083,
+      "step": 28360
+    },
+    {
+      "epoch": 0.5456267639832485,
+      "grad_norm": 424.87954375263115,
+      "learning_rate": 8.71428005944173e-06,
+      "loss": 237.5431,
+      "step": 28370
+    },
+    {
+      "epoch": 0.545819089243729,
+      "grad_norm": 414.1600756892164,
+      "learning_rate": 8.708227811671112e-06,
+      "loss": 233.075,
+      "step": 28380
+    },
+    {
+      "epoch": 0.5460114145042095,
+      "grad_norm": 435.18702464708923,
+      "learning_rate": 8.702176045064296e-06,
+      "loss": 237.1588,
+      "step": 28390
+    },
+    {
+      "epoch": 0.54620373976469,
+      "grad_norm": 398.9502647931613,
+      "learning_rate": 8.696124761875467e-06,
+      "loss": 239.082,
+      "step": 28400
+    },
+    {
+      "epoch": 0.5463960650251706,
+      "grad_norm": 385.8835734502926,
+      "learning_rate": 8.690073964358635e-06,
+      "loss": 244.7883,
+      "step": 28410
+    },
+    {
+      "epoch": 0.5465883902856511,
+      "grad_norm": 502.0638298680671,
+      "learning_rate": 8.684023654767613e-06,
+      "loss": 238.5478,
+      "step": 28420
+    },
+    {
+      "epoch": 0.5467807155461316,
+      "grad_norm": 420.8500437506724,
+      "learning_rate": 8.677973835356048e-06,
+      "loss": 232.1797,
+      "step": 28430
+    },
+    {
+      "epoch": 0.5469730408066121,
+      "grad_norm": 393.7392482397256,
+      "learning_rate": 8.671924508377392e-06,
+      "loss": 234.0055,
+      "step": 28440
+    },
+    {
+      "epoch": 0.5471653660670927,
+      "grad_norm": 434.21303276379894,
+      "learning_rate": 8.665875676084927e-06,
+      "loss": 237.1666,
+      "step": 28450
+    },
+    {
+      "epoch": 0.5473576913275732,
+      "grad_norm": 452.9313713458954,
+      "learning_rate": 8.659827340731738e-06,
+      "loss": 241.5489,
+      "step": 28460
+    },
+    {
+      "epoch": 0.5475500165880537,
+      "grad_norm": 455.40813596529097,
+      "learning_rate": 8.653779504570728e-06,
+      "loss": 237.8429,
+      "step": 28470
+    },
+    {
+      "epoch": 0.5477423418485342,
+      "grad_norm": 455.1225522038088,
+      "learning_rate": 8.647732169854622e-06,
+      "loss": 239.701,
+      "step": 28480
+    },
+    {
+      "epoch": 0.5479346671090147,
+      "grad_norm": 394.37135795117234,
+      "learning_rate": 8.641685338835947e-06,
+      "loss": 237.3779,
+      "step": 28490
+    },
+    {
+      "epoch": 0.5481269923694952,
+      "grad_norm": 429.59407273639226,
+      "learning_rate": 8.635639013767053e-06,
+      "loss": 237.9776,
+      "step": 28500
+    },
+    {
+      "epoch": 0.5483193176299758,
+      "grad_norm": 455.9025364678148,
+      "learning_rate": 8.629593196900088e-06,
+      "loss": 235.7486,
+      "step": 28510
+    },
+    {
+      "epoch": 0.5485116428904563,
+      "grad_norm": 473.7664791092251,
+      "learning_rate": 8.62354789048703e-06,
+      "loss": 236.7516,
+      "step": 28520
+    },
+    {
+      "epoch": 0.5487039681509369,
+      "grad_norm": 440.4126155208757,
+      "learning_rate": 8.617503096779648e-06,
+      "loss": 244.196,
+      "step": 28530
+    },
+    {
+      "epoch": 0.5488962934114174,
+      "grad_norm": 440.3030451777663,
+      "learning_rate": 8.61145881802953e-06,
+      "loss": 240.0682,
+      "step": 28540
+    },
+    {
+      "epoch": 0.549088618671898,
+      "grad_norm": 391.99883130957693,
+      "learning_rate": 8.605415056488067e-06,
+      "loss": 237.6824,
+      "step": 28550
+    },
+    {
+      "epoch": 0.5492809439323785,
+      "grad_norm": 417.1824535424529,
+      "learning_rate": 8.599371814406465e-06,
+      "loss": 232.3002,
+      "step": 28560
+    },
+    {
+      "epoch": 0.549473269192859,
+      "grad_norm": 440.18805675726577,
+      "learning_rate": 8.59332909403573e-06,
+      "loss": 259.531,
+      "step": 28570
+    },
+    {
+      "epoch": 0.5496655944533395,
+      "grad_norm": 459.376879692458,
+      "learning_rate": 8.587286897626672e-06,
+      "loss": 237.1712,
+      "step": 28580
+    },
+    {
+      "epoch": 0.54985791971382,
+      "grad_norm": 423.95785600861467,
+      "learning_rate": 8.581245227429918e-06,
+      "loss": 241.3493,
+      "step": 28590
+    },
+    {
+      "epoch": 0.5500502449743006,
+      "grad_norm": 404.24463532965217,
+      "learning_rate": 8.575204085695887e-06,
+      "loss": 230.9917,
+      "step": 28600
+    },
+    {
+      "epoch": 0.5502425702347811,
+      "grad_norm": 449.0925299991616,
+      "learning_rate": 8.5691634746748e-06,
+      "loss": 241.6851,
+      "step": 28610
+    },
+    {
+      "epoch": 0.5504348954952616,
+      "grad_norm": 437.8093108272611,
+      "learning_rate": 8.563123396616683e-06,
+      "loss": 230.9252,
+      "step": 28620
+    },
+    {
+      "epoch": 0.5506272207557421,
+      "grad_norm": 439.6946376244109,
+      "learning_rate": 8.557083853771377e-06,
+      "loss": 235.145,
+      "step": 28630
+    },
+    {
+      "epoch": 0.5508195460162226,
+      "grad_norm": 505.0433888629805,
+      "learning_rate": 8.551044848388502e-06,
+      "loss": 239.8914,
+      "step": 28640
+    },
+    {
+      "epoch": 0.5510118712767031,
+      "grad_norm": 414.46081005848254,
+      "learning_rate": 8.545006382717487e-06,
+      "loss": 237.5509,
+      "step": 28650
+    },
+    {
+      "epoch": 0.5512041965371837,
+      "grad_norm": 450.1853383823327,
+      "learning_rate": 8.538968459007569e-06,
+      "loss": 244.1873,
+      "step": 28660
+    },
+    {
+      "epoch": 0.5513965217976642,
+      "grad_norm": 429.76686210562667,
+      "learning_rate": 8.532931079507772e-06,
+      "loss": 236.1655,
+      "step": 28670
+    },
+    {
+      "epoch": 0.5515888470581447,
+      "grad_norm": 427.8557260174465,
+      "learning_rate": 8.526894246466916e-06,
+      "loss": 233.7107,
+      "step": 28680
+    },
+    {
+      "epoch": 0.5517811723186252,
+      "grad_norm": 408.41928612310375,
+      "learning_rate": 8.520857962133623e-06,
+      "loss": 233.7812,
+      "step": 28690
+    },
+    {
+      "epoch": 0.5519734975791057,
+      "grad_norm": 424.209194049445,
+      "learning_rate": 8.514822228756311e-06,
+      "loss": 240.2727,
+      "step": 28700
+    },
+    {
+      "epoch": 0.5521658228395863,
+      "grad_norm": 419.8385413745241,
+      "learning_rate": 8.508787048583191e-06,
+      "loss": 235.4688,
+      "step": 28710
+    },
+    {
+      "epoch": 0.5523581481000668,
+      "grad_norm": 436.7339005060254,
+      "learning_rate": 8.502752423862264e-06,
+      "loss": 239.427,
+      "step": 28720
+    },
+    {
+      "epoch": 0.5525504733605474,
+      "grad_norm": 404.90859921527954,
+      "learning_rate": 8.496718356841335e-06,
+      "loss": 231.2506,
+      "step": 28730
+    },
+    {
+      "epoch": 0.5527427986210279,
+      "grad_norm": 404.1225956730404,
+      "learning_rate": 8.49068484976799e-06,
+      "loss": 237.4833,
+      "step": 28740
+    },
+    {
+      "epoch": 0.5529351238815085,
+      "grad_norm": 421.7255506220515,
+      "learning_rate": 8.484651904889614e-06,
+      "loss": 232.7491,
+      "step": 28750
+    },
+    {
+      "epoch": 0.553127449141989,
+      "grad_norm": 447.86602621506404,
+      "learning_rate": 8.478619524453369e-06,
+      "loss": 234.2143,
+      "step": 28760
+    },
+    {
+      "epoch": 0.5533197744024695,
+      "grad_norm": 415.51295028594865,
+      "learning_rate": 8.472587710706232e-06,
+      "loss": 230.9796,
+      "step": 28770
+    },
+    {
+      "epoch": 0.55351209966295,
+      "grad_norm": 436.53604856654425,
+      "learning_rate": 8.466556465894942e-06,
+      "loss": 238.3564,
+      "step": 28780
+    },
+    {
+      "epoch": 0.5537044249234305,
+      "grad_norm": 439.25119264483914,
+      "learning_rate": 8.460525792266046e-06,
+      "loss": 241.4103,
+      "step": 28790
+    },
+    {
+      "epoch": 0.553896750183911,
+      "grad_norm": 445.6663629979755,
+      "learning_rate": 8.454495692065862e-06,
+      "loss": 243.7715,
+      "step": 28800
+    },
+    {
+      "epoch": 0.5540890754443916,
+      "grad_norm": 430.0362692599995,
+      "learning_rate": 8.448466167540514e-06,
+      "loss": 232.4944,
+      "step": 28810
+    },
+    {
+      "epoch": 0.5542814007048721,
+      "grad_norm": 430.04124912487634,
+      "learning_rate": 8.442437220935893e-06,
+      "loss": 244.3331,
+      "step": 28820
+    },
+    {
+      "epoch": 0.5544737259653526,
+      "grad_norm": 400.5778873177401,
+      "learning_rate": 8.436408854497679e-06,
+      "loss": 239.7221,
+      "step": 28830
+    },
+    {
+      "epoch": 0.5546660512258331,
+      "grad_norm": 464.9492929250041,
+      "learning_rate": 8.430381070471348e-06,
+      "loss": 239.5716,
+      "step": 28840
+    },
+    {
+      "epoch": 0.5548583764863136,
+      "grad_norm": 427.00728512860155,
+      "learning_rate": 8.424353871102144e-06,
+      "loss": 232.512,
+      "step": 28850
+    },
+    {
+      "epoch": 0.5550507017467942,
+      "grad_norm": 455.05319687920013,
+      "learning_rate": 8.4183272586351e-06,
+      "loss": 237.7265,
+      "step": 28860
+    },
+    {
+      "epoch": 0.5552430270072747,
+      "grad_norm": 430.9555904005068,
+      "learning_rate": 8.412301235315026e-06,
+      "loss": 245.7041,
+      "step": 28870
+    },
+    {
+      "epoch": 0.5554353522677552,
+      "grad_norm": 430.51117604369654,
+      "learning_rate": 8.406275803386525e-06,
+      "loss": 238.007,
+      "step": 28880
+    },
+    {
+      "epoch": 0.5556276775282357,
+      "grad_norm": 446.39510440851063,
+      "learning_rate": 8.400250965093968e-06,
+      "loss": 238.4255,
+      "step": 28890
+    },
+    {
+      "epoch": 0.5558200027887162,
+      "grad_norm": 465.8734504330715,
+      "learning_rate": 8.394226722681498e-06,
+      "loss": 242.5858,
+      "step": 28900
+    },
+    {
+      "epoch": 0.5560123280491968,
+      "grad_norm": 454.0886816478026,
+      "learning_rate": 8.38820307839306e-06,
+      "loss": 235.7706,
+      "step": 28910
+    },
+    {
+      "epoch": 0.5562046533096773,
+      "grad_norm": 392.2356594557625,
+      "learning_rate": 8.382180034472353e-06,
+      "loss": 227.9209,
+      "step": 28920
+    },
+    {
+      "epoch": 0.5563969785701578,
+      "grad_norm": 436.39915523910065,
+      "learning_rate": 8.376157593162867e-06,
+      "loss": 232.2681,
+      "step": 28930
+    },
+    {
+      "epoch": 0.5565893038306384,
+      "grad_norm": 480.64260870326405,
+      "learning_rate": 8.370135756707853e-06,
+      "loss": 236.8256,
+      "step": 28940
+    },
+    {
+      "epoch": 0.556781629091119,
+      "grad_norm": 417.1925180290578,
+      "learning_rate": 8.364114527350357e-06,
+      "loss": 234.1597,
+      "step": 28950
+    },
+    {
+      "epoch": 0.5569739543515995,
+      "grad_norm": 449.36298230254397,
+      "learning_rate": 8.358093907333182e-06,
+      "loss": 231.3657,
+      "step": 28960
+    },
+    {
+      "epoch": 0.55716627961208,
+      "grad_norm": 471.0711818909541,
+      "learning_rate": 8.35207389889891e-06,
+      "loss": 238.2488,
+      "step": 28970
+    },
+    {
+      "epoch": 0.5573586048725605,
+      "grad_norm": 435.1234091862365,
+      "learning_rate": 8.346054504289888e-06,
+      "loss": 240.501,
+      "step": 28980
+    },
+    {
+      "epoch": 0.557550930133041,
+      "grad_norm": 423.04558933868947,
+      "learning_rate": 8.34003572574825e-06,
+      "loss": 232.0768,
+      "step": 28990
+    },
+    {
+      "epoch": 0.5577432553935215,
+      "grad_norm": 438.87182140576124,
+      "learning_rate": 8.334017565515892e-06,
+      "loss": 238.6667,
+      "step": 29000
+    },
+    {
+      "epoch": 0.5579355806540021,
+      "grad_norm": 439.66266151022955,
+      "learning_rate": 8.328000025834472e-06,
+      "loss": 226.9736,
+      "step": 29010
+    },
+    {
+      "epoch": 0.5581279059144826,
+      "grad_norm": 436.7555100469507,
+      "learning_rate": 8.321983108945431e-06,
+      "loss": 236.197,
+      "step": 29020
+    },
+    {
+      "epoch": 0.5583202311749631,
+      "grad_norm": 439.97985744223655,
+      "learning_rate": 8.315966817089972e-06,
+      "loss": 241.0712,
+      "step": 29030
+    },
+    {
+      "epoch": 0.5585125564354436,
+      "grad_norm": 433.4090724994676,
+      "learning_rate": 8.309951152509057e-06,
+      "loss": 235.3943,
+      "step": 29040
+    },
+    {
+      "epoch": 0.5587048816959241,
+      "grad_norm": 436.90969806228907,
+      "learning_rate": 8.303936117443422e-06,
+      "loss": 242.7536,
+      "step": 29050
+    },
+    {
+      "epoch": 0.5588972069564047,
+      "grad_norm": 404.12443658387616,
+      "learning_rate": 8.297921714133576e-06,
+      "loss": 238.2426,
+      "step": 29060
+    },
+    {
+      "epoch": 0.5590895322168852,
+      "grad_norm": 468.93512599776744,
+      "learning_rate": 8.291907944819782e-06,
+      "loss": 232.6437,
+      "step": 29070
+    },
+    {
+      "epoch": 0.5592818574773657,
+      "grad_norm": 1199.7744068590969,
+      "learning_rate": 8.285894811742065e-06,
+      "loss": 226.6045,
+      "step": 29080
+    },
+    {
+      "epoch": 0.5594741827378462,
+      "grad_norm": 434.02002733519527,
+      "learning_rate": 8.279882317140224e-06,
+      "loss": 229.8454,
+      "step": 29090
+    },
+    {
+      "epoch": 0.5596665079983267,
+      "grad_norm": 452.2935112989161,
+      "learning_rate": 8.273870463253813e-06,
+      "loss": 235.6263,
+      "step": 29100
+    },
+    {
+      "epoch": 0.5598588332588073,
+      "grad_norm": 450.0834081645108,
+      "learning_rate": 8.267859252322144e-06,
+      "loss": 239.8458,
+      "step": 29110
+    },
+    {
+      "epoch": 0.5600511585192878,
+      "grad_norm": 446.1097247813684,
+      "learning_rate": 8.261848686584293e-06,
+      "loss": 230.9927,
+      "step": 29120
+    },
+    {
+      "epoch": 0.5602434837797683,
+      "grad_norm": 443.7628055153972,
+      "learning_rate": 8.255838768279106e-06,
+      "loss": 232.7486,
+      "step": 29130
+    },
+    {
+      "epoch": 0.5604358090402489,
+      "grad_norm": 433.5340503963023,
+      "learning_rate": 8.249829499645167e-06,
+      "loss": 235.3675,
+      "step": 29140
+    },
+    {
+      "epoch": 0.5606281343007294,
+      "grad_norm": 434.9334074012724,
+      "learning_rate": 8.243820882920837e-06,
+      "loss": 229.7794,
+      "step": 29150
+    },
+    {
+      "epoch": 0.56082045956121,
+      "grad_norm": 472.2739399899253,
+      "learning_rate": 8.23781292034422e-06,
+      "loss": 233.5955,
+      "step": 29160
+    },
+    {
+      "epoch": 0.5610127848216905,
+      "grad_norm": 417.91950790778156,
+      "learning_rate": 8.231805614153192e-06,
+      "loss": 232.6033,
+      "step": 29170
+    },
+    {
+      "epoch": 0.561205110082171,
+      "grad_norm": 432.3120678574319,
+      "learning_rate": 8.22579896658537e-06,
+      "loss": 231.9148,
+      "step": 29180
+    },
+    {
+      "epoch": 0.5613974353426515,
+      "grad_norm": 457.3705948304885,
+      "learning_rate": 8.219792979878126e-06,
+      "loss": 229.5217,
+      "step": 29190
+    },
+    {
+      "epoch": 0.561589760603132,
+      "grad_norm": 433.83089134028205,
+      "learning_rate": 8.213787656268599e-06,
+      "loss": 238.5942,
+      "step": 29200
+    },
+    {
+      "epoch": 0.5617820858636126,
+      "grad_norm": 414.6576865837906,
+      "learning_rate": 8.20778299799367e-06,
+      "loss": 234.8118,
+      "step": 29210
+    },
+    {
+      "epoch": 0.5619744111240931,
+      "grad_norm": 418.0767112332765,
+      "learning_rate": 8.201779007289975e-06,
+      "loss": 231.9389,
+      "step": 29220
+    },
+    {
+      "epoch": 0.5621667363845736,
+      "grad_norm": 458.975100774114,
+      "learning_rate": 8.195775686393898e-06,
+      "loss": 242.5277,
+      "step": 29230
+    },
+    {
+      "epoch": 0.5623590616450541,
+      "grad_norm": 455.0985297645168,
+      "learning_rate": 8.189773037541585e-06,
+      "loss": 236.6801,
+      "step": 29240
+    },
+    {
+      "epoch": 0.5625513869055346,
+      "grad_norm": 413.6803791152206,
+      "learning_rate": 8.183771062968917e-06,
+      "loss": 239.2418,
+      "step": 29250
+    },
+    {
+      "epoch": 0.5627437121660152,
+      "grad_norm": 479.33810263701804,
+      "learning_rate": 8.177769764911528e-06,
+      "loss": 236.2911,
+      "step": 29260
+    },
+    {
+      "epoch": 0.5629360374264957,
+      "grad_norm": 431.8818470494597,
+      "learning_rate": 8.171769145604812e-06,
+      "loss": 237.3531,
+      "step": 29270
+    },
+    {
+      "epoch": 0.5631283626869762,
+      "grad_norm": 393.4924122227779,
+      "learning_rate": 8.165769207283891e-06,
+      "loss": 231.8063,
+      "step": 29280
+    },
+    {
+      "epoch": 0.5633206879474567,
+      "grad_norm": 441.90694572883865,
+      "learning_rate": 8.15976995218365e-06,
+      "loss": 236.3916,
+      "step": 29290
+    },
+    {
+      "epoch": 0.5635130132079372,
+      "grad_norm": 443.6313797272376,
+      "learning_rate": 8.153771382538706e-06,
+      "loss": 227.402,
+      "step": 29300
+    },
+    {
+      "epoch": 0.5637053384684177,
+      "grad_norm": 463.69689906755025,
+      "learning_rate": 8.147773500583434e-06,
+      "loss": 236.4332,
+      "step": 29310
+    },
+    {
+      "epoch": 0.5638976637288983,
+      "grad_norm": 410.4716215607669,
+      "learning_rate": 8.141776308551942e-06,
+      "loss": 234.3254,
+      "step": 29320
+    },
+    {
+      "epoch": 0.5640899889893788,
+      "grad_norm": 444.71168832516685,
+      "learning_rate": 8.135779808678084e-06,
+      "loss": 236.9215,
+      "step": 29330
+    },
+    {
+      "epoch": 0.5642823142498593,
+      "grad_norm": 426.33049987883237,
+      "learning_rate": 8.129784003195458e-06,
+      "loss": 227.8894,
+      "step": 29340
+    },
+    {
+      "epoch": 0.5644746395103399,
+      "grad_norm": 438.2620539316175,
+      "learning_rate": 8.123788894337405e-06,
+      "loss": 232.5625,
+      "step": 29350
+    },
+    {
+      "epoch": 0.5646669647708205,
+      "grad_norm": 426.74264904474813,
+      "learning_rate": 8.117794484337003e-06,
+      "loss": 234.1855,
+      "step": 29360
+    },
+    {
+      "epoch": 0.564859290031301,
+      "grad_norm": 437.32046894423706,
+      "learning_rate": 8.111800775427066e-06,
+      "loss": 233.2375,
+      "step": 29370
+    },
+    {
+      "epoch": 0.5650516152917815,
+      "grad_norm": 428.3537798232405,
+      "learning_rate": 8.10580776984016e-06,
+      "loss": 232.6249,
+      "step": 29380
+    },
+    {
+      "epoch": 0.565243940552262,
+      "grad_norm": 392.1065371992853,
+      "learning_rate": 8.099815469808573e-06,
+      "loss": 232.3645,
+      "step": 29390
+    },
+    {
+      "epoch": 0.5654362658127425,
+      "grad_norm": 416.9722829901236,
+      "learning_rate": 8.093823877564343e-06,
+      "loss": 233.1531,
+      "step": 29400
+    },
+    {
+      "epoch": 0.565628591073223,
+      "grad_norm": 423.60496820830866,
+      "learning_rate": 8.087832995339236e-06,
+      "loss": 232.5088,
+      "step": 29410
+    },
+    {
+      "epoch": 0.5658209163337036,
+      "grad_norm": 429.36969833486455,
+      "learning_rate": 8.081842825364756e-06,
+      "loss": 228.5401,
+      "step": 29420
+    },
+    {
+      "epoch": 0.5660132415941841,
+      "grad_norm": 430.1413141416627,
+      "learning_rate": 8.075853369872149e-06,
+      "loss": 234.6433,
+      "step": 29430
+    },
+    {
+      "epoch": 0.5662055668546646,
+      "grad_norm": 408.5181320541568,
+      "learning_rate": 8.069864631092377e-06,
+      "loss": 234.2195,
+      "step": 29440
+    },
+    {
+      "epoch": 0.5663978921151451,
+      "grad_norm": 426.82140710699855,
+      "learning_rate": 8.063876611256158e-06,
+      "loss": 236.027,
+      "step": 29450
+    },
+    {
+      "epoch": 0.5665902173756256,
+      "grad_norm": 396.1152339466765,
+      "learning_rate": 8.057889312593924e-06,
+      "loss": 238.9753,
+      "step": 29460
+    },
+    {
+      "epoch": 0.5667825426361062,
+      "grad_norm": 1433.6892418418208,
+      "learning_rate": 8.051902737335847e-06,
+      "loss": 244.3272,
+      "step": 29470
+    },
+    {
+      "epoch": 0.5669748678965867,
+      "grad_norm": 454.6451528307559,
+      "learning_rate": 8.045916887711822e-06,
+      "loss": 237.8165,
+      "step": 29480
+    },
+    {
+      "epoch": 0.5671671931570672,
+      "grad_norm": 438.9242695035424,
+      "learning_rate": 8.039931765951485e-06,
+      "loss": 242.5767,
+      "step": 29490
+    },
+    {
+      "epoch": 0.5673595184175477,
+      "grad_norm": 430.1708938564893,
+      "learning_rate": 8.033947374284196e-06,
+      "loss": 229.8049,
+      "step": 29500
+    },
+    {
+      "epoch": 0.5675518436780282,
+      "grad_norm": 433.9020765366243,
+      "learning_rate": 8.02796371493904e-06,
+      "loss": 238.5978,
+      "step": 29510
+    },
+    {
+      "epoch": 0.5677441689385088,
+      "grad_norm": 457.179755585707,
+      "learning_rate": 8.021980790144828e-06,
+      "loss": 226.3361,
+      "step": 29520
+    },
+    {
+      "epoch": 0.5679364941989893,
+      "grad_norm": 401.10126184229546,
+      "learning_rate": 8.015998602130107e-06,
+      "loss": 233.015,
+      "step": 29530
+    },
+    {
+      "epoch": 0.5681288194594698,
+      "grad_norm": 419.8783587524023,
+      "learning_rate": 8.01001715312314e-06,
+      "loss": 230.0096,
+      "step": 29540
+    },
+    {
+      "epoch": 0.5683211447199504,
+      "grad_norm": 406.5105505968295,
+      "learning_rate": 8.00403644535191e-06,
+      "loss": 233.0671,
+      "step": 29550
+    },
+    {
+      "epoch": 0.568513469980431,
+      "grad_norm": 429.09092078313,
+      "learning_rate": 7.998056481044146e-06,
+      "loss": 235.6919,
+      "step": 29560
+    },
+    {
+      "epoch": 0.5687057952409115,
+      "grad_norm": 482.7599097446289,
+      "learning_rate": 7.992077262427282e-06,
+      "loss": 243.4911,
+      "step": 29570
+    },
+    {
+      "epoch": 0.568898120501392,
+      "grad_norm": 443.50735620417,
+      "learning_rate": 7.986098791728475e-06,
+      "loss": 231.8863,
+      "step": 29580
+    },
+    {
+      "epoch": 0.5690904457618725,
+      "grad_norm": 428.7615281032246,
+      "learning_rate": 7.9801210711746e-06,
+      "loss": 237.9173,
+      "step": 29590
+    },
+    {
+      "epoch": 0.569282771022353,
+      "grad_norm": 408.8363081666945,
+      "learning_rate": 7.974144102992273e-06,
+      "loss": 232.385,
+      "step": 29600
+    },
+    {
+      "epoch": 0.5694750962828335,
+      "grad_norm": 504.4538415192509,
+      "learning_rate": 7.968167889407813e-06,
+      "loss": 229.3145,
+      "step": 29610
+    },
+    {
+      "epoch": 0.5696674215433141,
+      "grad_norm": 407.0660969430595,
+      "learning_rate": 7.96219243264725e-06,
+      "loss": 236.7475,
+      "step": 29620
+    },
+    {
+      "epoch": 0.5698597468037946,
+      "grad_norm": 429.1831615564374,
+      "learning_rate": 7.956217734936353e-06,
+      "loss": 231.8069,
+      "step": 29630
+    },
+    {
+      "epoch": 0.5700520720642751,
+      "grad_norm": 422.59361402290205,
+      "learning_rate": 7.950243798500593e-06,
+      "loss": 229.0326,
+      "step": 29640
+    },
+    {
+      "epoch": 0.5702443973247556,
+      "grad_norm": 455.3871506965756,
+      "learning_rate": 7.94427062556517e-06,
+      "loss": 236.7605,
+      "step": 29650
+    },
+    {
+      "epoch": 0.5704367225852361,
+      "grad_norm": 413.4690621057088,
+      "learning_rate": 7.938298218354985e-06,
+      "loss": 231.2507,
+      "step": 29660
+    },
+    {
+      "epoch": 0.5706290478457167,
+      "grad_norm": 454.06650192224396,
+      "learning_rate": 7.932326579094665e-06,
+      "loss": 236.3325,
+      "step": 29670
+    },
+    {
+      "epoch": 0.5708213731061972,
+      "grad_norm": 424.6707673047249,
+      "learning_rate": 7.926355710008545e-06,
+      "loss": 233.2701,
+      "step": 29680
+    },
+    {
+      "epoch": 0.5710136983666777,
+      "grad_norm": 431.309560668897,
+      "learning_rate": 7.920385613320675e-06,
+      "loss": 230.0675,
+      "step": 29690
+    },
+    {
+      "epoch": 0.5712060236271582,
+      "grad_norm": 503.5816889167475,
+      "learning_rate": 7.914416291254817e-06,
+      "loss": 224.8446,
+      "step": 29700
+    },
+    {
+      "epoch": 0.5713983488876387,
+      "grad_norm": 418.8314412889683,
+      "learning_rate": 7.908447746034447e-06,
+      "loss": 228.1362,
+      "step": 29710
+    },
+    {
+      "epoch": 0.5715906741481193,
+      "grad_norm": 419.6478281748611,
+      "learning_rate": 7.902479979882749e-06,
+      "loss": 238.2103,
+      "step": 29720
+    },
+    {
+      "epoch": 0.5717829994085998,
+      "grad_norm": 447.38569161606716,
+      "learning_rate": 7.896512995022614e-06,
+      "loss": 234.149,
+      "step": 29730
+    },
+    {
+      "epoch": 0.5719753246690803,
+      "grad_norm": 387.5736557529939,
+      "learning_rate": 7.890546793676652e-06,
+      "loss": 231.6947,
+      "step": 29740
+    },
+    {
+      "epoch": 0.5721676499295609,
+      "grad_norm": 424.8616425416752,
+      "learning_rate": 7.884581378067171e-06,
+      "loss": 234.9736,
+      "step": 29750
+    },
+    {
+      "epoch": 0.5723599751900414,
+      "grad_norm": 409.924725771009,
+      "learning_rate": 7.878616750416186e-06,
+      "loss": 230.9598,
+      "step": 29760
+    },
+    {
+      "epoch": 0.572552300450522,
+      "grad_norm": 478.96694226206637,
+      "learning_rate": 7.872652912945426e-06,
+      "loss": 231.2302,
+      "step": 29770
+    },
+    {
+      "epoch": 0.5727446257110025,
+      "grad_norm": 478.47508680884954,
+      "learning_rate": 7.866689867876323e-06,
+      "loss": 235.5065,
+      "step": 29780
+    },
+    {
+      "epoch": 0.572936950971483,
+      "grad_norm": 481.5899600576842,
+      "learning_rate": 7.860727617430013e-06,
+      "loss": 232.1499,
+      "step": 29790
+    },
+    {
+      "epoch": 0.5731292762319635,
+      "grad_norm": 455.6893184536974,
+      "learning_rate": 7.85476616382733e-06,
+      "loss": 224.5898,
+      "step": 29800
+    },
+    {
+      "epoch": 0.573321601492444,
+      "grad_norm": 452.9548592326789,
+      "learning_rate": 7.848805509288824e-06,
+      "loss": 239.2905,
+      "step": 29810
+    },
+    {
+      "epoch": 0.5735139267529246,
+      "grad_norm": 443.6704667489931,
+      "learning_rate": 7.84284565603474e-06,
+      "loss": 230.783,
+      "step": 29820
+    },
+    {
+      "epoch": 0.5737062520134051,
+      "grad_norm": 429.7996519098853,
+      "learning_rate": 7.83688660628502e-06,
+      "loss": 233.0479,
+      "step": 29830
+    },
+    {
+      "epoch": 0.5738985772738856,
+      "grad_norm": 426.8290524830637,
+      "learning_rate": 7.830928362259313e-06,
+      "loss": 232.428,
+      "step": 29840
+    },
+    {
+      "epoch": 0.5740909025343661,
+      "grad_norm": 424.59455158508,
+      "learning_rate": 7.824970926176967e-06,
+      "loss": 235.2198,
+      "step": 29850
+    },
+    {
+      "epoch": 0.5742832277948466,
+      "grad_norm": 429.35119612704057,
+      "learning_rate": 7.819014300257033e-06,
+      "loss": 230.9038,
+      "step": 29860
+    },
+    {
+      "epoch": 0.5744755530553272,
+      "grad_norm": 403.15251024914807,
+      "learning_rate": 7.813058486718252e-06,
+      "loss": 235.8594,
+      "step": 29870
+    },
+    {
+      "epoch": 0.5746678783158077,
+      "grad_norm": 415.4461327359759,
+      "learning_rate": 7.80710348777906e-06,
+      "loss": 233.234,
+      "step": 29880
+    },
+    {
+      "epoch": 0.5748602035762882,
+      "grad_norm": 441.5589073334,
+      "learning_rate": 7.801149305657609e-06,
+      "loss": 233.0159,
+      "step": 29890
+    },
+    {
+      "epoch": 0.5750525288367687,
+      "grad_norm": 409.6369009422606,
+      "learning_rate": 7.795195942571722e-06,
+      "loss": 242.0823,
+      "step": 29900
+    },
+    {
+      "epoch": 0.5752448540972492,
+      "grad_norm": 413.24560529311753,
+      "learning_rate": 7.789243400738934e-06,
+      "loss": 231.3054,
+      "step": 29910
+    },
+    {
+      "epoch": 0.5754371793577298,
+      "grad_norm": 418.90371337350683,
+      "learning_rate": 7.783291682376465e-06,
+      "loss": 241.4429,
+      "step": 29920
+    },
+    {
+      "epoch": 0.5756295046182103,
+      "grad_norm": 433.6583464583209,
+      "learning_rate": 7.77734078970124e-06,
+      "loss": 240.439,
+      "step": 29930
+    },
+    {
+      "epoch": 0.5758218298786908,
+      "grad_norm": 447.94753747349483,
+      "learning_rate": 7.77139072492986e-06,
+      "loss": 230.878,
+      "step": 29940
+    },
+    {
+      "epoch": 0.5760141551391713,
+      "grad_norm": 422.5523702070567,
+      "learning_rate": 7.765441490278625e-06,
+      "loss": 231.5587,
+      "step": 29950
+    },
+    {
+      "epoch": 0.5762064803996519,
+      "grad_norm": 405.55562650065235,
+      "learning_rate": 7.759493087963535e-06,
+      "loss": 235.8433,
+      "step": 29960
+    },
+    {
+      "epoch": 0.5763988056601325,
+      "grad_norm": 371.66110667266116,
+      "learning_rate": 7.753545520200264e-06,
+      "loss": 236.9481,
+      "step": 29970
+    },
+    {
+      "epoch": 0.576591130920613,
+      "grad_norm": 416.8493988111023,
+      "learning_rate": 7.747598789204183e-06,
+      "loss": 230.5512,
+      "step": 29980
+    },
+    {
+      "epoch": 0.5767834561810935,
+      "grad_norm": 447.3445072834849,
+      "learning_rate": 7.741652897190355e-06,
+      "loss": 235.6195,
+      "step": 29990
+    },
+    {
+      "epoch": 0.576975781441574,
+      "grad_norm": 401.13914360154604,
+      "learning_rate": 7.735707846373527e-06,
+      "loss": 229.3117,
+      "step": 30000
+    },
+    {
+      "epoch": 0.5771681067020545,
+      "grad_norm": 443.54738499440026,
+      "learning_rate": 7.72976363896813e-06,
+      "loss": 233.2199,
+      "step": 30010
+    },
+    {
+      "epoch": 0.5773604319625351,
+      "grad_norm": 413.1042957854533,
+      "learning_rate": 7.723820277188278e-06,
+      "loss": 235.5748,
+      "step": 30020
+    },
+    {
+      "epoch": 0.5775527572230156,
+      "grad_norm": 423.718236725885,
+      "learning_rate": 7.717877763247787e-06,
+      "loss": 231.1978,
+      "step": 30030
+    },
+    {
+      "epoch": 0.5777450824834961,
+      "grad_norm": 455.63615005758646,
+      "learning_rate": 7.711936099360138e-06,
+      "loss": 236.8507,
+      "step": 30040
+    },
+    {
+      "epoch": 0.5779374077439766,
+      "grad_norm": 446.42643797268255,
+      "learning_rate": 7.705995287738498e-06,
+      "loss": 231.9344,
+      "step": 30050
+    },
+    {
+      "epoch": 0.5781297330044571,
+      "grad_norm": 388.7642847848498,
+      "learning_rate": 7.70005533059573e-06,
+      "loss": 233.6973,
+      "step": 30060
+    },
+    {
+      "epoch": 0.5783220582649377,
+      "grad_norm": 421.4703086973421,
+      "learning_rate": 7.694116230144368e-06,
+      "loss": 231.8808,
+      "step": 30070
+    },
+    {
+      "epoch": 0.5785143835254182,
+      "grad_norm": 407.19610921994956,
+      "learning_rate": 7.688177988596628e-06,
+      "loss": 234.6228,
+      "step": 30080
+    },
+    {
+      "epoch": 0.5787067087858987,
+      "grad_norm": 455.4291490158142,
+      "learning_rate": 7.682240608164401e-06,
+      "loss": 226.4843,
+      "step": 30090
+    },
+    {
+      "epoch": 0.5788990340463792,
+      "grad_norm": 441.8403519094396,
+      "learning_rate": 7.676304091059273e-06,
+      "loss": 231.8175,
+      "step": 30100
+    },
+    {
+      "epoch": 0.5790913593068597,
+      "grad_norm": 450.8278827515652,
+      "learning_rate": 7.670368439492495e-06,
+      "loss": 232.8311,
+      "step": 30110
+    },
+    {
+      "epoch": 0.5792836845673403,
+      "grad_norm": 405.8194780721003,
+      "learning_rate": 7.664433655674995e-06,
+      "loss": 219.8278,
+      "step": 30120
+    },
+    {
+      "epoch": 0.5794760098278208,
+      "grad_norm": 427.7900353555768,
+      "learning_rate": 7.658499741817383e-06,
+      "loss": 238.838,
+      "step": 30130
+    },
+    {
+      "epoch": 0.5796683350883013,
+      "grad_norm": 419.69001637825784,
+      "learning_rate": 7.65256670012995e-06,
+      "loss": 228.6759,
+      "step": 30140
+    },
+    {
+      "epoch": 0.5798606603487818,
+      "grad_norm": 456.5769461953817,
+      "learning_rate": 7.646634532822652e-06,
+      "loss": 233.5072,
+      "step": 30150
+    },
+    {
+      "epoch": 0.5800529856092624,
+      "grad_norm": 432.19645898671143,
+      "learning_rate": 7.64070324210512e-06,
+      "loss": 237.3528,
+      "step": 30160
+    },
+    {
+      "epoch": 0.580245310869743,
+      "grad_norm": 414.74355142617253,
+      "learning_rate": 7.634772830186668e-06,
+      "loss": 226.0943,
+      "step": 30170
+    },
+    {
+      "epoch": 0.5804376361302235,
+      "grad_norm": 463.38022675404983,
+      "learning_rate": 7.628843299276276e-06,
+      "loss": 238.0814,
+      "step": 30180
+    },
+    {
+      "epoch": 0.580629961390704,
+      "grad_norm": 458.96808206393257,
+      "learning_rate": 7.62291465158259e-06,
+      "loss": 228.016,
+      "step": 30190
+    },
+    {
+      "epoch": 0.5808222866511845,
+      "grad_norm": 388.6820310962333,
+      "learning_rate": 7.616986889313939e-06,
+      "loss": 226.2567,
+      "step": 30200
+    },
+    {
+      "epoch": 0.581014611911665,
+      "grad_norm": 420.74530790257467,
+      "learning_rate": 7.611060014678313e-06,
+      "loss": 228.9455,
+      "step": 30210
+    },
+    {
+      "epoch": 0.5812069371721456,
+      "grad_norm": 404.46120195011787,
+      "learning_rate": 7.605134029883381e-06,
+      "loss": 231.8223,
+      "step": 30220
+    },
+    {
+      "epoch": 0.5813992624326261,
+      "grad_norm": 427.0435056406804,
+      "learning_rate": 7.599208937136465e-06,
+      "loss": 230.0195,
+      "step": 30230
+    },
+    {
+      "epoch": 0.5815915876931066,
+      "grad_norm": 492.652091815168,
+      "learning_rate": 7.593284738644574e-06,
+      "loss": 236.554,
+      "step": 30240
+    },
+    {
+      "epoch": 0.5817839129535871,
+      "grad_norm": 405.2876325245797,
+      "learning_rate": 7.58736143661437e-06,
+      "loss": 231.29,
+      "step": 30250
+    },
+    {
+      "epoch": 0.5819762382140676,
+      "grad_norm": 400.54105191631095,
+      "learning_rate": 7.5814390332521824e-06,
+      "loss": 231.5176,
+      "step": 30260
+    },
+    {
+      "epoch": 0.5821685634745482,
+      "grad_norm": 452.9209422209049,
+      "learning_rate": 7.575517530764011e-06,
+      "loss": 237.1046,
+      "step": 30270
+    },
+    {
+      "epoch": 0.5823608887350287,
+      "grad_norm": 419.08458295175444,
+      "learning_rate": 7.569596931355517e-06,
+      "loss": 230.7505,
+      "step": 30280
+    },
+    {
+      "epoch": 0.5825532139955092,
+      "grad_norm": 404.9006220897538,
+      "learning_rate": 7.56367723723203e-06,
+      "loss": 229.8239,
+      "step": 30290
+    },
+    {
+      "epoch": 0.5827455392559897,
+      "grad_norm": 430.54315501259185,
+      "learning_rate": 7.557758450598534e-06,
+      "loss": 234.9123,
+      "step": 30300
+    },
+    {
+      "epoch": 0.5829378645164702,
+      "grad_norm": 439.26702337885746,
+      "learning_rate": 7.551840573659677e-06,
+      "loss": 233.0573,
+      "step": 30310
+    },
+    {
+      "epoch": 0.5831301897769507,
+      "grad_norm": 424.3910357431075,
+      "learning_rate": 7.5459236086197775e-06,
+      "loss": 235.5535,
+      "step": 30320
+    },
+    {
+      "epoch": 0.5833225150374313,
+      "grad_norm": 435.8422419126059,
+      "learning_rate": 7.5400075576828e-06,
+      "loss": 230.3998,
+      "step": 30330
+    },
+    {
+      "epoch": 0.5835148402979118,
+      "grad_norm": 467.59057678324507,
+      "learning_rate": 7.534092423052382e-06,
+      "loss": 229.3895,
+      "step": 30340
+    },
+    {
+      "epoch": 0.5837071655583923,
+      "grad_norm": 392.83174014064605,
+      "learning_rate": 7.5281782069318075e-06,
+      "loss": 236.8895,
+      "step": 30350
+    },
+    {
+      "epoch": 0.5838994908188728,
+      "grad_norm": 408.64504988459606,
+      "learning_rate": 7.522264911524031e-06,
+      "loss": 225.9544,
+      "step": 30360
+    },
+    {
+      "epoch": 0.5840918160793535,
+      "grad_norm": 458.60682984834364,
+      "learning_rate": 7.516352539031654e-06,
+      "loss": 231.1042,
+      "step": 30370
+    },
+    {
+      "epoch": 0.584284141339834,
+      "grad_norm": 442.10506943256286,
+      "learning_rate": 7.510441091656933e-06,
+      "loss": 230.3113,
+      "step": 30380
+    },
+    {
+      "epoch": 0.5844764666003145,
+      "grad_norm": 424.90061971921887,
+      "learning_rate": 7.504530571601792e-06,
+      "loss": 232.1466,
+      "step": 30390
+    },
+    {
+      "epoch": 0.584668791860795,
+      "grad_norm": 450.53099398000296,
+      "learning_rate": 7.498620981067799e-06,
+      "loss": 231.1732,
+      "step": 30400
+    },
+    {
+      "epoch": 0.5848611171212755,
+      "grad_norm": 444.83171206775637,
+      "learning_rate": 7.492712322256177e-06,
+      "loss": 235.6026,
+      "step": 30410
+    },
+    {
+      "epoch": 0.585053442381756,
+      "grad_norm": 408.7054957045418,
+      "learning_rate": 7.486804597367807e-06,
+      "loss": 227.3443,
+      "step": 30420
+    },
+    {
+      "epoch": 0.5852457676422366,
+      "grad_norm": 472.02099183170355,
+      "learning_rate": 7.480897808603219e-06,
+      "loss": 227.5788,
+      "step": 30430
+    },
+    {
+      "epoch": 0.5854380929027171,
+      "grad_norm": 399.4266232448373,
+      "learning_rate": 7.474991958162594e-06,
+      "loss": 228.8603,
+      "step": 30440
+    },
+    {
+      "epoch": 0.5856304181631976,
+      "grad_norm": 424.9655179956938,
+      "learning_rate": 7.469087048245758e-06,
+      "loss": 228.3165,
+      "step": 30450
+    },
+    {
+      "epoch": 0.5858227434236781,
+      "grad_norm": 492.2468964577878,
+      "learning_rate": 7.463183081052201e-06,
+      "loss": 246.5162,
+      "step": 30460
+    },
+    {
+      "epoch": 0.5860150686841586,
+      "grad_norm": 470.4055813416328,
+      "learning_rate": 7.457280058781049e-06,
+      "loss": 234.8857,
+      "step": 30470
+    },
+    {
+      "epoch": 0.5862073939446392,
+      "grad_norm": 391.8138703097309,
+      "learning_rate": 7.451377983631078e-06,
+      "loss": 240.1962,
+      "step": 30480
+    },
+    {
+      "epoch": 0.5863997192051197,
+      "grad_norm": 440.67473922789134,
+      "learning_rate": 7.445476857800717e-06,
+      "loss": 232.0519,
+      "step": 30490
+    },
+    {
+      "epoch": 0.5865920444656002,
+      "grad_norm": 441.44048588385436,
+      "learning_rate": 7.439576683488039e-06,
+      "loss": 236.3313,
+      "step": 30500
+    },
+    {
+      "epoch": 0.5867843697260807,
+      "grad_norm": 413.6776950795708,
+      "learning_rate": 7.4336774628907604e-06,
+      "loss": 233.8738,
+      "step": 30510
+    },
+    {
+      "epoch": 0.5869766949865612,
+      "grad_norm": 377.0394250130346,
+      "learning_rate": 7.427779198206238e-06,
+      "loss": 227.6462,
+      "step": 30520
+    },
+    {
+      "epoch": 0.5871690202470418,
+      "grad_norm": 417.4955303849164,
+      "learning_rate": 7.421881891631487e-06,
+      "loss": 230.2861,
+      "step": 30530
+    },
+    {
+      "epoch": 0.5873613455075223,
+      "grad_norm": 452.8235469494731,
+      "learning_rate": 7.415985545363152e-06,
+      "loss": 227.2522,
+      "step": 30540
+    },
+    {
+      "epoch": 0.5875536707680028,
+      "grad_norm": 405.2523553072771,
+      "learning_rate": 7.410090161597523e-06,
+      "loss": 229.1613,
+      "step": 30550
+    },
+    {
+      "epoch": 0.5877459960284833,
+      "grad_norm": 430.0011771345925,
+      "learning_rate": 7.404195742530533e-06,
+      "loss": 228.5989,
+      "step": 30560
+    },
+    {
+      "epoch": 0.587938321288964,
+      "grad_norm": 439.6467105663855,
+      "learning_rate": 7.398302290357763e-06,
+      "loss": 229.5614,
+      "step": 30570
+    },
+    {
+      "epoch": 0.5881306465494445,
+      "grad_norm": 436.81704954605465,
+      "learning_rate": 7.392409807274421e-06,
+      "loss": 226.3135,
+      "step": 30580
+    },
+    {
+      "epoch": 0.588322971809925,
+      "grad_norm": 474.10804028632,
+      "learning_rate": 7.386518295475355e-06,
+      "loss": 231.6304,
+      "step": 30590
+    },
+    {
+      "epoch": 0.5885152970704055,
+      "grad_norm": 467.3658345242713,
+      "learning_rate": 7.380627757155065e-06,
+      "loss": 233.0394,
+      "step": 30600
+    },
+    {
+      "epoch": 0.588707622330886,
+      "grad_norm": 399.486822025666,
+      "learning_rate": 7.374738194507675e-06,
+      "loss": 227.9455,
+      "step": 30610
+    },
+    {
+      "epoch": 0.5888999475913665,
+      "grad_norm": 454.2506372473318,
+      "learning_rate": 7.3688496097269494e-06,
+      "loss": 229.979,
+      "step": 30620
+    },
+    {
+      "epoch": 0.5890922728518471,
+      "grad_norm": 445.4615069161232,
+      "learning_rate": 7.362962005006286e-06,
+      "loss": 240.3873,
+      "step": 30630
+    },
+    {
+      "epoch": 0.5892845981123276,
+      "grad_norm": 402.9525346358271,
+      "learning_rate": 7.3570753825387275e-06,
+      "loss": 223.9229,
+      "step": 30640
+    },
+    {
+      "epoch": 0.5894769233728081,
+      "grad_norm": 417.43736079717536,
+      "learning_rate": 7.35118974451694e-06,
+      "loss": 223.1825,
+      "step": 30650
+    },
+    {
+      "epoch": 0.5896692486332886,
+      "grad_norm": 426.8909589760677,
+      "learning_rate": 7.345305093133226e-06,
+      "loss": 226.7633,
+      "step": 30660
+    },
+    {
+      "epoch": 0.5898615738937691,
+      "grad_norm": 408.00056600101846,
+      "learning_rate": 7.3394214305795175e-06,
+      "loss": 231.6228,
+      "step": 30670
+    },
+    {
+      "epoch": 0.5900538991542497,
+      "grad_norm": 418.52003638147767,
+      "learning_rate": 7.33353875904739e-06,
+      "loss": 228.9093,
+      "step": 30680
+    },
+    {
+      "epoch": 0.5902462244147302,
+      "grad_norm": 423.2060876939536,
+      "learning_rate": 7.327657080728032e-06,
+      "loss": 226.2645,
+      "step": 30690
+    },
+    {
+      "epoch": 0.5904385496752107,
+      "grad_norm": 431.7907857457317,
+      "learning_rate": 7.321776397812279e-06,
+      "loss": 228.9725,
+      "step": 30700
+    },
+    {
+      "epoch": 0.5906308749356912,
+      "grad_norm": 426.5910519938674,
+      "learning_rate": 7.315896712490584e-06,
+      "loss": 232.8618,
+      "step": 30710
+    },
+    {
+      "epoch": 0.5908232001961717,
+      "grad_norm": 415.5122035914214,
+      "learning_rate": 7.310018026953036e-06,
+      "loss": 230.8512,
+      "step": 30720
+    },
+    {
+      "epoch": 0.5910155254566523,
+      "grad_norm": 416.11562562473597,
+      "learning_rate": 7.304140343389348e-06,
+      "loss": 230.4957,
+      "step": 30730
+    },
+    {
+      "epoch": 0.5912078507171328,
+      "grad_norm": 413.03003196926915,
+      "learning_rate": 7.298263663988853e-06,
+      "loss": 236.0105,
+      "step": 30740
+    },
+    {
+      "epoch": 0.5914001759776133,
+      "grad_norm": 489.76692735114614,
+      "learning_rate": 7.292387990940526e-06,
+      "loss": 221.4764,
+      "step": 30750
+    },
+    {
+      "epoch": 0.5915925012380938,
+      "grad_norm": 442.67044421149234,
+      "learning_rate": 7.286513326432953e-06,
+      "loss": 229.0771,
+      "step": 30760
+    },
+    {
+      "epoch": 0.5917848264985743,
+      "grad_norm": 434.77302435504646,
+      "learning_rate": 7.2806396726543526e-06,
+      "loss": 239.3977,
+      "step": 30770
+    },
+    {
+      "epoch": 0.591977151759055,
+      "grad_norm": 487.6630040877233,
+      "learning_rate": 7.2747670317925625e-06,
+      "loss": 234.1094,
+      "step": 30780
+    },
+    {
+      "epoch": 0.5921694770195355,
+      "grad_norm": 530.3969122692031,
+      "learning_rate": 7.268895406035046e-06,
+      "loss": 227.5739,
+      "step": 30790
+    },
+    {
+      "epoch": 0.592361802280016,
+      "grad_norm": 449.93679596780515,
+      "learning_rate": 7.263024797568884e-06,
+      "loss": 236.7667,
+      "step": 30800
+    },
+    {
+      "epoch": 0.5925541275404965,
+      "grad_norm": 409.5199402061422,
+      "learning_rate": 7.257155208580778e-06,
+      "loss": 228.957,
+      "step": 30810
+    },
+    {
+      "epoch": 0.592746452800977,
+      "grad_norm": 423.41739669071717,
+      "learning_rate": 7.251286641257062e-06,
+      "loss": 232.9693,
+      "step": 30820
+    },
+    {
+      "epoch": 0.5929387780614576,
+      "grad_norm": 440.0582684109187,
+      "learning_rate": 7.245419097783674e-06,
+      "loss": 238.0952,
+      "step": 30830
+    },
+    {
+      "epoch": 0.5931311033219381,
+      "grad_norm": 413.04356947514185,
+      "learning_rate": 7.239552580346181e-06,
+      "loss": 222.2364,
+      "step": 30840
+    },
+    {
+      "epoch": 0.5933234285824186,
+      "grad_norm": 454.83446336969416,
+      "learning_rate": 7.233687091129757e-06,
+      "loss": 239.9237,
+      "step": 30850
+    },
+    {
+      "epoch": 0.5935157538428991,
+      "grad_norm": 454.8123010774503,
+      "learning_rate": 7.227822632319208e-06,
+      "loss": 240.0741,
+      "step": 30860
+    },
+    {
+      "epoch": 0.5937080791033796,
+      "grad_norm": 475.77692664060123,
+      "learning_rate": 7.221959206098945e-06,
+      "loss": 244.5912,
+      "step": 30870
+    },
+    {
+      "epoch": 0.5939004043638602,
+      "grad_norm": 453.8827590251814,
+      "learning_rate": 7.216096814652992e-06,
+      "loss": 242.6609,
+      "step": 30880
+    },
+    {
+      "epoch": 0.5940927296243407,
+      "grad_norm": 411.61717707446314,
+      "learning_rate": 7.210235460165002e-06,
+      "loss": 231.2817,
+      "step": 30890
+    },
+    {
+      "epoch": 0.5942850548848212,
+      "grad_norm": 404.19490631455994,
+      "learning_rate": 7.2043751448182275e-06,
+      "loss": 229.3745,
+      "step": 30900
+    },
+    {
+      "epoch": 0.5944773801453017,
+      "grad_norm": 396.702491867281,
+      "learning_rate": 7.198515870795542e-06,
+      "loss": 227.3233,
+      "step": 30910
+    },
+    {
+      "epoch": 0.5946697054057822,
+      "grad_norm": 417.1059933929704,
+      "learning_rate": 7.192657640279421e-06,
+      "loss": 235.2362,
+      "step": 30920
+    },
+    {
+      "epoch": 0.5948620306662628,
+      "grad_norm": 412.98431677687495,
+      "learning_rate": 7.18680045545197e-06,
+      "loss": 235.2402,
+      "step": 30930
+    },
+    {
+      "epoch": 0.5950543559267433,
+      "grad_norm": 522.2108226665455,
+      "learning_rate": 7.180944318494888e-06,
+      "loss": 234.2923,
+      "step": 30940
+    },
+    {
+      "epoch": 0.5952466811872238,
+      "grad_norm": 416.30528816933514,
+      "learning_rate": 7.175089231589485e-06,
+      "loss": 234.4751,
+      "step": 30950
+    },
+    {
+      "epoch": 0.5954390064477043,
+      "grad_norm": 430.34876434075846,
+      "learning_rate": 7.1692351969166905e-06,
+      "loss": 241.9152,
+      "step": 30960
+    },
+    {
+      "epoch": 0.5956313317081848,
+      "grad_norm": 432.6461229166163,
+      "learning_rate": 7.163382216657033e-06,
+      "loss": 227.4343,
+      "step": 30970
+    },
+    {
+      "epoch": 0.5958236569686655,
+      "grad_norm": 391.851122062776,
+      "learning_rate": 7.157530292990654e-06,
+      "loss": 228.7255,
+      "step": 30980
+    },
+    {
+      "epoch": 0.596015982229146,
+      "grad_norm": 416.6609842411124,
+      "learning_rate": 7.151679428097291e-06,
+      "loss": 221.7838,
+      "step": 30990
+    },
+    {
+      "epoch": 0.5962083074896265,
+      "grad_norm": 436.3104190053486,
+      "learning_rate": 7.145829624156304e-06,
+      "loss": 232.5095,
+      "step": 31000
+    },
+    {
+      "epoch": 0.596400632750107,
+      "grad_norm": 444.625772468422,
+      "learning_rate": 7.1399808833466445e-06,
+      "loss": 228.7998,
+      "step": 31010
+    },
+    {
+      "epoch": 0.5965929580105875,
+      "grad_norm": 408.19194897682854,
+      "learning_rate": 7.134133207846869e-06,
+      "loss": 231.0402,
+      "step": 31020
+    },
+    {
+      "epoch": 0.5967852832710681,
+      "grad_norm": 441.69043582928697,
+      "learning_rate": 7.128286599835139e-06,
+      "loss": 235.2315,
+      "step": 31030
+    },
+    {
+      "epoch": 0.5969776085315486,
+      "grad_norm": 429.1934329341106,
+      "learning_rate": 7.122441061489228e-06,
+      "loss": 220.99,
+      "step": 31040
+    },
+    {
+      "epoch": 0.5971699337920291,
+      "grad_norm": 436.46288146683776,
+      "learning_rate": 7.1165965949864934e-06,
+      "loss": 231.0572,
+      "step": 31050
+    },
+    {
+      "epoch": 0.5973622590525096,
+      "grad_norm": 413.3573834085358,
+      "learning_rate": 7.110753202503906e-06,
+      "loss": 229.231,
+      "step": 31060
+    },
+    {
+      "epoch": 0.5975545843129901,
+      "grad_norm": 429.7575862430089,
+      "learning_rate": 7.104910886218036e-06,
+      "loss": 231.5436,
+      "step": 31070
+    },
+    {
+      "epoch": 0.5977469095734707,
+      "grad_norm": 452.85150154079275,
+      "learning_rate": 7.0990696483050466e-06,
+      "loss": 230.5596,
+      "step": 31080
+    },
+    {
+      "epoch": 0.5979392348339512,
+      "grad_norm": 407.79552272283394,
+      "learning_rate": 7.093229490940704e-06,
+      "loss": 231.8558,
+      "step": 31090
+    },
+    {
+      "epoch": 0.5981315600944317,
+      "grad_norm": 410.29861285862563,
+      "learning_rate": 7.087390416300364e-06,
+      "loss": 223.9269,
+      "step": 31100
+    },
+    {
+      "epoch": 0.5983238853549122,
+      "grad_norm": 423.527518905979,
+      "learning_rate": 7.081552426558995e-06,
+      "loss": 237.3905,
+      "step": 31110
+    },
+    {
+      "epoch": 0.5985162106153927,
+      "grad_norm": 416.8528848057666,
+      "learning_rate": 7.075715523891146e-06,
+      "loss": 224.5927,
+      "step": 31120
+    },
+    {
+      "epoch": 0.5987085358758732,
+      "grad_norm": 435.14326974674015,
+      "learning_rate": 7.069879710470965e-06,
+      "loss": 226.3886,
+      "step": 31130
+    },
+    {
+      "epoch": 0.5989008611363538,
+      "grad_norm": 444.6357715018171,
+      "learning_rate": 7.064044988472204e-06,
+      "loss": 222.7304,
+      "step": 31140
+    },
+    {
+      "epoch": 0.5990931863968343,
+      "grad_norm": 435.06287248578263,
+      "learning_rate": 7.058211360068196e-06,
+      "loss": 226.1482,
+      "step": 31150
+    },
+    {
+      "epoch": 0.5992855116573148,
+      "grad_norm": 425.1248581785124,
+      "learning_rate": 7.052378827431871e-06,
+      "loss": 231.0841,
+      "step": 31160
+    },
+    {
+      "epoch": 0.5994778369177953,
+      "grad_norm": 460.4416805066328,
+      "learning_rate": 7.046547392735747e-06,
+      "loss": 224.6262,
+      "step": 31170
+    },
+    {
+      "epoch": 0.5996701621782758,
+      "grad_norm": 447.77334630031237,
+      "learning_rate": 7.040717058151945e-06,
+      "loss": 225.4802,
+      "step": 31180
+    },
+    {
+      "epoch": 0.5998624874387565,
+      "grad_norm": 414.48880746781487,
+      "learning_rate": 7.034887825852164e-06,
+      "loss": 229.6972,
+      "step": 31190
+    },
+    {
+      "epoch": 0.600054812699237,
+      "grad_norm": 396.412252112278,
+      "learning_rate": 7.029059698007699e-06,
+      "loss": 222.5981,
+      "step": 31200
+    },
+    {
+      "epoch": 0.6002471379597175,
+      "grad_norm": 462.33603927648323,
+      "learning_rate": 7.023232676789424e-06,
+      "loss": 238.546,
+      "step": 31210
+    },
+    {
+      "epoch": 0.600439463220198,
+      "grad_norm": 465.2785037641388,
+      "learning_rate": 7.01740676436782e-06,
+      "loss": 229.8366,
+      "step": 31220
+    },
+    {
+      "epoch": 0.6006317884806786,
+      "grad_norm": 411.1024134751334,
+      "learning_rate": 7.011581962912936e-06,
+      "loss": 232.3649,
+      "step": 31230
+    },
+    {
+      "epoch": 0.6008241137411591,
+      "grad_norm": 442.1245248314739,
+      "learning_rate": 7.005758274594412e-06,
+      "loss": 229.8007,
+      "step": 31240
+    },
+    {
+      "epoch": 0.6010164390016396,
+      "grad_norm": 415.9724184568621,
+      "learning_rate": 6.999935701581482e-06,
+      "loss": 228.002,
+      "step": 31250
+    },
+    {
+      "epoch": 0.6012087642621201,
+      "grad_norm": 422.07573573152905,
+      "learning_rate": 6.9941142460429555e-06,
+      "loss": 229.0496,
+      "step": 31260
+    },
+    {
+      "epoch": 0.6014010895226006,
+      "grad_norm": 411.8871133271173,
+      "learning_rate": 6.988293910147229e-06,
+      "loss": 229.6827,
+      "step": 31270
+    },
+    {
+      "epoch": 0.6015934147830811,
+      "grad_norm": 438.6845728167455,
+      "learning_rate": 6.982474696062278e-06,
+      "loss": 230.1404,
+      "step": 31280
+    },
+    {
+      "epoch": 0.6017857400435617,
+      "grad_norm": 421.96162422719874,
+      "learning_rate": 6.97665660595567e-06,
+      "loss": 229.8157,
+      "step": 31290
+    },
+    {
+      "epoch": 0.6019780653040422,
+      "grad_norm": 437.0338238568071,
+      "learning_rate": 6.970839641994545e-06,
+      "loss": 226.9675,
+      "step": 31300
+    },
+    {
+      "epoch": 0.6021703905645227,
+      "grad_norm": 429.8338738138326,
+      "learning_rate": 6.965023806345619e-06,
+      "loss": 226.8889,
+      "step": 31310
+    },
+    {
+      "epoch": 0.6023627158250032,
+      "grad_norm": 376.2411129408347,
+      "learning_rate": 6.959209101175206e-06,
+      "loss": 225.3643,
+      "step": 31320
+    },
+    {
+      "epoch": 0.6025550410854837,
+      "grad_norm": 420.77861340778287,
