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": 340256.8990635632,
+      "learning_rate": 3.846153846153847e-08,
+      "loss": 25905.9922,
+      "step": 1
+    },
+    {
+      "epoch": 0.00019232526048052465,
+      "grad_norm": 329547.13392761885,
+      "learning_rate": 3.846153846153847e-07,
+      "loss": 25785.191,
+      "step": 10
+    },
+    {
+      "epoch": 0.0003846505209610493,
+      "grad_norm": 288839.0962993139,
+      "learning_rate": 7.692307692307694e-07,
+      "loss": 23700.6562,
+      "step": 20
+    },
+    {
+      "epoch": 0.000576975781441574,
+      "grad_norm": 125626.36864796947,
+      "learning_rate": 1.153846153846154e-06,
+      "loss": 20190.3344,
+      "step": 30
+    },
+    {
+      "epoch": 0.0007693010419220986,
+      "grad_norm": 47322.671994916804,
+      "learning_rate": 1.5384615384615387e-06,
+      "loss": 16867.6031,
+      "step": 40
+    },
+    {
+      "epoch": 0.0009616263024026233,
+      "grad_norm": 22843.44389963286,
+      "learning_rate": 1.9230769230769234e-06,
+      "loss": 14599.7812,
+      "step": 50
+    },
+    {
+      "epoch": 0.001153951562883148,
+      "grad_norm": 23203.932058026054,
+      "learning_rate": 2.307692307692308e-06,
+      "loss": 12818.4688,
+      "step": 60
+    },
+    {
+      "epoch": 0.0013462768233636726,
+      "grad_norm": 12005.578799387531,
+      "learning_rate": 2.6923076923076923e-06,
+      "loss": 11844.6297,
+      "step": 70
+    },
+    {
+      "epoch": 0.0015386020838441972,
+      "grad_norm": 32818.750754299195,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 11162.5469,
+      "step": 80
+    },
+    {
+      "epoch": 0.0017309273443247219,
+      "grad_norm": 17435.49809016785,
+      "learning_rate": 3.4615384615384617e-06,
+      "loss": 10793.7641,
+      "step": 90
+    },
+    {
+      "epoch": 0.0019232526048052465,
+      "grad_norm": 12701.85980578269,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 10653.1812,
+      "step": 100
+    },
+    {
+      "epoch": 0.0021155778652857714,
+      "grad_norm": 23037.58616236622,
+      "learning_rate": 4.230769230769231e-06,
+      "loss": 10358.1984,
+      "step": 110
+    },
+    {
+      "epoch": 0.002307903125766296,
+      "grad_norm": 13372.193504961218,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 10103.9813,
+      "step": 120
+    },
+    {
+      "epoch": 0.0025002283862468207,
+      "grad_norm": 16968.767589773826,
+      "learning_rate": 5e-06,
+      "loss": 9850.3547,
+      "step": 130
+    },
+    {
+      "epoch": 0.002692553646727345,
+      "grad_norm": 17165.642730963893,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 9720.2469,
+      "step": 140
+    },
+    {
+      "epoch": 0.00288487890720787,
+      "grad_norm": 13265.190772082746,
+      "learning_rate": 5.769230769230769e-06,
+      "loss": 9546.9609,
+      "step": 150
+    },
+    {
+      "epoch": 0.0030772041676883945,
+      "grad_norm": 21469.975105247308,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 9415.2406,
+      "step": 160
+    },
+    {
+      "epoch": 0.0032695294281689193,
+      "grad_norm": 15056.4294636194,
+      "learning_rate": 6.538461538461539e-06,
+      "loss": 9271.8852,
+      "step": 170
+    },
+    {
+      "epoch": 0.0034618546886494438,
+      "grad_norm": 24981.413385330285,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 9084.2461,
+      "step": 180
+    },
+    {
+      "epoch": 0.0036541799491299687,
+      "grad_norm": 27927.895744899266,
+      "learning_rate": 7.307692307692308e-06,
+      "loss": 8983.4891,
+      "step": 190
+    },
+    {
+      "epoch": 0.003846505209610493,
+      "grad_norm": 10409.821548080958,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 8836.9023,
+      "step": 200
+    },
+    {
+      "epoch": 0.0040388304700910175,
+      "grad_norm": 11841.413899636149,