+      "learning_rate": 6.9533955286491805e-06,
+      "loss": 233.2466,
+      "step": 31330
+    },
+    {
+      "epoch": 0.6027473663459643,
+      "grad_norm": 375.10389459286114,
+      "learning_rate": 6.947583090933008e-06,
+      "loss": 222.4511,
+      "step": 31340
+    },
+    {
+      "epoch": 0.6029396916064448,
+      "grad_norm": 384.4566297787739,
+      "learning_rate": 6.941771790191716e-06,
+      "loss": 224.1897,
+      "step": 31350
+    },
+    {
+      "epoch": 0.6031320168669253,
+      "grad_norm": 422.8293228385227,
+      "learning_rate": 6.9359616285899266e-06,
+      "loss": 230.1716,
+      "step": 31360
+    },
+    {
+      "epoch": 0.6033243421274058,
+      "grad_norm": 473.8511483921668,
+      "learning_rate": 6.930152608291829e-06,
+      "loss": 236.4207,
+      "step": 31370
+    },
+    {
+      "epoch": 0.6035166673878863,
+      "grad_norm": 409.07980676046975,
+      "learning_rate": 6.924344731461179e-06,
+      "loss": 222.9459,
+      "step": 31380
+    },
+    {
+      "epoch": 0.603708992648367,
+      "grad_norm": 404.4186209304979,
+      "learning_rate": 6.918538000261325e-06,
+      "loss": 225.3304,
+      "step": 31390
+    },
+    {
+      "epoch": 0.6039013179088475,
+      "grad_norm": 470.42591099954507,
+      "learning_rate": 6.912732416855171e-06,
+      "loss": 232.8654,
+      "step": 31400
+    },
+    {
+      "epoch": 0.604093643169328,
+      "grad_norm": 446.1054671043929,
+      "learning_rate": 6.906927983405207e-06,
+      "loss": 234.0471,
+      "step": 31410
+    },
+    {
+      "epoch": 0.6042859684298085,
+      "grad_norm": 404.6723379679115,
+      "learning_rate": 6.901124702073481e-06,
+      "loss": 227.3843,
+      "step": 31420
+    },
+    {
+      "epoch": 0.604478293690289,
+      "grad_norm": 412.32031400244597,
+      "learning_rate": 6.895322575021628e-06,
+      "loss": 233.438,
+      "step": 31430
+    },
+    {
+      "epoch": 0.6046706189507696,
+      "grad_norm": 442.68633185419355,
+      "learning_rate": 6.88952160441084e-06,
+      "loss": 233.5317,
+      "step": 31440
+    },
+    {
+      "epoch": 0.6048629442112501,
+      "grad_norm": 428.652802850576,
+      "learning_rate": 6.8837217924018825e-06,
+      "loss": 227.9286,
+      "step": 31450
+    },
+    {
+      "epoch": 0.6050552694717306,
+      "grad_norm": 405.59024106209904,
+      "learning_rate": 6.877923141155087e-06,
+      "loss": 234.7568,
+      "step": 31460
+    },
+    {
+      "epoch": 0.6052475947322111,
+      "grad_norm": 429.74687944027386,
+      "learning_rate": 6.87212565283036e-06,
+      "loss": 223.7243,
+      "step": 31470
+    },
+    {
+      "epoch": 0.6054399199926916,
+      "grad_norm": 413.62916687397916,
+      "learning_rate": 6.8663293295871715e-06,
+      "loss": 224.6487,
+      "step": 31480
+    },
+    {
+      "epoch": 0.6056322452531722,
+      "grad_norm": 403.35266000421103,
+      "learning_rate": 6.86053417358455e-06,
+      "loss": 232.3997,
+      "step": 31490
+    },
+    {
+      "epoch": 0.6058245705136527,
+      "grad_norm": 418.18742835234315,
+      "learning_rate": 6.854740186981102e-06,
+      "loss": 228.476,
+      "step": 31500
+    },
+    {
+      "epoch": 0.6060168957741332,
+      "grad_norm": 413.3692710430336,
+      "learning_rate": 6.848947371934989e-06,
+      "loss": 232.8474,
+      "step": 31510
+    },
+    {
+      "epoch": 0.6062092210346137,
+      "grad_norm": 458.7192780655746,
+      "learning_rate": 6.843155730603939e-06,
+      "loss": 234.3798,
+      "step": 31520
+    },
+    {
+      "epoch": 0.6064015462950942,
+      "grad_norm": 410.4532868082447,
+      "learning_rate": 6.837365265145237e-06,
+      "loss": 224.8932,
+      "step": 31530
+    },
+    {
+      "epoch": 0.6065938715555748,
+      "grad_norm": 441.7061583346476,
+      "learning_rate": 6.831575977715745e-06,
+      "loss": 230.1555,
+      "step": 31540
+    },
+    {
+      "epoch": 0.6067861968160553,
+      "grad_norm": 392.5810126768839,
+      "learning_rate": 6.825787870471873e-06,
+      "loss": 225.8656,
+      "step": 31550
+    },
+    {
+      "epoch": 0.6069785220765358,
+      "grad_norm": 510.8174711552602,
+      "learning_rate": 6.820000945569592e-06,
+      "loss": 226.7284,
+      "step": 31560
+    },
+    {
+      "epoch": 0.6071708473370163,
+      "grad_norm": 479.66101806827083,
+      "learning_rate": 6.814215205164444e-06,
+      "loss": 237.7188,
+      "step": 31570
+    },
+    {
+      "epoch": 0.6073631725974968,
+      "grad_norm": 399.1802075372459,
+      "learning_rate": 6.808430651411518e-06,
+      "loss": 225.9488,
+      "step": 31580
+    },
+    {
+      "epoch": 0.6075554978579775,
+      "grad_norm": 428.7763591661641,
+      "learning_rate": 6.802647286465461e-06,
+      "loss": 228.123,
+      "step": 31590
+    },
+    {
+      "epoch": 0.607747823118458,
+      "grad_norm": 425.208346229042,
+      "learning_rate": 6.796865112480482e-06,
+      "loss": 235.3086,
+      "step": 31600
+    },
+    {
+      "epoch": 0.6079401483789385,
+      "grad_norm": 410.9462737738275,
+      "learning_rate": 6.79108413161035e-06,
+      "loss": 230.0048,
+      "step": 31610
+    },
+    {
+      "epoch": 0.608132473639419,
+      "grad_norm": 434.71890880773276,
+      "learning_rate": 6.785304346008381e-06,
+      "loss": 229.3023,
+      "step": 31620
+    },
+    {
+      "epoch": 0.6083247988998995,
+      "grad_norm": 402.8449693305862,
+      "learning_rate": 6.779525757827452e-06,
+      "loss": 224.3566,
+      "step": 31630
+    },
+    {
+      "epoch": 0.6085171241603801,
+      "grad_norm": 423.4436545904073,
+      "learning_rate": 6.773748369219986e-06,
+      "loss": 232.5006,
+      "step": 31640
+    },
+    {
+      "epoch": 0.6087094494208606,
+      "grad_norm": 416.3640244341056,
+      "learning_rate": 6.767972182337974e-06,
+      "loss": 236.615,
+      "step": 31650
+    },
+    {
+      "epoch": 0.6089017746813411,
+      "grad_norm": 407.73175172586537,
+      "learning_rate": 6.762197199332945e-06,
+      "loss": 227.2565,
+      "step": 31660
+    },
+    {
+      "epoch": 0.6090940999418216,
+      "grad_norm": 387.546886934667,
+      "learning_rate": 6.756423422355981e-06,
+      "loss": 223.1566,
+      "step": 31670
+    },
+    {
+      "epoch": 0.6092864252023021,
+      "grad_norm": 426.7701797203757,
+      "learning_rate": 6.750650853557728e-06,
+      "loss": 222.9499,
+      "step": 31680
+    },
+    {
+      "epoch": 0.6094787504627827,
+      "grad_norm": 401.4472983638562,
+      "learning_rate": 6.744879495088364e-06,
+      "loss": 231.3808,
+      "step": 31690
+    },
+    {
+      "epoch": 0.6096710757232632,
+      "grad_norm": 422.555188962367,
+      "learning_rate": 6.7391093490976285e-06,
+      "loss": 220.4079,
+      "step": 31700
+    },
+    {
+      "epoch": 0.6098634009837437,
+      "grad_norm": 422.1159712098415,
+      "learning_rate": 6.7333404177348036e-06,
+      "loss": 220.7019,
+      "step": 31710
+    },
+    {
+      "epoch": 0.6100557262442242,
+      "grad_norm": 401.1683902465003,
+      "learning_rate": 6.727572703148726e-06,
+      "loss": 223.8618,
+      "step": 31720
+    },
+    {
+      "epoch": 0.6102480515047047,
+      "grad_norm": 399.4895379726904,
+      "learning_rate": 6.721806207487769e-06,
+      "loss": 228.9129,
+      "step": 31730
+    },
+    {
+      "epoch": 0.6104403767651853,
+      "grad_norm": 410.35629069607234,
+      "learning_rate": 6.716040932899857e-06,
+      "loss": 235.2316,
+      "step": 31740
+    },
+    {
+      "epoch": 0.6106327020256658,
+      "grad_norm": 402.15512413169114,
+      "learning_rate": 6.710276881532463e-06,
+      "loss": 228.6174,
+      "step": 31750
+    },
+    {
+      "epoch": 0.6108250272861463,
+      "grad_norm": 404.59875865191776,
+      "learning_rate": 6.704514055532597e-06,
+      "loss": 230.2955,
+      "step": 31760
+    },
+    {
+      "epoch": 0.6110173525466268,
+      "grad_norm": 444.9015145453593,
+      "learning_rate": 6.698752457046822e-06,
+      "loss": 223.154,
+      "step": 31770
+    },
+    {
+      "epoch": 0.6112096778071073,
+      "grad_norm": 436.20073319166994,
+      "learning_rate": 6.692992088221231e-06,
+      "loss": 233.3152,
+      "step": 31780
+    },
+    {
+      "epoch": 0.6114020030675879,
+      "grad_norm": 405.4522546906219,
+      "learning_rate": 6.687232951201473e-06,
+      "loss": 222.9069,
+      "step": 31790
+    },
+    {
+      "epoch": 0.6115943283280685,
+      "grad_norm": 419.9575646012201,
+      "learning_rate": 6.681475048132729e-06,
+      "loss": 234.9125,
+      "step": 31800
+    },
+    {
+      "epoch": 0.611786653588549,
+      "grad_norm": 416.17713050775745,
+      "learning_rate": 6.675718381159719e-06,
+      "loss": 225.6077,
+      "step": 31810
+    },
+    {
+      "epoch": 0.6119789788490295,
+      "grad_norm": 418.3594856479087,
+      "learning_rate": 6.6699629524267114e-06,
+      "loss": 224.1185,
+      "step": 31820
+    },
+    {
+      "epoch": 0.61217130410951,
+      "grad_norm": 409.5014753040369,
+      "learning_rate": 6.664208764077507e-06,
+      "loss": 230.5952,
+      "step": 31830
+    },
+    {
+      "epoch": 0.6123636293699906,
+      "grad_norm": 409.58535223196526,
+      "learning_rate": 6.658455818255445e-06,
+      "loss": 221.038,
+      "step": 31840
+    },
+    {
+      "epoch": 0.6125559546304711,
+      "grad_norm": 401.4956852117134,
+      "learning_rate": 6.652704117103401e-06,
+      "loss": 222.4708,
+      "step": 31850
+    },
+    {
+      "epoch": 0.6127482798909516,
+      "grad_norm": 408.45207805125233,
+      "learning_rate": 6.646953662763796e-06,
+      "loss": 237.5248,
+      "step": 31860
+    },
+    {
+      "epoch": 0.6129406051514321,
+      "grad_norm": 424.513095277039,
+      "learning_rate": 6.6412044573785725e-06,
+      "loss": 227.6676,
+      "step": 31870
+    },
+    {
+      "epoch": 0.6131329304119126,
+      "grad_norm": 422.6860046503193,
+      "learning_rate": 6.635456503089217e-06,
+      "loss": 226.0225,
+      "step": 31880
+    },
+    {
+      "epoch": 0.6133252556723932,
+      "grad_norm": 457.4857511993487,
+      "learning_rate": 6.6297098020367435e-06,
+      "loss": 242.5258,
+      "step": 31890
+    },
+    {
+      "epoch": 0.6135175809328737,
+      "grad_norm": 426.9327896119648,
+      "learning_rate": 6.623964356361707e-06,
+      "loss": 240.1359,
+      "step": 31900
+    },
+    {
+      "epoch": 0.6137099061933542,
+      "grad_norm": 430.3015965170402,
+      "learning_rate": 6.618220168204193e-06,
+      "loss": 225.0844,
+      "step": 31910
+    },
+    {
+      "epoch": 0.6139022314538347,
+      "grad_norm": 388.5793558755709,
+      "learning_rate": 6.6124772397038115e-06,
+      "loss": 227.3012,
+      "step": 31920
+    },
+    {
+      "epoch": 0.6140945567143152,
+      "grad_norm": 436.1814936283576,
+      "learning_rate": 6.606735572999714e-06,
+      "loss": 234.0395,
+      "step": 31930
+    },
+    {
+      "epoch": 0.6142868819747958,
+      "grad_norm": 398.50637975193155,
+      "learning_rate": 6.600995170230575e-06,
+      "loss": 218.5695,
+      "step": 31940
+    },
+    {
+      "epoch": 0.6144792072352763,
+      "grad_norm": 464.0005666559156,
+      "learning_rate": 6.595256033534598e-06,
+      "loss": 223.3937,
+      "step": 31950
+    },
+    {
+      "epoch": 0.6146715324957568,
+      "grad_norm": 413.6983316480439,
+      "learning_rate": 6.589518165049514e-06,
+      "loss": 223.8607,
+      "step": 31960
+    },
+    {
+      "epoch": 0.6148638577562373,
+      "grad_norm": 424.0484035368132,
+      "learning_rate": 6.5837815669125906e-06,
+      "loss": 229.2501,
+      "step": 31970
+    },
+    {
+      "epoch": 0.6150561830167178,
+      "grad_norm": 399.37052022136805,
+      "learning_rate": 6.5780462412606124e-06,
+      "loss": 217.4186,
+      "step": 31980
+    },
+    {
+      "epoch": 0.6152485082771983,
+      "grad_norm": 414.7301268176305,
+      "learning_rate": 6.572312190229895e-06,
+      "loss": 224.6406,
+      "step": 31990
+    },
+    {
+      "epoch": 0.615440833537679,
+      "grad_norm": 407.4929905602582,
+      "learning_rate": 6.5665794159562734e-06,
+      "loss": 224.0125,
+      "step": 32000
+    },
+    {
+      "epoch": 0.6156331587981595,
+      "grad_norm": 445.25430596855256,
+      "learning_rate": 6.560847920575118e-06,
+      "loss": 220.6891,
+      "step": 32010
+    },
+    {
+      "epoch": 0.61582548405864,
+      "grad_norm": 404.02473259242674,
+      "learning_rate": 6.5551177062213126e-06,
+      "loss": 232.0424,
+      "step": 32020
+    },
+    {
+      "epoch": 0.6160178093191205,
+      "grad_norm": 401.4137244071718,
+      "learning_rate": 6.5493887750292616e-06,
+      "loss": 224.4488,
+      "step": 32030
+    },
+    {
+      "epoch": 0.616210134579601,
+      "grad_norm": 394.8810662544867,
+      "learning_rate": 6.54366112913291e-06,
+      "loss": 234.6445,
+      "step": 32040
+    },
+    {
+      "epoch": 0.6164024598400816,
+      "grad_norm": 441.38737658537883,
+      "learning_rate": 6.537934770665701e-06,
+      "loss": 233.0354,
+      "step": 32050
+    },
+    {
+      "epoch": 0.6165947851005621,
+      "grad_norm": 433.46738262206486,
+      "learning_rate": 6.532209701760615e-06,
+      "loss": 237.4941,
+      "step": 32060
+    },
+    {
+      "epoch": 0.6167871103610426,
+      "grad_norm": 421.3984738399414,
+      "learning_rate": 6.526485924550138e-06,
+      "loss": 232.097,
+      "step": 32070
+    },
+    {
+      "epoch": 0.6169794356215231,
+      "grad_norm": 568.8816481180494,
+      "learning_rate": 6.520763441166291e-06,
+      "loss": 230.0154,
+      "step": 32080
+    },
+    {
+      "epoch": 0.6171717608820037,
+      "grad_norm": 385.5136898476257,
+      "learning_rate": 6.515042253740601e-06,
+      "loss": 216.763,
+      "step": 32090
+    },
+    {
+      "epoch": 0.6173640861424842,
+      "grad_norm": 429.7704586362384,
+      "learning_rate": 6.509322364404112e-06,
+      "loss": 229.7286,
+      "step": 32100
+    },
+    {
+      "epoch": 0.6175564114029647,
+      "grad_norm": 397.18492624274734,
+      "learning_rate": 6.503603775287395e-06,
+      "loss": 227.5604,
+      "step": 32110
+    },
+    {
+      "epoch": 0.6177487366634452,
+      "grad_norm": 428.32726061371864,
+      "learning_rate": 6.497886488520524e-06,
+      "loss": 222.9204,
+      "step": 32120
+    },
+    {
+      "epoch": 0.6179410619239257,
+      "grad_norm": 452.0128461453846,
+      "learning_rate": 6.4921705062331e-06,
+      "loss": 234.3032,
+      "step": 32130
+    },
+    {
+      "epoch": 0.6181333871844062,
+      "grad_norm": 410.44888770343476,
+      "learning_rate": 6.486455830554224e-06,
+      "loss": 222.3085,
+      "step": 32140
+    },
+    {
+      "epoch": 0.6183257124448868,
+      "grad_norm": 485.15354076366333,
+      "learning_rate": 6.4807424636125285e-06,
+      "loss": 226.4709,
+      "step": 32150
+    },
+    {
+      "epoch": 0.6185180377053673,
+      "grad_norm": 430.227140917458,
+      "learning_rate": 6.475030407536141e-06,
+      "loss": 228.9367,
+      "step": 32160
+    },
+    {
+      "epoch": 0.6187103629658478,
+      "grad_norm": 400.26791701588655,
+      "learning_rate": 6.469319664452709e-06,
+      "loss": 227.1511,
+      "step": 32170
+    },
+    {
+      "epoch": 0.6189026882263283,
+      "grad_norm": 420.3334240601306,
+      "learning_rate": 6.463610236489391e-06,
+      "loss": 221.2679,
+      "step": 32180
+    },
+    {
+      "epoch": 0.6190950134868088,
+      "grad_norm": 415.8398682323889,
+      "learning_rate": 6.457902125772854e-06,
+      "loss": 226.6218,
+      "step": 32190
+    },
+    {
+      "epoch": 0.6192873387472894,
+      "grad_norm": 491.827731218485,
+      "learning_rate": 6.452195334429277e-06,
+      "loss": 229.5702,
+      "step": 32200
+    },
+    {
+      "epoch": 0.61947966400777,
+      "grad_norm": 424.2696377102414,
+      "learning_rate": 6.446489864584341e-06,
+      "loss": 224.0806,
+      "step": 32210
+    },
+    {
+      "epoch": 0.6196719892682505,
+      "grad_norm": 397.0421844748927,
+      "learning_rate": 6.440785718363245e-06,
+      "loss": 223.3903,
+      "step": 32220
+    },
+    {
+      "epoch": 0.619864314528731,
+      "grad_norm": 418.63420095871084,
+      "learning_rate": 6.435082897890688e-06,
+      "loss": 223.8234,
+      "step": 32230
+    },
+    {
+      "epoch": 0.6200566397892115,
+      "grad_norm": 414.6394142402439,
+      "learning_rate": 6.429381405290873e-06,
+      "loss": 223.4642,
+      "step": 32240
+    },
+    {
+      "epoch": 0.6202489650496921,
+      "grad_norm": 417.6273198855874,
+      "learning_rate": 6.4236812426875124e-06,
+      "loss": 228.2614,
+      "step": 32250
+    },
+    {
+      "epoch": 0.6204412903101726,
+      "grad_norm": 428.1871115389233,
+      "learning_rate": 6.4179824122038244e-06,
+      "loss": 226.9813,
+      "step": 32260
+    },
+    {
+      "epoch": 0.6206336155706531,
+      "grad_norm": 459.7291468508094,
+      "learning_rate": 6.412284915962532e-06,
+      "loss": 228.2366,
+      "step": 32270
+    },
+    {
+      "epoch": 0.6208259408311336,
+      "grad_norm": 406.90371089574955,
+      "learning_rate": 6.406588756085849e-06,
+      "loss": 222.4472,
+      "step": 32280
+    },
+    {
+      "epoch": 0.6210182660916141,
+      "grad_norm": 422.53534585686907,
+      "learning_rate": 6.400893934695514e-06,
+      "loss": 224.6374,
+      "step": 32290
+    },
+    {
+      "epoch": 0.6212105913520947,
+      "grad_norm": 424.33570838179855,
+      "learning_rate": 6.395200453912747e-06,
+      "loss": 230.1573,
+      "step": 32300
+    },
+    {
+      "epoch": 0.6214029166125752,
+      "grad_norm": 402.7245190674156,
+      "learning_rate": 6.389508315858272e-06,
+      "loss": 228.3169,
+      "step": 32310
+    },
+    {
+      "epoch": 0.6215952418730557,
+      "grad_norm": 430.8290670915051,
+      "learning_rate": 6.38381752265232e-06,
+      "loss": 224.8335,
+      "step": 32320
+    },
+    {
+      "epoch": 0.6217875671335362,
+      "grad_norm": 470.47378827039034,
+      "learning_rate": 6.378128076414619e-06,
+      "loss": 230.406,
+      "step": 32330
+    },
+    {
+      "epoch": 0.6219798923940167,
+      "grad_norm": 431.366732515625,
+      "learning_rate": 6.372439979264393e-06,
+      "loss": 229.6548,
+      "step": 32340
+    },
+    {
+      "epoch": 0.6221722176544973,
+      "grad_norm": 414.25317450164516,
+      "learning_rate": 6.3667532333203655e-06,
+      "loss": 221.0754,
+      "step": 32350
+    },
+    {
+      "epoch": 0.6223645429149778,
+      "grad_norm": 384.6862073261352,
+      "learning_rate": 6.361067840700747e-06,
+      "loss": 223.1284,
+      "step": 32360
+    },
+    {
+      "epoch": 0.6225568681754583,
+      "grad_norm": 403.96040133608227,
+      "learning_rate": 6.355383803523265e-06,
+      "loss": 224.7733,
+      "step": 32370
+    },
+    {
+      "epoch": 0.6227491934359388,
+      "grad_norm": 409.6805576624066,
+      "learning_rate": 6.349701123905123e-06,
+      "loss": 225.2049,
+      "step": 32380
+    },
+    {
+      "epoch": 0.6229415186964193,
+      "grad_norm": 398.9666378801448,
+      "learning_rate": 6.344019803963021e-06,
+      "loss": 227.3894,
+      "step": 32390
+    },
+    {
+      "epoch": 0.6231338439568999,
+      "grad_norm": 417.0033530660901,
+      "learning_rate": 6.338339845813164e-06,
+      "loss": 223.8857,
+      "step": 32400
+    },
+    {
+      "epoch": 0.6233261692173805,
+      "grad_norm": 406.4564969629636,
+      "learning_rate": 6.332661251571241e-06,
+      "loss": 222.1794,
+      "step": 32410
+    },
+    {
+      "epoch": 0.623518494477861,
+      "grad_norm": 409.63481695095373,
+      "learning_rate": 6.326984023352435e-06,
+      "loss": 229.1074,
+      "step": 32420
+    },
+    {
+      "epoch": 0.6237108197383415,
+      "grad_norm": 401.4548811884464,
+      "learning_rate": 6.321308163271413e-06,
+      "loss": 224.6141,
+      "step": 32430
+    },
+    {
+      "epoch": 0.623903144998822,
+      "grad_norm": 427.1591386049867,
+      "learning_rate": 6.315633673442349e-06,
+      "loss": 224.6687,
+      "step": 32440
+    },
+    {
+      "epoch": 0.6240954702593026,
+      "grad_norm": 414.42085249198607,
+      "learning_rate": 6.309960555978894e-06,
+      "loss": 221.5361,
+      "step": 32450
+    },
+    {
+      "epoch": 0.6242877955197831,
+      "grad_norm": 396.68173985942974,
+      "learning_rate": 6.304288812994183e-06,
+      "loss": 222.9928,
+      "step": 32460
+    },
+    {
+      "epoch": 0.6244801207802636,
+      "grad_norm": 405.2462982459528,
+      "learning_rate": 6.298618446600856e-06,
+      "loss": 232.3568,
+      "step": 32470
+    },
+    {
+      "epoch": 0.6246724460407441,
+      "grad_norm": 408.38965061786035,
+      "learning_rate": 6.292949458911029e-06,
+      "loss": 226.5822,
+      "step": 32480
+    },
+    {
+      "epoch": 0.6248647713012246,
+      "grad_norm": 406.85472408021496,
+      "learning_rate": 6.287281852036304e-06,
+      "loss": 219.727,
+      "step": 32490
+    },
+    {
+      "epoch": 0.6250570965617052,
+      "grad_norm": 406.3130271794487,
+      "learning_rate": 6.2816156280877675e-06,
+      "loss": 223.7757,
+      "step": 32500
+    },
+    {
+      "epoch": 0.6252494218221857,
+      "grad_norm": 368.8070749656955,
+      "learning_rate": 6.275950789176002e-06,
+      "loss": 226.879,
+      "step": 32510
+    },
+    {
+      "epoch": 0.6254417470826662,
+      "grad_norm": 399.78400131991526,
+      "learning_rate": 6.270287337411064e-06,
+      "loss": 222.7621,
+      "step": 32520
+    },
+    {
+      "epoch": 0.6256340723431467,
+      "grad_norm": 418.7943588434436,
+      "learning_rate": 6.264625274902492e-06,
+      "loss": 229.581,
+      "step": 32530
+    },
+    {
+      "epoch": 0.6258263976036272,
+      "grad_norm": 408.46925834619316,
+      "learning_rate": 6.258964603759311e-06,
+      "loss": 231.0025,
+      "step": 32540
+    },
+    {
+      "epoch": 0.6260187228641078,
+      "grad_norm": 425.725554948789,
+      "learning_rate": 6.2533053260900345e-06,
+      "loss": 223.7737,
+      "step": 32550
+    },
+    {
+      "epoch": 0.6262110481245883,
+      "grad_norm": 379.102124914087,
+      "learning_rate": 6.247647444002644e-06,
+      "loss": 227.2497,
+      "step": 32560
+    },
+    {
+      "epoch": 0.6264033733850688,
+      "grad_norm": 410.724015359739,
+      "learning_rate": 6.241990959604607e-06,
+      "loss": 232.2127,
+      "step": 32570
+    },
+    {
+      "epoch": 0.6265956986455493,
+      "grad_norm": 389.35958129429366,
+      "learning_rate": 6.2363358750028745e-06,
+      "loss": 225.1557,
+      "step": 32580
+    },
+    {
+      "epoch": 0.6267880239060298,
+      "grad_norm": 403.83741678591196,
+      "learning_rate": 6.23068219230387e-06,
+      "loss": 220.9148,
+      "step": 32590
+    },
+    {
+      "epoch": 0.6269803491665104,
+      "grad_norm": 449.17386156052123,
+      "learning_rate": 6.2250299136134925e-06,
+      "loss": 221.5463,
+      "step": 32600
+    },
+    {
+      "epoch": 0.6271726744269909,
+      "grad_norm": 411.41821537712275,
+      "learning_rate": 6.219379041037128e-06,
+      "loss": 227.4787,
+      "step": 32610
+    },
+    {
+      "epoch": 0.6273649996874715,
+      "grad_norm": 849.7831508922266,
+      "learning_rate": 6.213729576679632e-06,
+      "loss": 224.3808,
+      "step": 32620
+    },
+    {
+      "epoch": 0.627557324947952,
+      "grad_norm": 403.5203729518764,
+      "learning_rate": 6.208081522645339e-06,
+      "loss": 219.9894,
+      "step": 32630
+    },
+    {
+      "epoch": 0.6277496502084325,
+      "grad_norm": 443.87665459898096,
+      "learning_rate": 6.202434881038048e-06,
+      "loss": 224.0993,
+      "step": 32640
+    },
+    {
+      "epoch": 0.6279419754689131,
+      "grad_norm": 378.8980746815382,
+      "learning_rate": 6.196789653961048e-06,
+      "loss": 222.3341,
+      "step": 32650
+    },
+    {
+      "epoch": 0.6281343007293936,
+      "grad_norm": 375.0334287873214,
+      "learning_rate": 6.191145843517093e-06,
+      "loss": 222.9797,
+      "step": 32660
+    },
+    {
+      "epoch": 0.6283266259898741,
+      "grad_norm": 432.9549116883272,
+      "learning_rate": 6.185503451808401e-06,
+      "loss": 226.5957,
+      "step": 32670
+    },
+    {
+      "epoch": 0.6285189512503546,
+      "grad_norm": 405.67308004001814,
+      "learning_rate": 6.1798624809366755e-06,
+      "loss": 231.5721,
+      "step": 32680
+    },
+    {
+      "epoch": 0.6287112765108351,
+      "grad_norm": 428.6002921642038,
+      "learning_rate": 6.174222933003084e-06,
+      "loss": 226.5769,
+      "step": 32690
+    },
+    {
+      "epoch": 0.6289036017713157,
+      "grad_norm": 427.70493777899384,
+      "learning_rate": 6.168584810108269e-06,
+      "loss": 227.2838,
+      "step": 32700
+    },
+    {
+      "epoch": 0.6290959270317962,
+      "grad_norm": 451.3863609797896,
+      "learning_rate": 6.162948114352328e-06,
+      "loss": 232.0322,
+      "step": 32710
+    },
+    {
+      "epoch": 0.6292882522922767,
+      "grad_norm": 429.05655546731646,
+      "learning_rate": 6.157312847834848e-06,
+      "loss": 225.3299,
+      "step": 32720
+    },
+    {
+      "epoch": 0.6294805775527572,
+      "grad_norm": 392.0913172198605,
+      "learning_rate": 6.1516790126548695e-06,
+      "loss": 222.9261,
+      "step": 32730
+    },
+    {
+      "epoch": 0.6296729028132377,
+      "grad_norm": 448.13366902461837,
+      "learning_rate": 6.146046610910899e-06,
+      "loss": 226.53,
+      "step": 32740
+    },
+    {
+      "epoch": 0.6298652280737183,
+      "grad_norm": 419.7198476188548,
+      "learning_rate": 6.140415644700915e-06,
+      "loss": 235.4107,
+      "step": 32750
+    },
+    {
+      "epoch": 0.6300575533341988,
+      "grad_norm": 430.6483742843943,
+      "learning_rate": 6.13478611612236e-06,
+      "loss": 227.4435,
+      "step": 32760
+    },
+    {
+      "epoch": 0.6302498785946793,
+      "grad_norm": 384.4994474298553,
+      "learning_rate": 6.129158027272144e-06,
+      "loss": 224.8102,
+      "step": 32770
+    },
+    {
+      "epoch": 0.6304422038551598,
+      "grad_norm": 442.829800147313,
+      "learning_rate": 6.123531380246632e-06,
+      "loss": 221.9298,
+      "step": 32780
+    },
+    {
+      "epoch": 0.6306345291156403,
+      "grad_norm": 373.9780847542322,
+      "learning_rate": 6.117906177141657e-06,
+      "loss": 227.9078,
+      "step": 32790
+    },
+    {
+      "epoch": 0.6308268543761208,
+      "grad_norm": 447.63520768166217,
+      "learning_rate": 6.112282420052518e-06,
+      "loss": 222.5495,
+      "step": 32800
+    },
+    {
+      "epoch": 0.6310191796366014,
+      "grad_norm": 396.84245210532055,
+      "learning_rate": 6.106660111073969e-06,
+      "loss": 225.7975,
+      "step": 32810
+    },
+    {
+      "epoch": 0.631211504897082,
+      "grad_norm": 403.2168245481589,
+      "learning_rate": 6.101039252300227e-06,
+      "loss": 228.843,
+      "step": 32820
+    },
+    {
+      "epoch": 0.6314038301575625,
+      "grad_norm": 404.034799768103,
+      "learning_rate": 6.095419845824971e-06,
+      "loss": 223.8758,
+      "step": 32830
+    },
+    {
+      "epoch": 0.631596155418043,
+      "grad_norm": 434.55380619672695,
+      "learning_rate": 6.089801893741338e-06,
+      "loss": 225.8411,
+      "step": 32840
+    },
+    {
+      "epoch": 0.6317884806785236,
+      "grad_norm": 416.5709962472005,
+      "learning_rate": 6.084185398141921e-06,
+      "loss": 224.7379,
+      "step": 32850
+    },
+    {
+      "epoch": 0.6319808059390041,
+      "grad_norm": 410.53352332312016,
+      "learning_rate": 6.078570361118768e-06,
+      "loss": 227.0122,
+      "step": 32860
+    },
+    {
+      "epoch": 0.6321731311994846,
+      "grad_norm": 427.8234096788836,
+      "learning_rate": 6.072956784763393e-06,
+      "loss": 224.9295,
+      "step": 32870
+    },
+    {
+      "epoch": 0.6323654564599651,
+      "grad_norm": 387.1536452080818,
+      "learning_rate": 6.067344671166757e-06,
+      "loss": 227.618,
+      "step": 32880
+    },
+    {
+      "epoch": 0.6325577817204456,
+      "grad_norm": 431.11970604539044,
+      "learning_rate": 6.061734022419281e-06,
+      "loss": 224.875,
+      "step": 32890
+    },
+    {
+      "epoch": 0.6327501069809262,
+      "grad_norm": 378.35286774475384,
+      "learning_rate": 6.056124840610839e-06,
+      "loss": 215.4979,
+      "step": 32900
+    },
+    {
+      "epoch": 0.6329424322414067,
+      "grad_norm": 420.8170335069947,
+      "learning_rate": 6.050517127830761e-06,
+      "loss": 225.6952,
+      "step": 32910
+    },
+    {
+      "epoch": 0.6331347575018872,
+      "grad_norm": 433.34233228942406,
+      "learning_rate": 6.044910886167825e-06,
+      "loss": 232.8873,
+      "step": 32920
+    },
+    {
+      "epoch": 0.6333270827623677,
+      "grad_norm": 434.79753443137196,
+      "learning_rate": 6.03930611771026e-06,
+      "loss": 222.2902,
+      "step": 32930
+    },
+    {
+      "epoch": 0.6335194080228482,
+      "grad_norm": 437.6716142386955,
+      "learning_rate": 6.033702824545755e-06,
+      "loss": 219.4517,
+      "step": 32940
+    },
+    {
+      "epoch": 0.6337117332833287,
+      "grad_norm": 415.6155024311574,
+      "learning_rate": 6.028101008761445e-06,
+      "loss": 215.7741,
+      "step": 32950
+    },
+    {
+      "epoch": 0.6339040585438093,
+      "grad_norm": 413.1861406045952,
+      "learning_rate": 6.022500672443907e-06,
+      "loss": 222.5636,
+      "step": 32960
+    },
+    {
+      "epoch": 0.6340963838042898,
+      "grad_norm": 455.1841454265467,
+      "learning_rate": 6.016901817679177e-06,
+      "loss": 229.8508,
+      "step": 32970
+    },
+    {
+      "epoch": 0.6342887090647703,
+      "grad_norm": 406.19998642742985,
+      "learning_rate": 6.011304446552741e-06,
+      "loss": 220.1547,
+      "step": 32980
+    },
+    {
+      "epoch": 0.6344810343252508,
+      "grad_norm": 408.67321509130966,
+      "learning_rate": 6.005708561149523e-06,
+      "loss": 224.5256,
+      "step": 32990
+    },
+    {
+      "epoch": 0.6346733595857313,
+      "grad_norm": 454.1508846177071,
+      "learning_rate": 6.000114163553894e-06,
+      "loss": 227.7325,
+      "step": 33000
+    },
+    {
+      "epoch": 0.6348656848462119,
+      "grad_norm": 407.94469387292355,
+      "learning_rate": 5.994521255849684e-06,
+      "loss": 229.7664,
+      "step": 33010
+    },
+    {
+      "epoch": 0.6350580101066924,
+      "grad_norm": 383.1222949482725,
+      "learning_rate": 5.988929840120151e-06,
+      "loss": 226.5475,
+      "step": 33020
+    },
+    {
+      "epoch": 0.635250335367173,
+      "grad_norm": 430.6352316115025,
+      "learning_rate": 5.983339918448008e-06,
+      "loss": 225.6059,
+      "step": 33030
+    },
+    {
+      "epoch": 0.6354426606276535,
+      "grad_norm": 439.07054562109374,
+      "learning_rate": 5.977751492915404e-06,
+      "loss": 220.4708,
+      "step": 33040
+    },
+    {
+      "epoch": 0.635634985888134,
+      "grad_norm": 404.56295238553315,
+      "learning_rate": 5.972164565603944e-06,
+      "loss": 224.4825,
+      "step": 33050
+    },
+    {
+      "epoch": 0.6358273111486146,
+      "grad_norm": 406.995529099741,
+      "learning_rate": 5.966579138594661e-06,
+      "loss": 231.6047,
+      "step": 33060
+    },
+    {
+      "epoch": 0.6360196364090951,
+      "grad_norm": 400.01546417372987,
+      "learning_rate": 5.960995213968033e-06,
+      "loss": 226.5418,
+      "step": 33070
+    },
+    {
+      "epoch": 0.6362119616695756,
+      "grad_norm": 419.6904858661481,
+      "learning_rate": 5.955412793803983e-06,
+      "loss": 220.0837,
+      "step": 33080
+    },
+    {
+      "epoch": 0.6364042869300561,
+      "grad_norm": 485.6585027234312,
+      "learning_rate": 5.949831880181869e-06,
+      "loss": 237.2516,
+      "step": 33090
+    },
+    {
+      "epoch": 0.6365966121905366,
+      "grad_norm": 394.2503466781619,
+      "learning_rate": 5.944252475180487e-06,
+      "loss": 233.9495,
+      "step": 33100
+    },
+    {
+      "epoch": 0.6367889374510172,
+      "grad_norm": 395.8595153683831,
+      "learning_rate": 5.938674580878077e-06,
+      "loss": 224.2514,
+      "step": 33110
+    },
+    {
+      "epoch": 0.6369812627114977,
+      "grad_norm": 390.84158609300476,
+      "learning_rate": 5.933098199352307e-06,
+      "loss": 228.95,
+      "step": 33120
+    },
+    {
+      "epoch": 0.6371735879719782,
+      "grad_norm": 413.4804885818183,
+      "learning_rate": 5.927523332680296e-06,
+      "loss": 228.8225,
+      "step": 33130
+    },
+    {
+      "epoch": 0.6373659132324587,
+      "grad_norm": 408.0006026856194,
+      "learning_rate": 5.921949982938583e-06,
+      "loss": 224.0146,
+      "step": 33140
+    },
+    {
+      "epoch": 0.6375582384929392,
+      "grad_norm": 437.68887809019344,
+      "learning_rate": 5.916378152203146e-06,
+      "loss": 228.0215,
+      "step": 33150
+    },
+    {
+      "epoch": 0.6377505637534198,
+      "grad_norm": 386.16005900095394,
+      "learning_rate": 5.910807842549409e-06,
+      "loss": 222.9319,
+      "step": 33160
+    },
+    {
+      "epoch": 0.6379428890139003,
+      "grad_norm": 408.9391249353531,
+      "learning_rate": 5.905239056052212e-06,
+      "loss": 221.9154,
+      "step": 33170
+    },
+    {
+      "epoch": 0.6381352142743808,
+      "grad_norm": 429.83712295028107,
+      "learning_rate": 5.8996717947858395e-06,
+      "loss": 224.3552,
+      "step": 33180
+    },
+    {
+      "epoch": 0.6383275395348613,
+      "grad_norm": 424.9036419643019,
+      "learning_rate": 5.894106060824005e-06,
+      "loss": 224.208,
+      "step": 33190
+    },
+    {
+      "epoch": 0.6385198647953418,
+      "grad_norm": 393.39569978460133,
+      "learning_rate": 5.888541856239854e-06,
+      "loss": 219.0052,
+      "step": 33200
+    },
+    {
+      "epoch": 0.6387121900558224,
+      "grad_norm": 403.41128611404923,
+      "learning_rate": 5.882979183105959e-06,
+      "loss": 224.0252,
+      "step": 33210
+    },
+    {
+      "epoch": 0.6389045153163029,
+      "grad_norm": 375.57838578237636,
+      "learning_rate": 5.8774180434943184e-06,
+      "loss": 221.4567,
+      "step": 33220
+    },
+    {
+      "epoch": 0.6390968405767835,
+      "grad_norm": 400.0832045059267,
+      "learning_rate": 5.871858439476374e-06,
+      "loss": 224.6318,
+      "step": 33230
+    },
+    {
+      "epoch": 0.639289165837264,
+      "grad_norm": 403.3555899226912,
+      "learning_rate": 5.86630037312298e-06,
+      "loss": 222.7588,
+      "step": 33240
+    },
+    {
+      "epoch": 0.6394814910977445,
+      "grad_norm": 394.8468972266905,
+      "learning_rate": 5.860743846504429e-06,
+      "loss": 227.7456,
+      "step": 33250
+    },
+    {
+      "epoch": 0.6396738163582251,
+      "grad_norm": 390.4306289888978,
+      "learning_rate": 5.8551888616904305e-06,
+      "loss": 224.8804,
+      "step": 33260
+    },
+    {
+      "epoch": 0.6398661416187056,
+      "grad_norm": 402.6392375588884,
+      "learning_rate": 5.849635420750131e-06,
+      "loss": 233.8076,
+      "step": 33270
+    },
+    {
+      "epoch": 0.6400584668791861,
+      "grad_norm": 405.4030079861563,
+      "learning_rate": 5.84408352575209e-06,
+      "loss": 224.683,
+      "step": 33280
+    },
+    {
+      "epoch": 0.6402507921396666,
+      "grad_norm": 422.18051938551014,
+      "learning_rate": 5.8385331787642956e-06,
+      "loss": 216.603,
+      "step": 33290
+    },
+    {
+      "epoch": 0.6404431174001471,
+      "grad_norm": 414.40698456003713,
+      "learning_rate": 5.8329843818541665e-06,
+      "loss": 225.3564,
+      "step": 33300
+    },
+    {
+      "epoch": 0.6406354426606277,
+      "grad_norm": 4053.9080014364586,
+      "learning_rate": 5.827437137088535e-06,
+      "loss": 217.1704,
+      "step": 33310
+    },
+    {
+      "epoch": 0.6408277679211082,
+      "grad_norm": 443.65653348116734,
+      "learning_rate": 5.8218914465336585e-06,
+      "loss": 222.8102,
+      "step": 33320
+    },
+    {
+      "epoch": 0.6410200931815887,
+      "grad_norm": 407.5381632684021,
+      "learning_rate": 5.816347312255209e-06,
+      "loss": 220.9344,
+      "step": 33330
+    },
+    {
+      "epoch": 0.6412124184420692,
+      "grad_norm": 432.8726081584538,
+      "learning_rate": 5.810804736318295e-06,
+      "loss": 218.7138,
+      "step": 33340
+    },
+    {
+      "epoch": 0.6414047437025497,
+      "grad_norm": 409.6094247013621,
+      "learning_rate": 5.805263720787426e-06,
+      "loss": 215.6701,
+      "step": 33350
+    },
+    {
+      "epoch": 0.6415970689630303,
+      "grad_norm": 421.3961216263699,
+      "learning_rate": 5.799724267726547e-06,
+      "loss": 230.2704,
+      "step": 33360
+    },
+    {
+      "epoch": 0.6417893942235108,
+      "grad_norm": 436.8606416476537,
+      "learning_rate": 5.794186379199004e-06,
+      "loss": 224.7854,
+      "step": 33370
+    },
+    {
+      "epoch": 0.6419817194839913,
+      "grad_norm": 385.19126170909993,
+      "learning_rate": 5.7886500572675774e-06,
+      "loss": 229.8884,
+      "step": 33380
+    },
+    {
+      "epoch": 0.6421740447444718,
+      "grad_norm": 410.08083947779033,
+      "learning_rate": 5.783115303994451e-06,
+      "loss": 228.7794,
+      "step": 33390
+    },
+    {
+      "epoch": 0.6423663700049523,
+      "grad_norm": 492.054819820757,
+      "learning_rate": 5.777582121441227e-06,
+      "loss": 227.8315,
+      "step": 33400
+    },
+    {
+      "epoch": 0.6425586952654329,
+      "grad_norm": 445.95883835936723,
+      "learning_rate": 5.772050511668931e-06,
+      "loss": 224.836,
+      "step": 33410
+    },
+    {
+      "epoch": 0.6427510205259134,
+      "grad_norm": 435.3855244057773,
+      "learning_rate": 5.766520476737993e-06,
+      "loss": 227.5989,
+      "step": 33420
+    },
+    {
+      "epoch": 0.642943345786394,
+      "grad_norm": 389.5184874367246,
+      "learning_rate": 5.760992018708253e-06,
+      "loss": 221.791,
+      "step": 33430
+    },
+    {
+      "epoch": 0.6431356710468745,
+      "grad_norm": 414.25805197352906,
+      "learning_rate": 5.755465139638983e-06,
+      "loss": 212.4072,
+      "step": 33440
+    },
+    {
+      "epoch": 0.643327996307355,
+      "grad_norm": 454.88337487877595,
+      "learning_rate": 5.749939841588846e-06,
+      "loss": 233.1634,
+      "step": 33450
+    },
+    {
+      "epoch": 0.6435203215678356,
+      "grad_norm": 394.29976695267965,
+      "learning_rate": 5.744416126615926e-06,
+      "loss": 222.2909,
+      "step": 33460
+    },
+    {
+      "epoch": 0.6437126468283161,
+      "grad_norm": 420.64409327834727,
+      "learning_rate": 5.738893996777713e-06,
+      "loss": 227.7246,
+      "step": 33470
+    },
+    {
+      "epoch": 0.6439049720887966,
+      "grad_norm": 419.37745056673526,
+      "learning_rate": 5.7333734541311144e-06,
+      "loss": 218.7318,
+      "step": 33480
+    },
+    {
+      "epoch": 0.6440972973492771,
+      "grad_norm": 475.780322785805,
+      "learning_rate": 5.727854500732435e-06,
+      "loss": 227.803,
+      "step": 33490
+    },
+    {
+      "epoch": 0.6442896226097576,
+      "grad_norm": 376.13930068645567,
+      "learning_rate": 5.722337138637402e-06,
+      "loss": 218.5222,
+      "step": 33500
+    },
+    {
+      "epoch": 0.6444819478702382,
+      "grad_norm": 389.52465494080644,
+      "learning_rate": 5.716821369901131e-06,
+      "loss": 221.9775,
+      "step": 33510
+    },
+    {
+      "epoch": 0.6446742731307187,
+      "grad_norm": 417.33202921995843,
+      "learning_rate": 5.711307196578166e-06,
+      "loss": 218.3189,
+      "step": 33520
+    },
+    {
+      "epoch": 0.6448665983911992,
+      "grad_norm": 384.08840911503916,
+      "learning_rate": 5.70579462072244e-06,
+      "loss": 221.0843,
+      "step": 33530
+    },
+    {
+      "epoch": 0.6450589236516797,
+      "grad_norm": 422.5164232195794,
+      "learning_rate": 5.700283644387292e-06,
+      "loss": 223.5748,
+      "step": 33540
+    },
+    {
+      "epoch": 0.6452512489121602,
+      "grad_norm": 394.31213942669797,
+      "learning_rate": 5.6947742696254794e-06,
+      "loss": 227.8395,
+      "step": 33550
+    },
+    {
+      "epoch": 0.6454435741726408,
+      "grad_norm": 388.4902837593827,
+      "learning_rate": 5.689266498489148e-06,
+      "loss": 223.9905,
+      "step": 33560
+    },
+    {
+      "epoch": 0.6456358994331213,
+      "grad_norm": 389.7127812151404,
+      "learning_rate": 5.683760333029851e-06,
+      "loss": 227.3252,
+      "step": 33570
+    },
+    {
+      "epoch": 0.6458282246936018,
+      "grad_norm": 404.24183921875766,
+      "learning_rate": 5.678255775298542e-06,
+      "loss": 219.1205,
+      "step": 33580
+    },
+    {
+      "epoch": 0.6460205499540823,
+      "grad_norm": 479.49408614188127,
+      "learning_rate": 5.672752827345584e-06,
+      "loss": 219.6218,
+      "step": 33590
+    },
+    {
+      "epoch": 0.6462128752145628,
+      "grad_norm": 415.62002185370955,
+      "learning_rate": 5.667251491220731e-06,
+      "loss": 215.895,
+      "step": 33600
+    },
+    {
+      "epoch": 0.6464052004750434,
+      "grad_norm": 434.1394585893389,
+      "learning_rate": 5.661751768973136e-06,
+      "loss": 221.4117,
+      "step": 33610
+    },
+    {
+      "epoch": 0.6465975257355239,
+      "grad_norm": 493.94627005401446,
+      "learning_rate": 5.656253662651362e-06,
+      "loss": 227.1449,
+      "step": 33620
+    },
+    {
+      "epoch": 0.6467898509960044,
+      "grad_norm": 425.0606242239906,
+      "learning_rate": 5.650757174303356e-06,
+      "loss": 216.2708,
+      "step": 33630
+    },
+    {
+      "epoch": 0.646982176256485,
+      "grad_norm": 443.4465030931299,
+      "learning_rate": 5.645262305976476e-06,
+      "loss": 224.5514,
+      "step": 33640
+    },
+    {
+      "epoch": 0.6471745015169655,
+      "grad_norm": 408.6957233228005,
+      "learning_rate": 5.639769059717462e-06,
+      "loss": 219.8099,
+      "step": 33650
+    },
+    {
+      "epoch": 0.6473668267774461,
+      "grad_norm": 486.7217919823491,
+      "learning_rate": 5.634277437572466e-06,
+      "loss": 222.9049,
+      "step": 33660
+    },
+    {
+      "epoch": 0.6475591520379266,
+      "grad_norm": 450.18604970929465,
+      "learning_rate": 5.6287874415870225e-06,
+      "loss": 223.3881,
+      "step": 33670
+    },
+    {
+      "epoch": 0.6477514772984071,
+      "grad_norm": 423.56262269041434,
+      "learning_rate": 5.623299073806063e-06,
+      "loss": 219.3294,
+      "step": 33680
+    },
+    {
+      "epoch": 0.6479438025588876,
+      "grad_norm": 395.5691473922273,
+      "learning_rate": 5.617812336273912e-06,
+      "loss": 223.5912,
+      "step": 33690
+    },
+    {
+      "epoch": 0.6481361278193681,
+      "grad_norm": 421.742192253228,
+      "learning_rate": 5.612327231034296e-06,
+      "loss": 218.6008,
+      "step": 33700
+    },
+    {
+      "epoch": 0.6483284530798487,
+      "grad_norm": 390.25443910740637,
+      "learning_rate": 5.606843760130321e-06,
+      "loss": 219.9142,
+      "step": 33710
+    },
+    {
+      "epoch": 0.6485207783403292,
+      "grad_norm": 430.8700159313394,
+      "learning_rate": 5.601361925604485e-06,
+      "loss": 221.7454,
+      "step": 33720
+    },
+    {
+      "epoch": 0.6487131036008097,
+      "grad_norm": 436.781275015363,
+      "learning_rate": 5.595881729498691e-06,
+      "loss": 217.4805,
+      "step": 33730
+    },
+    {
+      "epoch": 0.6489054288612902,
+      "grad_norm": 408.3330101282725,
+      "learning_rate": 5.590403173854215e-06,
+      "loss": 224.3699,
+      "step": 33740
+    },
+    {
+      "epoch": 0.6490977541217707,
+      "grad_norm": 388.9927700273765,
+      "learning_rate": 5.584926260711732e-06,
+      "loss": 224.5737,
+      "step": 33750
+    },
+    {
+      "epoch": 0.6492900793822513,
+      "grad_norm": 385.55469692825466,
+      "learning_rate": 5.579450992111294e-06,
+      "loss": 225.4747,
+      "step": 33760
+    },
+    {
+      "epoch": 0.6494824046427318,
+      "grad_norm": 375.76089468010906,
+      "learning_rate": 5.573977370092358e-06,
+      "loss": 216.5367,
+      "step": 33770
+    },
+    {
+      "epoch": 0.6496747299032123,
+      "grad_norm": 421.7683933429201,
+      "learning_rate": 5.568505396693749e-06,
+      "loss": 224.6679,
+      "step": 33780
+    },
+    {
+      "epoch": 0.6498670551636928,
+      "grad_norm": 424.2931956692947,
+      "learning_rate": 5.563035073953691e-06,
+      "loss": 228.2854,
+      "step": 33790
+    },
+    {
+      "epoch": 0.6500593804241733,
+      "grad_norm": 428.7432092626859,
+      "learning_rate": 5.557566403909794e-06,
+      "loss": 216.1636,
+      "step": 33800
+    },
+    {
+      "epoch": 0.6502517056846538,
+      "grad_norm": 410.74148038217317,
+      "learning_rate": 5.552099388599042e-06,
+      "loss": 224.7191,
+      "step": 33810
+    },
+    {
+      "epoch": 0.6504440309451344,
+      "grad_norm": 434.5821354657664,
+      "learning_rate": 5.5466340300578095e-06,
+      "loss": 217.314,
+      "step": 33820
+    },
+    {
+      "epoch": 0.6506363562056149,
+      "grad_norm": 409.0372763908047,
+      "learning_rate": 5.541170330321845e-06,
+      "loss": 220.3839,
+      "step": 33830
+    },
+    {
+      "epoch": 0.6508286814660955,
+      "grad_norm": 366.64841615178324,
+      "learning_rate": 5.535708291426297e-06,
+      "loss": 222.9088,
+      "step": 33840
+    },
+    {
+      "epoch": 0.651021006726576,
+      "grad_norm": 397.5235155205105,
+      "learning_rate": 5.53024791540568e-06,
+      "loss": 218.9031,
+      "step": 33850
+    },
+    {
+      "epoch": 0.6512133319870566,
+      "grad_norm": 386.4828244785691,
+      "learning_rate": 5.5247892042938944e-06,
+      "loss": 216.198,
+      "step": 33860
+    },
+    {
+      "epoch": 0.6514056572475371,
+      "grad_norm": 404.84116459039456,
+      "learning_rate": 5.519332160124215e-06,
+      "loss": 221.2257,
+      "step": 33870
+    },
+    {
+      "epoch": 0.6515979825080176,
+      "grad_norm": 430.55704155310667,
+      "learning_rate": 5.513876784929311e-06,
+      "loss": 216.755,
+      "step": 33880
+    },
+    {
+      "epoch": 0.6517903077684981,
+      "grad_norm": 412.7398547381667,
+      "learning_rate": 5.5084230807412135e-06,
+      "loss": 215.7313,
+      "step": 33890
+    },
+    {
+      "epoch": 0.6519826330289786,
+      "grad_norm": 450.4316455206422,
+      "learning_rate": 5.502971049591332e-06,
+      "loss": 227.2195,
+      "step": 33900
+    },
+    {
+      "epoch": 0.6521749582894592,
+      "grad_norm": 388.22484360431287,
+      "learning_rate": 5.497520693510469e-06,
+      "loss": 221.7378,
+      "step": 33910
+    },
+    {
+      "epoch": 0.6523672835499397,
+      "grad_norm": 380.15952453851884,
+      "learning_rate": 5.492072014528783e-06,
+      "loss": 224.4312,
+      "step": 33920
+    },
+    {
+      "epoch": 0.6525596088104202,
+      "grad_norm": 416.64682782605706,
+      "learning_rate": 5.4866250146758235e-06,
+      "loss": 221.9234,
+      "step": 33930
+    },
+    {
+      "epoch": 0.6527519340709007,
+      "grad_norm": 412.32675595857177,
+      "learning_rate": 5.481179695980503e-06,
+      "loss": 222.2215,
+      "step": 33940
+    },
+    {
+      "epoch": 0.6529442593313812,
+      "grad_norm": 370.1401697001076,
+      "learning_rate": 5.475736060471117e-06,
+      "loss": 226.2021,
+      "step": 33950
+    },
+    {
+      "epoch": 0.6531365845918617,
+      "grad_norm": 406.84148359695627,
+      "learning_rate": 5.470294110175329e-06,
+      "loss": 222.4338,
+      "step": 33960
+    },
+    {
+      "epoch": 0.6533289098523423,
+      "grad_norm": 380.2284682536495,
+      "learning_rate": 5.464853847120169e-06,
+      "loss": 216.6895,
+      "step": 33970
+    },
+    {
+      "epoch": 0.6535212351128228,
+      "grad_norm": 408.10588551318745,
+      "learning_rate": 5.459415273332056e-06,
+      "loss": 222.4578,
+      "step": 33980
+    },
+    {
+      "epoch": 0.6537135603733033,
+      "grad_norm": 422.16612821356006,
+      "learning_rate": 5.453978390836763e-06,
+      "loss": 215.5359,
+      "step": 33990
+    },
+    {
+      "epoch": 0.6539058856337838,
+      "grad_norm": 383.1771770680644,
+      "learning_rate": 5.44854320165944e-06,
+      "loss": 227.0804,
+      "step": 34000
+    },
+    {
+      "epoch": 0.6540982108942643,
+      "grad_norm": 406.17556187166747,
+      "learning_rate": 5.443109707824599e-06,
+      "loss": 227.7723,
+      "step": 34010
+    },
+    {
+      "epoch": 0.6542905361547449,
+      "grad_norm": 397.91469372714204,
+      "learning_rate": 5.437677911356137e-06,
+      "loss": 220.6621,
+      "step": 34020
+    },
+    {
+      "epoch": 0.6544828614152254,
+      "grad_norm": 433.8150369675571,
+      "learning_rate": 5.432247814277305e-06,
+      "loss": 215.9695,
+      "step": 34030
+    },
+    {
+      "epoch": 0.6546751866757059,
+      "grad_norm": 445.71808778097335,
+      "learning_rate": 5.426819418610718e-06,
+      "loss": 226.9019,
+      "step": 34040
+    },
+    {
+      "epoch": 0.6548675119361865,
+      "grad_norm": 427.87873663873904,
+      "learning_rate": 5.4213927263783725e-06,
+      "loss": 215.9779,
+      "step": 34050
+    },
+    {
+      "epoch": 0.655059837196667,
+      "grad_norm": 420.4167125671266,
+      "learning_rate": 5.415967739601616e-06,
+      "loss": 214.3531,
+      "step": 34060
+    },
+    {
+      "epoch": 0.6552521624571476,
+      "grad_norm": 408.00528417635167,
+      "learning_rate": 5.41054446030117e-06,
+      "loss": 220.8566,
+      "step": 34070
+    },
+    {
+      "epoch": 0.6554444877176281,
+      "grad_norm": 460.8509257662566,
+      "learning_rate": 5.405122890497114e-06,
+      "loss": 229.1481,
+      "step": 34080
+    },
+    {
+      "epoch": 0.6556368129781086,
+      "grad_norm": 389.10711964885326,
+      "learning_rate": 5.399703032208896e-06,
+      "loss": 221.9489,
+      "step": 34090
+    },
+    {
+      "epoch": 0.6558291382385891,
+      "grad_norm": 374.6333246079594,
+      "learning_rate": 5.3942848874553235e-06,
+      "loss": 219.252,
+      "step": 34100
+    },
+    {
+      "epoch": 0.6560214634990696,
+      "grad_norm": 436.0036236312717,
+      "learning_rate": 5.388868458254565e-06,
+      "loss": 213.0718,
+      "step": 34110
+    },
+    {
+      "epoch": 0.6562137887595502,
+      "grad_norm": 429.8354606863401,
+      "learning_rate": 5.3834537466241455e-06,
+      "loss": 218.5837,
+      "step": 34120
+    },
+    {
+      "epoch": 0.6564061140200307,
+      "grad_norm": 424.2748165593473,
+      "learning_rate": 5.378040754580964e-06,
+      "loss": 213.8902,
+      "step": 34130
+    },
+    {
+      "epoch": 0.6565984392805112,
+      "grad_norm": 406.9017677770473,
+      "learning_rate": 5.37262948414127e-06,
+      "loss": 222.2124,
+      "step": 34140
+    },
+    {
+      "epoch": 0.6567907645409917,
+      "grad_norm": 434.01457954406874,
+      "learning_rate": 5.367219937320663e-06,
+      "loss": 215.3664,
+      "step": 34150
+    },
+    {
+      "epoch": 0.6569830898014722,
+      "grad_norm": 412.75081737851093,
+      "learning_rate": 5.361812116134122e-06,
+      "loss": 219.1794,
+      "step": 34160
+    },
+    {
+      "epoch": 0.6571754150619528,
+      "grad_norm": 415.4118554333351,
+      "learning_rate": 5.356406022595963e-06,
+      "loss": 224.351,
+      "step": 34170
+    },
+    {
+      "epoch": 0.6573677403224333,
+      "grad_norm": 429.41077008673665,
+      "learning_rate": 5.351001658719872e-06,
+      "loss": 222.8747,
+      "step": 34180
+    },
+    {
+      "epoch": 0.6575600655829138,
+      "grad_norm": 369.8348048500022,
+      "learning_rate": 5.345599026518877e-06,
+      "loss": 221.3077,
+      "step": 34190
+    },
+    {
+      "epoch": 0.6577523908433943,
+      "grad_norm": 426.6929621375697,
+      "learning_rate": 5.3401981280053745e-06,
+      "loss": 228.286,
+      "step": 34200
+    },
+    {
+      "epoch": 0.6579447161038748,
+      "grad_norm": 385.2522102795723,
+      "learning_rate": 5.334798965191115e-06,
+      "loss": 220.0016,
+      "step": 34210
+    },
+    {
+      "epoch": 0.6581370413643554,
+      "grad_norm": 414.4667890306486,
+      "learning_rate": 5.329401540087188e-06,
+      "loss": 216.608,
+      "step": 34220
+    },
+    {
+      "epoch": 0.6583293666248359,
+      "grad_norm": 386.13409626904894,
+      "learning_rate": 5.3240058547040525e-06,
+      "loss": 213.6617,
+      "step": 34230
+    },
+    {
+      "epoch": 0.6585216918853164,
+      "grad_norm": 395.2595710193424,
+      "learning_rate": 5.318611911051512e-06,
+      "loss": 217.3819,
+      "step": 34240
+    },
+    {
+      "epoch": 0.658714017145797,
+      "grad_norm": 382.1077538938697,
+      "learning_rate": 5.313219711138717e-06,
+      "loss": 220.277,
+      "step": 34250
+    },
+    {
+      "epoch": 0.6589063424062775,
+      "grad_norm": 410.2347197492905,
+      "learning_rate": 5.307829256974173e-06,
+      "loss": 215.7177,
+      "step": 34260
+    },
+    {
+      "epoch": 0.6590986676667581,
+      "grad_norm": 408.4589098291733,
+      "learning_rate": 5.302440550565739e-06,
+      "loss": 210.0758,
+      "step": 34270
+    },
+    {
+      "epoch": 0.6592909929272386,
+      "grad_norm": 389.65221267945395,
+      "learning_rate": 5.297053593920618e-06,
+      "loss": 212.2247,
+      "step": 34280
+    },
+    {
+      "epoch": 0.6594833181877191,
+      "grad_norm": 384.61484155685713,
+      "learning_rate": 5.291668389045362e-06,
+      "loss": 218.25,
+      "step": 34290
+    },
+    {
+      "epoch": 0.6596756434481996,
+      "grad_norm": 409.0366665702512,
+      "learning_rate": 5.286284937945866e-06,
+      "loss": 220.1443,
+      "step": 34300
+    },
+    {
+      "epoch": 0.6598679687086801,
+      "grad_norm": 407.88290789083067,
+      "learning_rate": 5.280903242627384e-06,
+      "loss": 223.4864,
+      "step": 34310
+    },
+    {
+      "epoch": 0.6600602939691607,
+      "grad_norm": 628.1510297138167,
+      "learning_rate": 5.2755233050945076e-06,
+      "loss": 221.4672,
+      "step": 34320
+    },
+    {
+      "epoch": 0.6602526192296412,
+      "grad_norm": 427.69772756075423,
+      "learning_rate": 5.270145127351168e-06,
+      "loss": 226.2193,
+      "step": 34330
+    },
+    {
+      "epoch": 0.6604449444901217,
+      "grad_norm": 434.90129355321193,
+      "learning_rate": 5.264768711400656e-06,
+      "loss": 219.5553,
+      "step": 34340
+    },
+    {
+      "epoch": 0.6606372697506022,
+      "grad_norm": 399.23687171864344,
+      "learning_rate": 5.25939405924559e-06,
+      "loss": 221.3278,
+      "step": 34350
+    },
+    {
+      "epoch": 0.6608295950110827,
+      "grad_norm": 411.3223621323265,
+      "learning_rate": 5.254021172887947e-06,
+      "loss": 226.4097,
+      "step": 34360
+    },
+    {
+      "epoch": 0.6610219202715633,
+      "grad_norm": 400.7998974581744,
+      "learning_rate": 5.248650054329032e-06,
+      "loss": 227.0978,
+      "step": 34370
+    },
+    {
+      "epoch": 0.6612142455320438,
+      "grad_norm": 399.3956126061256,
+      "learning_rate": 5.2432807055695035e-06,
+      "loss": 226.1593,
+      "step": 34380
+    },
+    {
+      "epoch": 0.6614065707925243,
+      "grad_norm": 431.821138441697,
+      "learning_rate": 5.237913128609352e-06,
+      "loss": 221.4863,
+      "step": 34390
+    },
+    {
+      "epoch": 0.6615988960530048,
+      "grad_norm": 387.98993645508483,
+      "learning_rate": 5.232547325447908e-06,
+      "loss": 228.9696,
+      "step": 34400
+    },
+    {
+      "epoch": 0.6617912213134853,
+      "grad_norm": 455.2606264976643,
+      "learning_rate": 5.227183298083854e-06,
+      "loss": 219.1108,
+      "step": 34410
+    },
+    {
+      "epoch": 0.6619835465739659,
+      "grad_norm": 434.27210616795634,
+      "learning_rate": 5.221821048515193e-06,
+      "loss": 220.9553,
+      "step": 34420
+    },
+    {
+      "epoch": 0.6621758718344464,
+      "grad_norm": 405.2208327967891,
+      "learning_rate": 5.216460578739278e-06,
+      "loss": 217.0133,
+      "step": 34430
+    },
+    {
+      "epoch": 0.6623681970949269,
+      "grad_norm": 417.8550831398497,
+      "learning_rate": 5.211101890752792e-06,
+      "loss": 216.4514,
+      "step": 34440
+    },
+    {
+      "epoch": 0.6625605223554074,
+      "grad_norm": 406.3149397970147,
+      "learning_rate": 5.205744986551763e-06,
+      "loss": 217.8611,
+      "step": 34450
+    },
+    {
+      "epoch": 0.662752847615888,
+      "grad_norm": 412.35480112393446,
+      "learning_rate": 5.200389868131547e-06,
+      "loss": 227.1226,
+      "step": 34460
+    },
+    {
+      "epoch": 0.6629451728763686,
+      "grad_norm": 390.8784820652284,
+      "learning_rate": 5.195036537486833e-06,
+      "loss": 225.5386,
+      "step": 34470
+    },
+    {
+      "epoch": 0.6631374981368491,
+      "grad_norm": 410.0052035346963,
+      "learning_rate": 5.189684996611657e-06,
+      "loss": 225.1919,
+      "step": 34480
+    },
+    {
+      "epoch": 0.6633298233973296,
+      "grad_norm": 371.4325178548429,
+      "learning_rate": 5.18433524749937e-06,
+      "loss": 216.1162,
+      "step": 34490
+    },
+    {
+      "epoch": 0.6635221486578101,
+      "grad_norm": 474.5518626843271,
+      "learning_rate": 5.178987292142674e-06,
+      "loss": 231.3805,
+      "step": 34500
+    },
+    {
+      "epoch": 0.6637144739182906,
+      "grad_norm": 389.5231699921982,
+      "learning_rate": 5.173641132533586e-06,
+      "loss": 219.1944,
+      "step": 34510
+    },
+    {
+      "epoch": 0.6639067991787712,
+      "grad_norm": 390.4161380012965,
+      "learning_rate": 5.16829677066347e-06,
+      "loss": 226.4411,
+      "step": 34520
+    },
+    {
+      "epoch": 0.6640991244392517,
+      "grad_norm": 426.3713495748812,
+      "learning_rate": 5.16295420852301e-06,
+      "loss": 218.0286,
+      "step": 34530
+    },
+    {
+      "epoch": 0.6642914496997322,
+      "grad_norm": 392.93237883825964,
+      "learning_rate": 5.157613448102219e-06,
+      "loss": 222.1857,
+      "step": 34540
+    },
+    {
+      "epoch": 0.6644837749602127,
+      "grad_norm": 367.1567326154693,
+      "learning_rate": 5.152274491390441e-06,
+      "loss": 220.2064,
+      "step": 34550
+    },
+    {
+      "epoch": 0.6646761002206932,
+      "grad_norm": 414.91270187754816,
+      "learning_rate": 5.1469373403763555e-06,
+      "loss": 222.7994,
+      "step": 34560
+    },
+    {
+      "epoch": 0.6648684254811738,
+      "grad_norm": 391.92457518505773,
+      "learning_rate": 5.14160199704796e-06,
+      "loss": 218.081,
+      "step": 34570
+    },
+    {
+      "epoch": 0.6650607507416543,
+      "grad_norm": 440.6149088282293,
+      "learning_rate": 5.136268463392578e-06,
+      "loss": 217.0255,
+      "step": 34580
+    },
+    {
+      "epoch": 0.6652530760021348,
+      "grad_norm": 428.32323036007244,
+      "learning_rate": 5.13093674139687e-06,
+      "loss": 219.5003,
+      "step": 34590
+    },
+    {
+      "epoch": 0.6654454012626153,
+      "grad_norm": 386.524370795735,
+      "learning_rate": 5.12560683304681e-06,
+      "loss": 215.506,
+      "step": 34600
+    },
+    {
+      "epoch": 0.6656377265230958,
+      "grad_norm": 389.87164413735854,
+      "learning_rate": 5.120278740327702e-06,
+      "loss": 217.1482,
+      "step": 34610
+    },
+    {
+      "epoch": 0.6658300517835763,
+      "grad_norm": 391.91386671069597,
+      "learning_rate": 5.114952465224168e-06,
+      "loss": 219.4571,
+      "step": 34620
+    },
+    {
+      "epoch": 0.6660223770440569,
+      "grad_norm": 419.52780364454793,
+      "learning_rate": 5.109628009720162e-06,
+      "loss": 226.8344,
+      "step": 34630
+    },
+    {
+      "epoch": 0.6662147023045374,
+      "grad_norm": 456.95884508220956,
+      "learning_rate": 5.104305375798958e-06,
+      "loss": 221.8309,
+      "step": 34640
+    },
+    {
+      "epoch": 0.6664070275650179,
+      "grad_norm": 418.5649634874358,
+      "learning_rate": 5.098984565443148e-06,
+      "loss": 224.2618,
+      "step": 34650
+    },
+    {
+      "epoch": 0.6665993528254985,
+      "grad_norm": 435.9904573132652,
+      "learning_rate": 5.093665580634639e-06,
+      "loss": 215.8238,
+      "step": 34660
+    },
+    {
+      "epoch": 0.6667916780859791,
+      "grad_norm": 410.1478997836662,
+      "learning_rate": 5.088348423354674e-06,
+      "loss": 215.894,
+      "step": 34670
+    },
+    {
+      "epoch": 0.6669840033464596,
+      "grad_norm": 399.34382371602294,
+      "learning_rate": 5.0830330955838045e-06,
+      "loss": 223.2651,
+      "step": 34680
+    },
+    {
+      "epoch": 0.6671763286069401,
+      "grad_norm": 408.08863746996576,
+      "learning_rate": 5.077719599301895e-06,
+      "loss": 220.5595,
+      "step": 34690
+    },
+    {
+      "epoch": 0.6673686538674206,
+      "grad_norm": 475.0752896157902,
+      "learning_rate": 5.072407936488145e-06,
+      "loss": 233.0844,
+      "step": 34700
+    },
+    {
+      "epoch": 0.6675609791279011,
+      "grad_norm": 383.36865412071796,
+      "learning_rate": 5.067098109121058e-06,
+      "loss": 220.24,
+      "step": 34710
+    },
+    {
+      "epoch": 0.6677533043883817,
+      "grad_norm": 397.5417620210139,
+      "learning_rate": 5.0617901191784536e-06,
+      "loss": 220.2389,
+      "step": 34720
+    },
+    {
+      "epoch": 0.6679456296488622,
+      "grad_norm": 400.3219953075971,
+      "learning_rate": 5.05648396863747e-06,
+      "loss": 219.4802,
+      "step": 34730
+    },
+    {
+      "epoch": 0.6681379549093427,
+      "grad_norm": 427.2479469307242,
+      "learning_rate": 5.051179659474568e-06,
+      "loss": 222.0553,
+      "step": 34740
+    },
+    {
+      "epoch": 0.6683302801698232,
+      "grad_norm": 405.8582281354667,
+      "learning_rate": 5.045877193665508e-06,
+      "loss": 221.493,
+      "step": 34750
+    },
+    {
+      "epoch": 0.6685226054303037,
+      "grad_norm": 434.3289414696171,
+      "learning_rate": 5.040576573185372e-06,
+      "loss": 219.562,
+      "step": 34760
+    },
+    {
+      "epoch": 0.6687149306907842,
+      "grad_norm": 428.99912820381775,
+      "learning_rate": 5.0352778000085565e-06,
+      "loss": 217.7414,
+      "step": 34770
+    },
+    {
+      "epoch": 0.6689072559512648,
+      "grad_norm": 379.5355414447412,
+      "learning_rate": 5.029980876108762e-06,
+      "loss": 221.4646,
+      "step": 34780
+    },
+    {
+      "epoch": 0.6690995812117453,
+      "grad_norm": 371.29290585646544,
+      "learning_rate": 5.024685803459013e-06,
+      "loss": 214.9381,
+      "step": 34790
+    },
+    {
+      "epoch": 0.6692919064722258,
+      "grad_norm": 370.2864344194036,
+      "learning_rate": 5.019392584031628e-06,
+      "loss": 213.0196,
+      "step": 34800
+    },
+    {
+      "epoch": 0.6694842317327063,
+      "grad_norm": 432.81366640852184,
+      "learning_rate": 5.0141012197982534e-06,
+      "loss": 227.414,
+      "step": 34810
+    },
+    {
+      "epoch": 0.6696765569931868,
+      "grad_norm": 421.46565519766983,
+      "learning_rate": 5.0088117127298285e-06,
+      "loss": 217.3553,
+      "step": 34820
+    },
+    {
+      "epoch": 0.6698688822536674,
+      "grad_norm": 384.9203523314455,
+      "learning_rate": 5.003524064796608e-06,
+      "loss": 220.0766,
+      "step": 34830
+    },
+    {
+      "epoch": 0.6700612075141479,
+      "grad_norm": 401.2230869106227,
+      "learning_rate": 4.998238277968149e-06,
+      "loss": 220.3158,
+      "step": 34840
+    },
+    {
+      "epoch": 0.6702535327746284,
+      "grad_norm": 399.2223901244871,
+      "learning_rate": 4.992954354213331e-06,
+      "loss": 215.849,
+      "step": 34850
+    },
+    {
+      "epoch": 0.6704458580351089,
+      "grad_norm": 456.7888903253584,
+      "learning_rate": 4.98767229550032e-06,
+      "loss": 224.2743,
+      "step": 34860
+    },
+    {
+      "epoch": 0.6706381832955896,
+      "grad_norm": 415.7765312303501,
+      "learning_rate": 4.982392103796595e-06,
+      "loss": 216.1302,
+      "step": 34870
+    },
+    {
+      "epoch": 0.6708305085560701,
+      "grad_norm": 422.25682451052126,
+      "learning_rate": 4.977113781068945e-06,
+      "loss": 224.4988,
+      "step": 34880
+    },
+    {
+      "epoch": 0.6710228338165506,
+      "grad_norm": 387.5028416458855,
+      "learning_rate": 4.971837329283458e-06,
+      "loss": 222.0692,
+      "step": 34890
+    },
+    {
+      "epoch": 0.6712151590770311,
+      "grad_norm": 377.7802855481398,
+      "learning_rate": 4.966562750405517e-06,
+      "loss": 227.673,
+      "step": 34900
+    },
+    {
+      "epoch": 0.6714074843375116,
+      "grad_norm": 424.1443555863365,
+      "learning_rate": 4.9612900463998274e-06,
+      "loss": 218.8016,
+      "step": 34910
+    },
+    {
+      "epoch": 0.6715998095979921,
+      "grad_norm": 445.07863344730316,
+      "learning_rate": 4.9560192192303735e-06,
+      "loss": 227.9261,
+      "step": 34920
+    },
+    {
+      "epoch": 0.6717921348584727,
+      "grad_norm": 419.1258770683227,
+      "learning_rate": 4.95075027086046e-06,
+      "loss": 216.7835,
+      "step": 34930
+    },
+    {
+      "epoch": 0.6719844601189532,
+      "grad_norm": 425.4784161543398,
+      "learning_rate": 4.9454832032526755e-06,
+      "loss": 225.8217,
+      "step": 34940
+    },
+    {
+      "epoch": 0.6721767853794337,
+      "grad_norm": 410.22499642384236,
+      "learning_rate": 4.940218018368924e-06,
+      "loss": 220.4243,
+      "step": 34950
+    },
+    {
+      "epoch": 0.6723691106399142,
+      "grad_norm": 381.72086856082564,
+      "learning_rate": 4.934954718170396e-06,
+      "loss": 218.1714,
+      "step": 34960
+    },
+    {
+      "epoch": 0.6725614359003947,
+      "grad_norm": 416.010977343979,
+      "learning_rate": 4.9296933046175834e-06,
+      "loss": 214.2865,
+      "step": 34970
+    },
+    {
+      "epoch": 0.6727537611608753,
+      "grad_norm": 402.40914639630927,
+      "learning_rate": 4.924433779670271e-06,
+      "loss": 214.6974,
+      "step": 34980
+    },
+    {
+      "epoch": 0.6729460864213558,
+      "grad_norm": 391.17333376277037,
+      "learning_rate": 4.9191761452875554e-06,
+      "loss": 227.203,
+      "step": 34990
+    },
+    {
+      "epoch": 0.6731384116818363,
+      "grad_norm": 404.8483203612921,
+      "learning_rate": 4.913920403427812e-06,
+      "loss": 217.7973,
+      "step": 35000
+    },
+    {
+      "epoch": 0.6733307369423168,
+      "grad_norm": 445.71152017928637,
+      "learning_rate": 4.908666556048719e-06,
+      "loss": 215.3907,
+      "step": 35010
+    },
+    {
+      "epoch": 0.6735230622027973,
+      "grad_norm": 406.2545068423606,
+      "learning_rate": 4.903414605107244e-06,
+      "loss": 217.489,
+      "step": 35020
+    },
+    {
+      "epoch": 0.6737153874632779,
+      "grad_norm": 390.5927867242238,
+      "learning_rate": 4.89816455255966e-06,
+      "loss": 219.0073,
+      "step": 35030
+    },
+    {
+      "epoch": 0.6739077127237584,
+      "grad_norm": 407.93375759457336,
+      "learning_rate": 4.892916400361516e-06,
+      "loss": 220.319,
+      "step": 35040
+    },
+    {
+      "epoch": 0.6741000379842389,
+      "grad_norm": 397.9938323002778,
+      "learning_rate": 4.887670150467671e-06,
+      "loss": 218.9028,
+      "step": 35050
+    },
+    {
+      "epoch": 0.6742923632447194,
+      "grad_norm": 378.5453163839792,
+      "learning_rate": 4.882425804832258e-06,
+      "loss": 218.4058,
+      "step": 35060
+    },
+    {
+      "epoch": 0.6744846885052,
+      "grad_norm": 1701.0645554894036,
+      "learning_rate": 4.8771833654087165e-06,
+      "loss": 219.2215,
+      "step": 35070
+    },
+    {
+      "epoch": 0.6746770137656806,
+      "grad_norm": 421.12677006161465,
+      "learning_rate": 4.8719428341497665e-06,
+      "loss": 219.8677,
+      "step": 35080
+    },
+    {
+      "epoch": 0.6748693390261611,
+      "grad_norm": 403.05812729571164,
+      "learning_rate": 4.866704213007413e-06,
+      "loss": 222.7858,
+      "step": 35090
+    },
+    {
+      "epoch": 0.6750616642866416,
+      "grad_norm": 400.47874375477545,
+      "learning_rate": 4.861467503932965e-06,
+      "loss": 220.94,
+      "step": 35100
+    },
+    {
+      "epoch": 0.6752539895471221,
+      "grad_norm": 371.1653158650625,
+      "learning_rate": 4.856232708877007e-06,
+      "loss": 213.8064,
+      "step": 35110
+    },
+    {
+      "epoch": 0.6754463148076026,
+      "grad_norm": 449.53977222491653,
+      "learning_rate": 4.8509998297894075e-06,
+      "loss": 225.4592,
+      "step": 35120
+    },
+    {
+      "epoch": 0.6756386400680832,
+      "grad_norm": 425.94483782608006,
+      "learning_rate": 4.845768868619338e-06,
+      "loss": 219.3791,
+      "step": 35130
+    },
+    {
+      "epoch": 0.6758309653285637,
+      "grad_norm": 426.54675977469947,
+      "learning_rate": 4.8405398273152405e-06,
+      "loss": 224.3054,
+      "step": 35140
+    },
+    {
+      "epoch": 0.6760232905890442,
+      "grad_norm": 410.1514685033484,
+      "learning_rate": 4.8353127078248455e-06,
+      "loss": 228.5359,
+      "step": 35150
+    },
+    {
+      "epoch": 0.6762156158495247,
+      "grad_norm": 404.99268776020017,
+      "learning_rate": 4.830087512095164e-06,
+      "loss": 215.9931,
+      "step": 35160
+    },
+    {
+      "epoch": 0.6764079411100052,
+      "grad_norm": 396.35087915909986,
+      "learning_rate": 4.824864242072506e-06,
+      "loss": 216.081,
+      "step": 35170
+    },
+    {
+      "epoch": 0.6766002663704858,
+      "grad_norm": 429.4034664446162,
+      "learning_rate": 4.819642899702447e-06,
+      "loss": 219.0898,
+      "step": 35180
+    },
+    {
+      "epoch": 0.6767925916309663,
+      "grad_norm": 413.4915093918844,
+      "learning_rate": 4.814423486929846e-06,
+      "loss": 218.1461,
+      "step": 35190
+    },
+    {
+      "epoch": 0.6769849168914468,
+      "grad_norm": 431.1310407809555,
+      "learning_rate": 4.809206005698856e-06,
+      "loss": 216.2095,
+      "step": 35200
+    },
+    {
+      "epoch": 0.6771772421519273,
+      "grad_norm": 391.036190075778,
+      "learning_rate": 4.803990457952903e-06,
+      "loss": 230.6294,
+      "step": 35210
+    },
+    {
+      "epoch": 0.6773695674124078,
+      "grad_norm": 437.8428504398393,
+      "learning_rate": 4.798776845634688e-06,
+      "loss": 215.9749,
+      "step": 35220
+    },
+    {
+      "epoch": 0.6775618926728884,
+      "grad_norm": 474.30686517047116,
+      "learning_rate": 4.793565170686196e-06,
+      "loss": 221.5434,
+      "step": 35230
+    },
+    {
+      "epoch": 0.6777542179333689,
+      "grad_norm": 391.9764584980503,
+      "learning_rate": 4.7883554350486925e-06,
+      "loss": 225.1253,
+      "step": 35240
+    },
+    {
+      "epoch": 0.6779465431938494,
+      "grad_norm": 390.4915306979907,
+      "learning_rate": 4.783147640662716e-06,
+      "loss": 214.2675,
+      "step": 35250
+    },
+    {
+      "epoch": 0.6781388684543299,
+      "grad_norm": 404.8724496864521,
+      "learning_rate": 4.777941789468086e-06,
+      "loss": 217.5677,
+      "step": 35260
+    },
+    {
+      "epoch": 0.6783311937148104,
+      "grad_norm": 410.4147782816225,
+      "learning_rate": 4.772737883403889e-06,
+      "loss": 225.9869,
+      "step": 35270
+    },
+    {
+      "epoch": 0.6785235189752911,
+      "grad_norm": 417.6396121262922,
+      "learning_rate": 4.767535924408504e-06,
+      "loss": 222.0417,
+      "step": 35280
+    },
+    {
+      "epoch": 0.6787158442357716,
+      "grad_norm": 398.7773521099608,
+      "learning_rate": 4.76233591441957e-06,
+      "loss": 220.6352,
+      "step": 35290
+    },
+    {
+      "epoch": 0.6789081694962521,
+      "grad_norm": 444.68245930806336,
+      "learning_rate": 4.757137855374001e-06,
+      "loss": 216.2313,
+      "step": 35300
+    },
+    {
+      "epoch": 0.6791004947567326,
+      "grad_norm": 417.2630411286579,
+      "learning_rate": 4.751941749207996e-06,
+      "loss": 222.229,
+      "step": 35310
+    },
+    {
+      "epoch": 0.6792928200172131,
+      "grad_norm": 380.1821619331443,
+      "learning_rate": 4.746747597857014e-06,
+      "loss": 217.6884,
+      "step": 35320
+    },
+    {
+      "epoch": 0.6794851452776937,
+      "grad_norm": 411.9494189243853,
+      "learning_rate": 4.74155540325579e-06,
+      "loss": 220.5652,
+      "step": 35330
+    },
+    {
+      "epoch": 0.6796774705381742,
+      "grad_norm": 446.72028111730015,
+      "learning_rate": 4.736365167338333e-06,
+      "loss": 225.6941,
+      "step": 35340
+    },
+    {
+      "epoch": 0.6798697957986547,
+      "grad_norm": 427.20146684548655,
+      "learning_rate": 4.731176892037918e-06,
+      "loss": 215.628,
+      "step": 35350
+    },
+    {
+      "epoch": 0.6800621210591352,
+      "grad_norm": 400.4752864673295,
+      "learning_rate": 4.725990579287097e-06,
+      "loss": 213.7563,
+      "step": 35360
+    },
+    {
+      "epoch": 0.6802544463196157,
+      "grad_norm": 480.5398254124123,
+      "learning_rate": 4.720806231017676e-06,
+      "loss": 221.9207,
+      "step": 35370
+    },
+    {
+      "epoch": 0.6804467715800963,
+      "grad_norm": 404.4653045016176,
+      "learning_rate": 4.7156238491607506e-06,
+      "loss": 219.9062,
+      "step": 35380
+    },
+    {
+      "epoch": 0.6806390968405768,
+      "grad_norm": 398.92275156228794,
+      "learning_rate": 4.710443435646666e-06,
+      "loss": 216.9354,
+      "step": 35390
+    },
+    {
+      "epoch": 0.6808314221010573,
+      "grad_norm": 396.44172072203673,
+      "learning_rate": 4.705264992405043e-06,
+      "loss": 216.3257,
+      "step": 35400
+    },
+    {
+      "epoch": 0.6810237473615378,
+      "grad_norm": 414.9739739372363,
+      "learning_rate": 4.700088521364761e-06,
+      "loss": 216.6547,
+      "step": 35410
+    },
+    {
+      "epoch": 0.6812160726220183,
+      "grad_norm": 402.9635413856429,
+      "learning_rate": 4.694914024453977e-06,
+      "loss": 221.0535,
+      "step": 35420
+    },
+    {
+      "epoch": 0.6814083978824989,
+      "grad_norm": 448.0603991656356,
+      "learning_rate": 4.689741503600103e-06,
+      "loss": 220.5415,
+      "step": 35430
+    },
+    {
+      "epoch": 0.6816007231429794,
+      "grad_norm": 468.4523285944009,
+      "learning_rate": 4.684570960729818e-06,
+      "loss": 219.5447,
+      "step": 35440
+    },
+    {
+      "epoch": 0.6817930484034599,
+      "grad_norm": 395.9661355731399,
+      "learning_rate": 4.679402397769057e-06,
+      "loss": 214.6342,
+      "step": 35450
+    },
+    {
+      "epoch": 0.6819853736639404,
+      "grad_norm": 442.422516066603,
+      "learning_rate": 4.674235816643035e-06,
+      "loss": 228.8218,
+      "step": 35460
+    },
+    {
+      "epoch": 0.6821776989244209,
+      "grad_norm": 365.55331513023737,
+      "learning_rate": 4.66907121927621e-06,
+      "loss": 218.3393,
+      "step": 35470
+    },
+    {
+      "epoch": 0.6823700241849016,
+      "grad_norm": 406.7754196266838,
+      "learning_rate": 4.663908607592316e-06,
+      "loss": 215.1052,
+      "step": 35480
+    },
+    {
+      "epoch": 0.6825623494453821,
+      "grad_norm": 385.4351186325653,
+      "learning_rate": 4.658747983514334e-06,
+      "loss": 209.8509,
+      "step": 35490
+    },
+    {
+      "epoch": 0.6827546747058626,
+      "grad_norm": 430.42020327718194,
+      "learning_rate": 4.653589348964517e-06,
+      "loss": 218.0502,
+      "step": 35500
+    },
+    {
+      "epoch": 0.6829469999663431,
+      "grad_norm": 417.76945690414715,
+      "learning_rate": 4.648432705864369e-06,
+      "loss": 237.4081,
+      "step": 35510
+    },
+    {
+      "epoch": 0.6831393252268236,
+      "grad_norm": 415.3004662423424,
+      "learning_rate": 4.64327805613465e-06,
+      "loss": 225.0952,
+      "step": 35520
+    },
+    {
+      "epoch": 0.6833316504873042,
+      "grad_norm": 461.4128726600178,
+      "learning_rate": 4.638125401695391e-06,
+      "loss": 216.5117,
+      "step": 35530
+    },
+    {
+      "epoch": 0.6835239757477847,
+      "grad_norm": 437.23718739371304,
+      "learning_rate": 4.632974744465865e-06,
+      "loss": 224.5861,
+      "step": 35540
+    },
+    {
+      "epoch": 0.6837163010082652,
+      "grad_norm": 421.5048453080918,
+      "learning_rate": 4.627826086364603e-06,
+      "loss": 226.2318,
+      "step": 35550
+    },
+    {
+      "epoch": 0.6839086262687457,
+      "grad_norm": 502.55783888187017,
+      "learning_rate": 4.622679429309404e-06,
+      "loss": 220.68,
+      "step": 35560
+    },
+    {
+      "epoch": 0.6841009515292262,
+      "grad_norm": 390.21991768610644,
+      "learning_rate": 4.617534775217307e-06,
+      "loss": 209.8581,
+      "step": 35570
+    },
+    {
+      "epoch": 0.6842932767897068,
+      "grad_norm": 393.66885736023653,
+      "learning_rate": 4.6123921260046135e-06,
+      "loss": 214.7617,
+      "step": 35580
+    },
+    {
+      "epoch": 0.6844856020501873,
+      "grad_norm": 393.9994217123655,
+      "learning_rate": 4.607251483586869e-06,
+      "loss": 223.121,
+      "step": 35590
+    },
+    {
+      "epoch": 0.6846779273106678,
+      "grad_norm": 394.6677257224646,
+      "learning_rate": 4.6021128498788855e-06,
+      "loss": 212.4204,
+      "step": 35600
+    },
+    {
+      "epoch": 0.6848702525711483,
+      "grad_norm": 390.0223437269778,
+      "learning_rate": 4.596976226794718e-06,
+      "loss": 213.0069,
+      "step": 35610
+    },
+    {
+      "epoch": 0.6850625778316288,
+      "grad_norm": 365.9931384181439,
+      "learning_rate": 4.591841616247669e-06,
+      "loss": 225.4255,
+      "step": 35620
+    },
+    {
+      "epoch": 0.6852549030921093,
+      "grad_norm": 425.7434583247429,
+      "learning_rate": 4.586709020150297e-06,
+      "loss": 216.6788,
+      "step": 35630
+    },
+    {
+      "epoch": 0.6854472283525899,
+      "grad_norm": 416.9298132211286,
+      "learning_rate": 4.581578440414417e-06,
+      "loss": 214.8177,
+      "step": 35640
+    },
+    {
+      "epoch": 0.6856395536130704,
+      "grad_norm": 391.189193703694,
+      "learning_rate": 4.576449878951079e-06,
+      "loss": 221.3543,
+      "step": 35650
+    },
+    {
+      "epoch": 0.6858318788735509,
+      "grad_norm": 456.13133359288577,
+      "learning_rate": 4.571323337670584e-06,
+      "loss": 218.6048,
+      "step": 35660
+    },
+    {
+      "epoch": 0.6860242041340314,
+      "grad_norm": 356.172617503518,
+      "learning_rate": 4.566198818482494e-06,
+      "loss": 219.5613,
+      "step": 35670
+    },
+    {
+      "epoch": 0.686216529394512,
+      "grad_norm": 377.8119828173668,
+      "learning_rate": 4.5610763232956e-06,
+      "loss": 221.6812,
+      "step": 35680
+    },
+    {
+      "epoch": 0.6864088546549926,
+      "grad_norm": 389.65658442797036,
+      "learning_rate": 4.55595585401795e-06,
+      "loss": 226.5807,
+      "step": 35690
+    },
+    {
+      "epoch": 0.6866011799154731,
+      "grad_norm": 425.29899536955804,
+      "learning_rate": 4.550837412556828e-06,
+      "loss": 229.349,
+      "step": 35700
+    },
+    {
+      "epoch": 0.6867935051759536,
+      "grad_norm": 401.015777030877,
+      "learning_rate": 4.545721000818778e-06,
+      "loss": 226.9609,
+      "step": 35710
+    },
+    {
+      "epoch": 0.6869858304364341,
+      "grad_norm": 399.04573981872664,
+      "learning_rate": 4.540606620709575e-06,
+      "loss": 234.3499,
+      "step": 35720
+    },
+    {
+      "epoch": 0.6871781556969146,
+      "grad_norm": 395.16422376910265,
+      "learning_rate": 4.535494274134236e-06,
+      "loss": 223.3513,
+      "step": 35730
+    },
+    {
+      "epoch": 0.6873704809573952,
+      "grad_norm": 412.30307874966866,
+      "learning_rate": 4.530383962997036e-06,
+      "loss": 220.3366,
+      "step": 35740
+    },
+    {
+      "epoch": 0.6875628062178757,
+      "grad_norm": 435.66609210857285,
+      "learning_rate": 4.525275689201476e-06,
+      "loss": 229.2069,
+      "step": 35750
+    },
+    {
+      "epoch": 0.6877551314783562,
+      "grad_norm": 460.42100774759336,
+      "learning_rate": 4.520169454650299e-06,
+      "loss": 222.6201,
+      "step": 35760
+    },
+    {
+      "epoch": 0.6879474567388367,
+      "grad_norm": 463.9467816745536,
+      "learning_rate": 4.515065261245498e-06,
+      "loss": 224.6992,
+      "step": 35770
+    },
+    {
+      "epoch": 0.6881397819993172,
+      "grad_norm": 400.3259681576388,
+      "learning_rate": 4.509963110888305e-06,
+      "loss": 218.8042,
+      "step": 35780
+    },
+    {
+      "epoch": 0.6883321072597978,
+      "grad_norm": 398.00445988717433,
+      "learning_rate": 4.504863005479182e-06,
+      "loss": 214.1886,
+      "step": 35790
+    },
+    {
+      "epoch": 0.6885244325202783,
+      "grad_norm": 453.40343361828576,
+      "learning_rate": 4.499764946917834e-06,
+      "loss": 222.4477,
+      "step": 35800
+    },
+    {
+      "epoch": 0.6887167577807588,
+      "grad_norm": 377.2414076529407,
+      "learning_rate": 4.494668937103201e-06,
+      "loss": 219.7592,
+      "step": 35810
+    },
+    {
+      "epoch": 0.6889090830412393,
+      "grad_norm": 384.9161514166962,
+      "learning_rate": 4.489574977933467e-06,
+      "loss": 216.2119,
+      "step": 35820
+    },
+    {
+      "epoch": 0.6891014083017198,
+      "grad_norm": 409.8341540854874,
+      "learning_rate": 4.484483071306048e-06,
+      "loss": 226.0465,
+      "step": 35830
+    },
+    {
+      "epoch": 0.6892937335622004,
+      "grad_norm": 366.05182434451,
+      "learning_rate": 4.479393219117588e-06,
+      "loss": 215.4577,
+      "step": 35840
+    },
+    {
+      "epoch": 0.6894860588226809,
+      "grad_norm": 379.5787758698706,
+      "learning_rate": 4.474305423263984e-06,
+      "loss": 216.8345,
+      "step": 35850
+    },
+    {
+      "epoch": 0.6896783840831614,
+      "grad_norm": 436.45034008295283,
+      "learning_rate": 4.469219685640348e-06,
+      "loss": 213.4985,
+      "step": 35860
+    },
+    {
+      "epoch": 0.6898707093436419,
+      "grad_norm": 438.91448163518584,
+      "learning_rate": 4.4641360081410356e-06,
+      "loss": 219.4,
+      "step": 35870
+    },
+    {
+      "epoch": 0.6900630346041224,
+      "grad_norm": 409.6116397181822,
+      "learning_rate": 4.4590543926596285e-06,
+      "loss": 220.7814,
+      "step": 35880
+    },
+    {
+      "epoch": 0.6902553598646031,
+      "grad_norm": 418.8105487840902,
+      "learning_rate": 4.453974841088953e-06,
+      "loss": 223.7794,
+      "step": 35890
+    },
+    {
+      "epoch": 0.6904476851250836,
+      "grad_norm": 398.5432586878791,
+      "learning_rate": 4.448897355321049e-06,
+      "loss": 218.6897,
+      "step": 35900
+    },
+    {
+      "epoch": 0.6906400103855641,
+      "grad_norm": 383.66018857637016,
+      "learning_rate": 4.443821937247205e-06,
+      "loss": 217.4083,
+      "step": 35910
+    },
+    {
+      "epoch": 0.6908323356460446,
+      "grad_norm": 414.73046501485646,
+      "learning_rate": 4.438748588757922e-06,
+      "loss": 218.7771,
+      "step": 35920
+    },
+    {
+      "epoch": 0.6910246609065251,
+      "grad_norm": 380.7592696692757,
+      "learning_rate": 4.433677311742945e-06,
+      "loss": 216.4235,
+      "step": 35930
+    },
+    {
+      "epoch": 0.6912169861670057,
+      "grad_norm": 383.20210251628,
+      "learning_rate": 4.428608108091241e-06,
+      "loss": 216.2294,
+      "step": 35940
+    },
+    {
+      "epoch": 0.6914093114274862,
+      "grad_norm": 389.83690133773337,
+      "learning_rate": 4.4235409796909965e-06,
+      "loss": 217.3402,
+      "step": 35950
+    },
+    {
+      "epoch": 0.6916016366879667,
+      "grad_norm": 409.7192019886682,
+      "learning_rate": 4.418475928429644e-06,
+      "loss": 218.1749,
+      "step": 35960
+    },
+    {
+      "epoch": 0.6917939619484472,
+      "grad_norm": 411.4669615637612,
+      "learning_rate": 4.413412956193826e-06,
+      "loss": 225.1929,
+      "step": 35970
+    },
+    {
+      "epoch": 0.6919862872089277,
+      "grad_norm": 440.62809127856247,
+      "learning_rate": 4.4083520648694156e-06,
+      "loss": 210.2017,
+      "step": 35980
+    },
+    {
+      "epoch": 0.6921786124694083,
+      "grad_norm": 386.5606776080072,
+      "learning_rate": 4.4032932563415075e-06,
+      "loss": 216.4489,
+      "step": 35990
+    },
+    {
+      "epoch": 0.6923709377298888,
+      "grad_norm": 389.3415345420483,
+      "learning_rate": 4.398236532494434e-06,
+      "loss": 217.0247,
+      "step": 36000
+    },
+    {
+      "epoch": 0.6925632629903693,
+      "grad_norm": 437.0736833560942,
+      "learning_rate": 4.393181895211735e-06,
+      "loss": 227.7811,
+      "step": 36010
+    },
+    {
+      "epoch": 0.6927555882508498,
+      "grad_norm": 411.9620831442087,
+      "learning_rate": 4.388129346376177e-06,
+      "loss": 217.2434,
+      "step": 36020
+    },
+    {
+      "epoch": 0.6929479135113303,
+      "grad_norm": 401.59662123749564,
+      "learning_rate": 4.383078887869759e-06,
+      "loss": 214.7082,
+      "step": 36030
+    },
+    {
+      "epoch": 0.6931402387718109,
+      "grad_norm": 366.91846840815407,
+      "learning_rate": 4.378030521573683e-06,
+      "loss": 214.2149,
+      "step": 36040
+    },
+    {
+      "epoch": 0.6933325640322914,
+      "grad_norm": 401.0252302703506,
+      "learning_rate": 4.372984249368393e-06,
+      "loss": 217.7893,
+      "step": 36050
+    },
+    {
+      "epoch": 0.6935248892927719,
+      "grad_norm": 370.61991209202967,
+      "learning_rate": 4.367940073133533e-06,
+      "loss": 215.8928,
+      "step": 36060
+    },
+    {
+      "epoch": 0.6937172145532524,
+      "grad_norm": 372.9909954708285,
+      "learning_rate": 4.362897994747982e-06,
+      "loss": 218.2172,
+      "step": 36070
+    },
+    {
+      "epoch": 0.6939095398137329,
+      "grad_norm": 377.8590749980655,
+      "learning_rate": 4.35785801608983e-06,
+      "loss": 210.704,
+      "step": 36080
+    },
+    {
+      "epoch": 0.6941018650742136,
+      "grad_norm": 443.0240719634579,
+      "learning_rate": 4.352820139036379e-06,
+      "loss": 217.6218,
+      "step": 36090
+    },
+    {
+      "epoch": 0.6942941903346941,
+      "grad_norm": 378.81805403458264,
+      "learning_rate": 4.347784365464163e-06,
+      "loss": 216.5826,
+      "step": 36100
+    },
+    {
+      "epoch": 0.6944865155951746,
+      "grad_norm": 377.0661857394148,
+      "learning_rate": 4.342750697248922e-06,
+      "loss": 223.8753,
+      "step": 36110
+    },
+    {
+      "epoch": 0.6946788408556551,
+      "grad_norm": 423.67280073615655,
+      "learning_rate": 4.337719136265614e-06,
+      "loss": 211.0271,
+      "step": 36120
+    },
+    {
+      "epoch": 0.6948711661161356,
+      "grad_norm": 360.0370892686145,
+      "learning_rate": 4.332689684388408e-06,
+      "loss": 214.1245,
+      "step": 36130
+    },
+    {
+      "epoch": 0.6950634913766162,
+      "grad_norm": 391.7009479640741,
+      "learning_rate": 4.327662343490701e-06,
+      "loss": 217.2823,
+      "step": 36140
+    },
+    {
+      "epoch": 0.6952558166370967,
+      "grad_norm": 419.38673821848687,
+      "learning_rate": 4.322637115445088e-06,
+      "loss": 224.3325,
+      "step": 36150
+    },
+    {
+      "epoch": 0.6954481418975772,
+      "grad_norm": 413.7845806064118,
+      "learning_rate": 4.3176140021233845e-06,
+      "loss": 215.707,
+      "step": 36160
+    },
+    {
+      "epoch": 0.6956404671580577,
+      "grad_norm": 390.1556774787163,
+      "learning_rate": 4.312593005396615e-06,
+      "loss": 218.6496,
+      "step": 36170
+    },
+    {
+      "epoch": 0.6958327924185382,
+      "grad_norm": 387.4228282520312,
+      "learning_rate": 4.307574127135022e-06,
+      "loss": 213.2394,
+      "step": 36180
+    },
+    {
+      "epoch": 0.6960251176790188,
+      "grad_norm": 436.92378932308515,
+      "learning_rate": 4.3025573692080516e-06,
+      "loss": 219.7877,
+      "step": 36190
+    },
+    {
+      "epoch": 0.6962174429394993,
+      "grad_norm": 398.60931934556857,
+      "learning_rate": 4.297542733484364e-06,
+      "loss": 214.5497,
+      "step": 36200
+    },
+    {
+      "epoch": 0.6964097681999798,
+      "grad_norm": 362.11414098938627,
+      "learning_rate": 4.292530221831832e-06,
+      "loss": 219.6255,
+      "step": 36210
+    },
+    {
+      "epoch": 0.6966020934604603,
+      "grad_norm": 389.04778788562544,
+      "learning_rate": 4.2875198361175305e-06,
+      "loss": 218.1958,
+      "step": 36220
+    },
+    {
+      "epoch": 0.6967944187209408,
+      "grad_norm": 382.16408108309616,
+      "learning_rate": 4.282511578207746e-06,
+      "loss": 215.1704,
+      "step": 36230
+    },
+    {
+      "epoch": 0.6969867439814214,
+      "grad_norm": 396.8392630552316,
+      "learning_rate": 4.277505449967967e-06,
+      "loss": 220.5081,
+      "step": 36240
+    },
+    {
+      "epoch": 0.6971790692419019,
+      "grad_norm": 378.9595549794043,
+      "learning_rate": 4.2725014532629015e-06,
+      "loss": 220.177,
+      "step": 36250
+    },
+    {
+      "epoch": 0.6973713945023824,
+      "grad_norm": 437.6280921220109,
+      "learning_rate": 4.267499589956453e-06,
+      "loss": 217.4724,
+      "step": 36260
+    },
+    {
+      "epoch": 0.6975637197628629,
+      "grad_norm": 397.7583302478522,
+      "learning_rate": 4.262499861911727e-06,
+      "loss": 223.5542,
+      "step": 36270
+    },
+    {
+      "epoch": 0.6977560450233434,
+      "grad_norm": 408.00787437210926,
+      "learning_rate": 4.257502270991048e-06,
+      "loss": 219.0948,
+      "step": 36280
+    },
+    {
+      "epoch": 0.697948370283824,
+      "grad_norm": 402.04654195431567,
+      "learning_rate": 4.252506819055934e-06,
+      "loss": 220.6372,
+      "step": 36290
+    },
+    {
+      "epoch": 0.6981406955443046,
+      "grad_norm": 414.8568135035925,
+      "learning_rate": 4.2475135079671045e-06,
+      "loss": 217.2164,
+      "step": 36300
+    },
+    {
+      "epoch": 0.6983330208047851,
+      "grad_norm": 368.07813678844695,
+      "learning_rate": 4.242522339584486e-06,
+      "loss": 213.7817,
+      "step": 36310
+    },
+    {
+      "epoch": 0.6985253460652656,
+      "grad_norm": 378.734022113604,
+      "learning_rate": 4.2375333157672114e-06,
+      "loss": 218.8537,
+      "step": 36320
+    },
+    {
+      "epoch": 0.6987176713257461,
+      "grad_norm": 377.1309120295078,
+      "learning_rate": 4.232546438373604e-06,
+      "loss": 217.3216,
+      "step": 36330
+    },
+    {
+      "epoch": 0.6989099965862267,
+      "grad_norm": 413.2698859907694,
+      "learning_rate": 4.227561709261198e-06,
+      "loss": 210.7284,
+      "step": 36340
+    },
+    {
+      "epoch": 0.6991023218467072,
+      "grad_norm": 401.79411990133167,
+      "learning_rate": 4.222579130286716e-06,
+      "loss": 220.0384,
+      "step": 36350
+    },
+    {
+      "epoch": 0.6992946471071877,
+      "grad_norm": 413.46076823933174,
+      "learning_rate": 4.217598703306095e-06,
+      "loss": 222.4217,
+      "step": 36360
+    },
+    {
+      "epoch": 0.6994869723676682,
+      "grad_norm": 388.77647912409503,
+      "learning_rate": 4.212620430174457e-06,
+      "loss": 222.7028,
+      "step": 36370
+    },
+    {
+      "epoch": 0.6996792976281487,
+      "grad_norm": 427.063085247276,
+      "learning_rate": 4.207644312746124e-06,
+      "loss": 217.6686,
+      "step": 36380
+    },
+    {
+      "epoch": 0.6998716228886293,
+      "grad_norm": 385.49363023837566,
+      "learning_rate": 4.202670352874625e-06,
+      "loss": 215.446,
+      "step": 36390
+    },
+    {
+      "epoch": 0.7000639481491098,
+      "grad_norm": 394.7322835256111,
+      "learning_rate": 4.197698552412672e-06,
+      "loss": 224.6516,
+      "step": 36400
+    },
+    {
+      "epoch": 0.7002562734095903,
+      "grad_norm": 388.8321210388741,
+      "learning_rate": 4.192728913212181e-06,
+      "loss": 220.3064,
+      "step": 36410
+    },
+    {
+      "epoch": 0.7004485986700708,
+      "grad_norm": 378.6688332614007,
+      "learning_rate": 4.187761437124256e-06,
+      "loss": 210.9347,
+      "step": 36420
+    },
+    {
+      "epoch": 0.7006409239305513,
+      "grad_norm": 508.0817570342048,
+      "learning_rate": 4.182796125999207e-06,
+      "loss": 222.0558,
+      "step": 36430
+    },
+    {
+      "epoch": 0.7008332491910318,
+      "grad_norm": 480.9940368722151,
+      "learning_rate": 4.177832981686526e-06,
+      "loss": 221.5692,
+      "step": 36440
+    },
+    {
+      "epoch": 0.7010255744515124,
+      "grad_norm": 415.9012148632597,
+      "learning_rate": 4.172872006034899e-06,
+      "loss": 218.7598,
+      "step": 36450
+    },
+    {
+      "epoch": 0.7012178997119929,
+      "grad_norm": 431.4645534519969,
+      "learning_rate": 4.167913200892217e-06,
+      "loss": 214.2937,
+      "step": 36460
+    },
+    {
+      "epoch": 0.7014102249724734,
+      "grad_norm": 381.06772321132337,
+      "learning_rate": 4.162956568105543e-06,
+      "loss": 225.915,
+      "step": 36470
+    },
+    {
+      "epoch": 0.7016025502329539,
+      "grad_norm": 388.60806706428485,
+      "learning_rate": 4.158002109521149e-06,
+      "loss": 214.21,
+      "step": 36480
+    },
+    {
+      "epoch": 0.7017948754934344,
+      "grad_norm": 408.77425431935796,
+      "learning_rate": 4.153049826984482e-06,
+      "loss": 218.154,
+      "step": 36490
+    },
+    {
+      "epoch": 0.7019872007539151,
+      "grad_norm": 369.10178348734564,
+      "learning_rate": 4.148099722340192e-06,
+      "loss": 225.4862,
+      "step": 36500
+    },
+    {
+      "epoch": 0.7021795260143956,
+      "grad_norm": 396.3483999643876,
+      "learning_rate": 4.143151797432109e-06,
+      "loss": 216.5708,
+      "step": 36510
+    },
+    {
+      "epoch": 0.7023718512748761,
+      "grad_norm": 411.0565684847517,
+      "learning_rate": 4.1382060541032505e-06,
+      "loss": 219.8197,
+      "step": 36520
+    },
+    {
+      "epoch": 0.7025641765353566,
+      "grad_norm": 401.13262557200215,
+      "learning_rate": 4.133262494195824e-06,
+      "loss": 217.943,
+      "step": 36530
+    },
+    {
+      "epoch": 0.7027565017958372,
+      "grad_norm": 364.3826246453327,
+      "learning_rate": 4.12832111955123e-06,
+      "loss": 209.2772,
+      "step": 36540
+    },
+    {
+      "epoch": 0.7029488270563177,
+      "grad_norm": 405.0066517413073,
+      "learning_rate": 4.123381932010044e-06,
+      "loss": 220.1532,
+      "step": 36550
+    },
+    {
+      "epoch": 0.7031411523167982,
+      "grad_norm": 373.58866199094336,
+      "learning_rate": 4.11844493341203e-06,
+      "loss": 218.792,
+      "step": 36560
+    },
+    {
+      "epoch": 0.7033334775772787,
+      "grad_norm": 382.6080284475787,
+      "learning_rate": 4.113510125596145e-06,
+      "loss": 220.5175,
+      "step": 36570
+    },
+    {
+      "epoch": 0.7035258028377592,
+      "grad_norm": 383.4123986557622,
+      "learning_rate": 4.1085775104005186e-06,
+      "loss": 210.9176,
+      "step": 36580
+    },
+    {
+      "epoch": 0.7037181280982397,
+      "grad_norm": 426.4322933050965,
+      "learning_rate": 4.10364708966247e-06,
+      "loss": 219.6772,
+      "step": 36590
+    },
+    {
+      "epoch": 0.7039104533587203,
+      "grad_norm": 379.7676666846126,
+      "learning_rate": 4.098718865218496e-06,
+      "loss": 210.9602,
+      "step": 36600
+    },
+    {
+      "epoch": 0.7041027786192008,
+      "grad_norm": 397.40373076708136,
+      "learning_rate": 4.0937928389042815e-06,
+      "loss": 224.6631,
+      "step": 36610
+    },
+    {
+      "epoch": 0.7042951038796813,
+      "grad_norm": 375.13173179317903,
+      "learning_rate": 4.088869012554694e-06,
+      "loss": 212.3754,
+      "step": 36620
+    },
+    {
+      "epoch": 0.7044874291401618,
+      "grad_norm": 401.3622644076648,
+      "learning_rate": 4.08394738800377e-06,
+      "loss": 217.7254,
+      "step": 36630
+    },
+    {
+      "epoch": 0.7046797544006423,
+      "grad_norm": 373.0768116647785,
+      "learning_rate": 4.07902796708474e-06,
+      "loss": 212.6043,
+      "step": 36640
+    },
+    {
+      "epoch": 0.7048720796611229,
+      "grad_norm": 407.3650275149077,
+      "learning_rate": 4.074110751630005e-06,
+      "loss": 223.1159,
+      "step": 36650
+    },
+    {
+      "epoch": 0.7050644049216034,
+      "grad_norm": 387.43968848525844,
+      "learning_rate": 4.0691957434711446e-06,
+      "loss": 214.1859,
+      "step": 36660
+    },
+    {
+      "epoch": 0.7052567301820839,
+      "grad_norm": 388.9674021574162,
+      "learning_rate": 4.0642829444389165e-06,
+      "loss": 215.2391,
+      "step": 36670
+    },
+    {
+      "epoch": 0.7054490554425644,
+      "grad_norm": 375.57479401836554,
+      "learning_rate": 4.059372356363263e-06,
+      "loss": 224.5012,
+      "step": 36680
+    },
+    {
+      "epoch": 0.7056413807030449,
+      "grad_norm": 369.53617259237427,
+      "learning_rate": 4.054463981073296e-06,
+      "loss": 216.8208,
+      "step": 36690
+    },
+    {
+      "epoch": 0.7058337059635255,
+      "grad_norm": 395.5861650392418,
+      "learning_rate": 4.049557820397297e-06,
+      "loss": 218.5601,
+      "step": 36700
+    },
+    {
+      "epoch": 0.7060260312240061,
+      "grad_norm": 428.8634696467282,
+      "learning_rate": 4.044653876162738e-06,
+      "loss": 209.5418,
+      "step": 36710
+    },
+    {
+      "epoch": 0.7062183564844866,
+      "grad_norm": 395.0019917828623,
+      "learning_rate": 4.039752150196257e-06,
+      "loss": 223.0564,
+      "step": 36720
+    },
+    {
+      "epoch": 0.7064106817449671,
+      "grad_norm": 384.39564430705565,
+      "learning_rate": 4.034852644323661e-06,
+      "loss": 220.2196,
+      "step": 36730
+    },
+    {
+      "epoch": 0.7066030070054476,
+      "grad_norm": 378.0906398002916,
+      "learning_rate": 4.029955360369935e-06,
+      "loss": 212.0748,
+      "step": 36740
+    },
+    {
+      "epoch": 0.7067953322659282,
+      "grad_norm": 412.94105531361157,
+      "learning_rate": 4.0250603001592416e-06,
+      "loss": 209.6829,
+      "step": 36750
+    },
+    {
+      "epoch": 0.7069876575264087,
+      "grad_norm": 367.7742189188349,
+      "learning_rate": 4.020167465514903e-06,
+      "loss": 212.8914,
+      "step": 36760
+    },
+    {
+      "epoch": 0.7071799827868892,
+      "grad_norm": 466.9708614206917,
+      "learning_rate": 4.015276858259427e-06,
+      "loss": 216.3343,
+      "step": 36770
+    },
+    {
+      "epoch": 0.7073723080473697,
+      "grad_norm": 417.7808912930527,
+      "learning_rate": 4.0103884802144775e-06,
+      "loss": 218.156,
+      "step": 36780
+    },
+    {
+      "epoch": 0.7075646333078502,
+      "grad_norm": 382.6623648613474,
+      "learning_rate": 4.0055023332009e-06,
+      "loss": 217.9552,
+      "step": 36790
+    },
+    {
+      "epoch": 0.7077569585683308,
+      "grad_norm": 380.1630706460532,
+      "learning_rate": 4.000618419038702e-06,
+      "loss": 205.2449,
+      "step": 36800
+    },
+    {
+      "epoch": 0.7079492838288113,
+      "grad_norm": 390.85828417221916,
+      "learning_rate": 3.9957367395470555e-06,
+      "loss": 220.6086,
+      "step": 36810
+    },
+    {
+      "epoch": 0.7081416090892918,
+      "grad_norm": 384.39487026876157,
+      "learning_rate": 3.990857296544315e-06,
+      "loss": 209.5735,
+      "step": 36820
+    },
+    {
+      "epoch": 0.7083339343497723,
+      "grad_norm": 394.3181175812435,
+      "learning_rate": 3.985980091847985e-06,
+      "loss": 213.1617,
+      "step": 36830
+    },
+    {
+      "epoch": 0.7085262596102528,
+      "grad_norm": 402.4696077077873,
+      "learning_rate": 3.981105127274748e-06,
+      "loss": 218.0245,
+      "step": 36840
+    },
+    {
+      "epoch": 0.7087185848707334,
+      "grad_norm": 418.98189710713734,
+      "learning_rate": 3.976232404640441e-06,
+      "loss": 208.4677,
+      "step": 36850
+    },
+    {
+      "epoch": 0.7089109101312139,
+      "grad_norm": 377.55295471670394,
+      "learning_rate": 3.971361925760081e-06,
+      "loss": 216.8333,
+      "step": 36860
+    },
+    {
+      "epoch": 0.7091032353916944,
+      "grad_norm": 399.90882363551344,
+      "learning_rate": 3.966493692447838e-06,
+      "loss": 218.0203,
+      "step": 36870
+    },
+    {
+      "epoch": 0.7092955606521749,
+      "grad_norm": 387.3373044280473,
+      "learning_rate": 3.961627706517044e-06,
+      "loss": 214.0818,
+      "step": 36880
+    },
+    {
+      "epoch": 0.7094878859126554,
+      "grad_norm": 380.9525935212469,
+      "learning_rate": 3.956763969780206e-06,
+      "loss": 214.3401,
+      "step": 36890
+    },
+    {
+      "epoch": 0.709680211173136,
+      "grad_norm": 372.8655952139652,
+      "learning_rate": 3.951902484048978e-06,
+      "loss": 212.3872,
+      "step": 36900
+    },
+    {
+      "epoch": 0.7098725364336166,
+      "grad_norm": 428.66956815989255,
+      "learning_rate": 3.94704325113419e-06,
+      "loss": 213.0268,
+      "step": 36910
+    },
+    {
+      "epoch": 0.7100648616940971,
+      "grad_norm": 371.1782345954982,
+      "learning_rate": 3.942186272845821e-06,
+      "loss": 213.4061,
+      "step": 36920
+    },
+    {
+      "epoch": 0.7102571869545776,
+      "grad_norm": 381.22663818214284,
+      "learning_rate": 3.937331550993021e-06,
+      "loss": 216.1139,
+      "step": 36930
+    },
+    {
+      "epoch": 0.7104495122150581,
+      "grad_norm": 423.725917758182,
+      "learning_rate": 3.932479087384089e-06,
+      "loss": 220.7802,
+      "step": 36940
+    },
+    {
+      "epoch": 0.7106418374755387,
+      "grad_norm": 385.8627263522587,
+      "learning_rate": 3.927628883826488e-06,
+      "loss": 215.2264,
+      "step": 36950
+    },
+    {
+      "epoch": 0.7108341627360192,
+      "grad_norm": 390.6378826256979,
+      "learning_rate": 3.922780942126837e-06,
+      "loss": 215.5406,
+      "step": 36960
+    },
+    {
+      "epoch": 0.7110264879964997,
+      "grad_norm": 366.64284324379037,
+      "learning_rate": 3.91793526409092e-06,
+      "loss": 215.8305,
+      "step": 36970
+    },
+    {
+      "epoch": 0.7112188132569802,
+      "grad_norm": 405.37768285901956,
+      "learning_rate": 3.913091851523667e-06,
+      "loss": 216.9105,
+      "step": 36980
+    },
+    {
+      "epoch": 0.7114111385174607,
+      "grad_norm": 381.79533253671025,
+      "learning_rate": 3.908250706229168e-06,
+      "loss": 211.014,
+      "step": 36990
+    },
+    {
+      "epoch": 0.7116034637779413,
+      "grad_norm": 375.0037377936083,
+      "learning_rate": 3.903411830010676e-06,
+      "loss": 221.757,
+      "step": 37000
+    },
+    {
+      "epoch": 0.7117957890384218,
+      "grad_norm": 392.06676748232667,
+      "learning_rate": 3.8985752246705885e-06,
+      "loss": 215.6115,
+      "step": 37010
+    },
+    {
+      "epoch": 0.7119881142989023,
+      "grad_norm": 398.8947573310577,
+      "learning_rate": 3.893740892010463e-06,
+      "loss": 216.497,
+      "step": 37020
+    },
+    {
+      "epoch": 0.7121804395593828,
+      "grad_norm": 384.7264355834237,
+      "learning_rate": 3.888908833831002e-06,
+      "loss": 220.2314,
+      "step": 37030
+    },
+    {
+      "epoch": 0.7123727648198633,
+      "grad_norm": 376.65721600794285,
+      "learning_rate": 3.884079051932073e-06,
+      "loss": 216.734,
+      "step": 37040
+    },
+    {
+      "epoch": 0.7125650900803439,
+      "grad_norm": 382.61379253815494,
+      "learning_rate": 3.879251548112692e-06,
+      "loss": 221.2815,
+      "step": 37050
+    },
+    {
+      "epoch": 0.7127574153408244,
+      "grad_norm": 448.9567243317206,
+      "learning_rate": 3.874426324171019e-06,
+      "loss": 228.9296,
+      "step": 37060
+    },
+    {
+      "epoch": 0.7129497406013049,
+      "grad_norm": 464.1160680085575,
+      "learning_rate": 3.869603381904377e-06,
+      "loss": 217.6615,
+      "step": 37070
+    },
+    {
+      "epoch": 0.7131420658617854,
+      "grad_norm": 385.99393433823667,
+      "learning_rate": 3.864782723109227e-06,
+      "loss": 210.8737,
+      "step": 37080
+    },
+    {
+      "epoch": 0.7133343911222659,
+      "grad_norm": 401.45338608556045,
+      "learning_rate": 3.859964349581187e-06,
+      "loss": 213.8944,
+      "step": 37090
+    },
+    {
+      "epoch": 0.7135267163827465,
+      "grad_norm": 405.72798247152645,
+      "learning_rate": 3.855148263115017e-06,
+      "loss": 214.8662,
+      "step": 37100
+    },
+    {
+      "epoch": 0.713719041643227,
+      "grad_norm": 386.78157222311626,
+      "learning_rate": 3.850334465504637e-06,
+      "loss": 217.1708,
+      "step": 37110
+    },
+    {
+      "epoch": 0.7139113669037076,
+      "grad_norm": 414.9903177788113,
+      "learning_rate": 3.845522958543104e-06,
+      "loss": 222.5964,
+      "step": 37120
+    },
+    {
+      "epoch": 0.7141036921641881,
+      "grad_norm": 445.21050251339625,
+      "learning_rate": 3.840713744022624e-06,
+      "loss": 217.331,
+      "step": 37130
+    },
+    {
+      "epoch": 0.7142960174246686,
+      "grad_norm": 396.368690572658,
+      "learning_rate": 3.835906823734548e-06,
+      "loss": 220.5609,
+      "step": 37140
+    },
+    {
+      "epoch": 0.7144883426851492,
+      "grad_norm": 403.12981001721914,
+      "learning_rate": 3.831102199469379e-06,
+      "loss": 230.0353,
+      "step": 37150
+    },
+    {
+      "epoch": 0.7146806679456297,
+      "grad_norm": 386.08035290109416,
+      "learning_rate": 3.826299873016758e-06,
+      "loss": 219.3719,
+      "step": 37160
+    },
+    {
+      "epoch": 0.7148729932061102,
+      "grad_norm": 402.72050886512335,
+      "learning_rate": 3.821499846165468e-06,
+      "loss": 208.788,
+      "step": 37170
+    },
+    {
+      "epoch": 0.7150653184665907,
+      "grad_norm": 383.4067361784764,
+      "learning_rate": 3.816702120703449e-06,
+      "loss": 215.059,
+      "step": 37180
+    },
+    {
+      "epoch": 0.7152576437270712,
+      "grad_norm": 376.40450746521014,
+      "learning_rate": 3.8119066984177654e-06,
+      "loss": 212.6686,
+      "step": 37190
+    },
+    {
+      "epoch": 0.7154499689875518,
+      "grad_norm": 413.62893387422844,
+      "learning_rate": 3.8071135810946415e-06,
+      "loss": 217.3548,
+      "step": 37200
+    },
+    {
+      "epoch": 0.7156422942480323,
+      "grad_norm": 412.3567630502925,
+      "learning_rate": 3.802322770519424e-06,
+      "loss": 211.1593,
+      "step": 37210
+    },
+    {
+      "epoch": 0.7158346195085128,
+      "grad_norm": 359.2289363382688,
+      "learning_rate": 3.7975342684766215e-06,
+      "loss": 211.115,
+      "step": 37220
+    },
+    {
+      "epoch": 0.7160269447689933,
+      "grad_norm": 448.77799802289206,
+      "learning_rate": 3.792748076749867e-06,
+      "loss": 217.7228,
+      "step": 37230
+    },
+    {
+      "epoch": 0.7162192700294738,
+      "grad_norm": 455.2804006658165,
+      "learning_rate": 3.787964197121934e-06,
+      "loss": 218.0768,
+      "step": 37240
+    },
+    {
+      "epoch": 0.7164115952899544,
+      "grad_norm": 395.4492883309248,
+      "learning_rate": 3.7831826313747454e-06,