+      "learning_rate": 8.076923076923077e-06,
+      "loss": 8635.6063,
+      "step": 210
+    },
+    {
+      "epoch": 0.004231155730571543,
+      "grad_norm": 16907.549285896923,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 8474.7594,
+      "step": 220
+    },
+    {
+      "epoch": 0.004423480991052067,
+      "grad_norm": 27817.25135796424,
+      "learning_rate": 8.846153846153847e-06,
+      "loss": 8325.1938,
+      "step": 230
+    },
+    {
+      "epoch": 0.004615806251532592,
+      "grad_norm": 11686.494480442112,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 8167.4578,
+      "step": 240
+    },
+    {
+      "epoch": 0.004808131512013117,
+      "grad_norm": 14611.295962037319,
+      "learning_rate": 9.615384615384616e-06,
+      "loss": 8038.5328,
+      "step": 250
+    },
+    {
+      "epoch": 0.0050004567724936414,
+      "grad_norm": 17857.53203550568,
+      "learning_rate": 1e-05,
+      "loss": 7909.875,
+      "step": 260
+    },
+    {
+      "epoch": 0.005192782032974166,
+      "grad_norm": 15809.087648960021,
+      "learning_rate": 1.0384615384615386e-05,
+      "loss": 7727.9883,
+      "step": 270
+    },
+    {
+      "epoch": 0.00538510729345469,
+      "grad_norm": 22914.004068667255,
+      "learning_rate": 1.076923076923077e-05,
+      "loss": 7617.4031,
+      "step": 280
+    },
+    {
+      "epoch": 0.005577432553935216,
+      "grad_norm": 19977.21406332264,
+      "learning_rate": 1.1153846153846154e-05,
+      "loss": 7466.7016,
+      "step": 290
+    },
+    {
+      "epoch": 0.00576975781441574,
+      "grad_norm": 15106.688610570774,
+      "learning_rate": 1.1538461538461538e-05,
+      "loss": 7329.4734,
+      "step": 300
+    },
+    {
+      "epoch": 0.0059620830748962645,
+      "grad_norm": 23038.7334332166,
+      "learning_rate": 1.1923076923076925e-05,
+      "loss": 7184.3031,
+      "step": 310
+    },
+    {
+      "epoch": 0.006154408335376789,
+      "grad_norm": 24002.17033078466,
+      "learning_rate": 1.230769230769231e-05,
+      "loss": 7171.5922,
+      "step": 320
+    },
+    {
+      "epoch": 0.006346733595857314,
+      "grad_norm": 18958.9256998183,
+      "learning_rate": 1.2692307692307693e-05,
+      "loss": 7098.3914,
+      "step": 330
+    },
+    {
+      "epoch": 0.006539058856337839,
+      "grad_norm": 21815.718108192887,
+      "learning_rate": 1.3076923076923078e-05,
+      "loss": 6975.9375,
+      "step": 340
+    },
+    {
+      "epoch": 0.006731384116818363,
+      "grad_norm": 13891.644195177581,
+      "learning_rate": 1.3461538461538463e-05,
+      "loss": 6760.9484,
+      "step": 350
+    },
+    {
+      "epoch": 0.0069237093772988876,
+      "grad_norm": 15659.636466050824,
+      "learning_rate": 1.3846153846153847e-05,
+      "loss": 6681.1844,
+      "step": 360
+    },
+    {
+      "epoch": 0.007116034637779413,
+      "grad_norm": 15354.468375663539,
+      "learning_rate": 1.4230769230769232e-05,
+      "loss": 6656.9016,
+      "step": 370
+    },
+    {
+      "epoch": 0.007308359898259937,
+      "grad_norm": 16678.45446714103,
+      "learning_rate": 1.4615384615384615e-05,
+      "loss": 6528.625,
+      "step": 380
+    },
+    {
+      "epoch": 0.007500685158740462,
+      "grad_norm": 14644.542359131423,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 6349.0469,
+      "step": 390
+    },
+    {
+      "epoch": 0.007693010419220986,
+      "grad_norm": 12701.420488118676,
+      "learning_rate": 1.5384615384615387e-05,
+      "loss": 6235.1816,
+      "step": 400
+    },
+    {
+      "epoch": 0.00788533567970151,
+      "grad_norm": 20701.149469365322,
+      "learning_rate": 1.576923076923077e-05,
+      "loss": 6173.8125,
+      "step": 410
+    },
+    {
+      "epoch": 0.008077660940182035,
+      "grad_norm": 20562.39806175966,
+      "learning_rate": 1.6153846153846154e-05,