+      "loss": 209.8682,
+      "step": 37250
+    },
+    {
+      "epoch": 0.7166039205504349,
+      "grad_norm": 411.3581220117286,
+      "learning_rate": 3.778403381289353e-06,
+      "loss": 210.5628,
+      "step": 37260
+    },
+    {
+      "epoch": 0.7167962458109154,
+      "grad_norm": 382.0133005085732,
+      "learning_rate": 3.7736264486459486e-06,
+      "loss": 212.949,
+      "step": 37270
+    },
+    {
+      "epoch": 0.7169885710713959,
+      "grad_norm": 416.7489572781986,
+      "learning_rate": 3.7688518352238555e-06,
+      "loss": 213.1978,
+      "step": 37280
+    },
+    {
+      "epoch": 0.7171808963318764,
+      "grad_norm": 405.33642727439207,
+      "learning_rate": 3.7640795428015462e-06,
+      "loss": 217.8278,
+      "step": 37290
+    },
+    {
+      "epoch": 0.717373221592357,
+      "grad_norm": 381.83188029984956,
+      "learning_rate": 3.7593095731566186e-06,
+      "loss": 212.8764,
+      "step": 37300
+    },
+    {
+      "epoch": 0.7175655468528375,
+      "grad_norm": 381.97402060451054,
+      "learning_rate": 3.7545419280658025e-06,
+      "loss": 215.6786,
+      "step": 37310
+    },
+    {
+      "epoch": 0.7177578721133181,
+      "grad_norm": 394.3325526817493,
+      "learning_rate": 3.749776609304975e-06,
+      "loss": 214.9202,
+      "step": 37320
+    },
+    {
+      "epoch": 0.7179501973737986,
+      "grad_norm": 409.54189789928955,
+      "learning_rate": 3.7450136186491315e-06,
+      "loss": 218.1853,
+      "step": 37330
+    },
+    {
+      "epoch": 0.7181425226342791,
+      "grad_norm": 407.07468646164665,
+      "learning_rate": 3.7402529578724134e-06,
+      "loss": 215.7882,
+      "step": 37340
+    },
+    {
+      "epoch": 0.7183348478947597,
+      "grad_norm": 377.22379337736726,
+      "learning_rate": 3.735494628748082e-06,
+      "loss": 214.9534,
+      "step": 37350
+    },
+    {
+      "epoch": 0.7185271731552402,
+      "grad_norm": 394.0257390962629,
+      "learning_rate": 3.730738633048543e-06,
+      "loss": 211.4239,
+      "step": 37360
+    },
+    {
+      "epoch": 0.7187194984157207,
+      "grad_norm": 389.58732067506895,
+      "learning_rate": 3.7259849725453225e-06,
+      "loss": 217.8116,
+      "step": 37370
+    },
+    {
+      "epoch": 0.7189118236762012,
+      "grad_norm": 377.0923953290185,
+      "learning_rate": 3.7212336490090815e-06,
+      "loss": 212.8401,
+      "step": 37380
+    },
+    {
+      "epoch": 0.7191041489366817,
+      "grad_norm": 403.96954872872465,
+      "learning_rate": 3.7164846642096053e-06,
+      "loss": 216.3146,
+      "step": 37390
+    },
+    {
+      "epoch": 0.7192964741971622,
+      "grad_norm": 385.1954453311772,
+      "learning_rate": 3.7117380199158204e-06,
+      "loss": 221.3708,
+      "step": 37400
+    },
+    {
+      "epoch": 0.7194887994576428,
+      "grad_norm": 382.8224178957188,
+      "learning_rate": 3.706993717895768e-06,
+      "loss": 212.1186,
+      "step": 37410
+    },
+    {
+      "epoch": 0.7196811247181233,
+      "grad_norm": 420.2076096408132,
+      "learning_rate": 3.70225175991662e-06,
+      "loss": 214.2058,
+      "step": 37420
+    },
+    {
+      "epoch": 0.7198734499786038,
+      "grad_norm": 418.72564680723696,
+      "learning_rate": 3.697512147744684e-06,
+      "loss": 220.3151,
+      "step": 37430
+    },
+    {
+      "epoch": 0.7200657752390843,
+      "grad_norm": 410.8615680402483,
+      "learning_rate": 3.6927748831453835e-06,
+      "loss": 210.4465,
+      "step": 37440
+    },
+    {
+      "epoch": 0.7202581004995648,
+      "grad_norm": 451.79272070794457,
+      "learning_rate": 3.688039967883269e-06,
+      "loss": 215.5568,
+      "step": 37450
+    },
+    {
+      "epoch": 0.7204504257600454,
+      "grad_norm": 383.8605944001642,
+      "learning_rate": 3.683307403722025e-06,
+      "loss": 212.7483,
+      "step": 37460
+    },
+    {
+      "epoch": 0.7206427510205259,
+      "grad_norm": 395.36973397526606,
+      "learning_rate": 3.678577192424445e-06,
+      "loss": 212.0434,
+      "step": 37470
+    },
+    {
+      "epoch": 0.7208350762810064,
+      "grad_norm": 453.8066554156848,
+      "learning_rate": 3.6738493357524628e-06,
+      "loss": 212.7815,
+      "step": 37480
+    },
+    {
+      "epoch": 0.7210274015414869,
+      "grad_norm": 380.37639243922393,
+      "learning_rate": 3.6691238354671233e-06,
+      "loss": 216.1363,
+      "step": 37490
+    },
+    {
+      "epoch": 0.7212197268019674,
+      "grad_norm": 441.8785987891232,
+      "learning_rate": 3.664400693328595e-06,
+      "loss": 216.0949,
+      "step": 37500
+    },
+    {
+      "epoch": 0.721412052062448,
+      "grad_norm": 390.01442903407167,
+      "learning_rate": 3.6596799110961746e-06,
+      "loss": 210.2427,
+      "step": 37510
+    },
+    {
+      "epoch": 0.7216043773229286,
+      "grad_norm": 402.5980550515958,
+      "learning_rate": 3.6549614905282724e-06,
+      "loss": 219.8088,
+      "step": 37520
+    },
+    {
+      "epoch": 0.7217967025834091,
+      "grad_norm": 390.03219720239906,
+      "learning_rate": 3.6502454333824224e-06,
+      "loss": 217.078,
+      "step": 37530
+    },
+    {
+      "epoch": 0.7219890278438896,
+      "grad_norm": 398.9283709306277,
+      "learning_rate": 3.6455317414152803e-06,
+      "loss": 215.0608,
+      "step": 37540
+    },
+    {
+      "epoch": 0.7221813531043701,
+      "grad_norm": 386.8608648754265,
+      "learning_rate": 3.640820416382618e-06,
+      "loss": 216.3629,
+      "step": 37550
+    },
+    {
+      "epoch": 0.7223736783648507,
+      "grad_norm": 365.46810760796393,
+      "learning_rate": 3.6361114600393242e-06,
+      "loss": 207.1098,
+      "step": 37560
+    },
+    {
+      "epoch": 0.7225660036253312,
+      "grad_norm": 409.28456355992006,
+      "learning_rate": 3.6314048741394057e-06,
+      "loss": 219.487,
+      "step": 37570
+    },
+    {
+      "epoch": 0.7227583288858117,
+      "grad_norm": 386.7037923787949,
+      "learning_rate": 3.6267006604359943e-06,
+      "loss": 216.4873,
+      "step": 37580
+    },
+    {
+      "epoch": 0.7229506541462922,
+      "grad_norm": 407.7254066263435,
+      "learning_rate": 3.6219988206813285e-06,
+      "loss": 221.3888,
+      "step": 37590
+    },
+    {
+      "epoch": 0.7231429794067727,
+      "grad_norm": 379.2419560029266,
+      "learning_rate": 3.6172993566267623e-06,
+      "loss": 211.0802,
+      "step": 37600
+    },
+    {
+      "epoch": 0.7233353046672533,
+      "grad_norm": 423.416286781411,
+      "learning_rate": 3.6126022700227715e-06,
+      "loss": 215.6808,
+      "step": 37610
+    },
+    {
+      "epoch": 0.7235276299277338,
+      "grad_norm": 392.0818348934516,
+      "learning_rate": 3.6079075626189476e-06,
+      "loss": 216.8167,
+      "step": 37620
+    },
+    {
+      "epoch": 0.7237199551882143,
+      "grad_norm": 375.6976536586021,
+      "learning_rate": 3.603215236163987e-06,
+      "loss": 209.7872,
+      "step": 37630
+    },
+    {
+      "epoch": 0.7239122804486948,
+      "grad_norm": 381.61006376092143,
+      "learning_rate": 3.5985252924057023e-06,
+      "loss": 211.1029,
+      "step": 37640
+    },
+    {
+      "epoch": 0.7241046057091753,
+      "grad_norm": 360.58643961335366,
+      "learning_rate": 3.5938377330910245e-06,
+      "loss": 214.0201,
+      "step": 37650
+    },
+    {
+      "epoch": 0.7242969309696559,
+      "grad_norm": 389.5572225164136,
+      "learning_rate": 3.5891525599659905e-06,
+      "loss": 214.1953,
+      "step": 37660
+    },
+    {
+      "epoch": 0.7244892562301364,
+      "grad_norm": 376.96841726163024,
+      "learning_rate": 3.5844697747757496e-06,
+      "loss": 211.0877,
+      "step": 37670
+    },
+    {
+      "epoch": 0.7246815814906169,
+      "grad_norm": 385.9190622298062,
+      "learning_rate": 3.5797893792645577e-06,
+      "loss": 214.5382,
+      "step": 37680
+    },
+    {
+      "epoch": 0.7248739067510974,
+      "grad_norm": 441.936096418896,
+      "learning_rate": 3.5751113751757925e-06,
+      "loss": 216.2822,
+      "step": 37690
+    },
+    {
+      "epoch": 0.7250662320115779,
+      "grad_norm": 371.7068358969222,
+      "learning_rate": 3.5704357642519295e-06,
+      "loss": 211.2753,
+      "step": 37700
+    },
+    {
+      "epoch": 0.7252585572720585,
+      "grad_norm": 439.6478301760872,
+      "learning_rate": 3.5657625482345526e-06,
+      "loss": 212.1111,
+      "step": 37710
+    },
+    {
+      "epoch": 0.725450882532539,
+      "grad_norm": 428.51900380435933,
+      "learning_rate": 3.5610917288643655e-06,
+      "loss": 213.3385,
+      "step": 37720
+    },
+    {
+      "epoch": 0.7256432077930196,
+      "grad_norm": 432.0612694325686,
+      "learning_rate": 3.556423307881167e-06,
+      "loss": 216.1817,
+      "step": 37730
+    },
+    {
+      "epoch": 0.7258355330535001,
+      "grad_norm": 419.2711745131861,
+      "learning_rate": 3.551757287023865e-06,
+      "loss": 216.0665,
+      "step": 37740
+    },
+    {
+      "epoch": 0.7260278583139806,
+      "grad_norm": 354.73624102156936,
+      "learning_rate": 3.547093668030479e-06,
+      "loss": 217.3852,
+      "step": 37750
+    },
+    {
+      "epoch": 0.7262201835744612,
+      "grad_norm": 411.31539865415704,
+      "learning_rate": 3.542432452638126e-06,
+      "loss": 215.8667,
+      "step": 37760
+    },
+    {
+      "epoch": 0.7264125088349417,
+      "grad_norm": 375.5391671951594,
+      "learning_rate": 3.5377736425830366e-06,
+      "loss": 219.6735,
+      "step": 37770
+    },
+    {
+      "epoch": 0.7266048340954222,
+      "grad_norm": 445.47570070130257,
+      "learning_rate": 3.5331172396005354e-06,
+      "loss": 209.6886,
+      "step": 37780
+    },
+    {
+      "epoch": 0.7267971593559027,
+      "grad_norm": 387.302323914234,
+      "learning_rate": 3.528463245425062e-06,
+      "loss": 220.6175,
+      "step": 37790
+    },
+    {
+      "epoch": 0.7269894846163832,
+      "grad_norm": 371.3444822515457,
+      "learning_rate": 3.5238116617901486e-06,
+      "loss": 213.4917,
+      "step": 37800
+    },
+    {
+      "epoch": 0.7271818098768638,
+      "grad_norm": 413.97165815655575,
+      "learning_rate": 3.519162490428433e-06,
+      "loss": 220.2051,
+      "step": 37810
+    },
+    {
+      "epoch": 0.7273741351373443,
+      "grad_norm": 400.79588450286514,
+      "learning_rate": 3.5145157330716516e-06,
+      "loss": 211.9624,
+      "step": 37820
+    },
+    {
+      "epoch": 0.7275664603978248,
+      "grad_norm": 382.4262973156571,
+      "learning_rate": 3.509871391450652e-06,
+      "loss": 220.5928,
+      "step": 37830
+    },
+    {
+      "epoch": 0.7277587856583053,
+      "grad_norm": 376.69697537133334,
+      "learning_rate": 3.505229467295371e-06,
+      "loss": 213.096,
+      "step": 37840
+    },
+    {
+      "epoch": 0.7279511109187858,
+      "grad_norm": 382.8646927653982,
+      "learning_rate": 3.5005899623348493e-06,
+      "loss": 224.9132,
+      "step": 37850
+    },
+    {
+      "epoch": 0.7281434361792664,
+      "grad_norm": 391.33572040404994,
+      "learning_rate": 3.495952878297221e-06,
+      "loss": 217.0161,
+      "step": 37860
+    },
+    {
+      "epoch": 0.7283357614397469,
+      "grad_norm": 386.14417599405647,
+      "learning_rate": 3.4913182169097315e-06,
+      "loss": 210.7214,
+      "step": 37870
+    },
+    {
+      "epoch": 0.7285280867002274,
+      "grad_norm": 372.181013680401,
+      "learning_rate": 3.4866859798987084e-06,
+      "loss": 212.2185,
+      "step": 37880
+    },
+    {
+      "epoch": 0.7287204119607079,
+      "grad_norm": 474.6264707198723,
+      "learning_rate": 3.4820561689895906e-06,
+      "loss": 215.612,
+      "step": 37890
+    },
+    {
+      "epoch": 0.7289127372211884,
+      "grad_norm": 416.08600144946297,
+      "learning_rate": 3.4774287859068988e-06,
+      "loss": 209.9914,
+      "step": 37900
+    },
+    {
+      "epoch": 0.729105062481669,
+      "grad_norm": 412.40189661700816,
+      "learning_rate": 3.472803832374263e-06,
+      "loss": 207.5467,
+      "step": 37910
+    },
+    {
+      "epoch": 0.7292973877421495,
+      "grad_norm": 378.48672886471013,
+      "learning_rate": 3.4681813101144e-06,
+      "loss": 215.2839,
+      "step": 37920
+    },
+    {
+      "epoch": 0.7294897130026301,
+      "grad_norm": 369.0745295431209,
+      "learning_rate": 3.4635612208491197e-06,
+      "loss": 217.8814,
+      "step": 37930
+    },
+    {
+      "epoch": 0.7296820382631106,
+      "grad_norm": 365.8571157845644,
+      "learning_rate": 3.458943566299334e-06,
+      "loss": 205.6632,
+      "step": 37940
+    },
+    {
+      "epoch": 0.7298743635235911,
+      "grad_norm": 396.1648681981586,
+      "learning_rate": 3.454328348185042e-06,
+      "loss": 206.6167,
+      "step": 37950
+    },
+    {
+      "epoch": 0.7300666887840717,
+      "grad_norm": 420.6619390886531,
+      "learning_rate": 3.4497155682253314e-06,
+      "loss": 220.5103,
+      "step": 37960
+    },
+    {
+      "epoch": 0.7302590140445522,
+      "grad_norm": 370.7848964152775,
+      "learning_rate": 3.4451052281383922e-06,
+      "loss": 208.4737,
+      "step": 37970
+    },
+    {
+      "epoch": 0.7304513393050327,
+      "grad_norm": 436.12107681847715,
+      "learning_rate": 3.440497329641499e-06,
+      "loss": 217.1044,
+      "step": 37980
+    },
+    {
+      "epoch": 0.7306436645655132,
+      "grad_norm": 405.2681207341782,
+      "learning_rate": 3.435891874451017e-06,
+      "loss": 211.4854,
+      "step": 37990
+    },
+    {
+      "epoch": 0.7308359898259937,
+      "grad_norm": 401.76486804390674,
+      "learning_rate": 3.431288864282398e-06,
+      "loss": 211.495,
+      "step": 38000
+    },
+    {
+      "epoch": 0.7310283150864743,
+      "grad_norm": 393.80215459274297,
+      "learning_rate": 3.4266883008501937e-06,
+      "loss": 208.8307,
+      "step": 38010
+    },
+    {
+      "epoch": 0.7312206403469548,
+      "grad_norm": 403.0547242021537,
+      "learning_rate": 3.4220901858680365e-06,
+      "loss": 219.4676,
+      "step": 38020
+    },
+    {
+      "epoch": 0.7314129656074353,
+      "grad_norm": 386.04037789868806,
+      "learning_rate": 3.4174945210486445e-06,
+      "loss": 216.0579,
+      "step": 38030
+    },
+    {
+      "epoch": 0.7316052908679158,
+      "grad_norm": 371.44900781854693,
+      "learning_rate": 3.4129013081038285e-06,
+      "loss": 216.7225,
+      "step": 38040
+    },
+    {
+      "epoch": 0.7317976161283963,
+      "grad_norm": 372.5616089910205,
+      "learning_rate": 3.40831054874449e-06,
+      "loss": 212.3076,
+      "step": 38050
+    },
+    {
+      "epoch": 0.7319899413888769,
+      "grad_norm": 390.34243525406276,
+      "learning_rate": 3.403722244680606e-06,
+      "loss": 211.6561,
+      "step": 38060
+    },
+    {
+      "epoch": 0.7321822666493574,
+      "grad_norm": 399.363061548625,
+      "learning_rate": 3.3991363976212423e-06,
+      "loss": 221.2642,
+      "step": 38070
+    },
+    {
+      "epoch": 0.7323745919098379,
+      "grad_norm": 390.86162224395724,
+      "learning_rate": 3.394553009274556e-06,
+      "loss": 208.7783,
+      "step": 38080
+    },
+    {
+      "epoch": 0.7325669171703184,
+      "grad_norm": 376.030082914757,
+      "learning_rate": 3.389972081347782e-06,
+      "loss": 211.4085,
+      "step": 38090
+    },
+    {
+      "epoch": 0.7327592424307989,
+      "grad_norm": 421.28527494798163,
+      "learning_rate": 3.385393615547239e-06,
+      "loss": 221.0359,
+      "step": 38100
+    },
+    {
+      "epoch": 0.7329515676912794,
+      "grad_norm": 385.6668608487081,
+      "learning_rate": 3.3808176135783276e-06,
+      "loss": 225.6684,
+      "step": 38110
+    },
+    {
+      "epoch": 0.73314389295176,
+      "grad_norm": 395.77734220139604,
+      "learning_rate": 3.3762440771455386e-06,
+      "loss": 217.6572,
+      "step": 38120
+    },
+    {
+      "epoch": 0.7333362182122405,
+      "grad_norm": 360.22850059152927,
+      "learning_rate": 3.371673007952435e-06,
+      "loss": 211.7321,
+      "step": 38130
+    },
+    {
+      "epoch": 0.7335285434727211,
+      "grad_norm": 451.19564657478725,
+      "learning_rate": 3.3671044077016634e-06,
+      "loss": 223.9502,
+      "step": 38140
+    },
+    {
+      "epoch": 0.7337208687332016,
+      "grad_norm": 377.72615175686684,
+      "learning_rate": 3.3625382780949576e-06,
+      "loss": 217.7665,
+      "step": 38150
+    },
+    {
+      "epoch": 0.7339131939936822,
+      "grad_norm": 415.1236291122088,
+      "learning_rate": 3.357974620833121e-06,
+      "loss": 216.6595,
+      "step": 38160
+    },
+    {
+      "epoch": 0.7341055192541627,
+      "grad_norm": 400.53581474755936,
+      "learning_rate": 3.353413437616039e-06,
+      "loss": 218.0245,
+      "step": 38170
+    },
+    {
+      "epoch": 0.7342978445146432,
+      "grad_norm": 404.45658003650885,
+      "learning_rate": 3.3488547301426786e-06,
+      "loss": 203.9033,
+      "step": 38180
+    },
+    {
+      "epoch": 0.7344901697751237,
+      "grad_norm": 405.42808882473145,
+      "learning_rate": 3.344298500111087e-06,
+      "loss": 213.4934,
+      "step": 38190
+    },
+    {
+      "epoch": 0.7346824950356042,
+      "grad_norm": 385.6353083140758,
+      "learning_rate": 3.3397447492183833e-06,
+      "loss": 213.109,
+      "step": 38200
+    },
+    {
+      "epoch": 0.7348748202960848,
+      "grad_norm": 395.50540149928486,
+      "learning_rate": 3.3351934791607576e-06,
+      "loss": 215.3671,
+      "step": 38210
+    },
+    {
+      "epoch": 0.7350671455565653,
+      "grad_norm": 396.4688833152432,
+      "learning_rate": 3.330644691633492e-06,
+      "loss": 213.5135,
+      "step": 38220
+    },
+    {
+      "epoch": 0.7352594708170458,
+      "grad_norm": 416.1352364224095,
+      "learning_rate": 3.3260983883309306e-06,
+      "loss": 220.9658,
+      "step": 38230
+    },
+    {
+      "epoch": 0.7354517960775263,
+      "grad_norm": 396.2827182497824,
+      "learning_rate": 3.321554570946497e-06,
+      "loss": 211.0146,
+      "step": 38240
+    },
+    {
+      "epoch": 0.7356441213380068,
+      "grad_norm": 392.59965837030563,
+      "learning_rate": 3.317013241172684e-06,
+      "loss": 213.7918,
+      "step": 38250
+    },
+    {
+      "epoch": 0.7358364465984873,
+      "grad_norm": 377.3783215380771,
+      "learning_rate": 3.3124744007010688e-06,
+      "loss": 211.642,
+      "step": 38260
+    },
+    {
+      "epoch": 0.7360287718589679,
+      "grad_norm": 407.9626386487798,
+      "learning_rate": 3.3079380512222904e-06,
+      "loss": 213.8329,
+      "step": 38270
+    },
+    {
+      "epoch": 0.7362210971194484,
+      "grad_norm": 399.81884197970874,
+      "learning_rate": 3.3034041944260654e-06,
+      "loss": 214.722,
+      "step": 38280
+    },
+    {
+      "epoch": 0.7364134223799289,
+      "grad_norm": 400.74149208559845,
+      "learning_rate": 3.2988728320011774e-06,
+      "loss": 211.8562,
+      "step": 38290
+    },
+    {
+      "epoch": 0.7366057476404094,
+      "grad_norm": 398.42200787287084,
+      "learning_rate": 3.294343965635489e-06,
+      "loss": 212.6863,
+      "step": 38300
+    },
+    {
+      "epoch": 0.7367980729008899,
+      "grad_norm": 360.8047174002981,
+      "learning_rate": 3.289817597015923e-06,
+      "loss": 206.4632,
+      "step": 38310
+    },
+    {
+      "epoch": 0.7369903981613705,
+      "grad_norm": 420.24950274864705,
+      "learning_rate": 3.2852937278284837e-06,
+      "loss": 216.2751,
+      "step": 38320
+    },
+    {
+      "epoch": 0.737182723421851,
+      "grad_norm": 413.1067219285103,
+      "learning_rate": 3.28077235975823e-06,
+      "loss": 212.4115,
+      "step": 38330
+    },
+    {
+      "epoch": 0.7373750486823316,
+      "grad_norm": 393.10881815630324,
+      "learning_rate": 3.2762534944893033e-06,
+      "loss": 216.4406,
+      "step": 38340
+    },
+    {
+      "epoch": 0.7375673739428121,
+      "grad_norm": 362.0353279532254,
+      "learning_rate": 3.271737133704904e-06,
+      "loss": 208.488,
+      "step": 38350
+    },
+    {
+      "epoch": 0.7377596992032927,
+      "grad_norm": 409.13149928697777,
+      "learning_rate": 3.2672232790872983e-06,
+      "loss": 214.1322,
+      "step": 38360
+    },
+    {
+      "epoch": 0.7379520244637732,
+      "grad_norm": 398.8009694634766,
+      "learning_rate": 3.262711932317828e-06,
+      "loss": 216.5977,
+      "step": 38370
+    },
+    {
+      "epoch": 0.7381443497242537,
+      "grad_norm": 397.3899555932812,
+      "learning_rate": 3.258203095076894e-06,
+      "loss": 212.8879,
+      "step": 38380
+    },
+    {
+      "epoch": 0.7383366749847342,
+      "grad_norm": 423.94228008383357,
+      "learning_rate": 3.2536967690439592e-06,
+      "loss": 222.6395,
+      "step": 38390
+    },
+    {
+      "epoch": 0.7385290002452147,
+      "grad_norm": 366.08395855529665,
+      "learning_rate": 3.249192955897562e-06,
+      "loss": 213.8753,
+      "step": 38400
+    },
+    {
+      "epoch": 0.7387213255056952,
+      "grad_norm": 382.8523485325907,
+      "learning_rate": 3.2446916573152955e-06,
+      "loss": 214.5685,
+      "step": 38410
+    },
+    {
+      "epoch": 0.7389136507661758,
+      "grad_norm": 370.92386982816544,
+      "learning_rate": 3.24019287497382e-06,
+      "loss": 210.0728,
+      "step": 38420
+    },
+    {
+      "epoch": 0.7391059760266563,
+      "grad_norm": 386.27770105026764,
+      "learning_rate": 3.235696610548852e-06,
+      "loss": 206.765,
+      "step": 38430
+    },
+    {
+      "epoch": 0.7392983012871368,
+      "grad_norm": 377.6579108413092,
+      "learning_rate": 3.231202865715184e-06,
+      "loss": 215.869,
+      "step": 38440
+    },
+    {
+      "epoch": 0.7394906265476173,
+      "grad_norm": 364.18110165419216,
+      "learning_rate": 3.226711642146655e-06,
+      "loss": 221.5153,
+      "step": 38450
+    },
+    {
+      "epoch": 0.7396829518080978,
+      "grad_norm": 366.34314623436023,
+      "learning_rate": 3.222222941516179e-06,
+      "loss": 211.9042,
+      "step": 38460
+    },
+    {
+      "epoch": 0.7398752770685784,
+      "grad_norm": 372.8199171418252,
+      "learning_rate": 3.2177367654957137e-06,
+      "loss": 205.8132,
+      "step": 38470
+    },
+    {
+      "epoch": 0.7400676023290589,
+      "grad_norm": 412.9839015956195,
+      "learning_rate": 3.213253115756295e-06,
+      "loss": 213.2388,
+      "step": 38480
+    },
+    {
+      "epoch": 0.7402599275895394,
+      "grad_norm": 384.3722887546716,
+      "learning_rate": 3.208771993968003e-06,
+      "loss": 213.9061,
+      "step": 38490
+    },
+    {
+      "epoch": 0.7404522528500199,
+      "grad_norm": 432.2626937087747,
+      "learning_rate": 3.2042934017999795e-06,
+      "loss": 209.8503,
+      "step": 38500
+    },
+    {
+      "epoch": 0.7406445781105004,
+      "grad_norm": 368.4995964807424,
+      "learning_rate": 3.1998173409204326e-06,
+      "loss": 211.2958,
+      "step": 38510
+    },
+    {
+      "epoch": 0.740836903370981,
+      "grad_norm": 411.90924050817864,
+      "learning_rate": 3.1953438129966175e-06,
+      "loss": 221.98,
+      "step": 38520
+    },
+    {
+      "epoch": 0.7410292286314615,
+      "grad_norm": 399.9467861844467,
+      "learning_rate": 3.190872819694849e-06,
+      "loss": 220.5172,
+      "step": 38530
+    },
+    {
+      "epoch": 0.741221553891942,
+      "grad_norm": 362.32565209940924,
+      "learning_rate": 3.1864043626804953e-06,
+      "loss": 207.5476,
+      "step": 38540
+    },
+    {
+      "epoch": 0.7414138791524226,
+      "grad_norm": 387.2613273555262,
+      "learning_rate": 3.18193844361799e-06,
+      "loss": 206.5205,
+      "step": 38550
+    },
+    {
+      "epoch": 0.7416062044129031,
+      "grad_norm": 381.62547798058756,
+      "learning_rate": 3.1774750641708095e-06,
+      "loss": 209.4551,
+      "step": 38560
+    },
+    {
+      "epoch": 0.7417985296733837,
+      "grad_norm": 400.43162824113284,
+      "learning_rate": 3.1730142260014875e-06,
+      "loss": 213.3418,
+      "step": 38570
+    },
+    {
+      "epoch": 0.7419908549338642,
+      "grad_norm": 399.7989303182746,
+      "learning_rate": 3.1685559307716187e-06,
+      "loss": 210.5069,
+      "step": 38580
+    },
+    {
+      "epoch": 0.7421831801943447,
+      "grad_norm": 404.96576419568987,
+      "learning_rate": 3.164100180141839e-06,
+      "loss": 212.9981,
+      "step": 38590
+    },
+    {
+      "epoch": 0.7423755054548252,
+      "grad_norm": 399.1935336954331,
+      "learning_rate": 3.159646975771842e-06,
+      "loss": 216.6827,
+      "step": 38600
+    },
+    {
+      "epoch": 0.7425678307153057,
+      "grad_norm": 414.73239676081874,
+      "learning_rate": 3.155196319320374e-06,
+      "loss": 220.2067,
+      "step": 38610
+    },
+    {
+      "epoch": 0.7427601559757863,
+      "grad_norm": 407.57386494826847,
+      "learning_rate": 3.1507482124452337e-06,
+      "loss": 217.2531,
+      "step": 38620
+    },
+    {
+      "epoch": 0.7429524812362668,
+      "grad_norm": 401.86931011984944,
+      "learning_rate": 3.146302656803266e-06,
+      "loss": 209.7045,
+      "step": 38630
+    },
+    {
+      "epoch": 0.7431448064967473,
+      "grad_norm": 388.65518628770843,
+      "learning_rate": 3.1418596540503653e-06,
+      "loss": 209.4193,
+      "step": 38640
+    },
+    {
+      "epoch": 0.7433371317572278,
+      "grad_norm": 404.19009529074054,
+      "learning_rate": 3.1374192058414755e-06,
+      "loss": 220.9851,
+      "step": 38650
+    },
+    {
+      "epoch": 0.7435294570177083,
+      "grad_norm": 392.35560514186375,
+      "learning_rate": 3.1329813138305944e-06,
+      "loss": 216.9903,
+      "step": 38660
+    },
+    {
+      "epoch": 0.7437217822781889,
+      "grad_norm": 378.99140078284177,
+      "learning_rate": 3.128545979670762e-06,
+      "loss": 205.9095,
+      "step": 38670
+    },
+    {
+      "epoch": 0.7439141075386694,
+      "grad_norm": 356.97991826053993,
+      "learning_rate": 3.124113205014063e-06,
+      "loss": 209.2848,
+      "step": 38680
+    },
+    {
+      "epoch": 0.7441064327991499,
+      "grad_norm": 409.0252085946919,
+      "learning_rate": 3.119682991511639e-06,
+      "loss": 214.9606,
+      "step": 38690
+    },
+    {
+      "epoch": 0.7442987580596304,
+      "grad_norm": 376.6255874008177,
+      "learning_rate": 3.1152553408136686e-06,
+      "loss": 211.097,
+      "step": 38700
+    },
+    {
+      "epoch": 0.7444910833201109,
+      "grad_norm": 423.33406546829696,
+      "learning_rate": 3.110830254569378e-06,
+      "loss": 214.8367,
+      "step": 38710
+    },
+    {
+      "epoch": 0.7446834085805915,
+      "grad_norm": 376.6874226133999,
+      "learning_rate": 3.106407734427037e-06,
+      "loss": 214.7831,
+      "step": 38720
+    },
+    {
+      "epoch": 0.744875733841072,
+      "grad_norm": 418.0899045821614,
+      "learning_rate": 3.101987782033966e-06,
+      "loss": 214.9129,
+      "step": 38730
+    },
+    {
+      "epoch": 0.7450680591015525,
+      "grad_norm": 400.30569583700344,
+      "learning_rate": 3.097570399036519e-06,
+      "loss": 213.6869,
+      "step": 38740
+    },
+    {
+      "epoch": 0.7452603843620331,
+      "grad_norm": 383.42602546053075,
+      "learning_rate": 3.0931555870801033e-06,
+      "loss": 206.9788,
+      "step": 38750
+    },
+    {
+      "epoch": 0.7454527096225136,
+      "grad_norm": 356.1900325540024,
+      "learning_rate": 3.0887433478091587e-06,
+      "loss": 210.5475,
+      "step": 38760
+    },
+    {
+      "epoch": 0.7456450348829942,
+      "grad_norm": 367.03826476126704,
+      "learning_rate": 3.0843336828671765e-06,
+      "loss": 211.6151,
+      "step": 38770
+    },
+    {
+      "epoch": 0.7458373601434747,
+      "grad_norm": 413.7632781708218,
+      "learning_rate": 3.079926593896683e-06,
+      "loss": 210.51,
+      "step": 38780
+    },
+    {
+      "epoch": 0.7460296854039552,
+      "grad_norm": 399.30631891094725,
+      "learning_rate": 3.0755220825392397e-06,
+      "loss": 208.5259,
+      "step": 38790
+    },
+    {
+      "epoch": 0.7462220106644357,
+      "grad_norm": 405.0131366510525,
+      "learning_rate": 3.0711201504354628e-06,
+      "loss": 212.0677,
+      "step": 38800
+    },
+    {
+      "epoch": 0.7464143359249162,
+      "grad_norm": 360.94292117850114,
+      "learning_rate": 3.0667207992249948e-06,
+      "loss": 214.5323,
+      "step": 38810
+    },
+    {
+      "epoch": 0.7466066611853968,
+      "grad_norm": 385.09193853354145,
+      "learning_rate": 3.062324030546523e-06,
+      "loss": 213.5694,
+      "step": 38820
+    },
+    {
+      "epoch": 0.7467989864458773,
+      "grad_norm": 398.3778364003248,
+      "learning_rate": 3.057929846037767e-06,
+      "loss": 209.2806,
+      "step": 38830
+    },
+    {
+      "epoch": 0.7469913117063578,
+      "grad_norm": 742.5675863119249,
+      "learning_rate": 3.0535382473354945e-06,
+      "loss": 214.8567,
+      "step": 38840
+    },
+    {
+      "epoch": 0.7471836369668383,
+      "grad_norm": 408.47781024558793,
+      "learning_rate": 3.0491492360755003e-06,
+      "loss": 214.0556,
+      "step": 38850
+    },
+    {
+      "epoch": 0.7473759622273188,
+      "grad_norm": 421.4223228558872,
+      "learning_rate": 3.0447628138926153e-06,
+      "loss": 205.1925,
+      "step": 38860
+    },
+    {
+      "epoch": 0.7475682874877994,
+      "grad_norm": 402.6709587511956,
+      "learning_rate": 3.0403789824207165e-06,
+      "loss": 206.5705,
+      "step": 38870
+    },
+    {
+      "epoch": 0.7477606127482799,
+      "grad_norm": 399.43451685824147,
+      "learning_rate": 3.0359977432927013e-06,
+      "loss": 209.7825,
+      "step": 38880
+    },
+    {
+      "epoch": 0.7479529380087604,
+      "grad_norm": 378.2326058062348,
+      "learning_rate": 3.0316190981405147e-06,
+      "loss": 209.3171,
+      "step": 38890
+    },
+    {
+      "epoch": 0.7481452632692409,
+      "grad_norm": 383.2054001914671,
+      "learning_rate": 3.0272430485951244e-06,
+      "loss": 205.747,
+      "step": 38900
+    },
+    {
+      "epoch": 0.7483375885297214,
+      "grad_norm": 381.85301956063654,
+      "learning_rate": 3.0228695962865438e-06,
+      "loss": 213.7018,
+      "step": 38910
+    },
+    {
+      "epoch": 0.748529913790202,
+      "grad_norm": 381.84347906420965,
+      "learning_rate": 3.018498742843806e-06,
+      "loss": 215.2569,
+      "step": 38920
+    },
+    {
+      "epoch": 0.7487222390506825,
+      "grad_norm": 389.72086948502,
+      "learning_rate": 3.014130489894982e-06,
+      "loss": 212.5456,
+      "step": 38930
+    },
+    {
+      "epoch": 0.748914564311163,
+      "grad_norm": 381.69047349620934,
+      "learning_rate": 3.0097648390671765e-06,
+      "loss": 205.8806,
+      "step": 38940
+    },
+    {
+      "epoch": 0.7491068895716435,
+      "grad_norm": 395.4374140124631,
+      "learning_rate": 3.005401791986522e-06,
+      "loss": 217.4661,
+      "step": 38950
+    },
+    {
+      "epoch": 0.7492992148321241,
+      "grad_norm": 412.10893285433343,
+      "learning_rate": 3.00104135027818e-06,
+      "loss": 216.992,
+      "step": 38960
+    },
+    {
+      "epoch": 0.7494915400926047,
+      "grad_norm": 384.5676629741645,
+      "learning_rate": 2.99668351556634e-06,
+      "loss": 215.7178,
+      "step": 38970
+    },
+    {
+      "epoch": 0.7496838653530852,
+      "grad_norm": 387.03736723957786,
+      "learning_rate": 2.99232828947423e-06,
+      "loss": 215.4647,
+      "step": 38980
+    },
+    {
+      "epoch": 0.7498761906135657,
+      "grad_norm": 394.12633978471496,
+      "learning_rate": 2.987975673624096e-06,
+      "loss": 208.3384,
+      "step": 38990
+    },
+    {
+      "epoch": 0.7500685158740462,
+      "grad_norm": 406.5301865142197,
+      "learning_rate": 2.9836256696372178e-06,
+      "loss": 216.6155,
+      "step": 39000
+    },
+    {
+      "epoch": 0.7502608411345267,
+      "grad_norm": 382.50716260252875,
+      "learning_rate": 2.9792782791338936e-06,
+      "loss": 213.395,
+      "step": 39010
+    },
+    {
+      "epoch": 0.7504531663950073,
+      "grad_norm": 385.7269047709961,
+      "learning_rate": 2.9749335037334604e-06,
+      "loss": 214.8669,
+      "step": 39020
+    },
+    {
+      "epoch": 0.7506454916554878,
+      "grad_norm": 375.7249782620801,
+      "learning_rate": 2.9705913450542777e-06,
+      "loss": 207.3446,
+      "step": 39030
+    },
+    {
+      "epoch": 0.7508378169159683,
+      "grad_norm": 395.05859468248786,
+      "learning_rate": 2.9662518047137214e-06,
+      "loss": 209.4555,
+      "step": 39040
+    },
+    {
+      "epoch": 0.7510301421764488,
+      "grad_norm": 436.94216794977086,
+      "learning_rate": 2.961914884328203e-06,
+      "loss": 213.5024,
+      "step": 39050
+    },
+    {
+      "epoch": 0.7512224674369293,
+      "grad_norm": 376.9192390926667,
+      "learning_rate": 2.9575805855131546e-06,
+      "loss": 211.8836,
+      "step": 39060
+    },
+    {
+      "epoch": 0.7514147926974099,
+      "grad_norm": 360.944377478836,
+      "learning_rate": 2.9532489098830274e-06,
+      "loss": 204.9103,
+      "step": 39070
+    },
+    {
+      "epoch": 0.7516071179578904,
+      "grad_norm": 391.2692282582815,
+      "learning_rate": 2.9489198590512967e-06,
+      "loss": 210.6938,
+      "step": 39080
+    },
+    {
+      "epoch": 0.7517994432183709,
+      "grad_norm": 380.3382717877565,
+      "learning_rate": 2.9445934346304706e-06,
+      "loss": 212.9648,
+      "step": 39090
+    },
+    {
+      "epoch": 0.7519917684788514,
+      "grad_norm": 375.75944537292645,
+      "learning_rate": 2.940269638232065e-06,
+      "loss": 211.1945,
+      "step": 39100
+    },
+    {
+      "epoch": 0.7521840937393319,
+      "grad_norm": 370.8254249722766,
+      "learning_rate": 2.935948471466622e-06,
+      "loss": 208.9388,
+      "step": 39110
+    },
+    {
+      "epoch": 0.7523764189998124,
+      "grad_norm": 375.93976811394225,
+      "learning_rate": 2.9316299359437085e-06,
+      "loss": 210.7232,
+      "step": 39120
+    },
+    {
+      "epoch": 0.752568744260293,
+      "grad_norm": 388.6608756366859,
+      "learning_rate": 2.9273140332719064e-06,
+      "loss": 209.3444,
+      "step": 39130
+    },
+    {
+      "epoch": 0.7527610695207735,
+      "grad_norm": 394.8411038032195,
+      "learning_rate": 2.923000765058818e-06,
+      "loss": 211.9417,
+      "step": 39140
+    },
+    {
+      "epoch": 0.752953394781254,
+      "grad_norm": 357.8108760822515,
+      "learning_rate": 2.9186901329110605e-06,
+      "loss": 206.0361,
+      "step": 39150
+    },
+    {
+      "epoch": 0.7531457200417346,
+      "grad_norm": 411.33338976079426,
+      "learning_rate": 2.9143821384342808e-06,
+      "loss": 208.3241,
+      "step": 39160
+    },
+    {
+      "epoch": 0.7533380453022152,
+      "grad_norm": 406.8763755050772,
+      "learning_rate": 2.9100767832331277e-06,
+      "loss": 209.0823,
+      "step": 39170
+    },
+    {
+      "epoch": 0.7535303705626957,
+      "grad_norm": 404.6217829369774,
+      "learning_rate": 2.9057740689112822e-06,
+      "loss": 216.5944,
+      "step": 39180
+    },
+    {
+      "epoch": 0.7537226958231762,
+      "grad_norm": 395.94485031572174,
+      "learning_rate": 2.901473997071428e-06,
+      "loss": 210.6494,
+      "step": 39190
+    },
+    {
+      "epoch": 0.7539150210836567,
+      "grad_norm": 364.8912713794418,
+      "learning_rate": 2.8971765693152767e-06,
+      "loss": 209.9009,
+      "step": 39200
+    },
+    {
+      "epoch": 0.7541073463441372,
+      "grad_norm": 368.61475603258856,
+      "learning_rate": 2.8928817872435465e-06,
+      "loss": 209.4373,
+      "step": 39210
+    },
+    {
+      "epoch": 0.7542996716046177,
+      "grad_norm": 383.7738753542534,
+      "learning_rate": 2.8885896524559696e-06,
+      "loss": 213.8918,
+      "step": 39220
+    },
+    {
+      "epoch": 0.7544919968650983,
+      "grad_norm": 396.9673238524041,
+      "learning_rate": 2.8843001665513016e-06,
+      "loss": 211.1548,
+      "step": 39230
+    },
+    {
+      "epoch": 0.7546843221255788,
+      "grad_norm": 398.0556002414846,
+      "learning_rate": 2.8800133311273016e-06,
+      "loss": 217.4401,
+      "step": 39240
+    },
+    {
+      "epoch": 0.7548766473860593,
+      "grad_norm": 407.99467701478494,
+      "learning_rate": 2.875729147780745e-06,
+      "loss": 214.796,
+      "step": 39250
+    },
+    {
+      "epoch": 0.7550689726465398,
+      "grad_norm": 430.7337783857894,
+      "learning_rate": 2.871447618107417e-06,
+      "loss": 210.8239,
+      "step": 39260
+    },
+    {
+      "epoch": 0.7552612979070203,
+      "grad_norm": 395.9840164375906,
+      "learning_rate": 2.867168743702122e-06,
+      "loss": 206.1601,
+      "step": 39270
+    },
+    {
+      "epoch": 0.7554536231675009,
+      "grad_norm": 409.001009160928,
+      "learning_rate": 2.8628925261586683e-06,
+      "loss": 216.4979,
+      "step": 39280
+    },
+    {
+      "epoch": 0.7556459484279814,
+      "grad_norm": 430.96511143901336,
+      "learning_rate": 2.8586189670698717e-06,
+      "loss": 212.8584,
+      "step": 39290
+    },
+    {
+      "epoch": 0.7558382736884619,
+      "grad_norm": 369.7596818222909,
+      "learning_rate": 2.854348068027568e-06,
+      "loss": 209.4182,
+      "step": 39300
+    },
+    {
+      "epoch": 0.7560305989489424,
+      "grad_norm": 402.3889795623061,
+      "learning_rate": 2.850079830622593e-06,
+      "loss": 215.6611,
+      "step": 39310
+    },
+    {
+      "epoch": 0.7562229242094229,
+      "grad_norm": 398.99672659650656,
+      "learning_rate": 2.845814256444799e-06,
+      "loss": 221.0187,
+      "step": 39320
+    },
+    {
+      "epoch": 0.7564152494699035,
+      "grad_norm": 367.41047924702616,
+      "learning_rate": 2.8415513470830357e-06,
+      "loss": 218.1719,
+      "step": 39330
+    },
+    {
+      "epoch": 0.756607574730384,
+      "grad_norm": 406.3223290528226,
+      "learning_rate": 2.837291104125174e-06,
+      "loss": 215.5918,
+      "step": 39340
+    },
+    {
+      "epoch": 0.7567998999908645,
+      "grad_norm": 363.89980218495265,
+      "learning_rate": 2.833033529158079e-06,
+      "loss": 208.7063,
+      "step": 39350
+    },
+    {
+      "epoch": 0.7569922252513451,
+      "grad_norm": 400.34670914874675,
+      "learning_rate": 2.8287786237676253e-06,
+      "loss": 219.3123,
+      "step": 39360
+    },
+    {
+      "epoch": 0.7571845505118256,
+      "grad_norm": 389.73127859276985,
+      "learning_rate": 2.824526389538701e-06,
+      "loss": 216.983,
+      "step": 39370
+    },
+    {
+      "epoch": 0.7573768757723062,
+      "grad_norm": 385.61119778823263,
+      "learning_rate": 2.8202768280551894e-06,
+      "loss": 208.8682,
+      "step": 39380
+    },
+    {
+      "epoch": 0.7575692010327867,
+      "grad_norm": 402.8467762543396,
+      "learning_rate": 2.8160299408999827e-06,
+      "loss": 219.3151,
+      "step": 39390
+    },
+    {
+      "epoch": 0.7577615262932672,
+      "grad_norm": 388.6695812007182,
+      "learning_rate": 2.811785729654972e-06,
+      "loss": 213.4402,
+      "step": 39400
+    },
+    {
+      "epoch": 0.7579538515537477,
+      "grad_norm": 433.95926371201625,
+      "learning_rate": 2.8075441959010628e-06,
+      "loss": 209.4541,
+      "step": 39410
+    },
+    {
+      "epoch": 0.7581461768142282,
+      "grad_norm": 389.98800289470756,
+      "learning_rate": 2.8033053412181543e-06,
+      "loss": 209.9189,
+      "step": 39420
+    },
+    {
+      "epoch": 0.7583385020747088,
+      "grad_norm": 350.80924189818757,
+      "learning_rate": 2.799069167185148e-06,
+      "loss": 199.8208,
+      "step": 39430
+    },
+    {
+      "epoch": 0.7585308273351893,
+      "grad_norm": 390.00383181463206,
+      "learning_rate": 2.7948356753799466e-06,
+      "loss": 219.3106,
+      "step": 39440
+    },
+    {
+      "epoch": 0.7587231525956698,
+      "grad_norm": 389.23729622559796,
+      "learning_rate": 2.7906048673794593e-06,
+      "loss": 209.4522,
+      "step": 39450
+    },
+    {
+      "epoch": 0.7589154778561503,
+      "grad_norm": 372.12571019976934,
+      "learning_rate": 2.7863767447595946e-06,
+      "loss": 215.2538,
+      "step": 39460
+    },
+    {
+      "epoch": 0.7591078031166308,
+      "grad_norm": 368.5555867845163,
+      "learning_rate": 2.7821513090952523e-06,
+      "loss": 208.9072,
+      "step": 39470
+    },
+    {
+      "epoch": 0.7593001283771114,
+      "grad_norm": 400.9093183377035,
+      "learning_rate": 2.7779285619603446e-06,
+      "loss": 211.3843,
+      "step": 39480
+    },
+    {
+      "epoch": 0.7594924536375919,
+      "grad_norm": 362.03232664261355,
+      "learning_rate": 2.77370850492777e-06,
+      "loss": 224.0575,
+      "step": 39490
+    },
+    {
+      "epoch": 0.7596847788980724,
+      "grad_norm": 412.70488746760253,
+      "learning_rate": 2.7694911395694324e-06,
+      "loss": 212.1221,
+      "step": 39500
+    },
+    {
+      "epoch": 0.7598771041585529,
+      "grad_norm": 383.4012591281291,
+      "learning_rate": 2.765276467456225e-06,
+      "loss": 215.0577,
+      "step": 39510
+    },
+    {
+      "epoch": 0.7600694294190334,
+      "grad_norm": 426.3263710762801,
+      "learning_rate": 2.761064490158052e-06,
+      "loss": 213.2636,
+      "step": 39520
+    },
+    {
+      "epoch": 0.760261754679514,
+      "grad_norm": 379.15038772281486,
+      "learning_rate": 2.7568552092438018e-06,
+      "loss": 211.3147,
+      "step": 39530
+    },
+    {
+      "epoch": 0.7604540799399945,
+      "grad_norm": 361.5405567790523,
+      "learning_rate": 2.7526486262813578e-06,
+      "loss": 209.4867,
+      "step": 39540
+    },
+    {
+      "epoch": 0.760646405200475,
+      "grad_norm": 406.0292213017384,
+      "learning_rate": 2.7484447428376094e-06,
+      "loss": 204.6452,
+      "step": 39550
+    },
+    {
+      "epoch": 0.7608387304609555,
+      "grad_norm": 396.3097126086856,
+      "learning_rate": 2.7442435604784313e-06,
+      "loss": 213.2015,
+      "step": 39560
+    },
+    {
+      "epoch": 0.7610310557214361,
+      "grad_norm": 418.29638892861743,
+      "learning_rate": 2.740045080768694e-06,
+      "loss": 211.9402,
+      "step": 39570
+    },
+    {
+      "epoch": 0.7612233809819167,
+      "grad_norm": 379.33138573699574,
+      "learning_rate": 2.7358493052722603e-06,
+      "loss": 208.9075,
+      "step": 39580
+    },
+    {
+      "epoch": 0.7614157062423972,
+      "grad_norm": 438.1353530896628,
+      "learning_rate": 2.7316562355519904e-06,
+      "loss": 209.6229,
+      "step": 39590
+    },
+    {
+      "epoch": 0.7616080315028777,
+      "grad_norm": 429.43310704191356,
+      "learning_rate": 2.727465873169729e-06,
+      "loss": 211.5358,
+      "step": 39600
+    },
+    {
+      "epoch": 0.7618003567633582,
+      "grad_norm": 375.30528783560396,
+      "learning_rate": 2.723278219686324e-06,
+      "loss": 211.6839,
+      "step": 39610
+    },
+    {
+      "epoch": 0.7619926820238387,
+      "grad_norm": 389.7426065337725,
+      "learning_rate": 2.7190932766615998e-06,
+      "loss": 213.2397,
+      "step": 39620
+    },
+    {
+      "epoch": 0.7621850072843193,
+      "grad_norm": 361.26565453677784,
+      "learning_rate": 2.714911045654385e-06,
+      "loss": 205.8367,
+      "step": 39630
+    },
+    {
+      "epoch": 0.7623773325447998,
+      "grad_norm": 410.2427715654697,
+      "learning_rate": 2.7107315282224878e-06,
+      "loss": 209.0607,
+      "step": 39640
+    },
+    {
+      "epoch": 0.7625696578052803,
+      "grad_norm": 389.6081665585904,
+      "learning_rate": 2.7065547259227078e-06,
+      "loss": 213.5682,
+      "step": 39650
+    },
+    {
+      "epoch": 0.7627619830657608,
+      "grad_norm": 414.38171683702114,
+      "learning_rate": 2.7023806403108397e-06,
+      "loss": 215.9959,
+      "step": 39660
+    },
+    {
+      "epoch": 0.7629543083262413,
+      "grad_norm": 373.52675098944684,
+      "learning_rate": 2.698209272941659e-06,
+      "loss": 206.277,
+      "step": 39670
+    },
+    {
+      "epoch": 0.7631466335867219,
+      "grad_norm": 427.9547862737436,
+      "learning_rate": 2.694040625368931e-06,
+      "loss": 220.976,
+      "step": 39680
+    },
+    {
+      "epoch": 0.7633389588472024,
+      "grad_norm": 394.99758024115886,
+      "learning_rate": 2.689874699145405e-06,
+      "loss": 208.6515,
+      "step": 39690
+    },
+    {
+      "epoch": 0.7635312841076829,
+      "grad_norm": 372.7943856847385,
+      "learning_rate": 2.685711495822827e-06,
+      "loss": 208.8916,
+      "step": 39700
+    },
+    {
+      "epoch": 0.7637236093681634,
+      "grad_norm": 395.39271770215197,
+      "learning_rate": 2.6815510169519164e-06,
+      "loss": 207.7454,
+      "step": 39710
+    },
+    {
+      "epoch": 0.7639159346286439,
+      "grad_norm": 386.7147699847869,
+      "learning_rate": 2.677393264082381e-06,
+      "loss": 220.2939,
+      "step": 39720
+    },
+    {
+      "epoch": 0.7641082598891245,
+      "grad_norm": 352.90171295598793,
+      "learning_rate": 2.673238238762921e-06,
+      "loss": 215.5709,
+      "step": 39730
+    },
+    {
+      "epoch": 0.764300585149605,
+      "grad_norm": 412.79674189734624,
+      "learning_rate": 2.6690859425412075e-06,
+      "loss": 211.8218,
+      "step": 39740
+    },
+    {
+      "epoch": 0.7644929104100855,
+      "grad_norm": 440.5740240739491,
+      "learning_rate": 2.6649363769639103e-06,
+      "loss": 221.7515,
+      "step": 39750
+    },
+    {
+      "epoch": 0.764685235670566,
+      "grad_norm": 401.37888801809254,
+      "learning_rate": 2.660789543576667e-06,
+      "loss": 211.859,
+      "step": 39760
+    },
+    {
+      "epoch": 0.7648775609310466,
+      "grad_norm": 369.5282264370912,
+      "learning_rate": 2.6566454439241107e-06,
+      "loss": 213.5541,
+      "step": 39770
+    },
+    {
+      "epoch": 0.7650698861915272,
+      "grad_norm": 397.52046623148067,
+      "learning_rate": 2.652504079549848e-06,
+      "loss": 218.0958,
+      "step": 39780
+    },
+    {
+      "epoch": 0.7652622114520077,
+      "grad_norm": 369.68359955230426,
+      "learning_rate": 2.648365451996466e-06,
+      "loss": 213.4677,
+      "step": 39790
+    },
+    {
+      "epoch": 0.7654545367124882,
+      "grad_norm": 358.15209748122936,
+      "learning_rate": 2.6442295628055346e-06,
+      "loss": 205.2239,
+      "step": 39800
+    },
+    {
+      "epoch": 0.7656468619729687,
+      "grad_norm": 382.9311218183089,
+      "learning_rate": 2.64009641351761e-06,
+      "loss": 208.5467,
+      "step": 39810
+    },
+    {
+      "epoch": 0.7658391872334492,
+      "grad_norm": 362.4926086620384,
+      "learning_rate": 2.635966005672218e-06,
+      "loss": 211.9969,
+      "step": 39820
+    },
+    {
+      "epoch": 0.7660315124939298,
+      "grad_norm": 391.845281686503,
+      "learning_rate": 2.631838340807865e-06,
+      "loss": 215.461,
+      "step": 39830
+    },
+    {
+      "epoch": 0.7662238377544103,
+      "grad_norm": 389.38735272616975,
+      "learning_rate": 2.6277134204620436e-06,
+      "loss": 211.6494,
+      "step": 39840
+    },
+    {
+      "epoch": 0.7664161630148908,
+      "grad_norm": 431.8984457834533,
+      "learning_rate": 2.6235912461712167e-06,
+      "loss": 218.689,
+      "step": 39850
+    },
+    {
+      "epoch": 0.7666084882753713,
+      "grad_norm": 402.3679808093642,
+      "learning_rate": 2.619471819470821e-06,
+      "loss": 209.1339,
+      "step": 39860
+    },
+    {
+      "epoch": 0.7668008135358518,
+      "grad_norm": 371.1726689792626,
+      "learning_rate": 2.6153551418952827e-06,
+      "loss": 206.0153,
+      "step": 39870
+    },
+    {
+      "epoch": 0.7669931387963324,
+      "grad_norm": 372.4524033856654,
+      "learning_rate": 2.6112412149779888e-06,
+      "loss": 212.4483,
+      "step": 39880
+    },
+    {
+      "epoch": 0.7671854640568129,
+      "grad_norm": 390.2470029755241,
+      "learning_rate": 2.6071300402513165e-06,
+      "loss": 213.3157,
+      "step": 39890
+    },
+    {
+      "epoch": 0.7673777893172934,
+      "grad_norm": 395.5268433720969,
+      "learning_rate": 2.603021619246604e-06,
+      "loss": 214.1548,
+      "step": 39900
+    },
+    {
+      "epoch": 0.7675701145777739,
+      "grad_norm": 374.38766170504414,
+      "learning_rate": 2.5989159534941768e-06,
+      "loss": 210.0371,
+      "step": 39910
+    },
+    {
+      "epoch": 0.7677624398382544,
+      "grad_norm": 348.97056465125155,
+      "learning_rate": 2.5948130445233232e-06,
+      "loss": 210.1506,
+      "step": 39920
+    },
+    {
+      "epoch": 0.767954765098735,
+      "grad_norm": 385.74405268303934,
+      "learning_rate": 2.5907128938623093e-06,
+      "loss": 211.9513,
+      "step": 39930
+    },
+    {
+      "epoch": 0.7681470903592155,
+      "grad_norm": 423.85196996392256,
+      "learning_rate": 2.5866155030383722e-06,
+      "loss": 217.6935,
+      "step": 39940
+    },
+    {
+      "epoch": 0.768339415619696,
+      "grad_norm": 407.2722872398317,
+      "learning_rate": 2.582520873577726e-06,
+      "loss": 206.3344,
+      "step": 39950
+    },
+    {
+      "epoch": 0.7685317408801765,
+      "grad_norm": 352.25993822206067,
+      "learning_rate": 2.578429007005552e-06,
+      "loss": 210.1658,
+      "step": 39960
+    },
+    {
+      "epoch": 0.768724066140657,
+      "grad_norm": 386.0638546789341,
+      "learning_rate": 2.5743399048460004e-06,
+      "loss": 215.3121,
+      "step": 39970
+    },
+    {
+      "epoch": 0.7689163914011377,
+      "grad_norm": 369.5798666043826,
+      "learning_rate": 2.570253568622193e-06,
+      "loss": 211.6251,
+      "step": 39980
+    },
+    {
+      "epoch": 0.7691087166616182,
+      "grad_norm": 374.4656310115537,
+      "learning_rate": 2.5661699998562286e-06,
+      "loss": 212.1195,
+      "step": 39990
+    },
+    {
+      "epoch": 0.7693010419220987,
+      "grad_norm": 409.6320917581006,
+      "learning_rate": 2.5620892000691643e-06,
+      "loss": 220.7276,
+      "step": 40000
+    },
+    {
+      "epoch": 0.7694933671825792,
+      "grad_norm": 399.7185184456794,
+      "learning_rate": 2.5580111707810296e-06,
+      "loss": 205.4715,
+      "step": 40010
+    },
+    {
+      "epoch": 0.7696856924430597,
+      "grad_norm": 417.77311432139237,
+      "learning_rate": 2.5539359135108244e-06,
+      "loss": 212.5975,
+      "step": 40020
+    },
+    {
+      "epoch": 0.7698780177035403,
+      "grad_norm": 357.71082793451444,
+      "learning_rate": 2.549863429776519e-06,
+      "loss": 207.3461,
+      "step": 40030
+    },
+    {
+      "epoch": 0.7700703429640208,
+      "grad_norm": 381.80855188560895,
+      "learning_rate": 2.5457937210950433e-06,
+      "loss": 211.3806,
+      "step": 40040
+    },
+    {
+      "epoch": 0.7702626682245013,
+      "grad_norm": 389.6639341252145,
+      "learning_rate": 2.541726788982294e-06,
+      "loss": 210.348,
+      "step": 40050
+    },
+    {
+      "epoch": 0.7704549934849818,
+      "grad_norm": 377.51986488485994,
+      "learning_rate": 2.5376626349531395e-06,
+      "loss": 211.5141,
+      "step": 40060
+    },
+    {
+      "epoch": 0.7706473187454623,
+      "grad_norm": 376.46020036282295,
+      "learning_rate": 2.53360126052141e-06,
+      "loss": 215.3596,
+      "step": 40070
+    },
+    {
+      "epoch": 0.7708396440059428,
+      "grad_norm": 391.33115229090686,
+      "learning_rate": 2.529542667199896e-06,
+      "loss": 215.8145,
+      "step": 40080
+    },
+    {
+      "epoch": 0.7710319692664234,
+      "grad_norm": 403.976553907104,
+      "learning_rate": 2.525486856500363e-06,
+      "loss": 208.0852,
+      "step": 40090
+    },
+    {
+      "epoch": 0.7712242945269039,
+      "grad_norm": 362.49665409997715,
+      "learning_rate": 2.5214338299335306e-06,
+      "loss": 211.3493,
+      "step": 40100
+    },
+    {
+      "epoch": 0.7714166197873844,
+      "grad_norm": 366.49316080226913,
+      "learning_rate": 2.5173835890090826e-06,
+      "loss": 214.2764,
+      "step": 40110
+    },
+    {
+      "epoch": 0.7716089450478649,
+      "grad_norm": 404.7037842540386,
+      "learning_rate": 2.5133361352356666e-06,
+      "loss": 209.8697,
+      "step": 40120
+    },
+    {
+      "epoch": 0.7718012703083454,
+      "grad_norm": 361.8774100387627,
+      "learning_rate": 2.5092914701208958e-06,
+      "loss": 204.0543,
+      "step": 40130
+    },
+    {
+      "epoch": 0.771993595568826,
+      "grad_norm": 356.3596220835635,
+      "learning_rate": 2.5052495951713406e-06,
+      "loss": 210.4549,
+      "step": 40140
+    },
+    {
+      "epoch": 0.7721859208293065,
+      "grad_norm": 358.213439768463,
+      "learning_rate": 2.5012105118925267e-06,
+      "loss": 208.4516,
+      "step": 40150
+    },
+    {
+      "epoch": 0.772378246089787,
+      "grad_norm": 399.55786379734417,
+      "learning_rate": 2.497174221788955e-06,
+      "loss": 208.3724,
+      "step": 40160
+    },
+    {
+      "epoch": 0.7725705713502675,
+      "grad_norm": 405.0462877160036,
+      "learning_rate": 2.4931407263640683e-06,
+      "loss": 210.5768,
+      "step": 40170
+    },
+    {
+      "epoch": 0.7727628966107482,
+      "grad_norm": 393.1726569261335,
+      "learning_rate": 2.489110027120285e-06,
+      "loss": 213.4158,
+      "step": 40180
+    },
+    {
+      "epoch": 0.7729552218712287,
+      "grad_norm": 374.1379140084526,
+      "learning_rate": 2.4850821255589664e-06,
+      "loss": 208.5962,
+      "step": 40190
+    },
+    {
+      "epoch": 0.7731475471317092,
+      "grad_norm": 353.74974249857166,
+      "learning_rate": 2.4810570231804463e-06,
+      "loss": 217.174,
+      "step": 40200
+    },
+    {
+      "epoch": 0.7733398723921897,
+      "grad_norm": 379.0423595891199,
+      "learning_rate": 2.4770347214840063e-06,
+      "loss": 208.6815,
+      "step": 40210
+    },
+    {
+      "epoch": 0.7735321976526702,
+      "grad_norm": 352.99146402263244,
+      "learning_rate": 2.473015221967886e-06,
+      "loss": 207.484,
+      "step": 40220
+    },
+    {
+      "epoch": 0.7737245229131507,
+      "grad_norm": 421.33861937517287,
+      "learning_rate": 2.4689985261292805e-06,
+      "loss": 208.2637,
+      "step": 40230
+    },
+    {
+      "epoch": 0.7739168481736313,
+      "grad_norm": 407.46918761914634,
+      "learning_rate": 2.464984635464348e-06,
+      "loss": 208.6252,
+      "step": 40240
+    },
+    {
+      "epoch": 0.7741091734341118,
+      "grad_norm": 354.16983107885034,
+      "learning_rate": 2.460973551468194e-06,
+      "loss": 198.5188,
+      "step": 40250
+    },
+    {
+      "epoch": 0.7743014986945923,
+      "grad_norm": 393.1502003953757,
+      "learning_rate": 2.456965275634878e-06,
+      "loss": 214.0779,
+      "step": 40260
+    },
+    {
+      "epoch": 0.7744938239550728,
+      "grad_norm": 370.7356071511445,
+      "learning_rate": 2.4529598094574226e-06,
+      "loss": 213.3693,
+      "step": 40270
+    },
+    {
+      "epoch": 0.7746861492155533,
+      "grad_norm": 380.25155597419234,
+      "learning_rate": 2.4489571544277944e-06,
+      "loss": 207.8812,
+      "step": 40280
+    },
+    {
+      "epoch": 0.7748784744760339,
+      "grad_norm": 384.6823808337117,
+      "learning_rate": 2.444957312036914e-06,
+      "loss": 216.7433,
+      "step": 40290
+    },
+    {
+      "epoch": 0.7750707997365144,
+      "grad_norm": 373.41206974329594,
+      "learning_rate": 2.4409602837746625e-06,
+      "loss": 206.5674,
+      "step": 40300
+    },
+    {
+      "epoch": 0.7752631249969949,
+      "grad_norm": 361.24037894315006,
+      "learning_rate": 2.4369660711298603e-06,
+      "loss": 206.8981,
+      "step": 40310
+    },
+    {
+      "epoch": 0.7754554502574754,
+      "grad_norm": 401.287058424949,
+      "learning_rate": 2.4329746755902917e-06,
+      "loss": 217.8212,
+      "step": 40320
+    },
+    {
+      "epoch": 0.7756477755179559,
+      "grad_norm": 396.7202974685465,
+      "learning_rate": 2.428986098642684e-06,
+      "loss": 207.4115,
+      "step": 40330
+    },
+    {
+      "epoch": 0.7758401007784365,
+      "grad_norm": 380.77805355964557,
+      "learning_rate": 2.425000341772711e-06,
+      "loss": 209.3615,
+      "step": 40340
+    },
+    {
+      "epoch": 0.776032426038917,
+      "grad_norm": 400.86785566184125,
+      "learning_rate": 2.4210174064650084e-06,
+      "loss": 203.576,
+      "step": 40350
+    },
+    {
+      "epoch": 0.7762247512993975,
+      "grad_norm": 416.89772600492324,
+      "learning_rate": 2.4170372942031506e-06,
+      "loss": 210.3003,
+      "step": 40360
+    },
+    {
+      "epoch": 0.776417076559878,
+      "grad_norm": 404.7180500533755,
+      "learning_rate": 2.4130600064696618e-06,
+      "loss": 209.1882,
+      "step": 40370
+    },
+    {
+      "epoch": 0.7766094018203585,
+      "grad_norm": 388.5533773879684,
+      "learning_rate": 2.4090855447460205e-06,
+      "loss": 213.25,
+      "step": 40380
+    },
+    {
+      "epoch": 0.7768017270808392,
+      "grad_norm": 393.5919657426778,
+      "learning_rate": 2.4051139105126463e-06,
+      "loss": 210.339,
+      "step": 40390
+    },
+    {
+      "epoch": 0.7769940523413197,
+      "grad_norm": 426.08640397184075,
+      "learning_rate": 2.4011451052489064e-06,
+      "loss": 214.112,
+      "step": 40400
+    },
+    {
+      "epoch": 0.7771863776018002,
+      "grad_norm": 399.0350070963163,
+      "learning_rate": 2.3971791304331125e-06,
+      "loss": 206.2459,
+      "step": 40410
+    },
+    {
+      "epoch": 0.7773787028622807,
+      "grad_norm": 366.80926515212064,
+      "learning_rate": 2.393215987542531e-06,
+      "loss": 209.712,
+      "step": 40420
+    },
+    {
+      "epoch": 0.7775710281227612,
+      "grad_norm": 386.78936525839777,
+      "learning_rate": 2.3892556780533606e-06,
+      "loss": 223.2839,
+      "step": 40430
+    },
+    {
+      "epoch": 0.7777633533832418,
+      "grad_norm": 386.6402615053517,
+      "learning_rate": 2.385298203440758e-06,
+      "loss": 208.233,
+      "step": 40440
+    },
+    {
+      "epoch": 0.7779556786437223,
+      "grad_norm": 367.298475936663,
+      "learning_rate": 2.3813435651788107e-06,
+      "loss": 209.0829,
+      "step": 40450
+    },
+    {
+      "epoch": 0.7781480039042028,
+      "grad_norm": 363.53060180816016,
+      "learning_rate": 2.377391764740562e-06,
+      "loss": 205.0176,
+      "step": 40460
+    },
+    {
+      "epoch": 0.7783403291646833,
+      "grad_norm": 373.2186635977586,
+      "learning_rate": 2.3734428035979883e-06,
+      "loss": 202.7703,
+      "step": 40470
+    },
+    {
+      "epoch": 0.7785326544251638,
+      "grad_norm": 352.4191148747799,
+      "learning_rate": 2.3694966832220123e-06,
+      "loss": 209.5993,
+      "step": 40480
+    },
+    {
+      "epoch": 0.7787249796856444,
+      "grad_norm": 392.4156549700369,
+      "learning_rate": 2.365553405082501e-06,
+      "loss": 212.0213,
+      "step": 40490
+    },
+    {
+      "epoch": 0.7789173049461249,
+      "grad_norm": 361.5778541776504,
+      "learning_rate": 2.3616129706482604e-06,
+      "loss": 211.7287,
+      "step": 40500
+    },
+    {
+      "epoch": 0.7791096302066054,
+      "grad_norm": 409.50114908696054,
+      "learning_rate": 2.357675381387036e-06,
+      "loss": 211.5296,
+      "step": 40510
+    },
+    {
+      "epoch": 0.7793019554670859,
+      "grad_norm": 370.6351263812637,
+      "learning_rate": 2.3537406387655114e-06,
+      "loss": 207.166,
+      "step": 40520
+    },
+    {
+      "epoch": 0.7794942807275664,
+      "grad_norm": 398.10616343751417,
+      "learning_rate": 2.349808744249321e-06,
+      "loss": 209.6109,
+      "step": 40530
+    },
+    {
+      "epoch": 0.779686605988047,
+      "grad_norm": 371.82780440909073,
+      "learning_rate": 2.345879699303025e-06,
+      "loss": 209.3387,
+      "step": 40540
+    },
+    {
+      "epoch": 0.7798789312485275,
+      "grad_norm": 409.6346665746718,
+      "learning_rate": 2.3419535053901264e-06,
+      "loss": 209.4224,
+      "step": 40550
+    },
+    {
+      "epoch": 0.780071256509008,
+      "grad_norm": 466.3919975552398,
+      "learning_rate": 2.338030163973073e-06,
+      "loss": 222.971,
+      "step": 40560
+    },
+    {
+      "epoch": 0.7802635817694885,
+      "grad_norm": 454.49359412655474,
+      "learning_rate": 2.334109676513242e-06,
+      "loss": 210.0865,
+      "step": 40570
+    },
+    {
+      "epoch": 0.780455907029969,
+      "grad_norm": 361.61006941299746,
+      "learning_rate": 2.330192044470948e-06,
+      "loss": 206.8723,
+      "step": 40580
+    },
+    {
+      "epoch": 0.7806482322904497,
+      "grad_norm": 379.20199257545136,
+      "learning_rate": 2.3262772693054457e-06,
+      "loss": 216.4956,
+      "step": 40590
+    },
+    {
+      "epoch": 0.7808405575509302,
+      "grad_norm": 388.46386274133926,
+      "learning_rate": 2.322365352474928e-06,
+      "loss": 208.4771,
+      "step": 40600
+    },
+    {
+      "epoch": 0.7810328828114107,
+      "grad_norm": 415.436739259405,
+      "learning_rate": 2.3184562954365153e-06,
+      "loss": 214.4146,
+      "step": 40610
+    },
+    {
+      "epoch": 0.7812252080718912,
+      "grad_norm": 398.5329126749291,
+      "learning_rate": 2.3145500996462656e-06,
+      "loss": 215.2695,
+      "step": 40620
+    },
+    {
+      "epoch": 0.7814175333323717,
+      "grad_norm": 438.33163439495354,
+      "learning_rate": 2.310646766559177e-06,
+      "loss": 217.8767,
+      "step": 40630
+    },
+    {
+      "epoch": 0.7816098585928523,
+      "grad_norm": 367.6509677730949,
+      "learning_rate": 2.3067462976291744e-06,
+      "loss": 221.5393,
+      "step": 40640
+    },
+    {
+      "epoch": 0.7818021838533328,
+      "grad_norm": 365.1621423978296,
+      "learning_rate": 2.302848694309118e-06,
+      "loss": 214.411,
+      "step": 40650
+    },
+    {
+      "epoch": 0.7819945091138133,
+      "grad_norm": 371.1656255314514,
+      "learning_rate": 2.2989539580507957e-06,
+      "loss": 212.0582,
+      "step": 40660
+    },
+    {
+      "epoch": 0.7821868343742938,
+      "grad_norm": 366.49136219272657,
+      "learning_rate": 2.2950620903049414e-06,
+      "loss": 213.4375,
+      "step": 40670
+    },
+    {
+      "epoch": 0.7823791596347743,
+      "grad_norm": 382.3267572102029,
+      "learning_rate": 2.2911730925212073e-06,
+      "loss": 212.9401,
+      "step": 40680
+    },
+    {
+      "epoch": 0.7825714848952549,
+      "grad_norm": 385.54656457172945,
+      "learning_rate": 2.2872869661481766e-06,
+      "loss": 205.6087,
+      "step": 40690
+    },
+    {
+      "epoch": 0.7827638101557354,
+      "grad_norm": 381.35934430726485,
+      "learning_rate": 2.283403712633375e-06,
+      "loss": 205.9314,
+      "step": 40700
+    },
+    {
+      "epoch": 0.7829561354162159,
+      "grad_norm": 395.2955811950514,
+      "learning_rate": 2.279523333423247e-06,
+      "loss": 217.6813,
+      "step": 40710
+    },
+    {
+      "epoch": 0.7831484606766964,
+      "grad_norm": 364.21103978588957,
+      "learning_rate": 2.2756458299631667e-06,
+      "loss": 203.4238,
+      "step": 40720
+    },
+    {
+      "epoch": 0.7833407859371769,
+      "grad_norm": 420.8925373105516,
+      "learning_rate": 2.271771203697445e-06,
+      "loss": 208.1244,
+      "step": 40730
+    },
+    {
+      "epoch": 0.7835331111976575,
+      "grad_norm": 402.48721611435695,
+      "learning_rate": 2.267899456069311e-06,
+      "loss": 211.8673,
+      "step": 40740
+    },
+    {
+      "epoch": 0.783725436458138,
+      "grad_norm": 387.2492381568478,
+      "learning_rate": 2.2640305885209336e-06,
+      "loss": 215.0771,
+      "step": 40750
+    },
+    {
+      "epoch": 0.7839177617186185,
+      "grad_norm": 490.7325616931793,
+      "learning_rate": 2.2601646024933976e-06,
+      "loss": 218.0952,
+      "step": 40760
+    },
+    {
+      "epoch": 0.784110086979099,
+      "grad_norm": 372.99874000973307,
+      "learning_rate": 2.256301499426716e-06,
+      "loss": 208.6124,
+      "step": 40770
+    },
+    {
+      "epoch": 0.7843024122395795,
+      "grad_norm": 381.8762460911402,
+      "learning_rate": 2.252441280759838e-06,
+      "loss": 208.2737,
+      "step": 40780
+    },
+    {
+      "epoch": 0.78449473750006,
+      "grad_norm": 373.99020109006455,
+      "learning_rate": 2.248583947930628e-06,
+      "loss": 204.0592,
+      "step": 40790
+    },
+    {
+      "epoch": 0.7846870627605407,
+      "grad_norm": 364.2311543412736,
+      "learning_rate": 2.2447295023758755e-06,
+      "loss": 208.1338,
+      "step": 40800
+    },
+    {
+      "epoch": 0.7848793880210212,
+      "grad_norm": 378.7443553582108,
+      "learning_rate": 2.2408779455313035e-06,
+      "loss": 209.0336,
+      "step": 40810
+    },
+    {
+      "epoch": 0.7850717132815017,
+      "grad_norm": 373.838788342394,
+      "learning_rate": 2.2370292788315505e-06,
+      "loss": 212.4266,
+      "step": 40820
+    },
+    {
+      "epoch": 0.7852640385419822,
+      "grad_norm": 388.422659611989,
+      "learning_rate": 2.2331835037101825e-06,
+      "loss": 203.205,
+      "step": 40830
+    },
+    {
+      "epoch": 0.7854563638024628,
+      "grad_norm": 379.6486678988751,
+      "learning_rate": 2.2293406215996814e-06,
+      "loss": 208.6698,
+      "step": 40840
+    },
+    {
+      "epoch": 0.7856486890629433,
+      "grad_norm": 358.08450003870985,
+      "learning_rate": 2.2255006339314667e-06,
+      "loss": 209.7238,
+      "step": 40850
+    },
+    {
+      "epoch": 0.7858410143234238,
+      "grad_norm": 426.21040775390503,
+      "learning_rate": 2.2216635421358623e-06,
+      "loss": 205.0207,
+      "step": 40860
+    },
+    {
+      "epoch": 0.7860333395839043,
+      "grad_norm": 388.51231251878716,
+      "learning_rate": 2.2178293476421276e-06,
+      "loss": 206.4274,
+      "step": 40870
+    },
+    {
+      "epoch": 0.7862256648443848,
+      "grad_norm": 465.4097706179838,
+      "learning_rate": 2.213998051878431e-06,
+      "loss": 207.1637,
+      "step": 40880
+    },
+    {
+      "epoch": 0.7864179901048653,
+      "grad_norm": 421.3127931274906,
+      "learning_rate": 2.2101696562718735e-06,
+      "loss": 203.6789,
+      "step": 40890
+    },
+    {
+      "epoch": 0.7866103153653459,
+      "grad_norm": 397.25827661302765,
+      "learning_rate": 2.206344162248466e-06,
+      "loss": 216.7128,
+      "step": 40900
+    },
+    {
+      "epoch": 0.7868026406258264,
+      "grad_norm": 381.4484702235938,
+      "learning_rate": 2.2025215712331383e-06,
+      "loss": 210.033,
+      "step": 40910
+    },
+    {
+      "epoch": 0.7869949658863069,
+      "grad_norm": 395.54586204684006,
+      "learning_rate": 2.1987018846497487e-06,
+      "loss": 205.8125,
+      "step": 40920
+    },
+    {
+      "epoch": 0.7871872911467874,
+      "grad_norm": 367.7789538166651,
+      "learning_rate": 2.194885103921064e-06,
+      "loss": 211.1847,
+      "step": 40930
+    },
+    {
+      "epoch": 0.787379616407268,
+      "grad_norm": 393.0109345617761,
+      "learning_rate": 2.191071230468772e-06,
+      "loss": 212.5428,
+      "step": 40940
+    },
+    {
+      "epoch": 0.7875719416677485,
+      "grad_norm": 416.3467328375286,
+      "learning_rate": 2.1872602657134757e-06,
+      "loss": 212.9899,
+      "step": 40950
+    },
+    {
+      "epoch": 0.787764266928229,
+      "grad_norm": 376.86730881825264,
+      "learning_rate": 2.1834522110747014e-06,
+      "loss": 209.9806,
+      "step": 40960
+    },
+    {
+      "epoch": 0.7879565921887095,
+      "grad_norm": 370.94172546040613,
+      "learning_rate": 2.179647067970885e-06,
+      "loss": 207.9988,
+      "step": 40970
+    },
+    {
+      "epoch": 0.78814891744919,
+      "grad_norm": 484.27605644855134,
+      "learning_rate": 2.1758448378193743e-06,
+      "loss": 214.4127,
+      "step": 40980
+    },
+    {
+      "epoch": 0.7883412427096705,
+      "grad_norm": 391.2194759694569,
+      "learning_rate": 2.1720455220364443e-06,
+      "loss": 208.5151,
+      "step": 40990
+    },
+    {
+      "epoch": 0.7885335679701512,
+      "grad_norm": 396.5174047992866,
+      "learning_rate": 2.168249122037275e-06,
+      "loss": 205.8971,
+      "step": 41000
+    },
+    {
+      "epoch": 0.7887258932306317,
+      "grad_norm": 380.90194724114434,
+      "learning_rate": 2.1644556392359583e-06,
+      "loss": 207.6405,
+      "step": 41010
+    },
+    {
+      "epoch": 0.7889182184911122,
+      "grad_norm": 375.352348977756,
+      "learning_rate": 2.160665075045508e-06,
+      "loss": 199.7787,
+      "step": 41020
+    },
+    {
+      "epoch": 0.7891105437515927,
+      "grad_norm": 372.62798697200947,
+      "learning_rate": 2.1568774308778494e-06,
+      "loss": 211.7347,
+      "step": 41030
+    },
+    {
+      "epoch": 0.7893028690120732,
+      "grad_norm": 406.53854028717467,
+      "learning_rate": 2.1530927081438148e-06,
+      "loss": 219.5607,
+      "step": 41040
+    },
+    {
+      "epoch": 0.7894951942725538,
+      "grad_norm": 423.68231539923755,
+      "learning_rate": 2.1493109082531473e-06,
+      "loss": 200.8829,
+      "step": 41050
+    },
+    {
+      "epoch": 0.7896875195330343,
+      "grad_norm": 396.1324553868352,
+      "learning_rate": 2.1455320326145103e-06,
+      "loss": 203.8986,
+      "step": 41060
+    },
+    {
+      "epoch": 0.7898798447935148,
+      "grad_norm": 364.4211746942873,
+      "learning_rate": 2.141756082635471e-06,
+      "loss": 206.797,
+      "step": 41070
+    },
+    {
+      "epoch": 0.7900721700539953,
+      "grad_norm": 389.649453891042,
+      "learning_rate": 2.137983059722507e-06,
+      "loss": 201.7376,
+      "step": 41080
+    },
+    {
+      "epoch": 0.7902644953144758,
+      "grad_norm": 387.0089249503018,
+      "learning_rate": 2.1342129652810063e-06,
+      "loss": 208.0186,
+      "step": 41090
+    },
+    {
+      "epoch": 0.7904568205749564,
+      "grad_norm": 394.4162726860938,
+      "learning_rate": 2.1304458007152694e-06,
+      "loss": 204.6994,
+      "step": 41100
+    },
+    {
+      "epoch": 0.7906491458354369,
+      "grad_norm": 397.59111280299146,
+      "learning_rate": 2.1266815674285026e-06,
+      "loss": 216.6823,
+      "step": 41110
+    },
+    {
+      "epoch": 0.7908414710959174,
+      "grad_norm": 384.82380167561735,
+      "learning_rate": 2.1229202668228197e-06,
+      "loss": 218.8221,
+      "step": 41120
+    },
+    {
+      "epoch": 0.7910337963563979,
+      "grad_norm": 381.19785425991427,
+      "learning_rate": 2.1191619002992405e-06,
+      "loss": 201.5279,
+      "step": 41130
+    },
+    {
+      "epoch": 0.7912261216168784,
+      "grad_norm": 462.92689135610516,
+      "learning_rate": 2.1154064692577e-06,
+      "loss": 207.5111,
+      "step": 41140
+    },
+    {
+      "epoch": 0.791418446877359,
+      "grad_norm": 381.3694790807442,
+      "learning_rate": 2.111653975097029e-06,
+      "loss": 206.0437,
+      "step": 41150
+    },
+    {
+      "epoch": 0.7916107721378395,
+      "grad_norm": 410.5500321929193,
+      "learning_rate": 2.1079044192149713e-06,
+      "loss": 208.7455,
+      "step": 41160
+    },
+    {
+      "epoch": 0.79180309739832,
+      "grad_norm": 379.1735991775821,
+      "learning_rate": 2.1041578030081777e-06,
+      "loss": 208.1734,
+      "step": 41170
+    },
+    {
+      "epoch": 0.7919954226588005,
+      "grad_norm": 374.2296064865967,
+      "learning_rate": 2.100414127872198e-06,
+      "loss": 203.8804,
+      "step": 41180
+    },
+    {
+      "epoch": 0.792187747919281,
+      "grad_norm": 368.2118493745568,
+      "learning_rate": 2.0966733952014904e-06,
+      "loss": 215.8449,
+      "step": 41190
+    },
+    {
+      "epoch": 0.7923800731797617,
+      "grad_norm": 336.3000961510837,
+      "learning_rate": 2.0929356063894125e-06,
+      "loss": 207.8467,
+      "step": 41200
+    },
+    {
+      "epoch": 0.7925723984402422,
+      "grad_norm": 417.828983799324,
+      "learning_rate": 2.089200762828234e-06,
+      "loss": 209.6514,
+      "step": 41210
+    },
+    {
+      "epoch": 0.7927647237007227,
+      "grad_norm": 371.89545464401465,
+      "learning_rate": 2.0854688659091203e-06,
+      "loss": 210.8139,
+      "step": 41220
+    },
+    {
+      "epoch": 0.7929570489612032,
+      "grad_norm": 379.7393062396534,
+      "learning_rate": 2.08173991702214e-06,
+      "loss": 201.6136,
+      "step": 41230
+    },
+    {
+      "epoch": 0.7931493742216837,
+      "grad_norm": 401.7372233329505,
+      "learning_rate": 2.0780139175562675e-06,
+      "loss": 210.716,
+      "step": 41240
+    },
+    {
+      "epoch": 0.7933416994821643,
+      "grad_norm": 382.9525792334095,
+      "learning_rate": 2.0742908688993746e-06,
+      "loss": 205.0247,
+      "step": 41250
+    },
+    {
+      "epoch": 0.7935340247426448,
+      "grad_norm": 373.84164533215096,
+      "learning_rate": 2.070570772438236e-06,
+      "loss": 204.1771,
+      "step": 41260
+    },
+    {
+      "epoch": 0.7937263500031253,
+      "grad_norm": 375.2773060248128,
+      "learning_rate": 2.066853629558524e-06,
+      "loss": 209.8484,
+      "step": 41270
+    },
+    {
+      "epoch": 0.7939186752636058,
+      "grad_norm": 358.09329970319953,
+      "learning_rate": 2.0631394416448157e-06,
+      "loss": 200.9646,
+      "step": 41280
+    },
+    {
+      "epoch": 0.7941110005240863,
+      "grad_norm": 363.4687126312382,
+      "learning_rate": 2.059428210080583e-06,
+      "loss": 202.5251,
+      "step": 41290
+    },
+    {
+      "epoch": 0.7943033257845669,
+      "grad_norm": 412.07035716220867,
+      "learning_rate": 2.0557199362482005e-06,
+      "loss": 210.6676,
+      "step": 41300
+    },
+    {
+      "epoch": 0.7944956510450474,
+      "grad_norm": 371.78941453302485,
+      "learning_rate": 2.052014621528935e-06,
+      "loss": 208.0328,
+      "step": 41310
+    },
+    {
+      "epoch": 0.7946879763055279,
+      "grad_norm": 367.2522493217452,
+      "learning_rate": 2.048312267302961e-06,
+      "loss": 206.5413,
+      "step": 41320
+    },
+    {
+      "epoch": 0.7948803015660084,
+      "grad_norm": 363.6215668802465,
+      "learning_rate": 2.044612874949341e-06,
+      "loss": 203.8802,
+      "step": 41330
+    },
+    {
+      "epoch": 0.7950726268264889,
+      "grad_norm": 388.5862279885824,
+      "learning_rate": 2.040916445846034e-06,
+      "loss": 214.138,
+      "step": 41340
+    },
+    {
+      "epoch": 0.7952649520869695,
+      "grad_norm": 478.54260791926976,
+      "learning_rate": 2.037222981369905e-06,
+      "loss": 204.9165,
+      "step": 41350
+    },
+    {
+      "epoch": 0.79545727734745,
+      "grad_norm": 396.8818735865609,
+      "learning_rate": 2.033532482896707e-06,
+      "loss": 206.2721,
+      "step": 41360
+    },
+    {
+      "epoch": 0.7956496026079305,
+      "grad_norm": 400.0979738904692,
+      "learning_rate": 2.0298449518010875e-06,
+      "loss": 206.8979,
+      "step": 41370
+    },
+    {
+      "epoch": 0.795841927868411,
+      "grad_norm": 430.2141975558431,
+      "learning_rate": 2.0261603894565897e-06,
+      "loss": 207.9649,
+      "step": 41380
+    },
+    {
+      "epoch": 0.7960342531288915,
+      "grad_norm": 374.930373576075,
+      "learning_rate": 2.0224787972356574e-06,
+      "loss": 211.1537,
+      "step": 41390
+    },
+    {
+      "epoch": 0.796226578389372,
+      "grad_norm": 378.1549978452382,
+      "learning_rate": 2.0188001765096198e-06,
+      "loss": 211.6395,
+      "step": 41400
+    },
+    {
+      "epoch": 0.7964189036498527,
+      "grad_norm": 383.3548792639214,
+      "learning_rate": 2.0151245286486998e-06,
+      "loss": 208.4597,
+      "step": 41410
+    },
+    {
+      "epoch": 0.7966112289103332,
+      "grad_norm": 374.38924242464293,
+      "learning_rate": 2.011451855022021e-06,
+      "loss": 211.3832,
+      "step": 41420
+    },
+    {
+      "epoch": 0.7968035541708137,
+      "grad_norm": 377.74745787461427,
+      "learning_rate": 2.0077821569975885e-06,
+      "loss": 207.8268,
+      "step": 41430
+    },
+    {
+      "epoch": 0.7969958794312942,
+      "grad_norm": 401.8075312843853,
+      "learning_rate": 2.0041154359423087e-06,
+      "loss": 214.7287,
+      "step": 41440
+    },
+    {
+      "epoch": 0.7971882046917748,
+      "grad_norm": 408.96477305382166,
+      "learning_rate": 2.000451693221971e-06,
+      "loss": 214.4973,
+      "step": 41450
+    },
+    {
+      "epoch": 0.7973805299522553,
+      "grad_norm": 390.1806889630553,
+      "learning_rate": 1.9967909302012635e-06,
+      "loss": 206.4686,
+      "step": 41460
+    },
+    {
+      "epoch": 0.7975728552127358,
+      "grad_norm": 374.1116299851007,
+      "learning_rate": 1.9931331482437553e-06,
+      "loss": 209.1153,
+      "step": 41470
+    },
+    {
+      "epoch": 0.7977651804732163,
+      "grad_norm": 365.5129898815175,
+      "learning_rate": 1.989478348711913e-06,
+      "loss": 206.2217,
+      "step": 41480
+    },
+    {
+      "epoch": 0.7979575057336968,
+      "grad_norm": 416.0808687396897,
+      "learning_rate": 1.9858265329670844e-06,
+      "loss": 205.5742,
+      "step": 41490
+    },
+    {
+      "epoch": 0.7981498309941774,
+      "grad_norm": 398.41175782165334,
+      "learning_rate": 1.9821777023695178e-06,
+      "loss": 206.7222,
+      "step": 41500
+    },
+    {
+      "epoch": 0.7983421562546579,
+      "grad_norm": 369.40730826166464,
+      "learning_rate": 1.9785318582783375e-06,
+      "loss": 205.7667,
+      "step": 41510
+    },
+    {
+      "epoch": 0.7985344815151384,
+      "grad_norm": 370.9989182735855,
+      "learning_rate": 1.9748890020515577e-06,
+      "loss": 209.4025,
+      "step": 41520
+    },
+    {
+      "epoch": 0.7987268067756189,
+      "grad_norm": 361.7503771232004,
+      "learning_rate": 1.9712491350460895e-06,
+      "loss": 202.9891,
+      "step": 41530
+    },
+    {
+      "epoch": 0.7989191320360994,
+      "grad_norm": 345.97152031372815,
+      "learning_rate": 1.967612258617718e-06,
+      "loss": 204.3035,
+      "step": 41540
+    },
+    {
+      "epoch": 0.79911145729658,
+      "grad_norm": 404.66309614884284,
+      "learning_rate": 1.9639783741211218e-06,
+      "loss": 209.4293,
+      "step": 41550
+    },
+    {
+      "epoch": 0.7993037825570605,
+      "grad_norm": 376.599238149299,
+      "learning_rate": 1.960347482909859e-06,
+      "loss": 209.0351,
+      "step": 41560
+    },
+    {
+      "epoch": 0.799496107817541,
+      "grad_norm": 426.6318711373734,
+      "learning_rate": 1.956719586336382e-06,
+      "loss": 211.9038,
+      "step": 41570
+    },
+    {
+      "epoch": 0.7996884330780215,
+      "grad_norm": 372.0986377358308,
+      "learning_rate": 1.953094685752017e-06,
+      "loss": 209.4806,
+      "step": 41580
+    },
+    {
+      "epoch": 0.799880758338502,
+      "grad_norm": 350.6518750560805,
+      "learning_rate": 1.949472782506984e-06,
+      "loss": 211.3456,
+      "step": 41590
+    },
+    {
+      "epoch": 0.8000730835989825,
+      "grad_norm": 392.1056829673286,
+      "learning_rate": 1.945853877950382e-06,
+      "loss": 218.7484,
+      "step": 41600
+    },
+    {
+      "epoch": 0.8002654088594632,
+      "grad_norm": 426.1488961164271,
+      "learning_rate": 1.942237973430192e-06,
+      "loss": 213.8978,
+      "step": 41610
+    },
+    {
+      "epoch": 0.8004577341199437,
+      "grad_norm": 393.5070429472998,
+      "learning_rate": 1.9386250702932784e-06,
+      "loss": 210.5322,
+      "step": 41620
+    },
+    {
+      "epoch": 0.8006500593804242,
+      "grad_norm": 375.03818369316474,
+      "learning_rate": 1.9350151698853857e-06,
+      "loss": 210.4921,
+      "step": 41630
+    },
+    {
+      "epoch": 0.8008423846409047,
+      "grad_norm": 360.15439987391665,
+      "learning_rate": 1.9314082735511475e-06,
+      "loss": 208.094,
+      "step": 41640
+    },
+    {
+      "epoch": 0.8010347099013853,
+      "grad_norm": 381.0743658570249,
+      "learning_rate": 1.92780438263407e-06,
+      "loss": 205.0781,
+      "step": 41650
+    },
+    {
+      "epoch": 0.8012270351618658,
+      "grad_norm": 353.8982612598771,
+      "learning_rate": 1.9242034984765436e-06,
+      "loss": 208.9497,
+      "step": 41660
+    },
+    {
+      "epoch": 0.8014193604223463,
+      "grad_norm": 369.89037903693855,
+      "learning_rate": 1.9206056224198346e-06,
+      "loss": 203.1918,
+      "step": 41670
+    },
+    {
+      "epoch": 0.8016116856828268,
+      "grad_norm": 392.1642696750914,
+      "learning_rate": 1.9170107558040983e-06,
+      "loss": 214.6508,
+      "step": 41680
+    },
+    {
+      "epoch": 0.8018040109433073,
+      "grad_norm": 364.8607112589798,
+      "learning_rate": 1.9134188999683613e-06,
+      "loss": 202.766,
+      "step": 41690
+    },
+    {
+      "epoch": 0.8019963362037879,
+      "grad_norm": 386.51606403550096,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 207.0134,
+      "step": 41700
+    },
+    {
+      "epoch": 0.8021886614642684,
+      "grad_norm": 364.748551191837,
+      "learning_rate": 1.9062442259873847e-06,
+      "loss": 211.1021,
+      "step": 41710
+    },
+    {
+      "epoch": 0.8023809867247489,
+      "grad_norm": 401.25946898896325,
+      "learning_rate": 1.9026614105145935e-06,
+      "loss": 216.4434,
+      "step": 41720
+    },
+    {
+      "epoch": 0.8025733119852294,
+      "grad_norm": 392.9856943967848,
+      "learning_rate": 1.8990816111666976e-06,
+      "loss": 208.1454,
+      "step": 41730
+    },
+    {
+      "epoch": 0.8027656372457099,
+      "grad_norm": 390.4493611998971,
+      "learning_rate": 1.8955048292771083e-06,
+      "loss": 205.0919,
+      "step": 41740
+    },
+    {
+      "epoch": 0.8029579625061904,
+      "grad_norm": 415.085700641082,
+      "learning_rate": 1.891931066178122e-06,
+      "loss": 200.7006,
+      "step": 41750
+    },
+    {
+      "epoch": 0.803150287766671,
+      "grad_norm": 339.2285846018839,
+      "learning_rate": 1.888360323200904e-06,
+      "loss": 199.2415,
+      "step": 41760
+    },
+    {
+      "epoch": 0.8033426130271515,
+      "grad_norm": 401.20788185343383,
+      "learning_rate": 1.8847926016754947e-06,
+      "loss": 206.1355,
+      "step": 41770
+    },
+    {
+      "epoch": 0.803534938287632,
+      "grad_norm": 412.1138494801861,
+      "learning_rate": 1.8812279029308177e-06,
+      "loss": 205.3744,
+      "step": 41780
+    },
+    {
+      "epoch": 0.8037272635481125,
+      "grad_norm": 376.8151987996318,
+      "learning_rate": 1.87766622829466e-06,
+      "loss": 208.3064,
+      "step": 41790
+    },
+    {
+      "epoch": 0.803919588808593,
+      "grad_norm": 382.2229894827946,
+      "learning_rate": 1.874107579093688e-06,
+      "loss": 219.2669,
+      "step": 41800
+    },
+    {
+      "epoch": 0.8041119140690736,
+      "grad_norm": 368.6116072457296,
+      "learning_rate": 1.870551956653437e-06,
+      "loss": 211.5302,
+      "step": 41810
+    },
+    {
+      "epoch": 0.8043042393295542,
+      "grad_norm": 385.280062826804,
+      "learning_rate": 1.8669993622983217e-06,
+      "loss": 212.2618,
+      "step": 41820
+    },
+    {
+      "epoch": 0.8044965645900347,
+      "grad_norm": 354.77223705156314,
+      "learning_rate": 1.863449797351624e-06,
+      "loss": 207.746,
+      "step": 41830
+    },
+    {
+      "epoch": 0.8046888898505152,
+      "grad_norm": 367.2867320331355,
+      "learning_rate": 1.8599032631354963e-06,
+      "loss": 210.4903,
+      "step": 41840
+    },
+    {
+      "epoch": 0.8048812151109958,
+      "grad_norm": 453.4886518814372,
+      "learning_rate": 1.8563597609709626e-06,
+      "loss": 216.1181,
+      "step": 41850
+    },
+    {
+      "epoch": 0.8050735403714763,
+      "grad_norm": 375.4274482466092,
+      "learning_rate": 1.852819292177922e-06,
+      "loss": 209.5629,
+      "step": 41860
+    },
+    {
+      "epoch": 0.8052658656319568,
+      "grad_norm": 412.8022239124099,
+      "learning_rate": 1.8492818580751414e-06,
+      "loss": 218.483,
+      "step": 41870
+    },
+    {
+      "epoch": 0.8054581908924373,
+      "grad_norm": 386.54951328148917,
+      "learning_rate": 1.8457474599802527e-06,
+      "loss": 203.4288,
+      "step": 41880
+    },
+    {
+      "epoch": 0.8056505161529178,
+      "grad_norm": 386.6269139117028,
+      "learning_rate": 1.842216099209767e-06,
+      "loss": 212.1619,
+      "step": 41890
+    },
+    {
+      "epoch": 0.8058428414133983,
+      "grad_norm": 362.9212830565881,
+      "learning_rate": 1.8386877770790524e-06,
+      "loss": 205.978,
+      "step": 41900
+    },
+    {
+      "epoch": 0.8060351666738789,
+      "grad_norm": 444.8007768212546,
+      "learning_rate": 1.8351624949023539e-06,
+      "loss": 214.794,
+      "step": 41910
+    },
+    {
+      "epoch": 0.8062274919343594,
+      "grad_norm": 374.81534673955224,
+      "learning_rate": 1.8316402539927757e-06,
+      "loss": 209.651,
+      "step": 41920
+    },
+    {
+      "epoch": 0.8064198171948399,
+      "grad_norm": 395.46132099994793,
+      "learning_rate": 1.8281210556623007e-06,
+      "loss": 202.8083,
+      "step": 41930
+    },
+    {
+      "epoch": 0.8066121424553204,
+      "grad_norm": 385.1851609238975,
+      "learning_rate": 1.8246049012217693e-06,
+      "loss": 203.7979,
+      "step": 41940
+    },
+    {
+      "epoch": 0.8068044677158009,
+      "grad_norm": 376.71585137162646,
+      "learning_rate": 1.8210917919808891e-06,
+      "loss": 213.5396,
+      "step": 41950
+    },
+    {
+      "epoch": 0.8069967929762815,
+      "grad_norm": 383.27855075608784,
+      "learning_rate": 1.817581729248239e-06,
+      "loss": 208.0222,
+      "step": 41960
+    },
+    {
+      "epoch": 0.807189118236762,
+      "grad_norm": 428.5429226169241,
+      "learning_rate": 1.8140747143312588e-06,
+      "loss": 206.4999,
+      "step": 41970
+    },
+    {
+      "epoch": 0.8073814434972425,
+      "grad_norm": 412.9275824586872,
+      "learning_rate": 1.8105707485362511e-06,
+      "loss": 201.0712,
+      "step": 41980
+    },
+    {
+      "epoch": 0.807573768757723,
+      "grad_norm": 357.2183740508322,
+      "learning_rate": 1.8070698331683844e-06,
+      "loss": 208.2365,
+      "step": 41990
+    },
+    {
+      "epoch": 0.8077660940182035,
+      "grad_norm": 371.07263761142406,
+      "learning_rate": 1.8035719695316955e-06,
+      "loss": 215.9635,
+      "step": 42000
+    },
+    {
+      "epoch": 0.8079584192786841,
+      "grad_norm": 370.5917801017485,
+      "learning_rate": 1.800077158929081e-06,
+      "loss": 208.0769,
+      "step": 42010
+    },
+    {
+      "epoch": 0.8081507445391647,
+      "grad_norm": 390.5821982735599,
+      "learning_rate": 1.7965854026622953e-06,
+      "loss": 205.1271,
+      "step": 42020
+    },
+    {
+      "epoch": 0.8083430697996452,
+      "grad_norm": 408.2207047930982,
+      "learning_rate": 1.7930967020319667e-06,
+      "loss": 211.125,
+      "step": 42030
+    },
+    {
+      "epoch": 0.8085353950601257,
+      "grad_norm": 387.5800192722426,
+      "learning_rate": 1.7896110583375747e-06,
+      "loss": 200.1964,
+      "step": 42040
+    },
+    {
+      "epoch": 0.8087277203206062,
+      "grad_norm": 339.61653075244897,
+      "learning_rate": 1.7861284728774652e-06,
+      "loss": 207.2948,
+      "step": 42050
+    },
+    {
+      "epoch": 0.8089200455810868,
+      "grad_norm": 374.41868452555894,
+      "learning_rate": 1.7826489469488395e-06,
+      "loss": 204.1716,
+      "step": 42060
+    },
+    {
+      "epoch": 0.8091123708415673,
+      "grad_norm": 389.567981799662,
+      "learning_rate": 1.7791724818477708e-06,
+      "loss": 204.8572,
+      "step": 42070
+    },
+    {
+      "epoch": 0.8093046961020478,
+      "grad_norm": 378.25298561747064,
+      "learning_rate": 1.7756990788691797e-06,
+      "loss": 203.1536,
+      "step": 42080
+    },
+    {
+      "epoch": 0.8094970213625283,
+      "grad_norm": 353.5955856533669,
+      "learning_rate": 1.772228739306854e-06,
+      "loss": 206.8032,
+      "step": 42090
+    },
+    {
+      "epoch": 0.8096893466230088,
+      "grad_norm": 405.7057134537505,
+      "learning_rate": 1.7687614644534333e-06,
+      "loss": 204.1162,
+      "step": 42100
+    },
+    {
+      "epoch": 0.8098816718834894,
+      "grad_norm": 417.5545628769309,
+      "learning_rate": 1.7652972556004267e-06,
+      "loss": 212.9713,
+      "step": 42110
+    },
+    {
+      "epoch": 0.8100739971439699,
+      "grad_norm": 400.07789244321077,
+      "learning_rate": 1.7618361140381922e-06,
+      "loss": 217.3094,
+      "step": 42120
+    },
+    {
+      "epoch": 0.8102663224044504,
+      "grad_norm": 399.8002995939122,
+      "learning_rate": 1.7583780410559449e-06,
+      "loss": 209.0708,
+      "step": 42130
+    },
+    {
+      "epoch": 0.8104586476649309,
+      "grad_norm": 375.89655749514503,
+      "learning_rate": 1.7549230379417636e-06,
+      "loss": 209.8407,
+      "step": 42140
+    },
+    {
+      "epoch": 0.8106509729254114,
+      "grad_norm": 407.82013037505374,
+      "learning_rate": 1.7514711059825773e-06,
+      "loss": 203.6781,
+      "step": 42150
+    },
+    {
+      "epoch": 0.810843298185892,
+      "grad_norm": 361.21082426808533,
+      "learning_rate": 1.7480222464641783e-06,
+      "loss": 219.0545,
+      "step": 42160
+    },
+    {
+      "epoch": 0.8110356234463725,
+      "grad_norm": 381.5994405974679,
+      "learning_rate": 1.7445764606712024e-06,
+      "loss": 204.033,
+      "step": 42170
+    },
+    {
+      "epoch": 0.811227948706853,
+      "grad_norm": 388.86811375164496,
+      "learning_rate": 1.7411337498871561e-06,
+      "loss": 205.6674,
+      "step": 42180
+    },
+    {
+      "epoch": 0.8114202739673335,
+      "grad_norm": 402.71771828181284,
+      "learning_rate": 1.737694115394387e-06,
+      "loss": 210.5151,
+      "step": 42190
+    },
+    {
+      "epoch": 0.811612599227814,
+      "grad_norm": 380.73564784427685,
+      "learning_rate": 1.7342575584741018e-06,
+      "loss": 206.2773,
+      "step": 42200
+    },
+    {
+      "epoch": 0.8118049244882946,
+      "grad_norm": 397.428208511356,
+      "learning_rate": 1.7308240804063648e-06,
+      "loss": 205.6393,
+      "step": 42210
+    },
+    {
+      "epoch": 0.8119972497487751,
+      "grad_norm": 351.08297928426407,
+      "learning_rate": 1.7273936824700888e-06,
+      "loss": 212.4962,
+      "step": 42220
+    },
+    {
+      "epoch": 0.8121895750092557,
+      "grad_norm": 348.61299077882325,
+      "learning_rate": 1.7239663659430384e-06,
+      "loss": 207.6494,
+      "step": 42230
+    },
+    {
+      "epoch": 0.8123819002697362,
+      "grad_norm": 404.09632401668375,
+      "learning_rate": 1.7205421321018312e-06,
+      "loss": 212.5592,
+      "step": 42240
+    },
+    {
+      "epoch": 0.8125742255302167,
+      "grad_norm": 353.0282682863881,
+      "learning_rate": 1.7171209822219427e-06,
+      "loss": 214.9213,
+      "step": 42250
+    },
+    {
+      "epoch": 0.8127665507906973,
+      "grad_norm": 374.0724173205953,
+      "learning_rate": 1.713702917577692e-06,
+      "loss": 210.6041,
+      "step": 42260
+    },
+    {
+      "epoch": 0.8129588760511778,
+      "grad_norm": 377.9211951109493,
+      "learning_rate": 1.71028793944225e-06,
+      "loss": 206.3823,
+      "step": 42270
+    },
+    {
+      "epoch": 0.8131512013116583,
+      "grad_norm": 369.4642192940653,
+      "learning_rate": 1.7068760490876425e-06,
+      "loss": 205.6524,
+      "step": 42280
+    },
+    {
+      "epoch": 0.8133435265721388,
+      "grad_norm": 364.6709525429443,
+      "learning_rate": 1.7034672477847402e-06,
+      "loss": 208.5839,
+      "step": 42290
+    },
+    {
+      "epoch": 0.8135358518326193,
+      "grad_norm": 363.41763280913244,
+      "learning_rate": 1.700061536803268e-06,
+      "loss": 204.9894,
+      "step": 42300
+    },
+    {
+      "epoch": 0.8137281770930999,
+      "grad_norm": 461.2436710840498,
+      "learning_rate": 1.696658917411793e-06,
+      "loss": 205.032,
+      "step": 42310
+    },
+    {
+      "epoch": 0.8139205023535804,
+      "grad_norm": 350.8760346611292,
+      "learning_rate": 1.6932593908777394e-06,
+      "loss": 209.3596,
+      "step": 42320
+    },
+    {
+      "epoch": 0.8141128276140609,
+      "grad_norm": 368.2903131814151,
+      "learning_rate": 1.689862958467372e-06,
+      "loss": 204.6411,
+      "step": 42330
+    },
+    {
+      "epoch": 0.8143051528745414,
+      "grad_norm": 399.36844148503786,
+      "learning_rate": 1.6864696214458065e-06,
+      "loss": 207.4226,
+      "step": 42340
+    },
+    {
+      "epoch": 0.8144974781350219,
+      "grad_norm": 371.03722013969127,
+      "learning_rate": 1.683079381077003e-06,
+      "loss": 211.7935,
+      "step": 42350
+    },
+    {
+      "epoch": 0.8146898033955025,
+      "grad_norm": 372.1826920492772,
+      "learning_rate": 1.6796922386237724e-06,
+      "loss": 214.8903,
+      "step": 42360
+    },
+    {
+      "epoch": 0.814882128655983,
+      "grad_norm": 399.4672720882924,
+      "learning_rate": 1.67630819534777e-06,
+      "loss": 208.9046,
+      "step": 42370
+    },
+    {
+      "epoch": 0.8150744539164635,
+      "grad_norm": 415.16717761610704,
+      "learning_rate": 1.6729272525094908e-06,
+      "loss": 208.0337,
+      "step": 42380
+    },
+    {
+      "epoch": 0.815266779176944,
+      "grad_norm": 389.78413077117796,
+      "learning_rate": 1.6695494113682874e-06,
+      "loss": 210.0916,
+      "step": 42390
+    },
+    {
+      "epoch": 0.8154591044374245,
+      "grad_norm": 388.33934171479245,
+      "learning_rate": 1.6661746731823458e-06,
+      "loss": 205.351,
+      "step": 42400
+    },
+    {
+      "epoch": 0.815651429697905,
+      "grad_norm": 367.4188093087156,
+      "learning_rate": 1.6628030392087001e-06,
+      "loss": 209.2656,
+      "step": 42410
+    },
+    {
+      "epoch": 0.8158437549583856,
+      "grad_norm": 369.82074700429,
+      "learning_rate": 1.6594345107032273e-06,
+      "loss": 212.7064,
+      "step": 42420
+    },
+    {
+      "epoch": 0.8160360802188662,
+      "grad_norm": 362.7657596873365,
+      "learning_rate": 1.6560690889206499e-06,
+      "loss": 210.4991,
+      "step": 42430
+    },
+    {
+      "epoch": 0.8162284054793467,
+      "grad_norm": 365.01707799881785,
+      "learning_rate": 1.6527067751145354e-06,
+      "loss": 201.3554,
+      "step": 42440
+    },
+    {
+      "epoch": 0.8164207307398272,
+      "grad_norm": 375.0802696652425,
+      "learning_rate": 1.6493475705372863e-06,
+      "loss": 207.5702,
+      "step": 42450
+    },
+    {
+      "epoch": 0.8166130560003078,
+      "grad_norm": 403.0493934491084,
+      "learning_rate": 1.6459914764401497e-06,
+      "loss": 213.1153,
+      "step": 42460
+    },
+    {
+      "epoch": 0.8168053812607883,
+      "grad_norm": 396.8085935361489,
+      "learning_rate": 1.64263849407322e-06,
+      "loss": 207.8117,
+      "step": 42470
+    },
+    {
+      "epoch": 0.8169977065212688,
+      "grad_norm": 360.9792495982528,
+      "learning_rate": 1.6392886246854234e-06,
+      "loss": 197.5292,
+      "step": 42480
+    },
+    {
+      "epoch": 0.8171900317817493,
+      "grad_norm": 399.14962520313526,
+      "learning_rate": 1.6359418695245311e-06,
+      "loss": 204.9792,
+      "step": 42490
+    },
+    {
+      "epoch": 0.8173823570422298,
+      "grad_norm": 387.88111820914077,
+      "learning_rate": 1.632598229837158e-06,
+      "loss": 207.9681,
+      "step": 42500
+    },
+    {
+      "epoch": 0.8175746823027104,
+      "grad_norm": 369.77251215534267,
+      "learning_rate": 1.629257706868751e-06,
+      "loss": 206.1137,
+      "step": 42510
+    },
+    {
+      "epoch": 0.8177670075631909,
+      "grad_norm": 360.0090283721679,
+      "learning_rate": 1.6259203018636016e-06,
+      "loss": 204.9569,
+      "step": 42520
+    },
+    {
+      "epoch": 0.8179593328236714,
+      "grad_norm": 392.34529343451965,
+      "learning_rate": 1.6225860160648343e-06,
+      "loss": 213.567,
+      "step": 42530
+    },
+    {
+      "epoch": 0.8181516580841519,
+      "grad_norm": 367.09983865534264,
+      "learning_rate": 1.6192548507144213e-06,
+      "loss": 210.2467,
+      "step": 42540
+    },
+    {
+      "epoch": 0.8183439833446324,
+      "grad_norm": 361.2438051846816,
+      "learning_rate": 1.6159268070531642e-06,
+      "loss": 203.0745,
+      "step": 42550
+    },
+    {
+      "epoch": 0.818536308605113,
+      "grad_norm": 415.82254238437633,
+      "learning_rate": 1.6126018863207005e-06,
+      "loss": 215.2535,
+      "step": 42560
+    },
+    {
+      "epoch": 0.8187286338655935,
+      "grad_norm": 374.52827279819735,
+      "learning_rate": 1.609280089755515e-06,
+      "loss": 208.4515,
+      "step": 42570
+    },
+    {
+      "epoch": 0.818920959126074,
+      "grad_norm": 394.0702665850043,
+      "learning_rate": 1.6059614185949157e-06,
+      "loss": 209.876,
+      "step": 42580
+    },
+    {
+      "epoch": 0.8191132843865545,
+      "grad_norm": 388.4257304191313,
+      "learning_rate": 1.6026458740750584e-06,
+      "loss": 198.6361,
+      "step": 42590
+    },
+    {
+      "epoch": 0.819305609647035,
+      "grad_norm": 358.9323469595191,
+      "learning_rate": 1.5993334574309238e-06,
+      "loss": 210.6616,
+      "step": 42600
+    },
+    {
+      "epoch": 0.8194979349075155,
+      "grad_norm": 368.07816743601427,
+      "learning_rate": 1.5960241698963374e-06,
+      "loss": 201.8158,
+      "step": 42610
+    },
+    {
+      "epoch": 0.8196902601679961,
+      "grad_norm": 371.98819391646555,
+      "learning_rate": 1.592718012703951e-06,
+      "loss": 206.5568,
+      "step": 42620
+    },
+    {
+      "epoch": 0.8198825854284766,
+      "grad_norm": 366.2847503045154,
+      "learning_rate": 1.589414987085255e-06,
+      "loss": 198.923,
+      "step": 42630
+    },
+    {
+      "epoch": 0.8200749106889572,
+      "grad_norm": 377.38426657334304,
+      "learning_rate": 1.5861150942705672e-06,
+      "loss": 204.7531,
+      "step": 42640
+    },
+    {
+      "epoch": 0.8202672359494377,
+      "grad_norm": 345.66962036636977,
+      "learning_rate": 1.5828183354890504e-06,
+      "loss": 199.6333,
+      "step": 42650
+    },
+    {
+      "epoch": 0.8204595612099183,
+      "grad_norm": 366.6233608059497,
+      "learning_rate": 1.5795247119686885e-06,
+      "loss": 200.4812,
+      "step": 42660
+    },
+    {
+      "epoch": 0.8206518864703988,
+      "grad_norm": 390.75195589742253,
+      "learning_rate": 1.5762342249363006e-06,
+      "loss": 212.3442,
+      "step": 42670
+    },
+    {
+      "epoch": 0.8208442117308793,
+      "grad_norm": 357.6144815084936,
+      "learning_rate": 1.5729468756175426e-06,
+      "loss": 199.1743,
+      "step": 42680
+    },
+    {
+      "epoch": 0.8210365369913598,
+      "grad_norm": 391.48648376608594,
+      "learning_rate": 1.5696626652368973e-06,
+      "loss": 203.6902,
+      "step": 42690
+    },
+    {
+      "epoch": 0.8212288622518403,
+      "grad_norm": 411.47456657443206,
+      "learning_rate": 1.5663815950176742e-06,
+      "loss": 206.812,
+      "step": 42700
+    },
+    {
+      "epoch": 0.8214211875123208,
+      "grad_norm": 403.3799944984759,
+      "learning_rate": 1.5631036661820232e-06,
+      "loss": 212.0593,
+      "step": 42710
+    },
+    {
+      "epoch": 0.8216135127728014,
+      "grad_norm": 352.70857191156745,
+      "learning_rate": 1.5598288799509153e-06,
+      "loss": 203.5993,
+      "step": 42720
+    },
+    {
+      "epoch": 0.8218058380332819,
+      "grad_norm": 399.7315480094786,
+      "learning_rate": 1.5565572375441573e-06,
+      "loss": 205.3522,
+      "step": 42730
+    },
+    {
+      "epoch": 0.8219981632937624,
+      "grad_norm": 420.6868483884786,
+      "learning_rate": 1.5532887401803787e-06,
+      "loss": 208.7667,
+      "step": 42740
+    },
+    {
+      "epoch": 0.8221904885542429,
+      "grad_norm": 385.3181638305602,
+      "learning_rate": 1.5500233890770434e-06,
+      "loss": 206.6352,
+      "step": 42750
+    },
+    {
+      "epoch": 0.8223828138147234,
+      "grad_norm": 377.90850155617335,
+      "learning_rate": 1.5467611854504406e-06,
+      "loss": 201.4559,
+      "step": 42760
+    },
+    {
+      "epoch": 0.822575139075204,
+      "grad_norm": 389.8834038356547,
+      "learning_rate": 1.5435021305156862e-06,
+      "loss": 208.0062,
+      "step": 42770
+    },
+    {
+      "epoch": 0.8227674643356845,
+      "grad_norm": 381.4462336169545,
+      "learning_rate": 1.5402462254867222e-06,
+      "loss": 213.6258,
+      "step": 42780
+    },
+    {
+      "epoch": 0.822959789596165,
+      "grad_norm": 405.7552092019749,
+      "learning_rate": 1.5369934715763235e-06,
+      "loss": 207.9729,
+      "step": 42790
+    },
+    {
+      "epoch": 0.8231521148566455,
+      "grad_norm": 361.5114360632329,
+      "learning_rate": 1.533743869996086e-06,
+      "loss": 202.8256,
+      "step": 42800
+    },
+    {
+      "epoch": 0.823344440117126,
+      "grad_norm": 377.7792467971178,
+      "learning_rate": 1.5304974219564318e-06,
+      "loss": 205.7259,
+      "step": 42810
+    },
+    {
+      "epoch": 0.8235367653776066,
+      "grad_norm": 391.09395136335803,
+      "learning_rate": 1.5272541286666075e-06,
+      "loss": 202.7261,
+      "step": 42820
+    },
+    {
+      "epoch": 0.8237290906380871,
+      "grad_norm": 411.43715038388666,
+      "learning_rate": 1.5240139913346906e-06,
+      "loss": 207.5069,
+      "step": 42830
+    },
+    {
+      "epoch": 0.8239214158985677,
+      "grad_norm": 418.0467421549725,
+      "learning_rate": 1.5207770111675735e-06,
+      "loss": 213.0044,
+      "step": 42840
+    },
+    {
+      "epoch": 0.8241137411590482,
+      "grad_norm": 348.3304180145024,
+      "learning_rate": 1.5175431893709836e-06,
+      "loss": 208.9385,
+      "step": 42850
+    },
+    {
+      "epoch": 0.8243060664195287,
+      "grad_norm": 382.98146330525566,
+      "learning_rate": 1.5143125271494607e-06,
+      "loss": 204.573,
+      "step": 42860
+    },
+    {
+      "epoch": 0.8244983916800093,
+      "grad_norm": 380.2728364984045,
+      "learning_rate": 1.511085025706378e-06,
+      "loss": 204.909,
+      "step": 42870
+    },
+    {
+      "epoch": 0.8246907169404898,
+      "grad_norm": 364.8850353229922,
+      "learning_rate": 1.5078606862439248e-06,
+      "loss": 203.9125,
+      "step": 42880
+    },
+    {
+      "epoch": 0.8248830422009703,
+      "grad_norm": 385.4312039366786,
+      "learning_rate": 1.5046395099631106e-06,
+      "loss": 207.736,
+      "step": 42890
+    },
+    {
+      "epoch": 0.8250753674614508,
+      "grad_norm": 415.9649915079839,
+      "learning_rate": 1.5014214980637754e-06,
+      "loss": 204.9748,
+      "step": 42900
+    },
+    {
+      "epoch": 0.8252676927219313,
+      "grad_norm": 382.41490357263086,
+      "learning_rate": 1.4982066517445748e-06,
+      "loss": 202.9691,
+      "step": 42910
+    },
+    {
+      "epoch": 0.8254600179824119,
+      "grad_norm": 425.19013791079897,
+      "learning_rate": 1.4949949722029811e-06,
+      "loss": 205.6067,
+      "step": 42920
+    },
+    {
+      "epoch": 0.8256523432428924,
+      "grad_norm": 389.78754848592973,
+      "learning_rate": 1.4917864606352983e-06,
+      "loss": 215.3066,
+      "step": 42930
+    },
+    {
+      "epoch": 0.8258446685033729,
+      "grad_norm": 378.0103159459695,
+      "learning_rate": 1.4885811182366406e-06,
+      "loss": 211.2018,
+      "step": 42940
+    },
+    {
+      "epoch": 0.8260369937638534,
+      "grad_norm": 374.5487429698657,
+      "learning_rate": 1.485378946200946e-06,
+      "loss": 206.4354,
+      "step": 42950
+    },
+    {
+      "epoch": 0.8262293190243339,
+      "grad_norm": 337.45494276169416,
+      "learning_rate": 1.4821799457209684e-06,
+      "loss": 203.6319,
+      "step": 42960
+    },
+    {
+      "epoch": 0.8264216442848145,
+      "grad_norm": 395.82782939758397,
+      "learning_rate": 1.478984117988287e-06,
+      "loss": 208.2151,
+      "step": 42970
+    },
+    {
+      "epoch": 0.826613969545295,
+      "grad_norm": 366.9216357376794,
+      "learning_rate": 1.4757914641932924e-06,
+      "loss": 202.5619,
+      "step": 42980
+    },
+    {
+      "epoch": 0.8268062948057755,
+      "grad_norm": 355.8624239101601,
+      "learning_rate": 1.4726019855251928e-06,
+      "loss": 208.0176,
+      "step": 42990
+    },
+    {
+      "epoch": 0.826998620066256,
+      "grad_norm": 354.09193575255347,
+      "learning_rate": 1.4694156831720185e-06,
+      "loss": 204.8916,
+      "step": 43000
+    },
+    {
+      "epoch": 0.8271909453267365,
+      "grad_norm": 378.4216589868182,
+      "learning_rate": 1.4662325583206172e-06,
+      "loss": 210.9614,
+      "step": 43010
+    },
+    {
+      "epoch": 0.8273832705872171,
+      "grad_norm": 354.49938820199424,
+      "learning_rate": 1.463052612156649e-06,
+      "loss": 208.6167,
+      "step": 43020
+    },
+    {
+      "epoch": 0.8275755958476976,
+      "grad_norm": 396.37355740819686,
+      "learning_rate": 1.4598758458645878e-06,
+      "loss": 208.938,
+      "step": 43030
+    },
+    {
+      "epoch": 0.8277679211081782,
+      "grad_norm": 395.3336639522541,
+      "learning_rate": 1.4567022606277314e-06,
+      "loss": 205.7786,
+      "step": 43040
+    },
+    {
+      "epoch": 0.8279602463686587,
+      "grad_norm": 386.8890530766187,
+      "learning_rate": 1.4535318576281854e-06,
+      "loss": 202.6189,
+      "step": 43050
+    },
+    {
+      "epoch": 0.8281525716291392,
+      "grad_norm": 377.7512675996208,
+      "learning_rate": 1.4503646380468729e-06,
+      "loss": 202.4282,
+      "step": 43060
+    },
+    {
+      "epoch": 0.8283448968896198,
+      "grad_norm": 409.4644380711561,
+      "learning_rate": 1.4472006030635288e-06,
+      "loss": 208.351,
+      "step": 43070
+    },
+    {
+      "epoch": 0.8285372221501003,
+      "grad_norm": 428.1367689688268,
+      "learning_rate": 1.4440397538567086e-06,
+      "loss": 210.8609,
+      "step": 43080
+    },
+    {
+      "epoch": 0.8287295474105808,
+      "grad_norm": 352.9025028590039,
+      "learning_rate": 1.4408820916037735e-06,
+      "loss": 204.6995,
+      "step": 43090
+    },
+    {
+      "epoch": 0.8289218726710613,
+      "grad_norm": 463.0241299362715,
+      "learning_rate": 1.4377276174808984e-06,
+      "loss": 204.3954,
+      "step": 43100
+    },
+    {
+      "epoch": 0.8291141979315418,
+      "grad_norm": 409.4859384209472,
+      "learning_rate": 1.4345763326630768e-06,
+      "loss": 206.5103,
+      "step": 43110
+    },
+    {
+      "epoch": 0.8293065231920224,
+      "grad_norm": 377.0778820136871,
+      "learning_rate": 1.4314282383241097e-06,
+      "loss": 204.2329,
+      "step": 43120
+    },
+    {
+      "epoch": 0.8294988484525029,
+      "grad_norm": 389.2982591056965,
+      "learning_rate": 1.4282833356366066e-06,
+      "loss": 204.242,
+      "step": 43130
+    },
+    {
+      "epoch": 0.8296911737129834,
+      "grad_norm": 351.0782059364485,
+      "learning_rate": 1.4251416257719962e-06,
+      "loss": 203.7137,
+      "step": 43140
+    },
+    {
+      "epoch": 0.8298834989734639,
+      "grad_norm": 411.833262302244,
+      "learning_rate": 1.4220031099005094e-06,
+      "loss": 206.6646,
+      "step": 43150
+    },
+    {
+      "epoch": 0.8300758242339444,
+      "grad_norm": 354.1078997029452,
+      "learning_rate": 1.4188677891911961e-06,
+      "loss": 198.2409,
+      "step": 43160
+    },
+    {
+      "epoch": 0.830268149494425,
+      "grad_norm": 385.26701070746,
+      "learning_rate": 1.4157356648119103e-06,
+      "loss": 201.8019,
+      "step": 43170
+    },
+    {
+      "epoch": 0.8304604747549055,
+      "grad_norm": 398.0017416010933,
+      "learning_rate": 1.412606737929313e-06,
+      "loss": 209.9437,
+      "step": 43180
+    },
+    {
+      "epoch": 0.830652800015386,
+      "grad_norm": 350.38366159555943,
+      "learning_rate": 1.4094810097088817e-06,
+      "loss": 206.9258,
+      "step": 43190
+    },
+    {
+      "epoch": 0.8308451252758665,
+      "grad_norm": 376.1112166408293,
+      "learning_rate": 1.4063584813148979e-06,
+      "loss": 204.1622,
+      "step": 43200
+    },
+    {
+      "epoch": 0.831037450536347,
+      "grad_norm": 381.8069347895772,
+      "learning_rate": 1.4032391539104484e-06,
+      "loss": 209.0268,
+      "step": 43210
+    },
+    {
+      "epoch": 0.8312297757968276,
+      "grad_norm": 366.5592143026922,
+      "learning_rate": 1.4001230286574363e-06,
+      "loss": 208.1606,
+      "step": 43220
+    },
+    {
+      "epoch": 0.8314221010573081,
+      "grad_norm": 399.8774281314007,
+      "learning_rate": 1.3970101067165642e-06,
+      "loss": 206.0998,
+      "step": 43230
+    },
+    {
+      "epoch": 0.8316144263177886,
+      "grad_norm": 377.794880820037,
+      "learning_rate": 1.3939003892473446e-06,
+      "loss": 203.2025,
+      "step": 43240
+    },
+    {
+      "epoch": 0.8318067515782692,
+      "grad_norm": 430.77880711968334,
+      "learning_rate": 1.390793877408093e-06,
+      "loss": 207.3121,
+      "step": 43250
+    },
+    {
+      "epoch": 0.8319990768387497,
+      "grad_norm": 365.30208837721693,
+      "learning_rate": 1.3876905723559397e-06,
+      "loss": 202.6982,
+      "step": 43260
+    },
+    {
+      "epoch": 0.8321914020992303,
+      "grad_norm": 348.6124368238359,
+      "learning_rate": 1.3845904752468075e-06,
+      "loss": 207.9971,
+      "step": 43270
+    },
+    {
+      "epoch": 0.8323837273597108,
+      "grad_norm": 426.8109912064802,
+      "learning_rate": 1.3814935872354385e-06,
+      "loss": 203.388,
+      "step": 43280
+    },
+    {
+      "epoch": 0.8325760526201913,
+      "grad_norm": 370.86422185206516,
+      "learning_rate": 1.3783999094753653e-06,
+      "loss": 205.4613,
+      "step": 43290
+    },
+    {
+      "epoch": 0.8327683778806718,
+      "grad_norm": 391.34146476262,
+      "learning_rate": 1.3753094431189385e-06,
+      "loss": 210.4081,
+      "step": 43300
+    },
+    {
+      "epoch": 0.8329607031411523,
+      "grad_norm": 363.34214430757095,
+      "learning_rate": 1.3722221893173027e-06,
+      "loss": 199.529,
+      "step": 43310
+    },
+    {
+      "epoch": 0.8331530284016329,
+      "grad_norm": 358.37837525956473,
+      "learning_rate": 1.369138149220407e-06,
+      "loss": 205.2364,
+      "step": 43320
+    },
+    {
+      "epoch": 0.8333453536621134,
+      "grad_norm": 407.2503065166098,
+      "learning_rate": 1.3660573239770091e-06,
+      "loss": 210.4062,
+      "step": 43330
+    },
+    {
+      "epoch": 0.8335376789225939,
+      "grad_norm": 381.8922203649801,
+      "learning_rate": 1.3629797147346635e-06,
+      "loss": 214.432,
+      "step": 43340
+    },
+    {
+      "epoch": 0.8337300041830744,
+      "grad_norm": 377.12143221474616,
+      "learning_rate": 1.3599053226397275e-06,
+      "loss": 209.5975,
+      "step": 43350
+    },
+    {
+      "epoch": 0.8339223294435549,
+      "grad_norm": 412.8418477082048,
+      "learning_rate": 1.3568341488373637e-06,
+      "loss": 208.5336,
+      "step": 43360
+    },
+    {
+      "epoch": 0.8341146547040355,
+      "grad_norm": 365.8775906786246,
+      "learning_rate": 1.3537661944715342e-06,
+      "loss": 204.7445,
+      "step": 43370
+    },
+    {
+      "epoch": 0.834306979964516,
+      "grad_norm": 354.70476910530317,
+      "learning_rate": 1.350701460685e-06,
+      "loss": 210.1551,
+      "step": 43380
+    },
+    {
+      "epoch": 0.8344993052249965,
+      "grad_norm": 393.22456570496587,
+      "learning_rate": 1.3476399486193214e-06,
+      "loss": 203.1828,
+      "step": 43390
+    },
+    {
+      "epoch": 0.834691630485477,
+      "grad_norm": 408.1931928165359,
+      "learning_rate": 1.3445816594148654e-06,
+      "loss": 214.0573,
+      "step": 43400
+    },
+    {
+      "epoch": 0.8348839557459575,
+      "grad_norm": 427.5986714768565,
+      "learning_rate": 1.3415265942107925e-06,
+      "loss": 210.9196,
+      "step": 43410
+    },
+    {
+      "epoch": 0.835076281006438,
+      "grad_norm": 345.37005481901633,
+      "learning_rate": 1.3384747541450615e-06,
+      "loss": 199.947,
+      "step": 43420
+    },
+    {
+      "epoch": 0.8352686062669186,
+      "grad_norm": 400.3090468199459,
+      "learning_rate": 1.3354261403544345e-06,
+      "loss": 206.9206,
+      "step": 43430
+    },
+    {
+      "epoch": 0.8354609315273991,
+      "grad_norm": 396.8117352224526,
+      "learning_rate": 1.3323807539744726e-06,
+      "loss": 215.036,
+      "step": 43440
+    },
+    {
+      "epoch": 0.8356532567878797,
+      "grad_norm": 388.14309009985914,
+      "learning_rate": 1.32933859613953e-06,
+      "loss": 207.7481,
+      "step": 43450
+    },
+    {
+      "epoch": 0.8358455820483602,
+      "grad_norm": 400.6428246283045,
+      "learning_rate": 1.3262996679827567e-06,
+      "loss": 208.6612,
+      "step": 43460
+    },
+    {
+      "epoch": 0.8360379073088408,
+      "grad_norm": 393.72973428537125,
+      "learning_rate": 1.3232639706361083e-06,
+      "loss": 207.2187,
+      "step": 43470
+    },
+    {
+      "epoch": 0.8362302325693213,
+      "grad_norm": 424.8615231714765,
+      "learning_rate": 1.3202315052303304e-06,
+      "loss": 197.9963,
+      "step": 43480
+    },
+    {
+      "epoch": 0.8364225578298018,
+      "grad_norm": 366.87618959401823,
+      "learning_rate": 1.3172022728949651e-06,
+      "loss": 206.0512,
+      "step": 43490
+    },
+    {
+      "epoch": 0.8366148830902823,
+      "grad_norm": 385.2714230521603,
+      "learning_rate": 1.3141762747583498e-06,
+      "loss": 200.3243,
+      "step": 43500
+    },
+    {
+      "epoch": 0.8368072083507628,
+      "grad_norm": 379.18669169817093,
+      "learning_rate": 1.3111535119476237e-06,
+      "loss": 206.8235,
+      "step": 43510
+    },
+    {
+      "epoch": 0.8369995336112434,
+      "grad_norm": 393.2839572386095,
+      "learning_rate": 1.3081339855887133e-06,
+      "loss": 212.3775,
+      "step": 43520
+    },
+    {
+      "epoch": 0.8371918588717239,
+      "grad_norm": 453.1823512102,
+      "learning_rate": 1.3051176968063407e-06,
+      "loss": 209.7109,
+      "step": 43530
+    },
+    {
+      "epoch": 0.8373841841322044,
+      "grad_norm": 389.20289934192317,
+      "learning_rate": 1.302104646724026e-06,
+      "loss": 205.8315,
+      "step": 43540
+    },
+    {
+      "epoch": 0.8375765093926849,
+      "grad_norm": 378.9919446636174,
+      "learning_rate": 1.299094836464081e-06,
+      "loss": 208.582,
+      "step": 43550
+    },
+    {
+      "epoch": 0.8377688346531654,
+      "grad_norm": 384.9649496654457,
+      "learning_rate": 1.2960882671476062e-06,
+      "loss": 203.4462,
+      "step": 43560
+    },
+    {
+      "epoch": 0.837961159913646,
+      "grad_norm": 346.4524821498244,
+      "learning_rate": 1.2930849398945033e-06,
+      "loss": 203.5101,
+      "step": 43570
+    },
+    {
+      "epoch": 0.8381534851741265,
+      "grad_norm": 363.42465926197104,
+      "learning_rate": 1.2900848558234625e-06,
+      "loss": 206.3717,
+      "step": 43580
+    },
+    {
+      "epoch": 0.838345810434607,
+      "grad_norm": 358.66666701920167,
+      "learning_rate": 1.2870880160519628e-06,
+      "loss": 206.9612,
+      "step": 43590
+    },
+    {
+      "epoch": 0.8385381356950875,
+      "grad_norm": 384.55230391340456,
+      "learning_rate": 1.2840944216962802e-06,
+      "loss": 209.1919,
+      "step": 43600
+    },
+    {
+      "epoch": 0.838730460955568,
+      "grad_norm": 364.057246624706,
+      "learning_rate": 1.2811040738714742e-06,
+      "loss": 204.3007,
+      "step": 43610
+    },
+    {
+      "epoch": 0.8389227862160485,
+      "grad_norm": 357.7533222636988,
+      "learning_rate": 1.2781169736914067e-06,
+      "loss": 202.8367,
+      "step": 43620
+    },
+    {
+      "epoch": 0.8391151114765291,
+      "grad_norm": 365.36288961938175,
+      "learning_rate": 1.275133122268719e-06,
+      "loss": 199.7843,
+      "step": 43630
+    },
+    {
+      "epoch": 0.8393074367370096,
+      "grad_norm": 372.39789154120996,
+      "learning_rate": 1.2721525207148456e-06,
+      "loss": 205.1193,
+      "step": 43640
+    },
+    {
+      "epoch": 0.8394997619974901,
+      "grad_norm": 392.8532414713598,
+      "learning_rate": 1.2691751701400145e-06,
+      "loss": 206.081,
+      "step": 43650
+    },
+    {
+      "epoch": 0.8396920872579707,
+      "grad_norm": 374.00260622159925,
+      "learning_rate": 1.2662010716532392e-06,
+      "loss": 206.8555,
+      "step": 43660
+    },
+    {
+      "epoch": 0.8398844125184513,
+      "grad_norm": 391.1200221624074,
+      "learning_rate": 1.2632302263623198e-06,
+      "loss": 200.8548,
+      "step": 43670
+    },
+    {
+      "epoch": 0.8400767377789318,
+      "grad_norm": 360.55748336623765,
+      "learning_rate": 1.2602626353738479e-06,
+      "loss": 207.4963,
+      "step": 43680
+    },
+    {
+      "epoch": 0.8402690630394123,
+      "grad_norm": 407.6834107541414,
+      "learning_rate": 1.2572982997932037e-06,
+      "loss": 210.2553,
+      "step": 43690
+    },
+    {
+      "epoch": 0.8404613882998928,
+      "grad_norm": 400.4503739308167,
+      "learning_rate": 1.2543372207245508e-06,
+      "loss": 205.8775,
+      "step": 43700
+    },
+    {
+      "epoch": 0.8406537135603733,
+      "grad_norm": 360.2130206354292,
+      "learning_rate": 1.2513793992708467e-06,
+      "loss": 210.1699,
+      "step": 43710
+    },
+    {
+      "epoch": 0.8408460388208538,
+      "grad_norm": 395.0485487885795,
+      "learning_rate": 1.2484248365338248e-06,
+      "loss": 206.1793,
+      "step": 43720
+    },
+    {
+      "epoch": 0.8410383640813344,
+      "grad_norm": 367.2410192807635,
+      "learning_rate": 1.2454735336140167e-06,
+      "loss": 207.9479,
+      "step": 43730
+    },
+    {
+      "epoch": 0.8412306893418149,
+      "grad_norm": 363.75268193113,
+      "learning_rate": 1.2425254916107321e-06,
+      "loss": 205.6623,
+      "step": 43740
+    },
+    {
+      "epoch": 0.8414230146022954,
+      "grad_norm": 371.9467390396526,
+      "learning_rate": 1.2395807116220648e-06,
+      "loss": 207.7912,
+      "step": 43750
+    },
+    {
+      "epoch": 0.8416153398627759,
+      "grad_norm": 387.1542990519759,
+      "learning_rate": 1.236639194744902e-06,
+      "loss": 210.2121,
+      "step": 43760
+    },
+    {
+      "epoch": 0.8418076651232564,
+      "grad_norm": 406.63128959488984,
+      "learning_rate": 1.233700942074907e-06,
+      "loss": 205.1486,
+      "step": 43770
+    },
+    {
+      "epoch": 0.841999990383737,
+      "grad_norm": 345.39095881535553,
+      "learning_rate": 1.230765954706531e-06,
+      "loss": 202.5575,
+      "step": 43780
+    },
+    {
+      "epoch": 0.8421923156442175,
+      "grad_norm": 400.62383242473885,
+      "learning_rate": 1.227834233733005e-06,
+      "loss": 209.2507,
+      "step": 43790
+    },
+    {
+      "epoch": 0.842384640904698,
+      "grad_norm": 363.9228370647116,
+      "learning_rate": 1.2249057802463527e-06,
+      "loss": 206.0571,
+      "step": 43800
+    },
+    {
+      "epoch": 0.8425769661651785,
+      "grad_norm": 356.91017574922256,
+      "learning_rate": 1.221980595337372e-06,
+      "loss": 204.8864,
+      "step": 43810
+    },
+    {
+      "epoch": 0.842769291425659,
+      "grad_norm": 373.6667604329769,
+      "learning_rate": 1.219058680095644e-06,
+      "loss": 207.9622,
+      "step": 43820
+    },
+    {
+      "epoch": 0.8429616166861396,
+      "grad_norm": 367.0553791020593,
+      "learning_rate": 1.2161400356095376e-06,
+      "loss": 200.6157,
+      "step": 43830
+    },
+    {
+      "epoch": 0.8431539419466201,
+      "grad_norm": 359.1084804345651,
+      "learning_rate": 1.2132246629661948e-06,
+      "loss": 204.2739,
+      "step": 43840
+    },
+    {
+      "epoch": 0.8433462672071006,
+      "grad_norm": 355.012140131159,
+      "learning_rate": 1.21031256325155e-06,
+      "loss": 206.1288,
+      "step": 43850
+    },
+    {
+      "epoch": 0.8435385924675812,
+      "grad_norm": 409.81794687657094,
+      "learning_rate": 1.2074037375503056e-06,
+      "loss": 198.1729,
+      "step": 43860
+    },
+    {
+      "epoch": 0.8437309177280617,
+      "grad_norm": 379.0431068053593,
+      "learning_rate": 1.2044981869459571e-06,
+      "loss": 210.0953,
+      "step": 43870
+    },
+    {
+      "epoch": 0.8439232429885423,
+      "grad_norm": 377.71797648753795,
+      "learning_rate": 1.201595912520771e-06,
+      "loss": 208.7903,
+      "step": 43880
+    },
+    {
+      "epoch": 0.8441155682490228,
+      "grad_norm": 379.4799052424159,
+      "learning_rate": 1.198696915355796e-06,
+      "loss": 215.2234,
+      "step": 43890
+    },
+    {
+      "epoch": 0.8443078935095033,
+      "grad_norm": 395.77615052504984,
+      "learning_rate": 1.1958011965308624e-06,
+      "loss": 210.4527,
+      "step": 43900
+    },
+    {
+      "epoch": 0.8445002187699838,
+      "grad_norm": 392.10683098479126,
+      "learning_rate": 1.192908757124578e-06,
+      "loss": 206.6592,
+      "step": 43910
+    },
+    {
+      "epoch": 0.8446925440304643,
+      "grad_norm": 362.38496475020617,
+      "learning_rate": 1.190019598214327e-06,
+      "loss": 208.6194,
+      "step": 43920
+    },
+    {
+      "epoch": 0.8448848692909449,
+      "grad_norm": 409.81221917113635,
+      "learning_rate": 1.1871337208762723e-06,
+      "loss": 202.5024,
+      "step": 43930
+    },
+    {
+      "epoch": 0.8450771945514254,
+      "grad_norm": 417.1247358705236,
+      "learning_rate": 1.1842511261853596e-06,
+      "loss": 210.2868,
+      "step": 43940
+    },
+    {
+      "epoch": 0.8452695198119059,
+      "grad_norm": 372.9501483526634,
+      "learning_rate": 1.1813718152153054e-06,
+      "loss": 202.5984,
+      "step": 43950
+    },
+    {
+      "epoch": 0.8454618450723864,
+      "grad_norm": 389.8741194615128,
+      "learning_rate": 1.1784957890386051e-06,
+      "loss": 204.0407,
+      "step": 43960
+    },
+    {
+      "epoch": 0.8456541703328669,
+      "grad_norm": 363.2134290907783,
+      "learning_rate": 1.1756230487265296e-06,
+      "loss": 203.0969,
+      "step": 43970
+    },
+    {
+      "epoch": 0.8458464955933475,
+      "grad_norm": 392.96502388654324,
+      "learning_rate": 1.1727535953491308e-06,
+      "loss": 204.6773,
+      "step": 43980
+    },
+    {
+      "epoch": 0.846038820853828,
+      "grad_norm": 367.1321391536074,
+      "learning_rate": 1.1698874299752293e-06,
+      "loss": 199.7048,
+      "step": 43990
+    },
+    {
+      "epoch": 0.8462311461143085,
+      "grad_norm": 354.87337064746697,
+      "learning_rate": 1.1670245536724267e-06,
+      "loss": 205.9254,
+      "step": 44000
+    },
+    {
+      "epoch": 0.846423471374789,
+      "grad_norm": 367.77289151048666,
+      "learning_rate": 1.1641649675070975e-06,
+      "loss": 207.6062,
+      "step": 44010
+    },
+    {
+      "epoch": 0.8466157966352695,
+      "grad_norm": 412.32926517504535,
+      "learning_rate": 1.161308672544389e-06,
+      "loss": 219.878,
+      "step": 44020
+    },
+    {
+      "epoch": 0.84680812189575,
+      "grad_norm": 381.3936188019946,
+      "learning_rate": 1.1584556698482252e-06,
+      "loss": 204.3105,
+      "step": 44030
+    },
+    {
+      "epoch": 0.8470004471562306,
+      "grad_norm": 393.4578562868416,
+      "learning_rate": 1.1556059604812985e-06,
+      "loss": 203.6952,
+      "step": 44040
+    },
+    {
+      "epoch": 0.8471927724167111,
+      "grad_norm": 378.85770801966174,
+      "learning_rate": 1.1527595455050844e-06,
+      "loss": 206.0731,
+      "step": 44050
+    },
+    {
+      "epoch": 0.8473850976771916,
+      "grad_norm": 428.11879829537617,
+      "learning_rate": 1.1499164259798223e-06,
+      "loss": 206.6966,
+      "step": 44060
+    },
+    {
+      "epoch": 0.8475774229376722,
+      "grad_norm": 344.52758866783785,
+      "learning_rate": 1.1470766029645253e-06,
+      "loss": 209.6004,
+      "step": 44070
+    },
+    {
+      "epoch": 0.8477697481981528,
+      "grad_norm": 391.69909105604853,
+      "learning_rate": 1.1442400775169849e-06,
+      "loss": 208.7681,
+      "step": 44080
+    },
+    {
+      "epoch": 0.8479620734586333,
+      "grad_norm": 350.48260471174507,
+      "learning_rate": 1.141406850693757e-06,
+      "loss": 199.2871,
+      "step": 44090
+    },
+    {
+      "epoch": 0.8481543987191138,
+      "grad_norm": 360.88210862046384,
+      "learning_rate": 1.1385769235501742e-06,
+      "loss": 209.238,
+      "step": 44100
+    },
+    {
+      "epoch": 0.8483467239795943,
+      "grad_norm": 404.2304100493653,
+      "learning_rate": 1.1357502971403335e-06,
+      "loss": 208.8373,
+      "step": 44110
+    },
+    {
+      "epoch": 0.8485390492400748,
+      "grad_norm": 382.13975041926614,
+      "learning_rate": 1.132926972517111e-06,
+      "loss": 203.786,
+      "step": 44120
+    },
+    {
+      "epoch": 0.8487313745005554,
+      "grad_norm": 425.13146544979304,
+      "learning_rate": 1.130106950732145e-06,
+      "loss": 201.2365,
+      "step": 44130
+    },
+    {
+      "epoch": 0.8489236997610359,
+      "grad_norm": 377.39230241799714,
+      "learning_rate": 1.1272902328358514e-06,
+      "loss": 205.3801,
+      "step": 44140
+    },
+    {
+      "epoch": 0.8491160250215164,
+      "grad_norm": 427.3905961534483,
+      "learning_rate": 1.1244768198774047e-06,
+      "loss": 212.7839,
+      "step": 44150
+    },
+    {
+      "epoch": 0.8493083502819969,
+      "grad_norm": 414.7022196347994,
+      "learning_rate": 1.121666712904762e-06,
+      "loss": 223.0878,
+      "step": 44160
+    },
+    {
+      "epoch": 0.8495006755424774,
+      "grad_norm": 351.3713145163187,
+      "learning_rate": 1.1188599129646382e-06,
+      "loss": 205.506,
+      "step": 44170
+    },
+    {
+      "epoch": 0.849693000802958,
+      "grad_norm": 415.7091690217121,
+      "learning_rate": 1.116056421102517e-06,
+      "loss": 213.8812,
+      "step": 44180
+    },
+    {
+      "epoch": 0.8498853260634385,
+      "grad_norm": 367.5481271184139,
+      "learning_rate": 1.113256238362659e-06,
+      "loss": 210.2667,
+      "step": 44190
+    },
+    {
+      "epoch": 0.850077651323919,
+      "grad_norm": 405.32896873771574,
+      "learning_rate": 1.1104593657880812e-06,
+      "loss": 205.0123,
+      "step": 44200
+    },
+    {
+      "epoch": 0.8502699765843995,
+      "grad_norm": 404.8301095268004,
+      "learning_rate": 1.1076658044205746e-06,
+      "loss": 212.5438,
+      "step": 44210
+    },
+    {
+      "epoch": 0.85046230184488,
+      "grad_norm": 378.41032147574657,
+      "learning_rate": 1.1048755553006928e-06,
+      "loss": 211.2558,
+      "step": 44220
+    },
+    {
+      "epoch": 0.8506546271053606,
+      "grad_norm": 365.29816154303506,
+      "learning_rate": 1.1020886194677605e-06,
+      "loss": 201.2551,
+      "step": 44230
+    },
+    {
+      "epoch": 0.8508469523658411,
+      "grad_norm": 381.32882627359896,
+      "learning_rate": 1.0993049979598635e-06,
+      "loss": 206.3111,
+      "step": 44240
+    },
+    {
+      "epoch": 0.8510392776263216,
+      "grad_norm": 372.40198755307824,
+      "learning_rate": 1.0965246918138529e-06,
+      "loss": 213.805,
+      "step": 44250
+    },
+    {
+      "epoch": 0.8512316028868021,
+      "grad_norm": 374.1485402508953,
+      "learning_rate": 1.093747702065351e-06,
+      "loss": 204.924,
+      "step": 44260
+    },
+    {
+      "epoch": 0.8514239281472827,
+      "grad_norm": 378.64906319014085,
+      "learning_rate": 1.090974029748736e-06,
+      "loss": 207.0282,
+      "step": 44270
+    },
+    {
+      "epoch": 0.8516162534077633,
+      "grad_norm": 369.9100250219373,
+      "learning_rate": 1.0882036758971592e-06,
+      "loss": 205.483,
+      "step": 44280
+    },
+    {
+      "epoch": 0.8518085786682438,
+      "grad_norm": 377.76284828261345,
+      "learning_rate": 1.0854366415425289e-06,
+      "loss": 214.42,
+      "step": 44290
+    },
+    {
+      "epoch": 0.8520009039287243,
+      "grad_norm": 332.81151595857335,
+      "learning_rate": 1.0826729277155224e-06,
+      "loss": 205.6704,
+      "step": 44300
+    },
+    {
+      "epoch": 0.8521932291892048,
+      "grad_norm": 392.98028017869154,
+      "learning_rate": 1.0799125354455752e-06,
+      "loss": 213.0135,
+      "step": 44310
+    },
+    {
+      "epoch": 0.8523855544496853,
+      "grad_norm": 373.1936930199224,
+      "learning_rate": 1.0771554657608896e-06,
+      "loss": 206.1825,
+      "step": 44320
+    },
+    {
+      "epoch": 0.8525778797101659,
+      "grad_norm": 410.2686245364778,
+      "learning_rate": 1.0744017196884248e-06,
+      "loss": 204.4684,
+      "step": 44330
+    },
+    {
+      "epoch": 0.8527702049706464,
+      "grad_norm": 374.00561560123856,
+      "learning_rate": 1.0716512982539106e-06,
+      "loss": 201.9522,
+      "step": 44340
+    },
+    {
+      "epoch": 0.8529625302311269,
+      "grad_norm": 362.53881544070924,
+      "learning_rate": 1.0689042024818307e-06,
+      "loss": 205.2763,
+      "step": 44350
+    },
+    {
+      "epoch": 0.8531548554916074,
+      "grad_norm": 438.3463551343566,
+      "learning_rate": 1.0661604333954312e-06,
+      "loss": 210.8474,
+      "step": 44360
+    },
+    {
+      "epoch": 0.8533471807520879,
+      "grad_norm": 345.4678441637618,
+      "learning_rate": 1.0634199920167255e-06,
+      "loss": 202.4046,
+      "step": 44370
+    },
+    {
+      "epoch": 0.8535395060125684,
+      "grad_norm": 372.98725728652727,
+      "learning_rate": 1.0606828793664804e-06,
+      "loss": 204.8702,
+      "step": 44380
+    },
+    {
+      "epoch": 0.853731831273049,
+      "grad_norm": 403.59533892318296,
+      "learning_rate": 1.0579490964642247e-06,
+      "loss": 207.6489,
+      "step": 44390
+    },
+    {
+      "epoch": 0.8539241565335295,
+      "grad_norm": 406.1277687126905,
+      "learning_rate": 1.0552186443282464e-06,
+      "loss": 205.9058,
+      "step": 44400
+    },
+    {
+      "epoch": 0.85411648179401,
+      "grad_norm": 359.94142885986656,
+      "learning_rate": 1.0524915239755939e-06,
+      "loss": 200.386,
+      "step": 44410
+    },
+    {
+      "epoch": 0.8543088070544905,
+      "grad_norm": 402.329057345141,
+      "learning_rate": 1.0497677364220792e-06,
+      "loss": 205.3322,
+      "step": 44420
+    },
+    {
+      "epoch": 0.854501132314971,
+      "grad_norm": 350.8751641643714,
+      "learning_rate": 1.047047282682262e-06,
+      "loss": 210.0956,
+      "step": 44430
+    },
+    {
+      "epoch": 0.8546934575754516,
+      "grad_norm": 387.1363118151032,
+      "learning_rate": 1.0443301637694713e-06,
+      "loss": 212.7379,
+      "step": 44440
+    },
+    {
+      "epoch": 0.8548857828359321,
+      "grad_norm": 390.1386210830029,
+      "learning_rate": 1.0416163806957857e-06,
+      "loss": 197.3423,
+      "step": 44450
+    },
+    {
+      "epoch": 0.8550781080964126,
+      "grad_norm": 386.67706049782305,
+      "learning_rate": 1.0389059344720475e-06,
+      "loss": 206.0977,
+      "step": 44460
+    },
+    {
+      "epoch": 0.8552704333568931,
+      "grad_norm": 361.4399360569785,
+      "learning_rate": 1.0361988261078482e-06,
+      "loss": 202.2557,
+      "step": 44470
+    },
+    {
+      "epoch": 0.8554627586173738,
+      "grad_norm": 374.41295180218486,
+      "learning_rate": 1.0334950566115466e-06,
+      "loss": 200.1122,
+      "step": 44480
+    },
+    {
+      "epoch": 0.8556550838778543,
+      "grad_norm": 375.90329301969854,
+      "learning_rate": 1.0307946269902492e-06,
+      "loss": 206.7902,
+      "step": 44490
+    },
+    {
+      "epoch": 0.8558474091383348,
+      "grad_norm": 358.0379686092457,
+      "learning_rate": 1.0280975382498225e-06,
+      "loss": 207.3438,
+      "step": 44500
+    },
+    {
+      "epoch": 0.8560397343988153,
+      "grad_norm": 370.3599343400562,
+      "learning_rate": 1.0254037913948845e-06,
+      "loss": 195.9171,
+      "step": 44510
+    },
+    {
+      "epoch": 0.8562320596592958,
+      "grad_norm": 384.37643309280776,
+      "learning_rate": 1.0227133874288152e-06,
+      "loss": 210.6145,
+      "step": 44520
+    },
+    {
+      "epoch": 0.8564243849197763,
+      "grad_norm": 361.6373646381273,
+      "learning_rate": 1.0200263273537458e-06,
+      "loss": 205.4525,
+      "step": 44530
+    },
+    {
+      "epoch": 0.8566167101802569,
+      "grad_norm": 368.8958150630338,
+      "learning_rate": 1.0173426121705577e-06,
+      "loss": 209.8104,
+      "step": 44540
+    },
+    {
+      "epoch": 0.8568090354407374,
+      "grad_norm": 382.9865762354701,
+      "learning_rate": 1.0146622428788943e-06,
+      "loss": 206.8092,
+      "step": 44550
+    },
+    {
+      "epoch": 0.8570013607012179,
+      "grad_norm": 393.93271411196594,
+      "learning_rate": 1.0119852204771463e-06,
+      "loss": 213.0676,
+      "step": 44560
+    },
+    {
+      "epoch": 0.8571936859616984,
+      "grad_norm": 358.78545290721587,
+      "learning_rate": 1.0093115459624637e-06,
+      "loss": 203.6277,
+      "step": 44570
+    },
+    {
+      "epoch": 0.857386011222179,
+      "grad_norm": 353.05483385192855,
+      "learning_rate": 1.0066412203307419e-06,
+      "loss": 210.3207,
+      "step": 44580
+    },
+    {
+      "epoch": 0.8575783364826595,
+      "grad_norm": 397.2383677595859,
+      "learning_rate": 1.0039742445766376e-06,
+      "loss": 204.5051,
+      "step": 44590
+    },
+    {
+      "epoch": 0.85777066174314,
+      "grad_norm": 382.3735376840602,
+      "learning_rate": 1.0013106196935528e-06,
+      "loss": 210.2912,
+      "step": 44600
+    },
+    {
+      "epoch": 0.8579629870036205,
+      "grad_norm": 410.49734290936505,
+      "learning_rate": 9.986503466736419e-07,
+      "loss": 202.3068,
+      "step": 44610
+    },
+    {
+      "epoch": 0.858155312264101,
+      "grad_norm": 373.4030813457647,
+      "learning_rate": 9.959934265078176e-07,
+      "loss": 212.7593,
+      "step": 44620
+    },
+    {
+      "epoch": 0.8583476375245815,
+      "grad_norm": 413.0165296247096,
+      "learning_rate": 9.933398601857347e-07,
+      "loss": 203.608,
+      "step": 44630
+    },
+    {
+      "epoch": 0.8585399627850621,
+      "grad_norm": 373.5339538200506,
+      "learning_rate": 9.90689648695804e-07,
+      "loss": 202.5312,
+      "step": 44640
+    },
+    {
+      "epoch": 0.8587322880455426,
+      "grad_norm": 390.02380358076636,
+      "learning_rate": 9.880427930251834e-07,
+      "loss": 203.2612,
+      "step": 44650
+    },
+    {
+      "epoch": 0.8589246133060231,
+      "grad_norm": 356.10087692230826,
+      "learning_rate": 9.853992941597878e-07,
+      "loss": 203.1021,
+      "step": 44660
+    },
+    {
+      "epoch": 0.8591169385665036,
+      "grad_norm": 364.3438617370437,
+      "learning_rate": 9.827591530842729e-07,
+      "loss": 201.0466,
+      "step": 44670
+    },
+    {
+      "epoch": 0.8593092638269842,
+      "grad_norm": 343.3440077623596,
+      "learning_rate": 9.801223707820484e-07,
+      "loss": 202.8622,
+      "step": 44680
+    },
+    {
+      "epoch": 0.8595015890874648,
+      "grad_norm": 363.3457369517634,
+      "learning_rate": 9.774889482352735e-07,
+      "loss": 200.0119,
+      "step": 44690
+    },
+    {
+      "epoch": 0.8596939143479453,
+      "grad_norm": 338.7743387512505,
+      "learning_rate": 9.74858886424852e-07,
+      "loss": 202.2917,
+      "step": 44700
+    },
+    {
+      "epoch": 0.8598862396084258,
+      "grad_norm": 404.3062313829463,
+      "learning_rate": 9.722321863304418e-07,
+      "loss": 208.2456,
+      "step": 44710
+    },
+    {
+      "epoch": 0.8600785648689063,
+      "grad_norm": 396.1293156302092,
+      "learning_rate": 9.696088489304412e-07,
+      "loss": 208.8752,
+      "step": 44720
+    },
+    {
+      "epoch": 0.8602708901293868,
+      "grad_norm": 367.6745241601311,
+      "learning_rate": 9.669888752020061e-07,
+      "loss": 204.5631,
+      "step": 44730
+    },
+    {
+      "epoch": 0.8604632153898674,
+      "grad_norm": 349.8437209042745,
+      "learning_rate": 9.643722661210285e-07,
+      "loss": 197.7485,
+      "step": 44740
+    },
+    {
+      "epoch": 0.8606555406503479,
+      "grad_norm": 354.87852525496646,
+      "learning_rate": 9.617590226621543e-07,
+      "loss": 202.4761,
+      "step": 44750
+    },
+    {
+      "epoch": 0.8608478659108284,
+      "grad_norm": 366.24691164947046,
+      "learning_rate": 9.5914914579877e-07,
+      "loss": 200.2327,
+      "step": 44760
+    },
+    {
+      "epoch": 0.8610401911713089,
+      "grad_norm": 375.11075292850165,
+      "learning_rate": 9.565426365030172e-07,
+      "loss": 206.2503,
+      "step": 44770
+    },
+    {
+      "epoch": 0.8612325164317894,
+      "grad_norm": 346.9034725974352,
+      "learning_rate": 9.539394957457737e-07,
+      "loss": 212.3791,
+      "step": 44780
+    },
+    {
+      "epoch": 0.86142484169227,
+      "grad_norm": 334.45724346266996,
+      "learning_rate": 9.51339724496666e-07,
+      "loss": 203.4135,
+      "step": 44790
+    },
+    {
+      "epoch": 0.8616171669527505,
+      "grad_norm": 333.1165354642953,
+      "learning_rate": 9.487433237240695e-07,
+      "loss": 200.082,
+      "step": 44800
+    },
+    {
+      "epoch": 0.861809492213231,
+      "grad_norm": 376.32224355143666,
+      "learning_rate": 9.461502943950973e-07,
+      "loss": 210.6318,
+      "step": 44810
+    },
+    {
+      "epoch": 0.8620018174737115,
+      "grad_norm": 322.53950362849884,
+      "learning_rate": 9.435606374756123e-07,
+      "loss": 197.1338,
+      "step": 44820
+    },
+    {
+      "epoch": 0.862194142734192,
+      "grad_norm": 367.08690999989363,
+      "learning_rate": 9.409743539302152e-07,
+      "loss": 197.6756,
+      "step": 44830
+    },
+    {
+      "epoch": 0.8623864679946726,
+      "grad_norm": 389.54716928078994,
+      "learning_rate": 9.383914447222576e-07,
+      "loss": 206.0583,
+      "step": 44840
+    },
+    {
+      "epoch": 0.8625787932551531,
+      "grad_norm": 395.36836908466665,
+      "learning_rate": 9.358119108138309e-07,
+      "loss": 207.0507,
+      "step": 44850
+    },
+    {
+      "epoch": 0.8627711185156336,
+      "grad_norm": 364.116585749919,
+      "learning_rate": 9.332357531657644e-07,
+      "loss": 197.6354,
+      "step": 44860
+    },
+    {
+      "epoch": 0.8629634437761141,
+      "grad_norm": 351.18462138856256,
+      "learning_rate": 9.306629727376404e-07,
+      "loss": 203.6384,
+      "step": 44870
+    },
+    {
+      "epoch": 0.8631557690365947,
+      "grad_norm": 383.54192358548283,
+      "learning_rate": 9.280935704877736e-07,
+      "loss": 200.413,
+      "step": 44880
+    },
+    {
+      "epoch": 0.8633480942970753,
+      "grad_norm": 401.6479467350847,
+      "learning_rate": 9.255275473732239e-07,
+      "loss": 207.4342,
+      "step": 44890
+    },
+    {
+      "epoch": 0.8635404195575558,
+      "grad_norm": 382.1092868960059,
+      "learning_rate": 9.229649043497924e-07,
+      "loss": 207.5034,
+      "step": 44900
+    },
+    {
+      "epoch": 0.8637327448180363,
+      "grad_norm": 425.518572987113,
+      "learning_rate": 9.20405642372022e-07,
+      "loss": 208.9183,
+      "step": 44910
+    },
+    {
+      "epoch": 0.8639250700785168,
+      "grad_norm": 422.7285676914632,
+      "learning_rate": 9.178497623931959e-07,
+      "loss": 206.2202,
+      "step": 44920
+    },
+    {
+      "epoch": 0.8641173953389973,
+      "grad_norm": 383.11994673804685,
+      "learning_rate": 9.152972653653369e-07,
+      "loss": 202.4289,
+      "step": 44930
+    },
+    {
+      "epoch": 0.8643097205994779,
+      "grad_norm": 384.1125325853366,
+      "learning_rate": 9.127481522392068e-07,
+      "loss": 206.058,
+      "step": 44940
+    },
+    {
+      "epoch": 0.8645020458599584,
+      "grad_norm": 366.9270733253421,
+      "learning_rate": 9.102024239643092e-07,
+      "loss": 204.4223,
+      "step": 44950
+    },
+    {
+      "epoch": 0.8646943711204389,
+      "grad_norm": 424.3447502841698,
+      "learning_rate": 9.076600814888869e-07,
+      "loss": 206.8118,
+      "step": 44960
+    },
+    {
+      "epoch": 0.8648866963809194,
+      "grad_norm": 367.60663290787295,
+      "learning_rate": 9.051211257599169e-07,
+      "loss": 207.4621,
+      "step": 44970
+    },
+    {
+      "epoch": 0.8650790216413999,
+      "grad_norm": 345.05789977402355,
+      "learning_rate": 9.025855577231224e-07,
+      "loss": 206.2592,
+      "step": 44980
+    },
+    {
+      "epoch": 0.8652713469018805,
+      "grad_norm": 379.141354890545,
+      "learning_rate": 9.000533783229581e-07,
+      "loss": 207.8849,
+      "step": 44990
+    },
+    {
+      "epoch": 0.865463672162361,
+      "grad_norm": 349.9815884882839,
+      "learning_rate": 8.975245885026207e-07,
+      "loss": 209.9808,
+      "step": 45000
+    },
+    {
+      "epoch": 0.8656559974228415,
+      "grad_norm": 432.743954605365,
+      "learning_rate": 8.949991892040399e-07,
+      "loss": 207.3274,
+      "step": 45010
+    },
+    {
+      "epoch": 0.865848322683322,
+      "grad_norm": 402.31527873343225,
+      "learning_rate": 8.92477181367889e-07,
+      "loss": 201.7501,
+      "step": 45020
+    },
+    {
+      "epoch": 0.8660406479438025,
+      "grad_norm": 358.62038236501905,
+      "learning_rate": 8.899585659335719e-07,
+      "loss": 203.6723,
+      "step": 45030
+    },
+    {
+      "epoch": 0.866232973204283,
+      "grad_norm": 414.9560503106033,
+      "learning_rate": 8.874433438392305e-07,
+      "loss": 219.7011,
+      "step": 45040
+    },
+    {
+      "epoch": 0.8664252984647636,
+      "grad_norm": 365.7653499690012,
+      "learning_rate": 8.849315160217465e-07,
+      "loss": 204.9315,
+      "step": 45050
+    },
+    {
+      "epoch": 0.8666176237252441,
+      "grad_norm": 401.5348786535948,
+      "learning_rate": 8.824230834167325e-07,
+      "loss": 204.5615,
+      "step": 45060
+    },
+    {
+      "epoch": 0.8668099489857246,
+      "grad_norm": 359.1950619519764,
+      "learning_rate": 8.799180469585378e-07,
+      "loss": 216.4338,
+      "step": 45070
+    },
+    {
+      "epoch": 0.8670022742462051,
+      "grad_norm": 368.6054977733908,
+      "learning_rate": 8.77416407580246e-07,
+      "loss": 204.5957,
+      "step": 45080
+    },
+    {
+      "epoch": 0.8671945995066858,
+      "grad_norm": 390.0618423988173,
+      "learning_rate": 8.749181662136785e-07,
+      "loss": 210.4847,
+      "step": 45090
+    },
+    {
+      "epoch": 0.8673869247671663,
+      "grad_norm": 386.23430742941576,
+      "learning_rate": 8.724233237893897e-07,
+      "loss": 203.7169,
+      "step": 45100
+    },
+    {
+      "epoch": 0.8675792500276468,
+      "grad_norm": 381.93347106147223,
+      "learning_rate": 8.699318812366641e-07,
+      "loss": 206.1659,
+      "step": 45110
+    },
+    {
+      "epoch": 0.8677715752881273,
+      "grad_norm": 377.03377994121126,
+      "learning_rate": 8.67443839483526e-07,
+      "loss": 209.6646,
+      "step": 45120
+    },
+    {
+      "epoch": 0.8679639005486078,
+      "grad_norm": 361.7981862240369,
+      "learning_rate": 8.649591994567275e-07,
+      "loss": 202.7763,
+      "step": 45130
+    },
+    {
+      "epoch": 0.8681562258090884,
+      "grad_norm": 347.0297282655106,
+      "learning_rate": 8.62477962081758e-07,
+      "loss": 204.6304,
+      "step": 45140
+    },
+    {
+      "epoch": 0.8683485510695689,
+      "grad_norm": 401.05624166746236,
+      "learning_rate": 8.600001282828341e-07,
+      "loss": 211.8197,
+      "step": 45150
+    },
+    {
+      "epoch": 0.8685408763300494,
+      "grad_norm": 389.49478016128694,
+      "learning_rate": 8.57525698982914e-07,
+      "loss": 206.5129,
+      "step": 45160
+    },
+    {
+      "epoch": 0.8687332015905299,
+      "grad_norm": 355.5411996169538,
+      "learning_rate": 8.550546751036759e-07,
+      "loss": 211.5079,
+      "step": 45170
+    },
+    {
+      "epoch": 0.8689255268510104,
+      "grad_norm": 397.39041976636537,
+      "learning_rate": 8.525870575655393e-07,
+      "loss": 205.2016,
+      "step": 45180
+    },
+    {
+      "epoch": 0.869117852111491,
+      "grad_norm": 378.08690407276947,
+      "learning_rate": 8.501228472876466e-07,
+      "loss": 208.413,
+      "step": 45190
+    },
+    {
+      "epoch": 0.8693101773719715,
+      "grad_norm": 367.74648454741344,
+      "learning_rate": 8.476620451878803e-07,
+      "loss": 203.0439,
+      "step": 45200
+    },
+    {
+      "epoch": 0.869502502632452,
+      "grad_norm": 362.45996382295584,
+      "learning_rate": 8.45204652182846e-07,
+      "loss": 207.6312,
+      "step": 45210
+    },
+    {
+      "epoch": 0.8696948278929325,
+      "grad_norm": 368.74715038315435,
+      "learning_rate": 8.427506691878806e-07,
+      "loss": 207.3251,
+      "step": 45220
+    },
+    {
+      "epoch": 0.869887153153413,
+      "grad_norm": 444.66132386606756,
+      "learning_rate": 8.403000971170561e-07,
+      "loss": 202.5711,
+      "step": 45230
+    },
+    {
+      "epoch": 0.8700794784138935,
+      "grad_norm": 368.5699517376908,
+      "learning_rate": 8.378529368831667e-07,
+      "loss": 205.0931,
+      "step": 45240
+    },
+    {
+      "epoch": 0.8702718036743741,
+      "grad_norm": 373.52988715555995,
+      "learning_rate": 8.354091893977401e-07,
+      "loss": 202.8805,
+      "step": 45250
+    },
+    {
+      "epoch": 0.8704641289348546,
+      "grad_norm": 330.02472461193537,
+      "learning_rate": 8.329688555710336e-07,
+      "loss": 205.5159,
+      "step": 45260
+    },
+    {
+      "epoch": 0.8706564541953351,
+      "grad_norm": 369.8401631096334,
+      "learning_rate": 8.305319363120279e-07,
+      "loss": 209.6239,
+      "step": 45270
+    },
+    {
+      "epoch": 0.8708487794558156,
+      "grad_norm": 371.93020182746807,
+      "learning_rate": 8.280984325284392e-07,
+      "loss": 204.057,
+      "step": 45280
+    },
+    {
+      "epoch": 0.8710411047162963,
+      "grad_norm": 366.4124863602406,
+      "learning_rate": 8.256683451267044e-07,
+      "loss": 202.2217,
+      "step": 45290
+    },
+    {
+      "epoch": 0.8712334299767768,
+      "grad_norm": 401.3745714268526,
+      "learning_rate": 8.232416750119921e-07,
+      "loss": 204.0654,
+      "step": 45300
+    },
+    {
+      "epoch": 0.8714257552372573,
+      "grad_norm": 438.6099221597723,
+      "learning_rate": 8.208184230881966e-07,
+      "loss": 204.2088,
+      "step": 45310
+    },
+    {
+      "epoch": 0.8716180804977378,
+      "grad_norm": 382.21974873877775,
+      "learning_rate": 8.183985902579405e-07,
+      "loss": 202.3825,
+      "step": 45320
+    },
+    {
+      "epoch": 0.8718104057582183,
+      "grad_norm": 349.6362483656875,
+      "learning_rate": 8.159821774225685e-07,
+      "loss": 197.3225,
+      "step": 45330
+    },
+    {
+      "epoch": 0.8720027310186989,
+      "grad_norm": 373.74093574235343,
+      "learning_rate": 8.13569185482157e-07,
+      "loss": 201.8258,
+      "step": 45340
+    },
+    {
+      "epoch": 0.8721950562791794,
+      "grad_norm": 344.8620381836919,
+      "learning_rate": 8.111596153355061e-07,
+      "loss": 198.8958,
+      "step": 45350
+    },
+    {
+      "epoch": 0.8723873815396599,
+      "grad_norm": 363.25728616523907,
+      "learning_rate": 8.08753467880139e-07,
+      "loss": 205.9329,
+      "step": 45360
+    },
+    {
+      "epoch": 0.8725797068001404,
+      "grad_norm": 361.92353161253294,
+      "learning_rate": 8.063507440123052e-07,
+      "loss": 200.0677,
+      "step": 45370
+    },
+    {
+      "epoch": 0.8727720320606209,
+      "grad_norm": 402.9977827066541,
+      "learning_rate": 8.039514446269836e-07,
+      "loss": 209.2742,
+      "step": 45380
+    },
+    {
+      "epoch": 0.8729643573211014,
+      "grad_norm": 377.60017382719764,
+      "learning_rate": 8.015555706178702e-07,
+      "loss": 203.6551,
+      "step": 45390
+    },
+    {
+      "epoch": 0.873156682581582,
+      "grad_norm": 390.44910722744527,
+      "learning_rate": 7.991631228773889e-07,
+      "loss": 207.5807,
+      "step": 45400
+    },
+    {
+      "epoch": 0.8733490078420625,
+      "grad_norm": 416.6786635853311,
+      "learning_rate": 7.967741022966857e-07,
+      "loss": 200.2847,
+      "step": 45410
+    },
+    {
+      "epoch": 0.873541333102543,
+      "grad_norm": 377.07772320469735,
+      "learning_rate": 7.943885097656356e-07,
+      "loss": 198.2747,
+      "step": 45420
+    },
+    {
+      "epoch": 0.8737336583630235,
+      "grad_norm": 365.40415920328473,
+      "learning_rate": 7.920063461728311e-07,
+      "loss": 200.7008,
+      "step": 45430
+    },
+    {
+      "epoch": 0.873925983623504,
+      "grad_norm": 369.26419427899185,
+      "learning_rate": 7.896276124055846e-07,
+      "loss": 202.8512,
+      "step": 45440
+    },
+    {
+      "epoch": 0.8741183088839846,
+      "grad_norm": 382.19082452593113,
+      "learning_rate": 7.872523093499396e-07,
+      "loss": 206.8361,
+      "step": 45450
+    },
+    {
+      "epoch": 0.8743106341444651,
+      "grad_norm": 368.6169331143544,
+      "learning_rate": 7.848804378906561e-07,
+      "loss": 200.6371,
+      "step": 45460
+    },
+    {
+      "epoch": 0.8745029594049456,
+      "grad_norm": 332.67222186556825,
+      "learning_rate": 7.825119989112173e-07,
+      "loss": 198.5944,
+      "step": 45470
+    },
+    {
+      "epoch": 0.8746952846654261,
+      "grad_norm": 375.18432048767204,
+      "learning_rate": 7.801469932938255e-07,
+      "loss": 205.6325,
+      "step": 45480
+    },
+    {
+      "epoch": 0.8748876099259066,
+      "grad_norm": 345.5457286430055,
+      "learning_rate": 7.777854219194092e-07,
+      "loss": 199.7954,
+      "step": 45490
+    },
+    {
+      "epoch": 0.8750799351863873,
+      "grad_norm": 358.8387028070915,
+      "learning_rate": 7.754272856676126e-07,
+      "loss": 208.0702,
+      "step": 45500
+    },
+    {
+      "epoch": 0.8752722604468678,
+      "grad_norm": 411.84363556380816,
+      "learning_rate": 7.73072585416802e-07,
+      "loss": 208.7903,
+      "step": 45510
+    },
+    {
+      "epoch": 0.8754645857073483,
+      "grad_norm": 372.36643076886344,
+      "learning_rate": 7.707213220440679e-07,
+      "loss": 198.5439,
+      "step": 45520
+    },
+    {
+      "epoch": 0.8756569109678288,
+      "grad_norm": 354.6240118019231,
+      "learning_rate": 7.683734964252143e-07,
+      "loss": 203.2871,
+      "step": 45530
+    },
+    {
+      "epoch": 0.8758492362283093,
+      "grad_norm": 400.4324176672373,
+      "learning_rate": 7.66029109434766e-07,
+      "loss": 207.1067,
+      "step": 45540
+    },
+    {
+      "epoch": 0.8760415614887899,
+      "grad_norm": 361.84110404032936,
+      "learning_rate": 7.636881619459724e-07,
+      "loss": 204.4386,
+      "step": 45550
+    },
+    {
+      "epoch": 0.8762338867492704,
+      "grad_norm": 413.3732866995702,
+      "learning_rate": 7.613506548307936e-07,
+      "loss": 205.0402,
+      "step": 45560
+    },
+    {
+      "epoch": 0.8764262120097509,
+      "grad_norm": 389.5679749808248,
+      "learning_rate": 7.590165889599166e-07,
+      "loss": 202.9151,
+      "step": 45570
+    },
+    {
+      "epoch": 0.8766185372702314,
+      "grad_norm": 345.7123025795795,
+      "learning_rate": 7.566859652027381e-07,
+      "loss": 200.5478,
+      "step": 45580
+    },
+    {
+      "epoch": 0.8768108625307119,
+      "grad_norm": 393.9553002767725,
+      "learning_rate": 7.543587844273814e-07,
+      "loss": 204.4718,
+      "step": 45590
+    },
+    {
+      "epoch": 0.8770031877911925,
+      "grad_norm": 380.5869976110789,
+      "learning_rate": 7.52035047500681e-07,
+      "loss": 199.6537,
+      "step": 45600
+    },
+    {
+      "epoch": 0.877195513051673,
+      "grad_norm": 379.71790600150086,
+      "learning_rate": 7.497147552881901e-07,
+      "loss": 205.6837,
+      "step": 45610
+    },
+    {
+      "epoch": 0.8773878383121535,
+      "grad_norm": 391.56070334823016,
+      "learning_rate": 7.473979086541772e-07,
+      "loss": 205.2985,
+      "step": 45620
+    },
+    {
+      "epoch": 0.877580163572634,
+      "grad_norm": 375.6810724889429,
+      "learning_rate": 7.450845084616332e-07,
+      "loss": 201.014,
+      "step": 45630
+    },
+    {
+      "epoch": 0.8777724888331145,
+      "grad_norm": 355.06306727775114,
+      "learning_rate": 7.427745555722598e-07,
+      "loss": 211.5624,
+      "step": 45640
+    },
+    {
+      "epoch": 0.8779648140935951,
+      "grad_norm": 382.6134881670004,
+      "learning_rate": 7.404680508464767e-07,
+      "loss": 207.2711,
+      "step": 45650
+    },
+    {
+      "epoch": 0.8781571393540756,
+      "grad_norm": 415.84241390706507,
+      "learning_rate": 7.381649951434167e-07,
+      "loss": 206.2194,
+      "step": 45660
+    },
+    {
+      "epoch": 0.8783494646145561,
+      "grad_norm": 390.43837367641703,
+      "learning_rate": 7.358653893209333e-07,
+      "loss": 211.0992,
+      "step": 45670
+    },
+    {
+      "epoch": 0.8785417898750366,
+      "grad_norm": 345.89120578020845,
+      "learning_rate": 7.335692342355882e-07,
+      "loss": 204.2241,
+      "step": 45680
+    },
+    {
+      "epoch": 0.8787341151355171,
+      "grad_norm": 358.7017660828885,
+      "learning_rate": 7.312765307426662e-07,
+      "loss": 199.9989,
+      "step": 45690
+    },
+    {
+      "epoch": 0.8789264403959978,
+      "grad_norm": 365.00133067731844,
+      "learning_rate": 7.28987279696155e-07,
+      "loss": 206.2718,
+      "step": 45700
+    },
+    {
+      "epoch": 0.8791187656564783,
+      "grad_norm": 372.8764711544401,
+      "learning_rate": 7.267014819487695e-07,
+      "loss": 209.8032,
+      "step": 45710
+    },
+    {
+      "epoch": 0.8793110909169588,
+      "grad_norm": 371.093758204294,
+      "learning_rate": 7.244191383519272e-07,
+      "loss": 202.6339,
+      "step": 45720
+    },
+    {
+      "epoch": 0.8795034161774393,
+      "grad_norm": 394.87972330526776,
+      "learning_rate": 7.221402497557629e-07,
+      "loss": 200.9425,
+      "step": 45730
+    },
+    {
+      "epoch": 0.8796957414379198,
+      "grad_norm": 376.815066501531,
+      "learning_rate": 7.198648170091294e-07,
+      "loss": 199.555,
+      "step": 45740
+    },
+    {
+      "epoch": 0.8798880666984004,
+      "grad_norm": 385.33257332086646,
+      "learning_rate": 7.175928409595844e-07,
+      "loss": 203.5715,
+      "step": 45750
+    },
+    {
+      "epoch": 0.8800803919588809,
+      "grad_norm": 378.2450231973383,
+      "learning_rate": 7.153243224534001e-07,
+      "loss": 207.6875,
+      "step": 45760
+    },
+    {
+      "epoch": 0.8802727172193614,
+      "grad_norm": 382.7940984393643,
+      "learning_rate": 7.130592623355659e-07,
+      "loss": 201.5336,
+      "step": 45770
+    },
+    {
+      "epoch": 0.8804650424798419,
+      "grad_norm": 349.33252287289235,
+      "learning_rate": 7.10797661449778e-07,
+      "loss": 203.0433,
+      "step": 45780
+    },
+    {
+      "epoch": 0.8806573677403224,
+      "grad_norm": 384.9729025551196,
+      "learning_rate": 7.085395206384449e-07,
+      "loss": 204.3773,
+      "step": 45790
+    },
+    {
+      "epoch": 0.880849693000803,
+      "grad_norm": 352.9406174020772,
+      "learning_rate": 7.062848407426859e-07,
+      "loss": 200.2767,
+      "step": 45800
+    },
+    {
+      "epoch": 0.8810420182612835,
+      "grad_norm": 380.2513907905553,
+      "learning_rate": 7.040336226023336e-07,
+      "loss": 199.0211,
+      "step": 45810
+    },
+    {
+      "epoch": 0.881234343521764,
+      "grad_norm": 355.21992081418955,
+      "learning_rate": 7.017858670559274e-07,
+      "loss": 206.7003,
+      "step": 45820
+    },
+    {
+      "epoch": 0.8814266687822445,
+      "grad_norm": 389.61976290166365,
+      "learning_rate": 6.99541574940722e-07,
+      "loss": 208.8777,
+      "step": 45830
+    },
+    {
+      "epoch": 0.881618994042725,
+      "grad_norm": 374.2619308356377,
+      "learning_rate": 6.973007470926774e-07,
+      "loss": 204.061,
+      "step": 45840
+    },
+    {
+      "epoch": 0.8818113193032056,
+      "grad_norm": 372.2038190396962,
+      "learning_rate": 6.95063384346466e-07,
+      "loss": 204.5984,
+      "step": 45850
+    },
+    {
+      "epoch": 0.8820036445636861,
+      "grad_norm": 366.9288915472056,
+      "learning_rate": 6.92829487535468e-07,
+      "loss": 207.5679,
+      "step": 45860
+    },
+    {
+      "epoch": 0.8821959698241666,
+      "grad_norm": 368.095059287682,
+      "learning_rate": 6.905990574917709e-07,
+      "loss": 206.0688,
+      "step": 45870
+    },
+    {
+      "epoch": 0.8823882950846471,
+      "grad_norm": 367.48039929585497,
+      "learning_rate": 6.88372095046177e-07,
+      "loss": 199.513,
+      "step": 45880
+    },
+    {
+      "epoch": 0.8825806203451276,
+      "grad_norm": 375.3178781970457,
+      "learning_rate": 6.861486010281915e-07,
+      "loss": 210.4137,
+      "step": 45890
+    },
+    {
+      "epoch": 0.8827729456056082,
+      "grad_norm": 367.39012211890025,
+      "learning_rate": 6.839285762660275e-07,
+      "loss": 198.4013,
+      "step": 45900
+    },
+    {
+      "epoch": 0.8829652708660888,
+      "grad_norm": 351.84173165269743,
+      "learning_rate": 6.81712021586608e-07,
+      "loss": 203.327,
+      "step": 45910
+    },
+    {
+      "epoch": 0.8831575961265693,
+      "grad_norm": 353.607786444627,
+      "learning_rate": 6.794989378155659e-07,
+      "loss": 206.263,
+      "step": 45920
+    },
+    {
+      "epoch": 0.8833499213870498,
+      "grad_norm": 355.1080079478987,
+      "learning_rate": 6.772893257772361e-07,
+      "loss": 209.7235,
+      "step": 45930
+    },
+    {
+      "epoch": 0.8835422466475303,
+      "grad_norm": 379.0470728426872,
+      "learning_rate": 6.750831862946605e-07,
+      "loss": 214.05,
+      "step": 45940
+    },
+    {
+      "epoch": 0.8837345719080109,
+      "grad_norm": 378.9822573013641,
+      "learning_rate": 6.728805201895949e-07,
+      "loss": 208.5714,
+      "step": 45950
+    },
+    {
+      "epoch": 0.8839268971684914,
+      "grad_norm": 379.19435655325555,
+      "learning_rate": 6.70681328282492e-07,
+      "loss": 193.3402,
+      "step": 45960
+    },
+    {
+      "epoch": 0.8841192224289719,
+      "grad_norm": 365.1627720191045,
+      "learning_rate": 6.684856113925143e-07,
+      "loss": 204.8902,
+      "step": 45970
+    },
+    {
+      "epoch": 0.8843115476894524,
+      "grad_norm": 390.8553183688744,
+      "learning_rate": 6.662933703375307e-07,
+      "loss": 210.1179,
+      "step": 45980
+    },
+    {
+      "epoch": 0.8845038729499329,
+      "grad_norm": 381.18685966428393,
+      "learning_rate": 6.641046059341171e-07,
+      "loss": 197.886,
+      "step": 45990
+    },
+    {
+      "epoch": 0.8846961982104135,
+      "grad_norm": 414.52283383046336,
+      "learning_rate": 6.619193189975515e-07,
+      "loss": 206.7159,
+      "step": 46000
+    },
+    {
+      "epoch": 0.884888523470894,
+      "grad_norm": 364.0085577681013,
+      "learning_rate": 6.597375103418135e-07,
+      "loss": 202.9025,
+      "step": 46010
+    },
+    {
+      "epoch": 0.8850808487313745,
+      "grad_norm": 335.3442443427356,
+      "learning_rate": 6.575591807795944e-07,
+      "loss": 196.0612,
+      "step": 46020
+    },
+    {
+      "epoch": 0.885273173991855,
+      "grad_norm": 366.0912740882515,
+      "learning_rate": 6.553843311222863e-07,
+      "loss": 200.8568,
+      "step": 46030
+    },
+    {
+      "epoch": 0.8854654992523355,
+      "grad_norm": 348.7560643861536,
+      "learning_rate": 6.532129621799832e-07,
+      "loss": 204.3942,
+      "step": 46040
+    },
+    {
+      "epoch": 0.885657824512816,
+      "grad_norm": 402.2090307279005,
+      "learning_rate": 6.510450747614816e-07,
+      "loss": 209.5204,
+      "step": 46050
+    },
+    {
+      "epoch": 0.8858501497732966,
+      "grad_norm": 355.3617180768231,
+      "learning_rate": 6.488806696742889e-07,
+      "loss": 212.6424,
+      "step": 46060
+    },
+    {
+      "epoch": 0.8860424750337771,
+      "grad_norm": 369.40341690346924,
+      "learning_rate": 6.46719747724609e-07,
+      "loss": 216.6485,
+      "step": 46070
+    },
+    {
+      "epoch": 0.8862348002942576,
+      "grad_norm": 383.3276577951005,
+      "learning_rate": 6.44562309717347e-07,
+      "loss": 210.4688,
+      "step": 46080
+    },
+    {
+      "epoch": 0.8864271255547381,
+      "grad_norm": 403.9067351721234,
+      "learning_rate": 6.424083564561134e-07,
+      "loss": 207.8883,
+      "step": 46090
+    },
+    {
+      "epoch": 0.8866194508152186,
+      "grad_norm": 360.23802338025024,
+      "learning_rate": 6.402578887432232e-07,
+      "loss": 200.543,
+      "step": 46100
+    },
+    {
+      "epoch": 0.8868117760756993,
+      "grad_norm": 404.9765070823738,
+      "learning_rate": 6.381109073796865e-07,
+      "loss": 207.1947,
+      "step": 46110
+    },
+    {
+      "epoch": 0.8870041013361798,
+      "grad_norm": 377.127081222995,
+      "learning_rate": 6.359674131652204e-07,
+      "loss": 199.1138,
+      "step": 46120
+    },
+    {
+      "epoch": 0.8871964265966603,
+      "grad_norm": 419.97193095252965,
+      "learning_rate": 6.338274068982408e-07,
+      "loss": 205.2515,
+      "step": 46130
+    },
+    {
+      "epoch": 0.8873887518571408,
+      "grad_norm": 373.6805519913914,
+      "learning_rate": 6.316908893758656e-07,
+      "loss": 202.9596,
+      "step": 46140
+    },
+    {
+      "epoch": 0.8875810771176214,
+      "grad_norm": 377.849801341338,
+      "learning_rate": 6.295578613939113e-07,
+      "loss": 204.4356,
+      "step": 46150
+    },
+    {
+      "epoch": 0.8877734023781019,
+      "grad_norm": 371.3649719915459,
+      "learning_rate": 6.274283237468948e-07,
+      "loss": 198.1371,
+      "step": 46160
+    },
+    {
+      "epoch": 0.8879657276385824,
+      "grad_norm": 366.7234511654086,
+      "learning_rate": 6.25302277228036e-07,
+      "loss": 207.253,
+      "step": 46170
+    },
+    {
+      "epoch": 0.8881580528990629,
+      "grad_norm": 398.98612201339967,
+      "learning_rate": 6.231797226292502e-07,
+      "loss": 209.8231,
+      "step": 46180
+    },
+    {
+      "epoch": 0.8883503781595434,
+      "grad_norm": 365.6950231288033,
+      "learning_rate": 6.210606607411529e-07,
+      "loss": 198.6315,
+      "step": 46190
+    },
+    {
+      "epoch": 0.888542703420024,
+      "grad_norm": 356.54294065426336,
+      "learning_rate": 6.189450923530627e-07,
+      "loss": 199.4474,
+      "step": 46200
+    },
+    {
+      "epoch": 0.8887350286805045,
+      "grad_norm": 355.00000362213683,
+      "learning_rate": 6.168330182529924e-07,
+      "loss": 208.0717,
+      "step": 46210
+    },
+    {
+      "epoch": 0.888927353940985,
+      "grad_norm": 365.0013600609673,
+      "learning_rate": 6.147244392276541e-07,
+      "loss": 204.4221,
+      "step": 46220
+    },
+    {
+      "epoch": 0.8891196792014655,
+      "grad_norm": 333.9679464941792,
+      "learning_rate": 6.126193560624583e-07,
+      "loss": 207.7982,
+      "step": 46230
+    },
+    {
+      "epoch": 0.889312004461946,
+      "grad_norm": 368.20901956424706,
+      "learning_rate": 6.105177695415165e-07,
+      "loss": 203.0013,
+      "step": 46240
+    },
+    {
+      "epoch": 0.8895043297224265,
+      "grad_norm": 349.5582846764632,
+      "learning_rate": 6.084196804476317e-07,
+      "loss": 201.766,
+      "step": 46250
+    },
+    {
+      "epoch": 0.8896966549829071,
+      "grad_norm": 362.5981944507473,
+      "learning_rate": 6.063250895623096e-07,
+      "loss": 200.6178,
+      "step": 46260
+    },
+    {
+      "epoch": 0.8898889802433876,
+      "grad_norm": 468.65567792156855,
+      "learning_rate": 6.042339976657486e-07,
+      "loss": 210.9137,
+      "step": 46270
+    },
+    {
+      "epoch": 0.8900813055038681,
+      "grad_norm": 374.7715979663631,
+      "learning_rate": 6.021464055368498e-07,
+      "loss": 204.2966,
+      "step": 46280
+    },
+    {
+      "epoch": 0.8902736307643486,
+      "grad_norm": 342.5629581474716,
+      "learning_rate": 6.000623139532036e-07,
+      "loss": 210.7192,
+      "step": 46290
+    },
+    {
+      "epoch": 0.8904659560248291,
+      "grad_norm": 347.0291314072588,
+      "learning_rate": 5.979817236910979e-07,
+      "loss": 204.352,
+      "step": 46300
+    },
+    {
+      "epoch": 0.8906582812853097,
+      "grad_norm": 359.55541967698986,
+      "learning_rate": 5.959046355255238e-07,
+      "loss": 215.4892,
+      "step": 46310
+    },
+    {
+      "epoch": 0.8908506065457903,
+      "grad_norm": 367.9546075898875,
+      "learning_rate": 5.93831050230158e-07,
+      "loss": 203.4867,
+      "step": 46320
+    },
+    {
+      "epoch": 0.8910429318062708,
+      "grad_norm": 369.75157000110255,
+      "learning_rate": 5.917609685773784e-07,
+      "loss": 201.9396,
+      "step": 46330
+    },
+    {
+      "epoch": 0.8912352570667513,
+      "grad_norm": 365.94807170332615,
+      "learning_rate": 5.896943913382547e-07,
+      "loss": 197.4694,
+      "step": 46340
+    },
+    {
+      "epoch": 0.8914275823272318,
+      "grad_norm": 346.1240882698529,
+      "learning_rate": 5.876313192825544e-07,
+      "loss": 199.2389,
+      "step": 46350
+    },
+    {
+      "epoch": 0.8916199075877124,
+      "grad_norm": 388.23454574162827,
+      "learning_rate": 5.855717531787375e-07,
+      "loss": 216.136,
+      "step": 46360
+    },
+    {
+      "epoch": 0.8918122328481929,
+      "grad_norm": 482.01798595457234,
+      "learning_rate": 5.835156937939568e-07,
+      "loss": 209.7548,
+      "step": 46370
+    },
+    {
+      "epoch": 0.8920045581086734,
+      "grad_norm": 383.9791955451826,
+      "learning_rate": 5.814631418940641e-07,
+      "loss": 202.5437,
+      "step": 46380
+    },
+    {
+      "epoch": 0.8921968833691539,
+      "grad_norm": 366.6715312043861,
+      "learning_rate": 5.794140982435981e-07,
+      "loss": 203.0349,
+      "step": 46390
+    },
+    {
+      "epoch": 0.8923892086296344,
+      "grad_norm": 377.4855170659537,
+      "learning_rate": 5.773685636057924e-07,
+      "loss": 204.4214,
+      "step": 46400
+    },
+    {
+      "epoch": 0.892581533890115,
+      "grad_norm": 356.8191381889658,
+      "learning_rate": 5.753265387425777e-07,
+      "loss": 213.0747,
+      "step": 46410
+    },
+    {
+      "epoch": 0.8927738591505955,
+      "grad_norm": 445.6339628561458,
+      "learning_rate": 5.732880244145744e-07,
+      "loss": 200.147,
+      "step": 46420
+    },
+    {
+      "epoch": 0.892966184411076,
+      "grad_norm": 346.3965130011756,
+      "learning_rate": 5.712530213810951e-07,
+      "loss": 205.3982,
+      "step": 46430
+    },
+    {
+      "epoch": 0.8931585096715565,
+      "grad_norm": 359.86112888369723,
+      "learning_rate": 5.692215304001447e-07,
+      "loss": 200.0912,
+      "step": 46440
+    },
+    {
+      "epoch": 0.893350834932037,
+      "grad_norm": 371.2252252217051,
+      "learning_rate": 5.671935522284177e-07,
+      "loss": 207.2773,
+      "step": 46450
+    },
+    {
+      "epoch": 0.8935431601925176,
+      "grad_norm": 387.7891305057454,
+      "learning_rate": 5.651690876213067e-07,
+      "loss": 209.9808,
+      "step": 46460
+    },
+    {
+      "epoch": 0.8937354854529981,
+      "grad_norm": 376.3163557617063,
+      "learning_rate": 5.631481373328895e-07,
+      "loss": 210.1483,
+      "step": 46470
+    },
+    {
+      "epoch": 0.8939278107134786,
+      "grad_norm": 374.02670898027696,
+      "learning_rate": 5.61130702115934e-07,
+      "loss": 207.9196,
+      "step": 46480
+    },
+    {
+      "epoch": 0.8941201359739591,
+      "grad_norm": 395.47811864916605,
+      "learning_rate": 5.591167827219057e-07,
+      "loss": 204.4367,
+      "step": 46490
+    },
+    {
+      "epoch": 0.8943124612344396,
+      "grad_norm": 418.67294003082634,
+      "learning_rate": 5.571063799009546e-07,
+      "loss": 205.7631,
+      "step": 46500
+    },
+    {
+      "epoch": 0.8945047864949202,
+      "grad_norm": 358.3868145255379,
+      "learning_rate": 5.550994944019216e-07,
+      "loss": 204.9341,
+      "step": 46510
+    },
+    {
+      "epoch": 0.8946971117554008,
+      "grad_norm": 385.46057471016724,
+      "learning_rate": 5.53096126972339e-07,
+      "loss": 206.154,
+      "step": 46520
+    },
+    {
+      "epoch": 0.8948894370158813,
+      "grad_norm": 379.22927767318134,
+      "learning_rate": 5.510962783584295e-07,
+      "loss": 206.324,
+      "step": 46530
+    },
+    {
+      "epoch": 0.8950817622763618,
+      "grad_norm": 391.7849643196316,
+      "learning_rate": 5.490999493051008e-07,
+      "loss": 202.4588,
+      "step": 46540
+    },
+    {
+      "epoch": 0.8952740875368423,
+      "grad_norm": 367.13564177532015,
+      "learning_rate": 5.471071405559547e-07,
+      "loss": 204.6717,
+      "step": 46550
+    },
+    {
+      "epoch": 0.8954664127973229,
+      "grad_norm": 367.52771561929217,
+      "learning_rate": 5.451178528532786e-07,
+      "loss": 205.6467,
+      "step": 46560
+    },
+    {
+      "epoch": 0.8956587380578034,
+      "grad_norm": 370.97247477686295,
+      "learning_rate": 5.431320869380519e-07,
+      "loss": 195.571,
+      "step": 46570
+    },
+    {
+      "epoch": 0.8958510633182839,
+      "grad_norm": 354.1037625399484,
+      "learning_rate": 5.411498435499363e-07,
+      "loss": 208.7288,
+      "step": 46580
+    },
+    {
+      "epoch": 0.8960433885787644,
+      "grad_norm": 370.04010756323106,
+      "learning_rate": 5.391711234272856e-07,
+      "loss": 200.8438,
+      "step": 46590
+    },
+    {
+      "epoch": 0.8962357138392449,
+      "grad_norm": 342.60862362183866,
+      "learning_rate": 5.371959273071414e-07,
+      "loss": 205.324,
+      "step": 46600
+    },
+    {
+      "epoch": 0.8964280390997255,
+      "grad_norm": 340.18217719968425,
+      "learning_rate": 5.352242559252308e-07,
+      "loss": 196.882,
+      "step": 46610
+    },
+    {
+      "epoch": 0.896620364360206,
+      "grad_norm": 354.5573567298894,
+      "learning_rate": 5.332561100159683e-07,
+      "loss": 205.8079,
+      "step": 46620
+    },
+    {
+      "epoch": 0.8968126896206865,
+      "grad_norm": 373.54553948249486,
+      "learning_rate": 5.312914903124566e-07,
+      "loss": 201.3235,
+      "step": 46630
+    },
+    {
+      "epoch": 0.897005014881167,
+      "grad_norm": 373.0912869085149,
+      "learning_rate": 5.293303975464836e-07,
+      "loss": 203.0003,
+      "step": 46640
+    },
+    {
+      "epoch": 0.8971973401416475,
+      "grad_norm": 387.36782777616816,
+      "learning_rate": 5.273728324485261e-07,
+      "loss": 206.0036,
+      "step": 46650
+    },
+    {
+      "epoch": 0.8973896654021281,
+      "grad_norm": 374.1344936069173,
+      "learning_rate": 5.254187957477397e-07,
+      "loss": 215.6528,
+      "step": 46660
+    },
+    {
+      "epoch": 0.8975819906626086,
+      "grad_norm": 366.80684221497444,
+      "learning_rate": 5.234682881719766e-07,
+      "loss": 205.1975,
+      "step": 46670
+    },
+    {
+      "epoch": 0.8977743159230891,
+      "grad_norm": 354.88616736075994,
+      "learning_rate": 5.215213104477645e-07,
+      "loss": 202.4825,
+      "step": 46680
+    },
+    {
+      "epoch": 0.8979666411835696,
+      "grad_norm": 361.8222081322325,
+      "learning_rate": 5.195778633003223e-07,
+      "loss": 199.2351,
+      "step": 46690
+    },
+    {
+      "epoch": 0.8981589664440501,
+      "grad_norm": 352.66703581608317,
+      "learning_rate": 5.176379474535509e-07,
+      "loss": 201.5392,
+      "step": 46700
+    },
+    {
+      "epoch": 0.8983512917045307,
+      "grad_norm": 381.60741418949686,
+      "learning_rate": 5.1570156363004e-07,
+      "loss": 207.155,
+      "step": 46710
+    },
+    {
+      "epoch": 0.8985436169650112,
+      "grad_norm": 370.25547878484844,
+      "learning_rate": 5.13768712551057e-07,
+      "loss": 210.8079,
+      "step": 46720
+    },
+    {
+      "epoch": 0.8987359422254918,
+      "grad_norm": 347.8823398235995,
+      "learning_rate": 5.118393949365574e-07,
+      "loss": 203.7626,
+      "step": 46730
+    },
+    {
+      "epoch": 0.8989282674859723,
+      "grad_norm": 447.6125248578692,
+      "learning_rate": 5.099136115051829e-07,
+      "loss": 204.6365,
+      "step": 46740
+    },
+    {
+      "epoch": 0.8991205927464528,
+      "grad_norm": 364.81190244947885,
+      "learning_rate": 5.079913629742539e-07,
+      "loss": 205.7409,
+      "step": 46750
+    },
+    {
+      "epoch": 0.8993129180069334,
+      "grad_norm": 402.75434587675386,
+      "learning_rate": 5.060726500597768e-07,
+      "loss": 210.9729,
+      "step": 46760
+    },
+    {
+      "epoch": 0.8995052432674139,
+      "grad_norm": 371.0447523532784,
+      "learning_rate": 5.041574734764376e-07,
+      "loss": 203.5738,
+      "step": 46770
+    },
+    {
+      "epoch": 0.8996975685278944,
+      "grad_norm": 379.84309738134664,
+      "learning_rate": 5.022458339376124e-07,
+      "loss": 202.9529,
+      "step": 46780
+    },
+    {
+      "epoch": 0.8998898937883749,
+      "grad_norm": 373.46681249710264,
+      "learning_rate": 5.003377321553538e-07,
+      "loss": 208.3639,
+      "step": 46790
+    },
+    {
+      "epoch": 0.9000822190488554,
+      "grad_norm": 360.8372071765622,
+      "learning_rate": 4.984331688403976e-07,
+      "loss": 206.3781,
+      "step": 46800
+    },
+    {
+      "epoch": 0.900274544309336,
+      "grad_norm": 367.41910152285726,
+      "learning_rate": 4.96532144702162e-07,
+      "loss": 201.8339,
+      "step": 46810
+    },
+    {
+      "epoch": 0.9004668695698165,
+      "grad_norm": 367.18844600287684,
+      "learning_rate": 4.946346604487462e-07,
+      "loss": 214.1873,
+      "step": 46820
+    },
+    {
+      "epoch": 0.900659194830297,
+      "grad_norm": 362.82922440388523,
+      "learning_rate": 4.927407167869346e-07,
+      "loss": 202.1727,
+      "step": 46830
+    },
+    {
+      "epoch": 0.9008515200907775,
+      "grad_norm": 377.3317954957144,
+      "learning_rate": 4.908503144221877e-07,
+      "loss": 203.4602,
+      "step": 46840
+    },
+    {
+      "epoch": 0.901043845351258,
+      "grad_norm": 362.0899497662295,
+      "learning_rate": 4.889634540586518e-07,
+      "loss": 204.7197,
+      "step": 46850
+    },
+    {
+      "epoch": 0.9012361706117386,
+      "grad_norm": 380.6867409610158,
+      "learning_rate": 4.870801363991484e-07,
+      "loss": 211.6256,
+      "step": 46860
+    },
+    {
+      "epoch": 0.9014284958722191,
+      "grad_norm": 353.2190297394774,
+      "learning_rate": 4.852003621451829e-07,
+      "loss": 200.2173,
+      "step": 46870
+    },
+    {
+      "epoch": 0.9016208211326996,
+      "grad_norm": 383.00260016165333,
+      "learning_rate": 4.833241319969395e-07,
+      "loss": 199.2871,
+      "step": 46880
+    },
+    {
+      "epoch": 0.9018131463931801,
+      "grad_norm": 352.3503283548052,
+      "learning_rate": 4.814514466532849e-07,
+      "loss": 203.7786,
+      "step": 46890
+    },
+    {
+      "epoch": 0.9020054716536606,
+      "grad_norm": 392.9533483299655,
+      "learning_rate": 4.795823068117622e-07,
+      "loss": 207.2969,
+      "step": 46900
+    },
+    {
+      "epoch": 0.9021977969141411,
+      "grad_norm": 369.9887388449985,
+      "learning_rate": 4.777167131685945e-07,
+      "loss": 198.2329,
+      "step": 46910
+    },
+    {
+      "epoch": 0.9023901221746217,
+      "grad_norm": 350.70354203666636,
+      "learning_rate": 4.7585466641868696e-07,
+      "loss": 209.4631,
+      "step": 46920
+    },
+    {
+      "epoch": 0.9025824474351023,
+      "grad_norm": 336.65393823123486,
+      "learning_rate": 4.7399616725561925e-07,
+      "loss": 200.1722,
+      "step": 46930
+    },
+    {
+      "epoch": 0.9027747726955828,
+      "grad_norm": 399.165053561979,
+      "learning_rate": 4.721412163716521e-07,
+      "loss": 203.2035,
+      "step": 46940
+    },
+    {
+      "epoch": 0.9029670979560633,
+      "grad_norm": 387.47311179419717,
+      "learning_rate": 4.702898144577228e-07,
+      "loss": 215.1573,
+      "step": 46950
+    },
+    {
+      "epoch": 0.9031594232165439,
+      "grad_norm": 364.63146834015566,
+      "learning_rate": 4.6844196220345086e-07,
+      "loss": 200.2567,
+      "step": 46960
+    },
+    {
+      "epoch": 0.9033517484770244,
+      "grad_norm": 370.6505007186757,
+      "learning_rate": 4.665976602971278e-07,
+      "loss": 207.1559,
+      "step": 46970
+    },
+    {
+      "epoch": 0.9035440737375049,
+      "grad_norm": 418.33878762694326,
+      "learning_rate": 4.647569094257276e-07,
+      "loss": 209.9307,
+      "step": 46980
+    },
+    {
+      "epoch": 0.9037363989979854,
+      "grad_norm": 354.09666090788795,
+      "learning_rate": 4.629197102748984e-07,
+      "loss": 199.1513,
+      "step": 46990
+    },
+    {
+      "epoch": 0.9039287242584659,
+      "grad_norm": 379.9613313164776,
+      "learning_rate": 4.610860635289671e-07,
+      "loss": 205.9776,
+      "step": 47000
+    },
+    {
+      "epoch": 0.9041210495189465,
+      "grad_norm": 342.8654608475949,
+      "learning_rate": 4.592559698709387e-07,
+      "loss": 204.1239,
+      "step": 47010
+    },
+    {
+      "epoch": 0.904313374779427,
+      "grad_norm": 365.03178984759643,
+      "learning_rate": 4.5742942998248774e-07,
+      "loss": 203.5788,
+      "step": 47020
+    },
+    {
+      "epoch": 0.9045057000399075,
+      "grad_norm": 400.44008202355786,
+      "learning_rate": 4.5560644454397563e-07,
+      "loss": 204.3993,
+      "step": 47030
+    },
+    {
+      "epoch": 0.904698025300388,
+      "grad_norm": 357.6281828460192,
+      "learning_rate": 4.537870142344314e-07,
+      "loss": 205.6249,
+      "step": 47040
+    },
+    {
+      "epoch": 0.9048903505608685,
+      "grad_norm": 356.1279600757491,
+      "learning_rate": 4.5197113973156403e-07,
+      "loss": 202.473,
+      "step": 47050
+    },
+    {
+      "epoch": 0.905082675821349,
+      "grad_norm": 392.38275587321226,
+      "learning_rate": 4.5015882171175476e-07,
+      "loss": 209.4659,
+      "step": 47060
+    },
+    {
+      "epoch": 0.9052750010818296,
+      "grad_norm": 440.67633214524574,
+      "learning_rate": 4.483500608500657e-07,
+      "loss": 202.5479,
+      "step": 47070
+    },
+    {
+      "epoch": 0.9054673263423101,
+      "grad_norm": 405.8475253122133,
+      "learning_rate": 4.4654485782022697e-07,
+      "loss": 203.5308,
+      "step": 47080
+    },
+    {
+      "epoch": 0.9056596516027906,
+      "grad_norm": 367.5869952051399,
+      "learning_rate": 4.447432132946472e-07,
+      "loss": 208.0369,
+      "step": 47090
+    },
+    {
+      "epoch": 0.9058519768632711,
+      "grad_norm": 407.84748177285456,
+      "learning_rate": 4.429451279444119e-07,
+      "loss": 205.149,
+      "step": 47100
+    },
+    {
+      "epoch": 0.9060443021237516,
+      "grad_norm": 370.4969781493726,
+      "learning_rate": 4.411506024392753e-07,
+      "loss": 208.2056,
+      "step": 47110
+    },
+    {
+      "epoch": 0.9062366273842322,
+      "grad_norm": 432.42100237714845,
+      "learning_rate": 4.393596374476705e-07,
+      "loss": 198.6828,
+      "step": 47120
+    },
+    {
+      "epoch": 0.9064289526447128,
+      "grad_norm": 352.0118656013363,
+      "learning_rate": 4.3757223363670055e-07,
+      "loss": 196.4775,
+      "step": 47130
+    },
+    {
+      "epoch": 0.9066212779051933,
+      "grad_norm": 353.0874115951717,
+      "learning_rate": 4.3578839167214505e-07,
+      "loss": 207.3963,
+      "step": 47140
+    },
+    {
+      "epoch": 0.9068136031656738,
+      "grad_norm": 346.2629225427659,
+      "learning_rate": 4.3400811221845693e-07,
+      "loss": 206.5523,
+      "step": 47150
+    },
+    {
+      "epoch": 0.9070059284261544,
+      "grad_norm": 356.06970196521286,
+      "learning_rate": 4.322313959387592e-07,
+      "loss": 204.9614,
+      "step": 47160
+    },
+    {
+      "epoch": 0.9071982536866349,
+      "grad_norm": 352.3578156247154,
+      "learning_rate": 4.304582434948479e-07,
+      "loss": 203.4813,
+      "step": 47170
+    },
+    {
+      "epoch": 0.9073905789471154,
+      "grad_norm": 331.86270353744106,
+      "learning_rate": 4.2868865554719583e-07,
+      "loss": 204.4491,
+      "step": 47180
+    },
+    {
+      "epoch": 0.9075829042075959,
+      "grad_norm": 345.3433093695026,
+      "learning_rate": 4.269226327549447e-07,
+      "loss": 206.9256,
+      "step": 47190
+    },
+    {
+      "epoch": 0.9077752294680764,
+      "grad_norm": 381.57903165582337,
+      "learning_rate": 4.251601757759061e-07,
+      "loss": 197.6213,
+      "step": 47200
+    },
+    {
+      "epoch": 0.907967554728557,
+      "grad_norm": 377.9686273076321,
+      "learning_rate": 4.234012852665703e-07,
+      "loss": 210.6211,
+      "step": 47210
+    },
+    {
+      "epoch": 0.9081598799890375,
+      "grad_norm": 384.2630539010244,
+      "learning_rate": 4.2164596188209226e-07,
+      "loss": 208.5283,
+      "step": 47220
+    },
+    {
+      "epoch": 0.908352205249518,
+      "grad_norm": 404.399858694891,
+      "learning_rate": 4.198942062763023e-07,
+      "loss": 206.963,
+      "step": 47230
+    },
+    {
+      "epoch": 0.9085445305099985,
+      "grad_norm": 385.48323831190004,
+      "learning_rate": 4.181460191016984e-07,
+      "loss": 200.9335,
+      "step": 47240
+    },
+    {
+      "epoch": 0.908736855770479,
+      "grad_norm": 345.5274778341019,
+      "learning_rate": 4.1640140100945304e-07,
+      "loss": 208.171,
+      "step": 47250
+    },
+    {
+      "epoch": 0.9089291810309595,
+      "grad_norm": 400.21735481896087,
+      "learning_rate": 4.146603526494086e-07,
+      "loss": 203.9448,
+      "step": 47260
+    },
+    {
+      "epoch": 0.9091215062914401,
+      "grad_norm": 406.9966190864523,
+      "learning_rate": 4.129228746700742e-07,
+      "loss": 209.6477,
+      "step": 47270
+    },
+    {
+      "epoch": 0.9093138315519206,
+      "grad_norm": 457.7861940909638,
+      "learning_rate": 4.111889677186354e-07,
+      "loss": 204.5692,
+      "step": 47280
+    },
+    {
+      "epoch": 0.9095061568124011,
+      "grad_norm": 378.529792369946,
+      "learning_rate": 4.094586324409411e-07,
+      "loss": 200.7375,
+      "step": 47290
+    },
+    {
+      "epoch": 0.9096984820728816,
+      "grad_norm": 348.5234843857561,
+      "learning_rate": 4.0773186948151246e-07,
+      "loss": 207.2667,
+      "step": 47300
+    },
+    {
+      "epoch": 0.9098908073333621,
+      "grad_norm": 410.0679852549603,
+      "learning_rate": 4.060086794835405e-07,
+      "loss": 201.6419,
+      "step": 47310
+    },
+    {
+      "epoch": 0.9100831325938427,
+      "grad_norm": 354.9593348868804,
+      "learning_rate": 4.042890630888863e-07,
+      "loss": 201.8753,
+      "step": 47320
+    },
+    {
+      "epoch": 0.9102754578543232,
+      "grad_norm": 346.93318134298937,
+      "learning_rate": 4.025730209380774e-07,
+      "loss": 196.3145,
+      "step": 47330
+    },
+    {
+      "epoch": 0.9104677831148038,
+      "grad_norm": 383.25819779485136,
+      "learning_rate": 4.0086055367031027e-07,
+      "loss": 209.8385,
+      "step": 47340
+    },
+    {
+      "epoch": 0.9106601083752843,
+      "grad_norm": 339.2740254487882,
+      "learning_rate": 3.9915166192345365e-07,
+      "loss": 202.9469,
+      "step": 47350
+    },
+    {
+      "epoch": 0.9108524336357648,
+      "grad_norm": 356.6812401563461,
+      "learning_rate": 3.9744634633403944e-07,
+      "loss": 200.4444,
+      "step": 47360
+    },
+    {
+      "epoch": 0.9110447588962454,
+      "grad_norm": 355.92559105823244,
+      "learning_rate": 3.957446075372706e-07,
+      "loss": 198.258,
+      "step": 47370
+    },
+    {
+      "epoch": 0.9112370841567259,
+      "grad_norm": 344.25196126803843,
+      "learning_rate": 3.940464461670135e-07,
+      "loss": 208.3938,
+      "step": 47380
+    },
+    {
+      "epoch": 0.9114294094172064,
+      "grad_norm": 371.65925046265824,
+      "learning_rate": 3.923518628558087e-07,
+      "loss": 206.099,
+      "step": 47390
+    },
+    {
+      "epoch": 0.9116217346776869,
+      "grad_norm": 357.89725758710927,
+      "learning_rate": 3.9066085823485923e-07,
+      "loss": 204.5455,
+      "step": 47400
+    },
+    {
+      "epoch": 0.9118140599381674,
+      "grad_norm": 364.16976377382895,
+      "learning_rate": 3.8897343293403777e-07,
+      "loss": 210.3828,
+      "step": 47410
+    },
+    {
+      "epoch": 0.912006385198648,
+      "grad_norm": 387.65288304202727,
+      "learning_rate": 3.872895875818794e-07,
+      "loss": 203.672,
+      "step": 47420
+    },
+    {
+      "epoch": 0.9121987104591285,
+      "grad_norm": 381.59155738979234,
+      "learning_rate": 3.856093228055924e-07,
+      "loss": 200.3493,
+      "step": 47430
+    },
+    {
+      "epoch": 0.912391035719609,
+      "grad_norm": 366.5548177347879,
+      "learning_rate": 3.83932639231045e-07,
+      "loss": 205.937,
+      "step": 47440
+    },
+    {
+      "epoch": 0.9125833609800895,
+      "grad_norm": 377.39625723752096,
+      "learning_rate": 3.822595374827742e-07,
+      "loss": 201.2788,
+      "step": 47450
+    },
+    {
+      "epoch": 0.91277568624057,
+      "grad_norm": 347.20240379430714,
+      "learning_rate": 3.805900181839839e-07,
+      "loss": 206.0641,
+      "step": 47460
+    },
+    {
+      "epoch": 0.9129680115010506,
+      "grad_norm": 365.1487363771161,
+      "learning_rate": 3.789240819565432e-07,
+      "loss": 201.2513,
+      "step": 47470
+    },
+    {
+      "epoch": 0.9131603367615311,
+      "grad_norm": 343.1728654780718,
+      "learning_rate": 3.772617294209835e-07,
+      "loss": 203.1564,
+      "step": 47480
+    },
+    {
+      "epoch": 0.9133526620220116,
+      "grad_norm": 368.1403117998596,
+      "learning_rate": 3.7560296119650396e-07,
+      "loss": 203.5886,
+      "step": 47490
+    },
+    {
+      "epoch": 0.9135449872824921,
+      "grad_norm": 409.34288816991506,
+      "learning_rate": 3.739477779009704e-07,
+      "loss": 214.804,
+      "step": 47500
+    },
+    {
+      "epoch": 0.9137373125429726,
+      "grad_norm": 361.75045353472785,
+      "learning_rate": 3.7229618015091065e-07,
+      "loss": 206.2953,
+      "step": 47510
+    },
+    {
+      "epoch": 0.9139296378034532,
+      "grad_norm": 330.3557653212239,
+      "learning_rate": 3.7064816856151484e-07,
+      "loss": 201.3823,
+      "step": 47520
+    },
+    {
+      "epoch": 0.9141219630639337,
+      "grad_norm": 439.2700796390219,
+      "learning_rate": 3.6900374374664425e-07,
+      "loss": 209.4542,
+      "step": 47530
+    },
+    {
+      "epoch": 0.9143142883244143,
+      "grad_norm": 368.11447628322605,
+      "learning_rate": 3.6736290631881667e-07,
+      "loss": 206.7973,
+      "step": 47540
+    },
+    {
+      "epoch": 0.9145066135848948,
+      "grad_norm": 358.6895190021785,
+      "learning_rate": 3.657256568892187e-07,
+      "loss": 208.0302,
+      "step": 47550
+    },
+    {
+      "epoch": 0.9146989388453753,
+      "grad_norm": 384.87962908381985,
+      "learning_rate": 3.6409199606769806e-07,
+      "loss": 197.1002,
+      "step": 47560
+    },
+    {
+      "epoch": 0.9148912641058559,
+      "grad_norm": 377.1110216795921,
+      "learning_rate": 3.6246192446276694e-07,
+      "loss": 200.3165,
+      "step": 47570
+    },
+    {
+      "epoch": 0.9150835893663364,
+      "grad_norm": 365.9667458775916,
+      "learning_rate": 3.6083544268160077e-07,
+      "loss": 206.7293,
+      "step": 47580
+    },
+    {
+      "epoch": 0.9152759146268169,
+      "grad_norm": 356.25417010013473,
+      "learning_rate": 3.5921255133003483e-07,
+      "loss": 202.4675,
+      "step": 47590
+    },
+    {
+      "epoch": 0.9154682398872974,
+      "grad_norm": 383.3454193146976,
+      "learning_rate": 3.5759325101257013e-07,
+      "loss": 206.6678,
+      "step": 47600
+    },
+    {
+      "epoch": 0.9156605651477779,
+      "grad_norm": 425.3300779116341,
+      "learning_rate": 3.559775423323708e-07,
+      "loss": 203.6999,
+      "step": 47610
+    },
+    {
+      "epoch": 0.9158528904082585,
+      "grad_norm": 358.3871817832119,
+      "learning_rate": 3.54365425891261e-07,
+      "loss": 198.2385,
+      "step": 47620
+    },
+    {
+      "epoch": 0.916045215668739,
+      "grad_norm": 369.534451186069,
+      "learning_rate": 3.527569022897259e-07,
+      "loss": 208.2636,
+      "step": 47630
+    },
+    {
+      "epoch": 0.9162375409292195,
+      "grad_norm": 363.39713092435045,
+      "learning_rate": 3.511519721269163e-07,
+      "loss": 206.4843,
+      "step": 47640
+    },
+    {
+      "epoch": 0.9164298661897,
+      "grad_norm": 361.1899849085721,
+      "learning_rate": 3.4955063600064177e-07,
+      "loss": 199.0428,
+      "step": 47650
+    },
+    {
+      "epoch": 0.9166221914501805,
+      "grad_norm": 367.102044228421,
+      "learning_rate": 3.479528945073707e-07,
+      "loss": 202.6954,
+      "step": 47660
+    },
+    {
+      "epoch": 0.916814516710661,
+      "grad_norm": 370.2284489568778,
+      "learning_rate": 3.4635874824223924e-07,
+      "loss": 205.7312,
+      "step": 47670
+    },
+    {
+      "epoch": 0.9170068419711416,
+      "grad_norm": 1491.2089817496505,
+      "learning_rate": 3.4476819779903694e-07,
+      "loss": 203.5768,
+      "step": 47680
+    },
+    {
+      "epoch": 0.9171991672316221,
+      "grad_norm": 385.33103089603003,
+      "learning_rate": 3.43181243770222e-07,
+      "loss": 202.2294,
+      "step": 47690
+    },
+    {
+      "epoch": 0.9173914924921026,
+      "grad_norm": 386.7401731294933,
+      "learning_rate": 3.4159788674690386e-07,
+      "loss": 200.2038,
+      "step": 47700
+    },
+    {
+      "epoch": 0.9175838177525831,
+      "grad_norm": 360.4659660069398,
+      "learning_rate": 3.4001812731886077e-07,
+      "loss": 197.0495,
+      "step": 47710
+    },
+    {
+      "epoch": 0.9177761430130636,
+      "grad_norm": 338.9240260409679,
+      "learning_rate": 3.384419660745253e-07,
+      "loss": 198.7667,
+      "step": 47720
+    },
+    {
+      "epoch": 0.9179684682735442,
+      "grad_norm": 395.8833580109662,
+      "learning_rate": 3.368694036009923e-07,
+      "loss": 206.6854,
+      "step": 47730
+    },
+    {
+      "epoch": 0.9181607935340247,
+      "grad_norm": 359.3387069270951,
+      "learning_rate": 3.353004404840121e-07,
+      "loss": 195.6251,
+      "step": 47740
+    },
+    {
+      "epoch": 0.9183531187945053,
+      "grad_norm": 388.5566870010379,
+      "learning_rate": 3.3373507730800167e-07,
+      "loss": 214.3476,
+      "step": 47750
+    },
+    {
+      "epoch": 0.9185454440549858,
+      "grad_norm": 401.7933169326655,
+      "learning_rate": 3.321733146560324e-07,
+      "loss": 202.7705,
+      "step": 47760
+    },
+    {
+      "epoch": 0.9187377693154664,
+      "grad_norm": 395.89760442687924,
+      "learning_rate": 3.306151531098323e-07,
+      "loss": 206.8543,
+      "step": 47770
+    },
+    {
+      "epoch": 0.9189300945759469,
+      "grad_norm": 385.3185963686976,
+      "learning_rate": 3.2906059324979255e-07,
+      "loss": 203.3513,
+      "step": 47780
+    },
+    {
+      "epoch": 0.9191224198364274,
+      "grad_norm": 429.4162990212621,
+      "learning_rate": 3.275096356549612e-07,
+      "loss": 208.1671,
+      "step": 47790
+    },
+    {
+      "epoch": 0.9193147450969079,
+      "grad_norm": 356.6887472597652,
+      "learning_rate": 3.2596228090304496e-07,
+      "loss": 195.5573,
+      "step": 47800
+    },
+    {
+      "epoch": 0.9195070703573884,
+      "grad_norm": 366.6174730315061,
+      "learning_rate": 3.2441852957040607e-07,
+      "loss": 203.6505,
+      "step": 47810
+    },
+    {
+      "epoch": 0.919699395617869,
+      "grad_norm": 334.85261334029383,
+      "learning_rate": 3.228783822320669e-07,
+      "loss": 201.472,
+      "step": 47820
+    },
+    {
+      "epoch": 0.9198917208783495,
+      "grad_norm": 338.19620511147923,
+      "learning_rate": 3.213418394617085e-07,
+      "loss": 203.9608,
+      "step": 47830
+    },
+    {
+      "epoch": 0.92008404613883,
+      "grad_norm": 372.11493415277323,
+      "learning_rate": 3.1980890183166633e-07,
+      "loss": 202.6368,
+      "step": 47840
+    },
+    {
+      "epoch": 0.9202763713993105,
+      "grad_norm": 362.7566210990523,
+      "learning_rate": 3.1827956991293374e-07,
+      "loss": 205.0013,
+      "step": 47850
+    },
+    {
+      "epoch": 0.920468696659791,
+      "grad_norm": 356.19409593986904,
+      "learning_rate": 3.167538442751639e-07,
+      "loss": 205.7227,
+      "step": 47860
+    },
+    {
+      "epoch": 0.9206610219202715,
+      "grad_norm": 359.43992074056297,
+      "learning_rate": 3.1523172548666215e-07,
+      "loss": 204.0532,
+      "step": 47870
+    },
+    {
+      "epoch": 0.9208533471807521,
+      "grad_norm": 371.40706282104145,
+      "learning_rate": 3.1371321411439284e-07,
+      "loss": 208.5364,
+      "step": 47880
+    },
+    {
+      "epoch": 0.9210456724412326,
+      "grad_norm": 371.0894884672173,
+      "learning_rate": 3.1219831072397787e-07,
+      "loss": 204.8019,
+      "step": 47890
+    },
+    {
+      "epoch": 0.9212379977017131,
+      "grad_norm": 337.6790563896661,
+      "learning_rate": 3.1068701587969375e-07,
+      "loss": 211.4002,
+      "step": 47900
+    },
+    {
+      "epoch": 0.9214303229621936,
+      "grad_norm": 404.55132204350554,
+      "learning_rate": 3.091793301444701e-07,
+      "loss": 204.9296,
+      "step": 47910
+    },
+    {
+      "epoch": 0.9216226482226741,
+      "grad_norm": 366.7035027521802,
+      "learning_rate": 3.076752540798977e-07,
+      "loss": 204.2317,
+      "step": 47920
+    },
+    {
+      "epoch": 0.9218149734831547,
+      "grad_norm": 372.1984744229368,
+      "learning_rate": 3.061747882462185e-07,
+      "loss": 206.4927,
+      "step": 47930
+    },
+    {
+      "epoch": 0.9220072987436352,
+      "grad_norm": 356.53694210546337,
+      "learning_rate": 3.0467793320233306e-07,
+      "loss": 204.2181,
+      "step": 47940
+    },
+    {
+      "epoch": 0.9221996240041158,
+      "grad_norm": 344.61641376271024,
+      "learning_rate": 3.031846895057922e-07,
+      "loss": 206.777,
+      "step": 47950
+    },
+    {
+      "epoch": 0.9223919492645963,
+      "grad_norm": 376.2867652943092,
+      "learning_rate": 3.0169505771280747e-07,
+      "loss": 203.469,
+      "step": 47960
+    },
+    {
+      "epoch": 0.9225842745250769,
+      "grad_norm": 360.2214225164992,
+      "learning_rate": 3.002090383782408e-07,
+      "loss": 200.5217,
+      "step": 47970
+    },
+    {
+      "epoch": 0.9227765997855574,
+      "grad_norm": 403.49409568696825,
+      "learning_rate": 2.9872663205561035e-07,
+      "loss": 201.0238,
+      "step": 47980
+    },
+    {
+      "epoch": 0.9229689250460379,
+      "grad_norm": 381.81450013793,
+      "learning_rate": 2.972478392970857e-07,
+      "loss": 207.7948,
+      "step": 47990
+    },
+    {
+      "epoch": 0.9231612503065184,
+      "grad_norm": 411.2418845791998,
+      "learning_rate": 2.9577266065349716e-07,
+      "loss": 201.1859,
+      "step": 48000
+    },
+    {
+      "epoch": 0.9233535755669989,
+      "grad_norm": 349.44293308445873,
+      "learning_rate": 2.9430109667432096e-07,
+      "loss": 203.7939,
+      "step": 48010
+    },
+    {
+      "epoch": 0.9235459008274794,
+      "grad_norm": 404.8199214393542,
+      "learning_rate": 2.9283314790769177e-07,
+      "loss": 210.031,
+      "step": 48020
+    },
+    {
+      "epoch": 0.92373822608796,
+      "grad_norm": 360.9611356091389,
+      "learning_rate": 2.913688149003946e-07,
+      "loss": 201.0061,
+      "step": 48030
+    },
+    {
+      "epoch": 0.9239305513484405,
+      "grad_norm": 377.75875322922377,
+      "learning_rate": 2.899080981978719e-07,
+      "loss": 208.6834,
+      "step": 48040
+    },
+    {
+      "epoch": 0.924122876608921,
+      "grad_norm": 347.49444750421264,
+      "learning_rate": 2.8845099834421517e-07,
+      "loss": 198.3466,
+      "step": 48050
+    },
+    {
+      "epoch": 0.9243152018694015,
+      "grad_norm": 363.1898996353905,
+      "learning_rate": 2.869975158821681e-07,
+      "loss": 202.7241,
+      "step": 48060
+    },
+    {
+      "epoch": 0.924507527129882,
+      "grad_norm": 373.73072467323504,
+      "learning_rate": 2.8554765135313303e-07,
+      "loss": 210.3304,
+      "step": 48070
+    },
+    {
+      "epoch": 0.9246998523903626,
+      "grad_norm": 346.3678165713118,
+      "learning_rate": 2.8410140529715803e-07,
+      "loss": 198.656,
+      "step": 48080
+    },
+    {
+      "epoch": 0.9248921776508431,
+      "grad_norm": 408.8899856385577,
+      "learning_rate": 2.826587782529444e-07,
+      "loss": 197.3259,
+      "step": 48090
+    },
+    {
+      "epoch": 0.9250845029113236,
+      "grad_norm": 395.28756397129325,
+      "learning_rate": 2.812197707578501e-07,
+      "loss": 206.7726,
+      "step": 48100
+    },
+    {
+      "epoch": 0.9252768281718041,
+      "grad_norm": 374.73883939745434,
+      "learning_rate": 2.797843833478797e-07,
+      "loss": 203.0115,
+      "step": 48110
+    },
+    {
+      "epoch": 0.9254691534322846,
+      "grad_norm": 414.48937512172563,
+      "learning_rate": 2.7835261655769217e-07,
+      "loss": 198.1998,
+      "step": 48120
+    },
+    {
+      "epoch": 0.9256614786927652,
+      "grad_norm": 383.48483959778997,
+      "learning_rate": 2.769244709205976e-07,
+      "loss": 202.3205,
+      "step": 48130
+    },
+    {
+      "epoch": 0.9258538039532457,
+      "grad_norm": 354.60566734302694,
+      "learning_rate": 2.7549994696855376e-07,
+      "loss": 199.4333,
+      "step": 48140
+    },
+    {
+      "epoch": 0.9260461292137262,
+      "grad_norm": 351.7661442332772,
+      "learning_rate": 2.740790452321751e-07,
+      "loss": 197.2775,
+      "step": 48150
+    },
+    {
+      "epoch": 0.9262384544742068,
+      "grad_norm": 347.18242948752805,
+      "learning_rate": 2.726617662407238e-07,
+      "loss": 198.4915,
+      "step": 48160
+    },
+    {
+      "epoch": 0.9264307797346873,
+      "grad_norm": 363.7257243553536,
+      "learning_rate": 2.7124811052211097e-07,
+      "loss": 207.6766,
+      "step": 48170
+    },
+    {
+      "epoch": 0.9266231049951679,
+      "grad_norm": 373.9905698629852,
+      "learning_rate": 2.698380786029031e-07,
+      "loss": 199.6323,
+      "step": 48180
+    },
+    {
+      "epoch": 0.9268154302556484,
+      "grad_norm": 380.6972835135455,
+      "learning_rate": 2.6843167100831125e-07,
+      "loss": 204.1734,
+      "step": 48190
+    },
+    {
+      "epoch": 0.9270077555161289,
+      "grad_norm": 365.96935316520796,
+      "learning_rate": 2.6702888826219965e-07,
+      "loss": 205.8778,
+      "step": 48200
+    },
+    {
+      "epoch": 0.9272000807766094,
+      "grad_norm": 367.64405489146344,
+      "learning_rate": 2.6562973088708146e-07,
+      "loss": 201.8723,
+      "step": 48210
+    },
+    {
+      "epoch": 0.92739240603709,
+      "grad_norm": 361.5171836618392,
+      "learning_rate": 2.6423419940412086e-07,
+      "loss": 199.7775,
+      "step": 48220
+    },
+    {
+      "epoch": 0.9275847312975705,
+      "grad_norm": 360.6448170792559,
+      "learning_rate": 2.628422943331288e-07,
+      "loss": 206.8077,
+      "step": 48230
+    },
+    {
+      "epoch": 0.927777056558051,
+      "grad_norm": 355.49062067023806,
+      "learning_rate": 2.614540161925683e-07,
+      "loss": 207.698,
+      "step": 48240
+    },
+    {
+      "epoch": 0.9279693818185315,
+      "grad_norm": 359.92747535186066,
+      "learning_rate": 2.6006936549954784e-07,
+      "loss": 203.5888,
+      "step": 48250
+    },
+    {
+      "epoch": 0.928161707079012,
+      "grad_norm": 335.6376748689618,
+      "learning_rate": 2.5868834276983057e-07,
+      "loss": 206.4889,
+      "step": 48260
+    },
+    {
+      "epoch": 0.9283540323394925,
+      "grad_norm": 353.81266355058114,
+      "learning_rate": 2.573109485178216e-07,
+      "loss": 208.7306,
+      "step": 48270
+    },
+    {
+      "epoch": 0.9285463575999731,
+      "grad_norm": 342.592150067986,
+      "learning_rate": 2.5593718325657713e-07,
+      "loss": 208.251,
+      "step": 48280
+    },
+    {
+      "epoch": 0.9287386828604536,
+      "grad_norm": 387.7713118351166,
+      "learning_rate": 2.545670474978057e-07,
+      "loss": 197.3804,
+      "step": 48290
+    },
+    {
+      "epoch": 0.9289310081209341,
+      "grad_norm": 377.23164123419025,
+      "learning_rate": 2.532005417518568e-07,
+      "loss": 199.8187,
+      "step": 48300
+    },
+    {
+      "epoch": 0.9291233333814146,
+      "grad_norm": 360.85145185611924,
+      "learning_rate": 2.5183766652773336e-07,
+      "loss": 200.0176,
+      "step": 48310
+    },
+    {
+      "epoch": 0.9293156586418951,
+      "grad_norm": 376.82586839149235,
+      "learning_rate": 2.504784223330814e-07,
+      "loss": 200.3287,
+      "step": 48320
+    },
+    {
+      "epoch": 0.9295079839023757,
+      "grad_norm": 387.4454318593123,
+      "learning_rate": 2.4912280967419934e-07,
+      "loss": 204.5187,
+      "step": 48330
+    },
+    {
+      "epoch": 0.9297003091628562,
+      "grad_norm": 370.70729747122306,
+      "learning_rate": 2.477708290560299e-07,
+      "loss": 212.5554,
+      "step": 48340
+    },
+    {
+      "epoch": 0.9298926344233367,
+      "grad_norm": 358.11856272218097,
+      "learning_rate": 2.464224809821614e-07,
+      "loss": 214.3949,
+      "step": 48350
+    },
+    {
+      "epoch": 0.9300849596838173,
+      "grad_norm": 363.064909752825,
+      "learning_rate": 2.450777659548353e-07,
+      "loss": 206.2752,
+      "step": 48360
+    },
+    {
+      "epoch": 0.9302772849442978,
+      "grad_norm": 379.0780338120148,
+      "learning_rate": 2.4373668447493225e-07,
+      "loss": 208.6998,
+      "step": 48370
+    },
+    {
+      "epoch": 0.9304696102047784,
+      "grad_norm": 328.99901329137055,
+      "learning_rate": 2.4239923704198476e-07,
+      "loss": 201.4028,
+      "step": 48380
+    },
+    {
+      "epoch": 0.9306619354652589,
+      "grad_norm": 345.89962421584164,
+      "learning_rate": 2.410654241541688e-07,
+      "loss": 199.6826,
+      "step": 48390
+    },
+    {
+      "epoch": 0.9308542607257394,
+      "grad_norm": 359.66192706034724,
+      "learning_rate": 2.3973524630830804e-07,
+      "loss": 204.3283,
+      "step": 48400
+    },
+    {
+      "epoch": 0.9310465859862199,
+      "grad_norm": 363.39198243705334,
+      "learning_rate": 2.3840870399987283e-07,
+      "loss": 206.5529,
+      "step": 48410
+    },
+    {
+      "epoch": 0.9312389112467004,
+      "grad_norm": 341.8669495957109,
+      "learning_rate": 2.370857977229768e-07,
+      "loss": 206.5942,
+      "step": 48420
+    },
+    {
+      "epoch": 0.931431236507181,
+      "grad_norm": 349.1426673710083,
+      "learning_rate": 2.3576652797038247e-07,
+      "loss": 205.0635,
+      "step": 48430
+    },
+    {
+      "epoch": 0.9316235617676615,
+      "grad_norm": 373.8402796169833,
+      "learning_rate": 2.344508952334934e-07,
+      "loss": 201.2203,
+      "step": 48440
+    },
+    {
+      "epoch": 0.931815887028142,
+      "grad_norm": 358.8377869045671,
+      "learning_rate": 2.3313890000236316e-07,
+      "loss": 199.2215,
+      "step": 48450
+    },
+    {
+      "epoch": 0.9320082122886225,
+      "grad_norm": 351.36239148372766,
+      "learning_rate": 2.3183054276568752e-07,
+      "loss": 201.9332,
+      "step": 48460
+    },
+    {
+      "epoch": 0.932200537549103,
+      "grad_norm": 362.1154369119544,
+      "learning_rate": 2.305258240108077e-07,
+      "loss": 199.5552,
+      "step": 48470
+    },
+    {
+      "epoch": 0.9323928628095836,
+      "grad_norm": 375.7211434386111,
+      "learning_rate": 2.2922474422371166e-07,
+      "loss": 202.7204,
+      "step": 48480
+    },
+    {
+      "epoch": 0.9325851880700641,
+      "grad_norm": 347.4714011997012,
+      "learning_rate": 2.279273038890273e-07,
+      "loss": 205.913,
+      "step": 48490
+    },
+    {
+      "epoch": 0.9327775133305446,
+      "grad_norm": 369.7647960670378,
+      "learning_rate": 2.2663350349003134e-07,
+      "loss": 203.0409,
+      "step": 48500
+    },
+    {
+      "epoch": 0.9329698385910251,
+      "grad_norm": 363.10036707668166,
+      "learning_rate": 2.2534334350864274e-07,
+      "loss": 196.7114,
+      "step": 48510
+    },
+    {
+      "epoch": 0.9331621638515056,
+      "grad_norm": 407.3035070825069,
+      "learning_rate": 2.2405682442542487e-07,
+      "loss": 204.9385,
+      "step": 48520
+    },
+    {
+      "epoch": 0.9333544891119862,
+      "grad_norm": 344.4587572631576,
+      "learning_rate": 2.2277394671958442e-07,
+      "loss": 204.7839,
+      "step": 48530
+    },
+    {
+      "epoch": 0.9335468143724667,
+      "grad_norm": 367.51882233901233,
+      "learning_rate": 2.2149471086897355e-07,
+      "loss": 203.4901,
+      "step": 48540
+    },
+    {
+      "epoch": 0.9337391396329472,
+      "grad_norm": 430.9405742211201,
+      "learning_rate": 2.202191173500845e-07,
+      "loss": 207.0275,
+      "step": 48550
+    },
+    {
+      "epoch": 0.9339314648934277,
+      "grad_norm": 399.93485206691724,
+      "learning_rate": 2.1894716663805716e-07,
+      "loss": 206.4945,
+      "step": 48560
+    },
+    {
+      "epoch": 0.9341237901539083,
+      "grad_norm": 383.76664637446873,
+      "learning_rate": 2.176788592066692e-07,
+      "loss": 214.786,
+      "step": 48570
+    },
+    {
+      "epoch": 0.9343161154143889,
+      "grad_norm": 386.38907715344436,
+      "learning_rate": 2.164141955283472e-07,
+      "loss": 198.5718,
+      "step": 48580
+    },
+    {
+      "epoch": 0.9345084406748694,
+      "grad_norm": 359.70027705460774,
+      "learning_rate": 2.1515317607415654e-07,
+      "loss": 204.1084,
+      "step": 48590
+    },
+    {
+      "epoch": 0.9347007659353499,
+      "grad_norm": 345.1851583304933,
+      "learning_rate": 2.1389580131380373e-07,
+      "loss": 218.2766,
+      "step": 48600
+    },
+    {
+      "epoch": 0.9348930911958304,
+      "grad_norm": 396.48516202017555,
+      "learning_rate": 2.126420717156441e-07,
+      "loss": 213.2414,
+      "step": 48610
+    },
+    {
+      "epoch": 0.9350854164563109,
+      "grad_norm": 379.4163502223697,
+      "learning_rate": 2.113919877466686e-07,
+      "loss": 199.8932,
+      "step": 48620
+    },
+    {
+      "epoch": 0.9352777417167915,
+      "grad_norm": 431.10562221510736,
+      "learning_rate": 2.1014554987251356e-07,
+      "loss": 209.9968,
+      "step": 48630
+    },
+    {
+      "epoch": 0.935470066977272,
+      "grad_norm": 355.1958101296038,
+      "learning_rate": 2.0890275855745546e-07,
+      "loss": 202.85,
+      "step": 48640
+    },
+    {
+      "epoch": 0.9356623922377525,
+      "grad_norm": 361.6536819389774,
+      "learning_rate": 2.0766361426441505e-07,
+      "loss": 200.2586,
+      "step": 48650
+    },
+    {
+      "epoch": 0.935854717498233,
+      "grad_norm": 351.229851464096,
+      "learning_rate": 2.0642811745495206e-07,
+      "loss": 210.561,
+      "step": 48660
+    },
+    {
+      "epoch": 0.9360470427587135,
+      "grad_norm": 387.1118790965528,
+      "learning_rate": 2.0519626858926944e-07,
+      "loss": 203.5569,
+      "step": 48670
+    },
+    {
+      "epoch": 0.936239368019194,
+      "grad_norm": 362.0983303705198,
+      "learning_rate": 2.0396806812621018e-07,
+      "loss": 203.5939,
+      "step": 48680
+    },
+    {
+      "epoch": 0.9364316932796746,
+      "grad_norm": 372.9376111079694,
+      "learning_rate": 2.0274351652325942e-07,
+      "loss": 211.234,
+      "step": 48690
+    },
+    {
+      "epoch": 0.9366240185401551,
+      "grad_norm": 358.5530843818397,
+      "learning_rate": 2.0152261423654118e-07,
+      "loss": 207.7548,
+      "step": 48700
+    },
+    {
+      "epoch": 0.9368163438006356,
+      "grad_norm": 346.21485175206715,
+      "learning_rate": 2.003053617208217e-07,
+      "loss": 201.5005,
+      "step": 48710
+    },
+    {
+      "epoch": 0.9370086690611161,
+      "grad_norm": 380.7950642401895,
+      "learning_rate": 1.9909175942950832e-07,
+      "loss": 211.7329,
+      "step": 48720
+    },
+    {
+      "epoch": 0.9372009943215966,
+      "grad_norm": 409.8248611235089,
+      "learning_rate": 1.9788180781464716e-07,
+      "loss": 204.814,
+      "step": 48730
+    },
+    {
+      "epoch": 0.9373933195820772,
+      "grad_norm": 501.76856256447684,
+      "learning_rate": 1.9667550732692554e-07,
+      "loss": 206.4797,
+      "step": 48740
+    },
+    {
+      "epoch": 0.9375856448425577,
+      "grad_norm": 339.3430298952168,
+      "learning_rate": 1.9547285841566843e-07,
+      "loss": 198.2602,
+      "step": 48750
+    },
+    {
+      "epoch": 0.9377779701030382,
+      "grad_norm": 368.6174017289793,
+      "learning_rate": 1.9427386152884643e-07,
+      "loss": 202.4432,
+      "step": 48760
+    },
+    {
+      "epoch": 0.9379702953635188,
+      "grad_norm": 389.6407474801892,
+      "learning_rate": 1.9307851711306336e-07,
+      "loss": 205.2368,
+      "step": 48770
+    },
+    {
+      "epoch": 0.9381626206239994,
+      "grad_norm": 353.2704467807482,
+      "learning_rate": 1.918868256135653e-07,
+      "loss": 207.9949,
+      "step": 48780
+    },
+    {
+      "epoch": 0.9383549458844799,
+      "grad_norm": 344.1559176437801,
+      "learning_rate": 1.9069878747423943e-07,
+      "loss": 203.307,
+      "step": 48790
+    },
+    {
+      "epoch": 0.9385472711449604,
+      "grad_norm": 392.8586903856337,
+      "learning_rate": 1.8951440313760838e-07,
+      "loss": 204.1177,
+      "step": 48800
+    },
+    {
+      "epoch": 0.9387395964054409,
+      "grad_norm": 362.4137931939056,
+      "learning_rate": 1.8833367304483708e-07,
+      "loss": 206.1665,
+      "step": 48810
+    },
+    {
+      "epoch": 0.9389319216659214,
+      "grad_norm": 359.6413979311788,
+      "learning_rate": 1.8715659763572703e-07,
+      "loss": 203.5739,
+      "step": 48820
+    },
+    {
+      "epoch": 0.939124246926402,
+      "grad_norm": 366.1872098363556,
+      "learning_rate": 1.8598317734872086e-07,
+      "loss": 201.9677,
+      "step": 48830
+    },
+    {
+      "epoch": 0.9393165721868825,
+      "grad_norm": 381.0241662016051,
+      "learning_rate": 1.8481341262089668e-07,
+      "loss": 205.5354,
+      "step": 48840
+    },
+    {
+      "epoch": 0.939508897447363,
+      "grad_norm": 410.5420664912381,
+      "learning_rate": 1.836473038879727e-07,
+      "loss": 200.2537,
+      "step": 48850
+    },
+    {
+      "epoch": 0.9397012227078435,
+      "grad_norm": 416.3983696284498,
+      "learning_rate": 1.8248485158430696e-07,
+      "loss": 203.7448,
+      "step": 48860
+    },
+    {
+      "epoch": 0.939893547968324,
+      "grad_norm": 386.79674397757145,
+      "learning_rate": 1.813260561428909e-07,
+      "loss": 202.6938,
+      "step": 48870
+    },
+    {
+      "epoch": 0.9400858732288045,
+      "grad_norm": 390.39280922089506,
+      "learning_rate": 1.8017091799535924e-07,
+      "loss": 199.2614,
+      "step": 48880
+    },
+    {
+      "epoch": 0.9402781984892851,
+      "grad_norm": 350.1015938046084,
+      "learning_rate": 1.7901943757198003e-07,
+      "loss": 199.1779,
+      "step": 48890
+    },
+    {
+      "epoch": 0.9404705237497656,
+      "grad_norm": 355.49705519233663,
+      "learning_rate": 1.7787161530166242e-07,
+      "loss": 203.8312,
+      "step": 48900
+    },
+    {
+      "epoch": 0.9406628490102461,
+      "grad_norm": 386.0596600024102,
+      "learning_rate": 1.7672745161194992e-07,
+      "loss": 205.6768,
+      "step": 48910
+    },
+    {
+      "epoch": 0.9408551742707266,
+      "grad_norm": 362.82443817959654,
+      "learning_rate": 1.75586946929025e-07,
+      "loss": 201.8567,
+      "step": 48920
+    },
+    {
+      "epoch": 0.9410474995312071,
+      "grad_norm": 362.41809499919225,
+      "learning_rate": 1.7445010167770672e-07,
+      "loss": 206.5927,
+      "step": 48930
+    },
+    {
+      "epoch": 0.9412398247916877,
+      "grad_norm": 375.539960908434,
+      "learning_rate": 1.7331691628145076e-07,
+      "loss": 206.4405,
+      "step": 48940
+    },
+    {
+      "epoch": 0.9414321500521682,
+      "grad_norm": 382.0279265982589,
+      "learning_rate": 1.7218739116235061e-07,
+      "loss": 203.9179,
+      "step": 48950
+    },
+    {
+      "epoch": 0.9416244753126487,
+      "grad_norm": 344.5599209163815,
+      "learning_rate": 1.710615267411353e-07,
+      "loss": 198.841,
+      "step": 48960
+    },
+    {
+      "epoch": 0.9418168005731293,
+      "grad_norm": 382.27322880808947,
+      "learning_rate": 1.6993932343717158e-07,
+      "loss": 203.6364,
+      "step": 48970
+    },
+    {
+      "epoch": 0.9420091258336099,
+      "grad_norm": 361.94592765984186,
+      "learning_rate": 1.6882078166846173e-07,
+      "loss": 198.9499,
+      "step": 48980
+    },
+    {
+      "epoch": 0.9422014510940904,
+      "grad_norm": 359.8541416803671,
+      "learning_rate": 1.677059018516425e-07,
+      "loss": 199.0995,
+      "step": 48990
+    },
+    {
+      "epoch": 0.9423937763545709,
+      "grad_norm": 401.1738769568403,
+      "learning_rate": 1.6659468440198835e-07,
+      "loss": 206.322,
+      "step": 49000
+    },
+    {
+      "epoch": 0.9425861016150514,
+      "grad_norm": 330.51819131816546,
+      "learning_rate": 1.6548712973341152e-07,
+      "loss": 216.3232,
+      "step": 49010
+    },
+    {
+      "epoch": 0.9427784268755319,
+      "grad_norm": 376.554847946768,
+      "learning_rate": 1.6438323825845647e-07,
+      "loss": 200.087,
+      "step": 49020
+    },
+    {
+      "epoch": 0.9429707521360124,
+      "grad_norm": 368.0763010305208,
+      "learning_rate": 1.6328301038830429e-07,
+      "loss": 205.7377,
+      "step": 49030
+    },
+    {
+      "epoch": 0.943163077396493,
+      "grad_norm": 389.1818584560128,
+      "learning_rate": 1.621864465327716e-07,
+      "loss": 197.0627,
+      "step": 49040
+    },
+    {
+      "epoch": 0.9433554026569735,
+      "grad_norm": 407.1234660885087,
+      "learning_rate": 1.6109354710031167e-07,
+      "loss": 207.463,
+      "step": 49050
+    },
+    {
+      "epoch": 0.943547727917454,
+      "grad_norm": 337.339471767984,
+      "learning_rate": 1.6000431249800995e-07,
+      "loss": 194.6463,
+      "step": 49060
+    },
+    {
+      "epoch": 0.9437400531779345,
+      "grad_norm": 331.3329776157531,
+      "learning_rate": 1.5891874313158862e-07,
+      "loss": 194.299,
+      "step": 49070
+    },
+    {
+      "epoch": 0.943932378438415,
+      "grad_norm": 366.63202388553077,
+      "learning_rate": 1.5783683940540528e-07,
+      "loss": 200.2874,
+      "step": 49080
+    },
+    {
+      "epoch": 0.9441247036988956,
+      "grad_norm": 348.4823437808373,
+      "learning_rate": 1.5675860172244982e-07,
+      "loss": 199.1079,
+      "step": 49090
+    },
+    {
+      "epoch": 0.9443170289593761,
+      "grad_norm": 355.51891158510244,
+      "learning_rate": 1.5568403048434877e-07,
+      "loss": 198.4023,
+      "step": 49100
+    },
+    {
+      "epoch": 0.9445093542198566,
+      "grad_norm": 341.23308673530204,
+      "learning_rate": 1.5461312609136192e-07,
+      "loss": 201.2382,
+      "step": 49110
+    },
+    {
+      "epoch": 0.9447016794803371,
+      "grad_norm": 379.54409283688267,
+      "learning_rate": 1.5354588894238465e-07,
+      "loss": 198.9856,
+      "step": 49120
+    },
+    {
+      "epoch": 0.9448940047408176,
+      "grad_norm": 352.0549935720582,
+      "learning_rate": 1.524823194349434e-07,
+      "loss": 206.9746,
+      "step": 49130
+    },
+    {
+      "epoch": 0.9450863300012982,
+      "grad_norm": 379.5768404078221,
+      "learning_rate": 1.514224179652013e-07,
+      "loss": 206.9802,
+      "step": 49140
+    },
+    {
+      "epoch": 0.9452786552617787,
+      "grad_norm": 388.7938587044275,
+      "learning_rate": 1.5036618492795473e-07,
+      "loss": 204.4031,
+      "step": 49150
+    },
+    {
+      "epoch": 0.9454709805222592,
+      "grad_norm": 355.03687574565714,
+      "learning_rate": 1.4931362071663125e-07,
+      "loss": 202.7154,
+      "step": 49160
+    },
+    {
+      "epoch": 0.9456633057827397,
+      "grad_norm": 357.95611089396147,
+      "learning_rate": 1.4826472572329498e-07,
+      "loss": 202.3218,
+      "step": 49170
+    },
+    {
+      "epoch": 0.9458556310432203,
+      "grad_norm": 374.21875249515443,
+      "learning_rate": 1.4721950033864118e-07,
+      "loss": 197.0,
+      "step": 49180
+    },
+    {
+      "epoch": 0.9460479563037009,
+      "grad_norm": 349.2312173960439,
+      "learning_rate": 1.4617794495199956e-07,
+      "loss": 206.4769,
+      "step": 49190
+    },
+    {
+      "epoch": 0.9462402815641814,
+      "grad_norm": 374.2752050855024,
+      "learning_rate": 1.4514005995133197e-07,
+      "loss": 205.4025,
+      "step": 49200
+    },
+    {
+      "epoch": 0.9464326068246619,
+      "grad_norm": 360.80127601919077,
+      "learning_rate": 1.441058457232336e-07,
+      "loss": 199.2605,
+      "step": 49210
+    },
+    {
+      "epoch": 0.9466249320851424,
+      "grad_norm": 369.57335289896054,
+      "learning_rate": 1.43075302652933e-07,
+      "loss": 209.8724,
+      "step": 49220
+    },
+    {
+      "epoch": 0.9468172573456229,
+      "grad_norm": 356.4890944029722,
+      "learning_rate": 1.4204843112428867e-07,
+      "loss": 204.1399,
+      "step": 49230
+    },
+    {
+      "epoch": 0.9470095826061035,
+      "grad_norm": 356.8180844794629,
+      "learning_rate": 1.4102523151979574e-07,
+      "loss": 202.7515,
+      "step": 49240
+    },
+    {
+      "epoch": 0.947201907866584,
+      "grad_norm": 403.9491451758941,
+      "learning_rate": 1.40005704220576e-07,
+      "loss": 206.3632,
+      "step": 49250
+    },
+    {
+      "epoch": 0.9473942331270645,
+      "grad_norm": 362.0320227388929,
+      "learning_rate": 1.3898984960638907e-07,
+      "loss": 204.8017,
+      "step": 49260
+    },
+    {
+      "epoch": 0.947586558387545,
+      "grad_norm": 410.84140402433786,
+      "learning_rate": 1.3797766805562328e-07,
+      "loss": 207.2748,
+      "step": 49270
+    },
+    {
+      "epoch": 0.9477788836480255,
+      "grad_norm": 408.68346359505136,
+      "learning_rate": 1.3696915994530048e-07,
+      "loss": 203.7896,
+      "step": 49280
+    },
+    {
+      "epoch": 0.9479712089085061,
+      "grad_norm": 346.2945964479388,
+      "learning_rate": 1.359643256510701e-07,
+      "loss": 198.9533,
+      "step": 49290
+    },
+    {
+      "epoch": 0.9481635341689866,
+      "grad_norm": 364.7453645051926,
+      "learning_rate": 1.3496316554722056e-07,
+      "loss": 201.9824,
+      "step": 49300
+    },
+    {
+      "epoch": 0.9483558594294671,
+      "grad_norm": 373.742128369879,
+      "learning_rate": 1.3396568000666467e-07,
+      "loss": 204.5055,
+      "step": 49310
+    },
+    {
+      "epoch": 0.9485481846899476,
+      "grad_norm": 361.09834362350585,
+      "learning_rate": 1.3297186940094853e-07,
+      "loss": 207.0482,
+      "step": 49320
+    },
+    {
+      "epoch": 0.9487405099504281,
+      "grad_norm": 407.5412021337874,
+      "learning_rate": 1.3198173410025383e-07,
+      "loss": 200.9387,
+      "step": 49330
+    },
+    {
+      "epoch": 0.9489328352109087,
+      "grad_norm": 384.1957083876091,
+      "learning_rate": 1.3099527447338668e-07,
+      "loss": 205.2417,
+      "step": 49340
+    },
+    {
+      "epoch": 0.9491251604713892,
+      "grad_norm": 376.81351588748106,
+      "learning_rate": 1.3001249088778756e-07,
+      "loss": 208.6732,
+      "step": 49350
+    },
+    {
+      "epoch": 0.9493174857318697,
+      "grad_norm": 417.5765423310436,
+      "learning_rate": 1.29033383709527e-07,
+      "loss": 204.7751,
+      "step": 49360
+    },
+    {
+      "epoch": 0.9495098109923502,
+      "grad_norm": 389.2639735642371,
+      "learning_rate": 1.2805795330330774e-07,
+      "loss": 206.6264,
+      "step": 49370
+    },
+    {
+      "epoch": 0.9497021362528308,
+      "grad_norm": 355.58374485327874,
+      "learning_rate": 1.2708620003245908e-07,
+      "loss": 210.9091,
+      "step": 49380
+    },
+    {
+      "epoch": 0.9498944615133114,
+      "grad_norm": 358.71833706596266,
+      "learning_rate": 1.2611812425894487e-07,
+      "loss": 195.0542,
+      "step": 49390
+    },
+    {
+      "epoch": 0.9500867867737919,
+      "grad_norm": 366.26793443105026,
+      "learning_rate": 1.2515372634335666e-07,
+      "loss": 199.6854,
+      "step": 49400
+    },
+    {
+      "epoch": 0.9502791120342724,
+      "grad_norm": 370.2548974783838,
+      "learning_rate": 1.2419300664491707e-07,
+      "loss": 214.1474,
+      "step": 49410
+    },
+    {
+      "epoch": 0.9504714372947529,
+      "grad_norm": 339.2438335057985,
+      "learning_rate": 1.2323596552147876e-07,
+      "loss": 204.4429,
+      "step": 49420
+    },
+    {
+      "epoch": 0.9506637625552334,
+      "grad_norm": 364.2600805628213,
+      "learning_rate": 1.2228260332952213e-07,
+      "loss": 204.4185,
+      "step": 49430
+    },
+    {
+      "epoch": 0.950856087815714,
+      "grad_norm": 405.59084423019425,
+      "learning_rate": 1.213329204241609e-07,
+      "loss": 199.0191,
+      "step": 49440
+    },
+    {
+      "epoch": 0.9510484130761945,
+      "grad_norm": 389.25992145419013,
+      "learning_rate": 1.203869171591343e-07,
+      "loss": 198.2744,
+      "step": 49450
+    },
+    {
+      "epoch": 0.951240738336675,
+      "grad_norm": 418.872131031006,
+      "learning_rate": 1.1944459388681496e-07,
+      "loss": 211.5102,
+      "step": 49460
+    },
+    {
+      "epoch": 0.9514330635971555,
+      "grad_norm": 381.99020641868276,
+      "learning_rate": 1.1850595095820095e-07,
+      "loss": 205.557,
+      "step": 49470
+    },
+    {
+      "epoch": 0.951625388857636,
+      "grad_norm": 400.31684008483734,
+      "learning_rate": 1.175709887229215e-07,
+      "loss": 199.0176,
+      "step": 49480
+    },
+    {
+      "epoch": 0.9518177141181166,
+      "grad_norm": 390.2782834876259,
+      "learning_rate": 1.1663970752923581e-07,
+      "loss": 201.231,
+      "step": 49490
+    },
+    {
+      "epoch": 0.9520100393785971,
+      "grad_norm": 336.31515318587105,
+      "learning_rate": 1.1571210772402975e-07,
+      "loss": 203.0084,
+      "step": 49500
+    },
+    {
+      "epoch": 0.9522023646390776,
+      "grad_norm": 370.2327882528829,
+      "learning_rate": 1.1478818965281912e-07,
+      "loss": 200.8858,
+      "step": 49510
+    },
+    {
+      "epoch": 0.9523946898995581,
+      "grad_norm": 414.32139466194104,
+      "learning_rate": 1.1386795365974757e-07,
+      "loss": 208.771,
+      "step": 49520
+    },
+    {
+      "epoch": 0.9525870151600386,
+      "grad_norm": 373.1270070639388,
+      "learning_rate": 1.1295140008758864e-07,
+      "loss": 202.6968,
+      "step": 49530
+    },
+    {
+      "epoch": 0.9527793404205191,
+      "grad_norm": 485.10358612647025,
+      "learning_rate": 1.1203852927774372e-07,
+      "loss": 205.5088,
+      "step": 49540
+    },
+    {
+      "epoch": 0.9529716656809997,
+      "grad_norm": 397.7323628375016,
+      "learning_rate": 1.111293415702408e-07,
+      "loss": 201.8083,
+      "step": 49550
+    },
+    {
+      "epoch": 0.9531639909414802,
+      "grad_norm": 376.61355488506143,
+      "learning_rate": 1.1022383730373897e-07,
+      "loss": 205.5113,
+      "step": 49560
+    },
+    {
+      "epoch": 0.9533563162019607,
+      "grad_norm": 356.5223479198953,
+      "learning_rate": 1.093220168155218e-07,
+      "loss": 202.1081,
+      "step": 49570
+    },
+    {
+      "epoch": 0.9535486414624412,
+      "grad_norm": 383.8961525318942,
+      "learning_rate": 1.0842388044150387e-07,
+      "loss": 203.0571,
+      "step": 49580
+    },
+    {
+      "epoch": 0.9537409667229219,
+      "grad_norm": 373.77944159230293,
+      "learning_rate": 1.0752942851622649e-07,
+      "loss": 201.3082,
+      "step": 49590
+    },
+    {
+      "epoch": 0.9539332919834024,
+      "grad_norm": 363.3856229004409,
+      "learning_rate": 1.066386613728565e-07,
+      "loss": 200.6104,
+      "step": 49600
+    },
+    {
+      "epoch": 0.9541256172438829,
+      "grad_norm": 373.2443593283993,
+      "learning_rate": 1.0575157934319069e-07,
+      "loss": 206.9029,
+      "step": 49610
+    },
+    {
+      "epoch": 0.9543179425043634,
+      "grad_norm": 354.611394383355,
+      "learning_rate": 1.0486818275765364e-07,
+      "loss": 202.265,
+      "step": 49620
+    },
+    {
+      "epoch": 0.9545102677648439,
+      "grad_norm": 370.1579322913445,
+      "learning_rate": 1.0398847194529437e-07,
+      "loss": 205.4919,
+      "step": 49630
+    },
+    {
+      "epoch": 0.9547025930253245,
+      "grad_norm": 376.87278121082795,
+      "learning_rate": 1.0311244723379188e-07,
+      "loss": 203.0618,
+      "step": 49640
+    },
+    {
+      "epoch": 0.954894918285805,
+      "grad_norm": 360.9044723112168,
+      "learning_rate": 1.0224010894944958e-07,
+      "loss": 208.2209,
+      "step": 49650
+    },
+    {
+      "epoch": 0.9550872435462855,
+      "grad_norm": 356.3859438699043,
+      "learning_rate": 1.0137145741719867e-07,
+      "loss": 224.3437,
+      "step": 49660
+    },
+    {
+      "epoch": 0.955279568806766,
+      "grad_norm": 415.253070479042,
+      "learning_rate": 1.0050649296060033e-07,
+      "loss": 217.9474,
+      "step": 49670
+    },
+    {
+      "epoch": 0.9554718940672465,
+      "grad_norm": 382.4375103604002,
+      "learning_rate": 9.964521590183684e-08,
+      "loss": 206.8328,
+      "step": 49680
+    },
+    {
+      "epoch": 0.955664219327727,
+      "grad_norm": 382.3781089448624,
+      "learning_rate": 9.878762656172159e-08,
+      "loss": 203.4472,
+      "step": 49690
+    },
+    {
+      "epoch": 0.9558565445882076,
+      "grad_norm": 358.7459553032876,
+      "learning_rate": 9.793372525969125e-08,
+      "loss": 211.4204,
+      "step": 49700
+    },
+    {
+      "epoch": 0.9560488698486881,
+      "grad_norm": 388.9944664591722,
+      "learning_rate": 9.70835123138103e-08,
+      "loss": 206.2478,
+      "step": 49710
+    },
+    {
+      "epoch": 0.9562411951091686,
+      "grad_norm": 372.34198076520255,
+      "learning_rate": 9.623698804076875e-08,
+      "loss": 206.4913,
+      "step": 49720
+    },
+    {
+      "epoch": 0.9564335203696491,
+      "grad_norm": 382.5906584218293,
+      "learning_rate": 9.539415275588326e-08,
+      "loss": 212.3889,
+      "step": 49730
+    },
+    {
+      "epoch": 0.9566258456301296,
+      "grad_norm": 350.8513474884998,
+      "learning_rate": 9.455500677309603e-08,
+      "loss": 219.9304,
+      "step": 49740
+    },
+    {
+      "epoch": 0.9568181708906102,
+      "grad_norm": 388.96588130347004,
+      "learning_rate": 9.371955040497371e-08,
+      "loss": 202.6126,
+      "step": 49750
+    },
+    {
+      "epoch": 0.9570104961510907,
+      "grad_norm": 375.62725859907977,
+      "learning_rate": 9.288778396271292e-08,
+      "loss": 211.2534,
+      "step": 49760
+    },
+    {
+      "epoch": 0.9572028214115712,
+      "grad_norm": 378.96835722770936,
+      "learning_rate": 9.205970775613027e-08,
+      "loss": 204.1733,
+      "step": 49770
+    },
+    {
+      "epoch": 0.9573951466720517,
+      "grad_norm": 444.15361922264344,
+      "learning_rate": 9.123532209367237e-08,
+      "loss": 211.4552,
+      "step": 49780
+    },
+    {
+      "epoch": 0.9575874719325324,
+      "grad_norm": 356.41757443249867,
+      "learning_rate": 9.04146272824069e-08,
+      "loss": 195.6299,
+      "step": 49790
+    },
+    {
+      "epoch": 0.9577797971930129,
+      "grad_norm": 339.34410097238435,
+      "learning_rate": 8.959762362803159e-08,
+      "loss": 201.799,
+      "step": 49800
+    },
+    {
+      "epoch": 0.9579721224534934,
+      "grad_norm": 334.8532234197658,
+      "learning_rate": 8.87843114348652e-08,
+      "loss": 200.7965,
+      "step": 49810
+    },
+    {
+      "epoch": 0.9581644477139739,
+      "grad_norm": 369.6196269310955,
+      "learning_rate": 8.797469100585432e-08,
+      "loss": 208.1472,
+      "step": 49820
+    },
+    {
+      "epoch": 0.9583567729744544,
+      "grad_norm": 433.6968760664219,
+      "learning_rate": 8.716876264256768e-08,
+      "loss": 215.906,
+      "step": 49830
+    },
+    {
+      "epoch": 0.958549098234935,
+      "grad_norm": 359.4031513674069,
+      "learning_rate": 8.636652664520184e-08,
+      "loss": 205.0442,
+      "step": 49840
+    },
+    {
+      "epoch": 0.9587414234954155,
+      "grad_norm": 371.4292114478069,
+      "learning_rate": 8.556798331257555e-08,
+      "loss": 205.6551,
+      "step": 49850
+    },
+    {
+      "epoch": 0.958933748755896,
+      "grad_norm": 391.06737577632396,
+      "learning_rate": 8.477313294213307e-08,
+      "loss": 204.2198,
+      "step": 49860
+    },
+    {
+      "epoch": 0.9591260740163765,
+      "grad_norm": 359.97384413061303,
+      "learning_rate": 8.398197582994316e-08,
+      "loss": 197.127,
+      "step": 49870
+    },
+    {
+      "epoch": 0.959318399276857,
+      "grad_norm": 365.9605009515636,
+      "learning_rate": 8.319451227069897e-08,
+      "loss": 205.0339,
+      "step": 49880
+    },
+    {
+      "epoch": 0.9595107245373375,
+      "grad_norm": 357.00235980264677,
+      "learning_rate": 8.241074255771808e-08,
+      "loss": 200.3495,
+      "step": 49890
+    },
+    {
+      "epoch": 0.9597030497978181,
+      "grad_norm": 357.12807224336836,
+      "learning_rate": 8.163066698294031e-08,
+      "loss": 204.6414,
+      "step": 49900
+    },
+    {
+      "epoch": 0.9598953750582986,
+      "grad_norm": 358.2252431919054,
+      "learning_rate": 8.085428583693211e-08,
+      "loss": 201.7902,
+      "step": 49910
+    },
+    {
+      "epoch": 0.9600877003187791,
+      "grad_norm": 369.9463084300626,
+      "learning_rate": 8.008159940888216e-08,
+      "loss": 203.2062,
+      "step": 49920
+    },
+    {
+      "epoch": 0.9602800255792596,
+      "grad_norm": 358.4508390369491,
+      "learning_rate": 7.931260798660356e-08,
+      "loss": 206.8046,
+      "step": 49930
+    },
+    {
+      "epoch": 0.9604723508397401,
+      "grad_norm": 338.20180019751376,
+      "learning_rate": 7.854731185653386e-08,
+      "loss": 199.1347,
+      "step": 49940
+    },
+    {
+      "epoch": 0.9606646761002207,
+      "grad_norm": 346.69498247382717,
+      "learning_rate": 7.778571130373059e-08,
+      "loss": 202.7521,
+      "step": 49950
+    },
+    {
+      "epoch": 0.9608570013607012,
+      "grad_norm": 359.8435702581509,
+      "learning_rate": 7.702780661188014e-08,
+      "loss": 206.7363,
+      "step": 49960
+    },
+    {
+      "epoch": 0.9610493266211817,
+      "grad_norm": 353.89086273997054,
+      "learning_rate": 7.627359806328782e-08,
+      "loss": 205.0232,
+      "step": 49970
+    },
+    {
+      "epoch": 0.9612416518816622,
+      "grad_norm": 373.7622693781548,
+      "learning_rate": 7.552308593888558e-08,
+      "loss": 204.5835,
+      "step": 49980
+    },
+    {
+      "epoch": 0.9614339771421427,
+      "grad_norm": 373.6093164309671,
+      "learning_rate": 7.477627051822534e-08,
+      "loss": 209.0156,
+      "step": 49990
+    },
+    {
+      "epoch": 0.9616263024026234,
+      "grad_norm": 394.8648310465121,
+      "learning_rate": 7.403315207948236e-08,
+      "loss": 212.7868,
+      "step": 50000
+    },
+    {
+      "epoch": 0.9618186276631039,
+      "grad_norm": 389.73245897893435,
+      "learning_rate": 7.329373089945968e-08,
+      "loss": 197.4793,
+      "step": 50010
+    },
+    {
+      "epoch": 0.9620109529235844,
+      "grad_norm": 361.56549853055265,
+      "learning_rate": 7.255800725357586e-08,
+      "loss": 205.0992,
+      "step": 50020
+    },
+    {
+      "epoch": 0.9622032781840649,
+      "grad_norm": 366.5302465927919,
+      "learning_rate": 7.182598141587838e-08,
+      "loss": 205.1815,
+      "step": 50030
+    },
+    {
+      "epoch": 0.9623956034445454,
+      "grad_norm": 371.9301266193586,
+      "learning_rate": 7.109765365903243e-08,
+      "loss": 208.482,
+      "step": 50040
+    },
+    {
+      "epoch": 0.962587928705026,
+      "grad_norm": 357.2384001956751,
+      "learning_rate": 7.03730242543299e-08,
+      "loss": 205.4693,
+      "step": 50050
+    },
+    {
+      "epoch": 0.9627802539655065,
+      "grad_norm": 337.4818013276292,
+      "learning_rate": 6.965209347168156e-08,
+      "loss": 199.8143,
+      "step": 50060
+    },
+    {
+      "epoch": 0.962972579225987,
+      "grad_norm": 358.4629665251457,
+      "learning_rate": 6.89348615796237e-08,
+      "loss": 199.1667,
+      "step": 50070
+    },
+    {
+      "epoch": 0.9631649044864675,
+      "grad_norm": 357.08603315142034,
+      "learning_rate": 6.822132884531373e-08,
+      "loss": 201.4948,
+      "step": 50080
+    },
+    {
+      "epoch": 0.963357229746948,
+      "grad_norm": 400.5110726502833,
+      "learning_rate": 6.751149553452907e-08,
+      "loss": 208.5828,
+      "step": 50090
+    },
+    {
+      "epoch": 0.9635495550074286,
+      "grad_norm": 350.9263576414659,
+      "learning_rate": 6.680536191167263e-08,
+      "loss": 200.1319,
+      "step": 50100
+    },
+    {
+      "epoch": 0.9637418802679091,
+      "grad_norm": 359.42279037814063,
+      "learning_rate": 6.610292823976628e-08,
+      "loss": 199.5466,
+      "step": 50110
+    },
+    {
+      "epoch": 0.9639342055283896,
+      "grad_norm": 360.1887352280769,
+      "learning_rate": 6.540419478045623e-08,
+      "loss": 207.0207,
+      "step": 50120
+    },
+    {
+      "epoch": 0.9641265307888701,
+      "grad_norm": 378.47852751386193,
+      "learning_rate": 6.470916179400765e-08,
+      "loss": 198.0305,
+      "step": 50130
+    },
+    {
+      "epoch": 0.9643188560493506,
+      "grad_norm": 363.80575600351375,
+      "learning_rate": 6.401782953931013e-08,
+      "loss": 200.3281,
+      "step": 50140
+    },
+    {
+      "epoch": 0.9645111813098312,
+      "grad_norm": 380.6795736776294,
+      "learning_rate": 6.33301982738721e-08,
+      "loss": 201.7145,
+      "step": 50150
+    },
+    {
+      "epoch": 0.9647035065703117,
+      "grad_norm": 342.60556435465,
+      "learning_rate": 6.264626825382647e-08,
+      "loss": 209.3092,
+      "step": 50160
+    },
+    {
+      "epoch": 0.9648958318307922,
+      "grad_norm": 355.6897515699277,
+      "learning_rate": 6.196603973392501e-08,
+      "loss": 208.415,
+      "step": 50170
+    },
+    {
+      "epoch": 0.9650881570912727,
+      "grad_norm": 376.9843894696672,
+      "learning_rate": 6.12895129675406e-08,
+      "loss": 199.2474,
+      "step": 50180
+    },
+    {
+      "epoch": 0.9652804823517532,
+      "grad_norm": 378.4277548540897,
+      "learning_rate": 6.061668820667055e-08,
+      "loss": 206.1151,
+      "step": 50190
+    },
+    {
+      "epoch": 0.9654728076122339,
+      "grad_norm": 355.19572710632053,
+      "learning_rate": 5.994756570192994e-08,
+      "loss": 208.4154,
+      "step": 50200
+    },
+    {
+      "epoch": 0.9656651328727144,
+      "grad_norm": 361.552674747987,
+      "learning_rate": 5.928214570255497e-08,
+      "loss": 195.7584,
+      "step": 50210
+    },
+    {
+      "epoch": 0.9658574581331949,
+      "grad_norm": 386.79872658684997,
+      "learning_rate": 5.862042845640403e-08,
+      "loss": 201.5981,
+      "step": 50220
+    },
+    {
+      "epoch": 0.9660497833936754,
+      "grad_norm": 363.0551393693288,
+      "learning_rate": 5.796241420995663e-08,
+      "loss": 195.1777,
+      "step": 50230
+    },
+    {
+      "epoch": 0.9662421086541559,
+      "grad_norm": 349.4889589444486,
+      "learning_rate": 5.730810320831226e-08,
+      "loss": 204.0646,
+      "step": 50240
+    },
+    {
+      "epoch": 0.9664344339146365,
+      "grad_norm": 355.09980958283677,
+      "learning_rate": 5.66574956951893e-08,
+      "loss": 205.1864,
+      "step": 50250
+    },
+    {
+      "epoch": 0.966626759175117,
+      "grad_norm": 435.38762972989383,
+      "learning_rate": 5.6010591912930565e-08,
+      "loss": 208.1305,
+      "step": 50260
+    },
+    {
+      "epoch": 0.9668190844355975,
+      "grad_norm": 354.15731877745867,
+      "learning_rate": 5.5367392102495534e-08,
+      "loss": 204.5344,
+      "step": 50270
+    },
+    {
+      "epoch": 0.967011409696078,
+      "grad_norm": 377.60355751663684,
+      "learning_rate": 5.472789650346588e-08,
+      "loss": 205.7249,
+      "step": 50280
+    },
+    {
+      "epoch": 0.9672037349565585,
+      "grad_norm": 365.5187625608226,
+      "learning_rate": 5.4092105354043304e-08,
+      "loss": 205.6885,
+      "step": 50290
+    },
+    {
+      "epoch": 0.967396060217039,
+      "grad_norm": 363.4031489148836,
+      "learning_rate": 5.346001889104946e-08,
+      "loss": 197.1267,
+      "step": 50300
+    },
+    {
+      "epoch": 0.9675883854775196,
+      "grad_norm": 386.9760149331261,
+      "learning_rate": 5.2831637349926026e-08,
+      "loss": 201.4931,
+      "step": 50310
+    },
+    {
+      "epoch": 0.9677807107380001,
+      "grad_norm": 383.4784305510812,
+      "learning_rate": 5.220696096473465e-08,
+      "loss": 200.2673,
+      "step": 50320
+    },
+    {
+      "epoch": 0.9679730359984806,
+      "grad_norm": 450.6443804678607,
+      "learning_rate": 5.1585989968157e-08,
+      "loss": 215.3417,
+      "step": 50330
+    },
+    {
+      "epoch": 0.9681653612589611,
+      "grad_norm": 333.14887862669724,
+      "learning_rate": 5.096872459149471e-08,
+      "loss": 201.8255,
+      "step": 50340
+    },
+    {
+      "epoch": 0.9683576865194417,
+      "grad_norm": 375.73528120989096,
+      "learning_rate": 5.035516506466942e-08,
+      "loss": 212.2653,
+      "step": 50350
+    },
+    {
+      "epoch": 0.9685500117799222,
+      "grad_norm": 446.9680920916504,
+      "learning_rate": 4.9745311616220535e-08,
+      "loss": 213.5823,
+      "step": 50360
+    },
+    {
+      "epoch": 0.9687423370404027,
+      "grad_norm": 348.24251205490697,
+      "learning_rate": 4.913916447330858e-08,
+      "loss": 203.9064,
+      "step": 50370
+    },
+    {
+      "epoch": 0.9689346623008832,
+      "grad_norm": 361.1064584240234,
+      "learning_rate": 4.853672386171515e-08,
+      "loss": 209.2491,
+      "step": 50380
+    },
+    {
+      "epoch": 0.9691269875613637,
+      "grad_norm": 357.17661301751633,
+      "learning_rate": 4.793799000583743e-08,
+      "loss": 202.6957,
+      "step": 50390
+    },
+    {
+      "epoch": 0.9693193128218442,
+      "grad_norm": 389.8714792917872,
+      "learning_rate": 4.73429631286948e-08,
+      "loss": 202.3398,
+      "step": 50400
+    },
+    {
+      "epoch": 0.9695116380823249,
+      "grad_norm": 375.9401820395255,
+      "learning_rate": 4.675164345192551e-08,
+      "loss": 196.6517,
+      "step": 50410
+    },
+    {
+      "epoch": 0.9697039633428054,
+      "grad_norm": 364.1204881832129,
+      "learning_rate": 4.6164031195785606e-08,
+      "loss": 204.5044,
+      "step": 50420
+    },
+    {
+      "epoch": 0.9698962886032859,
+      "grad_norm": 340.01497015692127,
+      "learning_rate": 4.558012657915112e-08,
+      "loss": 203.8972,
+      "step": 50430
+    },
+    {
+      "epoch": 0.9700886138637664,
+      "grad_norm": 376.7845513146719,
+      "learning_rate": 4.4999929819515844e-08,
+      "loss": 209.5113,
+      "step": 50440
+    },
+    {
+      "epoch": 0.970280939124247,
+      "grad_norm": 392.9127299781355,
+      "learning_rate": 4.442344113299579e-08,
+      "loss": 204.1648,
+      "step": 50450
+    },
+    {
+      "epoch": 0.9704732643847275,
+      "grad_norm": 363.11778607954176,
+      "learning_rate": 4.385066073432143e-08,
+      "loss": 198.4782,
+      "step": 50460
+    },
+    {
+      "epoch": 0.970665589645208,
+      "grad_norm": 392.83199707541826,
+      "learning_rate": 4.3281588836844303e-08,
+      "loss": 202.6459,
+      "step": 50470
+    },
+    {
+      "epoch": 0.9708579149056885,
+      "grad_norm": 363.29121051422067,
+      "learning_rate": 4.271622565253486e-08,
+      "loss": 198.3532,
+      "step": 50480
+    },
+    {
+      "epoch": 0.971050240166169,
+      "grad_norm": 383.5253643785734,
+      "learning_rate": 4.2154571391982425e-08,
+      "loss": 201.4317,
+      "step": 50490
+    },
+    {
+      "epoch": 0.9712425654266496,
+      "grad_norm": 371.63736854051393,
+      "learning_rate": 4.159662626439187e-08,
+      "loss": 205.1226,
+      "step": 50500
+    },
+    {
+      "epoch": 0.9714348906871301,
+      "grad_norm": 413.13774609311355,
+      "learning_rate": 4.104239047758918e-08,
+      "loss": 207.9182,
+      "step": 50510
+    },
+    {
+      "epoch": 0.9716272159476106,
+      "grad_norm": 356.77074943930876,
+      "learning_rate": 4.049186423801921e-08,
+      "loss": 196.8607,
+      "step": 50520
+    },
+    {
+      "epoch": 0.9718195412080911,
+      "grad_norm": 398.50404199376686,
+      "learning_rate": 3.994504775074237e-08,
+      "loss": 198.2637,
+      "step": 50530
+    },
+    {
+      "epoch": 0.9720118664685716,
+      "grad_norm": 373.8703519544385,
+      "learning_rate": 3.9401941219440186e-08,
+      "loss": 206.9144,
+      "step": 50540
+    },
+    {
+      "epoch": 0.9722041917290521,
+      "grad_norm": 355.8067911671731,
+      "learning_rate": 3.8862544846409725e-08,
+      "loss": 200.8955,
+      "step": 50550
+    },
+    {
+      "epoch": 0.9723965169895327,
+      "grad_norm": 375.83505663357437,
+      "learning_rate": 3.832685883256915e-08,
+      "loss": 205.2627,
+      "step": 50560
+    },
+    {
+      "epoch": 0.9725888422500132,
+      "grad_norm": 349.45645297098065,
+      "learning_rate": 3.7794883377449966e-08,
+      "loss": 200.1641,
+      "step": 50570
+    },
+    {
+      "epoch": 0.9727811675104937,
+      "grad_norm": 348.5611055810571,
+      "learning_rate": 3.726661867920478e-08,
+      "loss": 203.0395,
+      "step": 50580
+    },
+    {
+      "epoch": 0.9729734927709742,
+      "grad_norm": 358.2669006617166,
+      "learning_rate": 3.674206493460508e-08,
+      "loss": 200.8857,
+      "step": 50590
+    },
+    {
+      "epoch": 0.9731658180314547,
+      "grad_norm": 351.54670609819493,
+      "learning_rate": 3.622122233903791e-08,
+      "loss": 199.7689,
+      "step": 50600
+    },
+    {
+      "epoch": 0.9733581432919354,
+      "grad_norm": 407.56600378938055,
+      "learning_rate": 3.5704091086508076e-08,
+      "loss": 207.6954,
+      "step": 50610
+    },
+    {
+      "epoch": 0.9735504685524159,
+      "grad_norm": 387.4352155478232,
+      "learning_rate": 3.519067136963705e-08,
+      "loss": 201.6304,
+      "step": 50620
+    },
+    {
+      "epoch": 0.9737427938128964,
+      "grad_norm": 390.14497586458174,
+      "learning_rate": 3.468096337966853e-08,
+      "loss": 203.7833,
+      "step": 50630
+    },
+    {
+      "epoch": 0.9739351190733769,
+      "grad_norm": 396.4177518961923,
+      "learning_rate": 3.417496730645731e-08,
+      "loss": 203.5137,
+      "step": 50640
+    },
+    {
+      "epoch": 0.9741274443338575,
+      "grad_norm": 359.62905979548225,
+      "learning_rate": 3.3672683338480396e-08,
+      "loss": 206.6287,
+      "step": 50650
+    },
+    {
+      "epoch": 0.974319769594338,
+      "grad_norm": 374.5785033485421,
+      "learning_rate": 3.317411166282813e-08,
+      "loss": 201.4154,
+      "step": 50660
+    },
+    {
+      "epoch": 0.9745120948548185,
+      "grad_norm": 344.7383211872308,
+      "learning_rate": 3.2679252465213085e-08,
+      "loss": 203.9449,
+      "step": 50670
+    },
+    {
+      "epoch": 0.974704420115299,
+      "grad_norm": 379.8235967438272,
+      "learning_rate": 3.218810592996113e-08,
+      "loss": 205.5725,
+      "step": 50680
+    },
+    {
+      "epoch": 0.9748967453757795,
+      "grad_norm": 437.5386935924966,
+      "learning_rate": 3.170067224001483e-08,
+      "loss": 214.244,
+      "step": 50690
+    },
+    {
+      "epoch": 0.97508907063626,
+      "grad_norm": 340.66102436358256,
+      "learning_rate": 3.1216951576936714e-08,
+      "loss": 194.8719,
+      "step": 50700
+    },
+    {
+      "epoch": 0.9752813958967406,
+      "grad_norm": 354.3046604099427,
+      "learning_rate": 3.073694412090489e-08,
+      "loss": 200.0747,
+      "step": 50710
+    },
+    {
+      "epoch": 0.9754737211572211,
+      "grad_norm": 404.6759521674685,
+      "learning_rate": 3.026065005071188e-08,
+      "loss": 205.5647,
+      "step": 50720
+    },
+    {
+      "epoch": 0.9756660464177016,
+      "grad_norm": 368.34396678368597,
+      "learning_rate": 2.9788069543772445e-08,
+      "loss": 208.5045,
+      "step": 50730
+    },
+    {
+      "epoch": 0.9758583716781821,
+      "grad_norm": 343.731348335998,
+      "learning_rate": 2.9319202776113553e-08,
+      "loss": 199.9978,
+      "step": 50740
+    },
+    {
+      "epoch": 0.9760506969386626,
+      "grad_norm": 358.37688261755903,
+      "learning_rate": 2.8854049922379946e-08,
+      "loss": 200.5989,
+      "step": 50750
+    },
+    {
+      "epoch": 0.9762430221991432,
+      "grad_norm": 398.3606329947295,
+      "learning_rate": 2.839261115583303e-08,
+      "loss": 206.8188,
+      "step": 50760
+    },
+    {
+      "epoch": 0.9764353474596237,
+      "grad_norm": 390.08735321288367,
+      "learning_rate": 2.793488664835309e-08,
+      "loss": 203.6234,
+      "step": 50770
+    },
+    {
+      "epoch": 0.9766276727201042,
+      "grad_norm": 373.6140125054845,
+      "learning_rate": 2.7480876570433746e-08,
+      "loss": 198.9102,
+      "step": 50780
+    },
+    {
+      "epoch": 0.9768199979805847,
+      "grad_norm": 369.86485291057227,
+      "learning_rate": 2.7030581091186393e-08,
+      "loss": 208.798,
+      "step": 50790
+    },
+    {
+      "epoch": 0.9770123232410652,
+      "grad_norm": 375.2328900326083,
+      "learning_rate": 2.658400037833686e-08,
+      "loss": 201.3384,
+      "step": 50800
+    },
+    {
+      "epoch": 0.9772046485015459,
+      "grad_norm": 351.08290942092873,
+      "learning_rate": 2.6141134598233197e-08,
+      "loss": 201.1264,
+      "step": 50810
+    },
+    {
+      "epoch": 0.9773969737620264,
+      "grad_norm": 407.0875589024692,
+      "learning_rate": 2.5701983915831232e-08,
+      "loss": 206.2367,
+      "step": 50820
+    },
+    {
+      "epoch": 0.9775892990225069,
+      "grad_norm": 335.8769076080468,
+      "learning_rate": 2.5266548494710108e-08,
+      "loss": 203.3587,
+      "step": 50830
+    },
+    {
+      "epoch": 0.9777816242829874,
+      "grad_norm": 347.6112306832183,
+      "learning_rate": 2.4834828497062315e-08,
+      "loss": 198.6425,
+      "step": 50840
+    },
+    {
+      "epoch": 0.977973949543468,
+      "grad_norm": 355.4252037339203,
+      "learning_rate": 2.4406824083694768e-08,
+      "loss": 206.2745,
+      "step": 50850
+    },
+    {
+      "epoch": 0.9781662748039485,
+      "grad_norm": 358.60436061549416,
+      "learning_rate": 2.398253541403217e-08,
+      "loss": 210.9274,
+      "step": 50860
+    },
+    {
+      "epoch": 0.978358600064429,
+      "grad_norm": 357.160490329138,
+      "learning_rate": 2.3561962646116988e-08,
+      "loss": 202.8984,
+      "step": 50870
+    },
+    {
+      "epoch": 0.9785509253249095,
+      "grad_norm": 365.29160886992554,
+      "learning_rate": 2.3145105936603906e-08,
+      "loss": 195.4823,
+      "step": 50880
+    },
+    {
+      "epoch": 0.97874325058539,
+      "grad_norm": 439.6208461726787,
+      "learning_rate": 2.273196544076539e-08,
+      "loss": 209.3478,
+      "step": 50890
+    },
+    {
+      "epoch": 0.9789355758458705,
+      "grad_norm": 342.78350491530574,
+      "learning_rate": 2.2322541312490565e-08,
+      "loss": 203.3775,
+      "step": 50900
+    },
+    {
+      "epoch": 0.9791279011063511,
+      "grad_norm": 371.94100727044844,
+      "learning_rate": 2.1916833704281882e-08,
+      "loss": 213.3107,
+      "step": 50910
+    },
+    {
+      "epoch": 0.9793202263668316,
+      "grad_norm": 374.65242784405365,
+      "learning_rate": 2.1514842767258458e-08,
+      "loss": 196.9479,
+      "step": 50920
+    },
+    {
+      "epoch": 0.9795125516273121,
+      "grad_norm": 362.6499251722214,
+      "learning_rate": 2.1116568651156076e-08,
+      "loss": 207.6976,
+      "step": 50930
+    },
+    {
+      "epoch": 0.9797048768877926,
+      "grad_norm": 359.99189821552505,
+      "learning_rate": 2.0722011504326066e-08,
+      "loss": 202.3579,
+      "step": 50940
+    },
+    {
+      "epoch": 0.9798972021482731,
+      "grad_norm": 358.3485630916027,
+      "learning_rate": 2.033117147373309e-08,
+      "loss": 199.6103,
+      "step": 50950
+    },
+    {
+      "epoch": 0.9800895274087537,
+      "grad_norm": 358.5260330266349,
+      "learning_rate": 1.9944048704959583e-08,
+      "loss": 204.0412,
+      "step": 50960
+    },
+    {
+      "epoch": 0.9802818526692342,
+      "grad_norm": 331.59994950907867,
+      "learning_rate": 1.956064334220131e-08,
+      "loss": 199.497,
+      "step": 50970
+    },
+    {
+      "epoch": 0.9804741779297147,
+      "grad_norm": 351.14588732668517,
+      "learning_rate": 1.9180955528270706e-08,
+      "loss": 207.2379,
+      "step": 50980
+    },
+    {
+      "epoch": 0.9806665031901952,
+      "grad_norm": 397.7205106483465,
+      "learning_rate": 1.8804985404595743e-08,
+      "loss": 206.9686,
+      "step": 50990
+    },
+    {
+      "epoch": 0.9808588284506757,
+      "grad_norm": 342.2036064924589,
+      "learning_rate": 1.843273311121885e-08,
+      "loss": 198.5449,
+      "step": 51000
+    },
+    {
+      "epoch": 0.9810511537111563,
+      "grad_norm": 973.6236059783471,
+      "learning_rate": 1.806419878679799e-08,
+      "loss": 207.533,
+      "step": 51010
+    },
+    {
+      "epoch": 0.9812434789716369,
+      "grad_norm": 353.3697998451452,
+      "learning_rate": 1.7699382568605595e-08,
+      "loss": 199.2419,
+      "step": 51020
+    },
+    {
+      "epoch": 0.9814358042321174,
+      "grad_norm": 358.8817119846803,
+      "learning_rate": 1.733828459253073e-08,
+      "loss": 205.8519,
+      "step": 51030
+    },
+    {
+      "epoch": 0.9816281294925979,
+      "grad_norm": 361.6388583905268,
+      "learning_rate": 1.6980904993075808e-08,
+      "loss": 201.8072,
+      "step": 51040
+    },
+    {
+      "epoch": 0.9818204547530784,
+      "grad_norm": 370.99855580721135,
+      "learning_rate": 1.6627243903357682e-08,
+      "loss": 216.963,
+      "step": 51050
+    },
+    {
+      "epoch": 0.982012780013559,
+      "grad_norm": 332.18711761413573,
+      "learning_rate": 1.6277301455110972e-08,
+      "loss": 201.4797,
+      "step": 51060
+    },
+    {
+      "epoch": 0.9822051052740395,
+      "grad_norm": 360.9403903695172,
+      "learning_rate": 1.5931077778682524e-08,
+      "loss": 205.7921,
+      "step": 51070
+    },
+    {
+      "epoch": 0.98239743053452,
+      "grad_norm": 405.309678195462,
+      "learning_rate": 1.5588573003035847e-08,
+      "loss": 205.7948,
+      "step": 51080
+    },
+    {
+      "epoch": 0.9825897557950005,
+      "grad_norm": 425.3253946702827,
+      "learning_rate": 1.5249787255747774e-08,
+      "loss": 205.9136,
+      "step": 51090
+    },
+    {
+      "epoch": 0.982782081055481,
+      "grad_norm": 342.8363078901211,
+      "learning_rate": 1.4914720663009585e-08,
+      "loss": 201.0298,
+      "step": 51100
+    },
+    {
+      "epoch": 0.9829744063159616,
+      "grad_norm": 364.46943806717525,
+      "learning_rate": 1.4583373349629226e-08,
+      "loss": 201.3648,
+      "step": 51110
+    },
+    {
+      "epoch": 0.9831667315764421,
+      "grad_norm": 356.8904847406518,
+      "learning_rate": 1.4255745439027968e-08,
+      "loss": 200.1133,
+      "step": 51120
+    },
+    {
+      "epoch": 0.9833590568369226,
+      "grad_norm": 382.6854737957742,
+      "learning_rate": 1.3931837053241526e-08,
+      "loss": 206.7351,
+      "step": 51130
+    },
+    {
+      "epoch": 0.9835513820974031,
+      "grad_norm": 365.6773762375214,
+      "learning_rate": 1.3611648312920057e-08,
+      "loss": 207.3315,
+      "step": 51140
+    },
+    {
+      "epoch": 0.9837437073578836,
+      "grad_norm": 375.61365436912155,
+      "learning_rate": 1.3295179337329267e-08,
+      "loss": 203.3566,
+      "step": 51150
+    },
+    {
+      "epoch": 0.9839360326183642,
+      "grad_norm": 361.52019708468623,
+      "learning_rate": 1.2982430244347088e-08,
+      "loss": 210.0051,
+      "step": 51160
+    },
+    {
+      "epoch": 0.9841283578788447,
+      "grad_norm": 400.3697977142987,
+      "learning_rate": 1.2673401150470333e-08,
+      "loss": 209.121,
+      "step": 51170
+    },
+    {
+      "epoch": 0.9843206831393252,
+      "grad_norm": 376.62403287433347,
+      "learning_rate": 1.2368092170804702e-08,
+      "loss": 204.1128,
+      "step": 51180
+    },
+    {
+      "epoch": 0.9845130083998057,
+      "grad_norm": 415.609338948867,
+      "learning_rate": 1.2066503419073672e-08,
+      "loss": 202.9911,
+      "step": 51190
+    },
+    {
+      "epoch": 0.9847053336602862,
+      "grad_norm": 351.03770142823123,
+      "learning_rate": 1.1768635007614049e-08,
+      "loss": 203.4782,
+      "step": 51200
+    },
+    {
+      "epoch": 0.9848976589207667,
+      "grad_norm": 348.4563186322979,
+      "learning_rate": 1.1474487047375971e-08,
+      "loss": 202.7223,
+      "step": 51210
+    },
+    {
+      "epoch": 0.9850899841812474,
+      "grad_norm": 468.1294713250719,
+      "learning_rate": 1.1184059647926238e-08,
+      "loss": 207.7692,
+      "step": 51220
+    },
+    {
+      "epoch": 0.9852823094417279,
+      "grad_norm": 401.7357054109847,
+      "learning_rate": 1.0897352917443871e-08,
+      "loss": 208.4456,
+      "step": 51230
+    },
+    {
+      "epoch": 0.9854746347022084,
+      "grad_norm": 379.80491056057997,
+      "learning_rate": 1.0614366962721223e-08,
+      "loss": 206.1894,
+      "step": 51240
+    },
+    {
+      "epoch": 0.9856669599626889,
+      "grad_norm": 332.7708771596052,
+      "learning_rate": 1.033510188916731e-08,
+      "loss": 194.6979,
+      "step": 51250
+    },
+    {
+      "epoch": 0.9858592852231695,
+      "grad_norm": 367.69120235667486,
+      "learning_rate": 1.0059557800802256e-08,
+      "loss": 203.0271,
+      "step": 51260
+    },
+    {
+      "epoch": 0.98605161048365,
+      "grad_norm": 354.47859012941774,
+      "learning_rate": 9.78773480026396e-09,
+      "loss": 197.9222,
+      "step": 51270
+    },
+    {
+      "epoch": 0.9862439357441305,
+      "grad_norm": 368.979646170427,
+      "learning_rate": 9.519632988800321e-09,
+      "loss": 207.5625,
+      "step": 51280
+    },
+    {
+      "epoch": 0.986436261004611,
+      "grad_norm": 350.6849494358281,
+      "learning_rate": 9.25525246627479e-09,
+      "loss": 206.7379,
+      "step": 51290
+    },
+    {
+      "epoch": 0.9866285862650915,
+      "grad_norm": 371.93165701237086,
+      "learning_rate": 8.994593331165257e-09,
+      "loss": 208.2924,
+      "step": 51300
+    },
+    {
+      "epoch": 0.986820911525572,
+      "grad_norm": 355.56485864819984,
+      "learning_rate": 8.737655680562951e-09,
+      "loss": 197.2067,
+      "step": 51310
+    },
+    {
+      "epoch": 0.9870132367860526,
+      "grad_norm": 387.0998506662651,
+      "learning_rate": 8.484439610172424e-09,
+      "loss": 202.9401,
+      "step": 51320
+    },
+    {
+      "epoch": 0.9872055620465331,
+      "grad_norm": 355.26681707954,
+      "learning_rate": 8.234945214312673e-09,
+      "loss": 205.5573,
+      "step": 51330
+    },
+    {
+      "epoch": 0.9873978873070136,
+      "grad_norm": 347.49241835232834,
+      "learning_rate": 7.989172585917138e-09,
+      "loss": 204.8943,
+      "step": 51340
+    },
+    {
+      "epoch": 0.9875902125674941,
+      "grad_norm": 366.33187406736545,
+      "learning_rate": 7.747121816530368e-09,
+      "loss": 202.9648,
+      "step": 51350
+    },
+    {
+      "epoch": 0.9877825378279746,
+      "grad_norm": 361.2236680092179,
+      "learning_rate": 7.508792996313573e-09,
+      "loss": 205.3263,
+      "step": 51360
+    },
+    {
+      "epoch": 0.9879748630884552,
+      "grad_norm": 346.94050202589585,
+      "learning_rate": 7.274186214040191e-09,
+      "loss": 199.8622,
+      "step": 51370
+    },
+    {
+      "epoch": 0.9881671883489357,
+      "grad_norm": 362.9462198362998,
+      "learning_rate": 7.043301557096982e-09,
+      "loss": 200.5334,
+      "step": 51380
+    },
+    {
+      "epoch": 0.9883595136094162,
+      "grad_norm": 366.0453311046353,
+      "learning_rate": 6.816139111484044e-09,
+      "loss": 202.2913,
+      "step": 51390
+    },
+    {
+      "epoch": 0.9885518388698967,
+      "grad_norm": 350.9302385734949,
+      "learning_rate": 6.592698961818134e-09,
+      "loss": 205.0694,
+      "step": 51400
+    },
+    {
+      "epoch": 0.9887441641303772,
+      "grad_norm": 334.7278843456203,
+      "learning_rate": 6.37298119132379e-09,
+      "loss": 197.3568,
+      "step": 51410
+    },
+    {
+      "epoch": 0.9889364893908578,
+      "grad_norm": 362.22652421114105,
+      "learning_rate": 6.156985881844435e-09,
+      "loss": 201.2631,
+      "step": 51420
+    },
+    {
+      "epoch": 0.9891288146513384,
+      "grad_norm": 361.4713180098413,
+      "learning_rate": 5.944713113833489e-09,
+      "loss": 202.7434,
+      "step": 51430
+    },
+    {
+      "epoch": 0.9893211399118189,
+      "grad_norm": 345.8168883809075,
+      "learning_rate": 5.736162966359926e-09,
+      "loss": 205.4553,
+      "step": 51440
+    },
+    {
+      "epoch": 0.9895134651722994,
+      "grad_norm": 357.7816003446847,
+      "learning_rate": 5.531335517104941e-09,
+      "loss": 201.2375,
+      "step": 51450
+    },
+    {
+      "epoch": 0.98970579043278,
+      "grad_norm": 376.926498796295,
+      "learning_rate": 5.3302308423641704e-09,
+      "loss": 201.9192,
+      "step": 51460
+    },
+    {
+      "epoch": 0.9898981156932605,
+      "grad_norm": 377.41667615426076,
+      "learning_rate": 5.132849017044361e-09,
+      "loss": 206.9932,
+      "step": 51470
+    },
+    {
+      "epoch": 0.990090440953741,
+      "grad_norm": 396.061243250311,
+      "learning_rate": 4.939190114666703e-09,
+      "loss": 210.7631,
+      "step": 51480
+    },
+    {
+      "epoch": 0.9902827662142215,
+      "grad_norm": 357.7638461008757,
+      "learning_rate": 4.749254207367937e-09,
+      "loss": 199.3093,
+      "step": 51490
+    },
+    {
+      "epoch": 0.990475091474702,
+      "grad_norm": 334.91060176109545,
+      "learning_rate": 4.563041365894805e-09,
+      "loss": 204.0571,
+      "step": 51500
+    },
+    {
+      "epoch": 0.9906674167351825,
+      "grad_norm": 412.1332012506797,
+      "learning_rate": 4.380551659608489e-09,
+      "loss": 205.031,
+      "step": 51510
+    },
+    {
+      "epoch": 0.9908597419956631,
+      "grad_norm": 354.85025322793365,
+      "learning_rate": 4.2017851564835065e-09,
+      "loss": 202.9292,
+      "step": 51520
+    },
+    {
+      "epoch": 0.9910520672561436,
+      "grad_norm": 362.5359086568774,
+      "learning_rate": 4.026741923107702e-09,
+      "loss": 199.1497,
+      "step": 51530
+    },
+    {
+      "epoch": 0.9912443925166241,
+      "grad_norm": 390.7978788203712,
+      "learning_rate": 3.855422024681144e-09,
+      "loss": 208.5175,
+      "step": 51540
+    },
+    {
+      "epoch": 0.9914367177771046,
+      "grad_norm": 319.8806252743432,
+      "learning_rate": 3.6878255250183406e-09,
+      "loss": 195.2692,
+      "step": 51550
+    },
+    {
+      "epoch": 0.9916290430375851,
+      "grad_norm": 329.742320677582,
+      "learning_rate": 3.5239524865460227e-09,
+      "loss": 199.9458,
+      "step": 51560
+    },
+    {
+      "epoch": 0.9918213682980657,
+      "grad_norm": 389.1102937204921,
+      "learning_rate": 3.363802970304253e-09,
+      "loss": 211.3452,
+      "step": 51570
+    },
+    {
+      "epoch": 0.9920136935585462,
+      "grad_norm": 363.6836465777458,
+      "learning_rate": 3.207377035946424e-09,
+      "loss": 205.9241,
+      "step": 51580
+    },
+    {
+      "epoch": 0.9922060188190267,
+      "grad_norm": 354.83362189362083,
+      "learning_rate": 3.05467474173704e-09,
+      "loss": 204.5754,
+      "step": 51590
+    },
+    {
+      "epoch": 0.9923983440795072,
+      "grad_norm": 380.2079650855752,
+      "learning_rate": 2.9056961445572686e-09,
+      "loss": 202.7342,
+      "step": 51600
+    },
+    {
+      "epoch": 0.9925906693399877,
+      "grad_norm": 368.47379931382955,
+      "learning_rate": 2.7604412998982754e-09,
+      "loss": 204.7153,
+      "step": 51610
+    },
+    {
+      "epoch": 0.9927829946004683,
+      "grad_norm": 389.45078787660464,
+      "learning_rate": 2.61891026186456e-09,
+      "loss": 200.7578,
+      "step": 51620
+    },
+    {
+      "epoch": 0.9929753198609489,
+      "grad_norm": 370.698779462313,
+      "learning_rate": 2.4811030831739525e-09,
+      "loss": 213.4457,
+      "step": 51630
+    },
+    {
+      "epoch": 0.9931676451214294,
+      "grad_norm": 379.0056945762546,
+      "learning_rate": 2.347019815158724e-09,
+      "loss": 206.9128,
+      "step": 51640
+    },
+    {
+      "epoch": 0.9933599703819099,
+      "grad_norm": 396.11509190807016,
+      "learning_rate": 2.216660507762258e-09,
+      "loss": 204.9282,
+      "step": 51650
+    },
+    {
+      "epoch": 0.9935522956423904,
+      "grad_norm": 368.6581037478492,
+      "learning_rate": 2.0900252095401583e-09,
+      "loss": 199.7001,
+      "step": 51660
+    },
+    {
+      "epoch": 0.993744620902871,
+      "grad_norm": 385.23305238769063,
+      "learning_rate": 1.9671139676624707e-09,
+      "loss": 202.9382,
+      "step": 51670
+    },
+    {
+      "epoch": 0.9939369461633515,
+      "grad_norm": 371.93326834522463,
+      "learning_rate": 1.8479268279125717e-09,
+      "loss": 205.3309,
+      "step": 51680
+    },
+    {
+      "epoch": 0.994129271423832,
+      "grad_norm": 357.21354124749047,
+      "learning_rate": 1.73246383468495e-09,
+      "loss": 198.5947,
+      "step": 51690
+    },
+    {
+      "epoch": 0.9943215966843125,
+      "grad_norm": 348.09372276413296,
+      "learning_rate": 1.6207250309874246e-09,
+      "loss": 203.0926,
+      "step": 51700
+    },
+    {
+      "epoch": 0.994513921944793,
+      "grad_norm": 401.7305325144985,
+      "learning_rate": 1.512710458442257e-09,
+      "loss": 206.9878,
+      "step": 51710
+    },
+    {
+      "epoch": 0.9947062472052736,
+      "grad_norm": 365.2095163742593,
+      "learning_rate": 1.408420157280599e-09,
+      "loss": 209.9957,
+      "step": 51720
+    },
+    {
+      "epoch": 0.9948985724657541,
+      "grad_norm": 355.9166980705559,
+      "learning_rate": 1.3078541663502642e-09,
+      "loss": 206.524,
+      "step": 51730
+    },
+    {
+      "epoch": 0.9950908977262346,
+      "grad_norm": 368.11151387147976,
+      "learning_rate": 1.2110125231112879e-09,
+      "loss": 208.0215,
+      "step": 51740
+    },
+    {
+      "epoch": 0.9952832229867151,
+      "grad_norm": 385.0085528021469,
+      "learning_rate": 1.117895263633706e-09,
+      "loss": 204.5201,
+      "step": 51750
+    },
+    {
+      "epoch": 0.9954755482471956,
+      "grad_norm": 390.7948995271977,
+      "learning_rate": 1.0285024226042162e-09,
+      "loss": 204.7518,
+      "step": 51760
+    },
+    {
+      "epoch": 0.9956678735076762,
+      "grad_norm": 344.1379499338581,
+      "learning_rate": 9.428340333184072e-10,
+      "loss": 199.1081,
+      "step": 51770
+    },
+    {
+      "epoch": 0.9958601987681567,
+      "grad_norm": 375.4426110016105,
+      "learning_rate": 8.608901276874193e-10,
+      "loss": 212.3385,
+      "step": 51780
+    },
+    {
+      "epoch": 0.9960525240286372,
+      "grad_norm": 397.50610033894924,
+      "learning_rate": 7.826707362335039e-10,
+      "loss": 208.1914,
+      "step": 51790
+    },
+    {
+      "epoch": 0.9962448492891177,
+      "grad_norm": 383.62146782858906,
+      "learning_rate": 7.081758880911338e-10,
+      "loss": 220.8641,
+      "step": 51800
+    },
+    {
+      "epoch": 0.9964371745495982,
+      "grad_norm": 348.1050821023941,
+      "learning_rate": 6.374056110103332e-10,
+      "loss": 197.8274,
+      "step": 51810
+    },
+    {
+      "epoch": 0.9966294998100788,
+      "grad_norm": 384.4545751015449,
+      "learning_rate": 5.703599313511277e-10,
+      "loss": 201.8037,
+      "step": 51820
+    },
+    {
+      "epoch": 0.9968218250705593,
+      "grad_norm": 360.0229515815529,
+      "learning_rate": 5.070388740868737e-10,
+      "loss": 205.0513,
+      "step": 51830
+    },
+    {
+      "epoch": 0.9970141503310399,
+      "grad_norm": 381.8984436367163,
+      "learning_rate": 4.474424628031493e-10,
+      "loss": 205.1911,
+      "step": 51840
+    },
+    {
+      "epoch": 0.9972064755915204,
+      "grad_norm": 345.9292320856205,
+      "learning_rate": 3.9157071969997407e-10,
+      "loss": 197.5689,
+      "step": 51850
+    },
+    {
+      "epoch": 0.9973988008520009,
+      "grad_norm": 391.7863656828509,
+      "learning_rate": 3.394236655873684e-10,
+      "loss": 206.6955,
+      "step": 51860
+    },
+    {
+      "epoch": 0.9975911261124815,
+      "grad_norm": 350.84394351790695,
+      "learning_rate": 2.910013198886841e-10,
+      "loss": 205.6138,
+      "step": 51870
+    },
+    {
+      "epoch": 0.997783451372962,
+      "grad_norm": 384.4586453573272,
+      "learning_rate": 2.4630370064171463e-10,
+      "loss": 205.7148,
+      "step": 51880
+    },
+    {
+      "epoch": 0.9979757766334425,
+      "grad_norm": 352.8663929488422,
+      "learning_rate": 2.0533082449647467e-10,
+      "loss": 202.1314,
+      "step": 51890
+    },
+    {
+      "epoch": 0.998168101893923,
+      "grad_norm": 359.50165438039494,
+      "learning_rate": 1.6808270671186954e-10,
+      "loss": 209.9645,
+      "step": 51900
+    },
+    {
+      "epoch": 0.9983604271544035,
+      "grad_norm": 415.44156285872293,
+      "learning_rate": 1.345593611645768e-10,
+      "loss": 200.2923,
+      "step": 51910
+    },
+    {
+      "epoch": 0.9985527524148841,
+      "grad_norm": 377.8012119963822,
+      "learning_rate": 1.0476080034016456e-10,
+      "loss": 212.1618,
+      "step": 51920
+    },
+    {
+      "epoch": 0.9987450776753646,
+      "grad_norm": 376.1355865143914,
+      "learning_rate": 7.868703533864264e-11,
+      "loss": 204.8333,
+      "step": 51930
+    },
+    {
+      "epoch": 0.9989374029358451,
+      "grad_norm": 396.76812887551205,
+      "learning_rate": 5.633807587224205e-11,
+      "loss": 205.867,
+      "step": 51940
+    },
+    {
+      "epoch": 0.9991297281963256,
+      "grad_norm": 367.44817221330374,
+      "learning_rate": 3.771393026541503e-11,
+      "loss": 201.6554,
+      "step": 51950
+    },
+    {
+      "epoch": 0.9993220534568061,
+      "grad_norm": 354.4649358181581,
+      "learning_rate": 2.281460545594527e-11,
+      "loss": 199.9556,
+      "step": 51960
+    },
+    {
+      "epoch": 0.9995143787172867,
+      "grad_norm": 360.9267721712571,
+      "learning_rate": 1.164010699272744e-11,
+      "loss": 201.5196,
+      "step": 51970
+    },
+    {
+      "epoch": 0.9997067039777672,
+      "grad_norm": 356.4807932466468,
+      "learning_rate": 4.190439037987659e-12,
+      "loss": 198.2291,
+      "step": 51980
+    },
+    {
+      "epoch": 0.9998990292382477,
+      "grad_norm": 368.023569934151,
+      "learning_rate": 4.656043661732668e-13,
+      "loss": 220.5766,
+      "step": 51990
+    },
+    {
+      "epoch": 0.999995191868488,
+      "eval_loss": 242.9601287841797,
+      "eval_runtime": 167.1518,
+      "eval_samples_per_second": 4.164,
+      "eval_steps_per_second": 0.132,
+      "step": 51995
+    },
+    {
+      "epoch": 0.999995191868488,
+      "step": 51995,
+      "total_flos": 7.252179007674778e+16,
+      "train_loss": 276.11489062246443,
+      "train_runtime": 988432.6936,
+      "train_samples_per_second": 3.367,
+      "train_steps_per_second": 0.053
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 51995,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 5000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.252179007674778e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}