+      "loss": 6166.375,
+      "step": 420
+    },
+    {
+      "epoch": 0.008269986200662561,
+      "grad_norm": 14584.105530767776,
+      "learning_rate": 1.653846153846154e-05,
+      "loss": 5970.6586,
+      "step": 430
+    },
+    {
+      "epoch": 0.008462311461143086,
+      "grad_norm": 16991.435993864558,
+      "learning_rate": 1.6923076923076924e-05,
+      "loss": 5935.7461,
+      "step": 440
+    },
+    {
+      "epoch": 0.00865463672162361,
+      "grad_norm": 14263.18538760599,
+      "learning_rate": 1.730769230769231e-05,
+      "loss": 5953.3781,
+      "step": 450
+    },
+    {
+      "epoch": 0.008846961982104135,
+      "grad_norm": 19145.49418149806,
+      "learning_rate": 1.7692307692307694e-05,
+      "loss": 5724.8188,
+      "step": 460
+    },
+    {
+      "epoch": 0.009039287242584659,
+      "grad_norm": 15563.56027844558,
+      "learning_rate": 1.807692307692308e-05,
+      "loss": 5690.4148,
+      "step": 470
+    },
+    {
+      "epoch": 0.009231612503065183,
+      "grad_norm": 13300.268934789272,
+      "learning_rate": 1.8461538461538465e-05,
+      "loss": 5494.1055,
+      "step": 480
+    },
+    {
+      "epoch": 0.009423937763545708,
+      "grad_norm": 12857.054925207442,
+      "learning_rate": 1.8846153846153846e-05,
+      "loss": 5543.3703,
+      "step": 490
+    },
+    {
+      "epoch": 0.009616263024026234,
+      "grad_norm": 13109.490361871232,
+      "learning_rate": 1.923076923076923e-05,
+      "loss": 5420.0109,
+      "step": 500
+    },
+    {
+      "epoch": 0.009808588284506758,
+      "grad_norm": 15974.71343804961,
+      "learning_rate": 1.9615384615384617e-05,
+      "loss": 5325.168,
+      "step": 510
+    },
+    {
+      "epoch": 0.010000913544987283,
+      "grad_norm": 12954.438121276595,
+      "learning_rate": 2e-05,
+      "loss": 5288.568,
+      "step": 520
+    },
+    {
+      "epoch": 0.010193238805467807,
+      "grad_norm": 11844.914072876223,
+      "learning_rate": 1.999999813758258e-05,
+      "loss": 5107.3727,
+      "step": 530
+    },
+    {
+      "epoch": 0.010385564065948332,
+      "grad_norm": 11648.785363103754,
+      "learning_rate": 1.9999992550331007e-05,
+      "loss": 5067.948,
+      "step": 540
+    },
+    {
+      "epoch": 0.010577889326428856,
+      "grad_norm": 11091.289174630127,
+      "learning_rate": 1.999998323824736e-05,
+      "loss": 4981.1875,
+      "step": 550
+    },
+    {
+      "epoch": 0.01077021458690938,
+      "grad_norm": 11335.166344142004,
+      "learning_rate": 1.999997020133512e-05,
+      "loss": 4876.9996,
+      "step": 560
+    },
+    {
+      "epoch": 0.010962539847389905,
+      "grad_norm": 11190.506875571928,
+      "learning_rate": 1.9999953439599132e-05,
+      "loss": 4795.6996,
+      "step": 570
+    },
+    {
+      "epoch": 0.011154865107870431,
+      "grad_norm": 12522.98405604213,
+      "learning_rate": 1.9999932953045638e-05,
+      "loss": 4721.6297,
+      "step": 580
+    },
+    {
+      "epoch": 0.011347190368350956,
+      "grad_norm": 10446.139123875913,
+      "learning_rate": 1.999990874168228e-05,
+      "loss": 4620.8328,
+      "step": 590
+    },
+    {
+      "epoch": 0.01153951562883148,
+      "grad_norm": 12858.000478299662,
+      "learning_rate": 1.9999880805518067e-05,
+      "loss": 4567.8957,
+      "step": 600
+    },
+    {
+      "epoch": 0.011731840889312005,
+      "grad_norm": 12490.711568372986,
+      "learning_rate": 1.9999849144563406e-05,
+      "loss": 4478.7648,
+      "step": 610
+    },
+    {
+      "epoch": 0.011924166149792529,
+      "grad_norm": 9637.057023518732,
+      "learning_rate": 1.9999813758830092e-05,
+      "loss": 4402.8023,
+      "step": 620
+    },
+    {
+      "epoch": 0.012116491410273053,
+      "grad_norm": 9604.54589302654,
+      "learning_rate": 1.9999774648331307e-05,
+      "loss": 4332.5977,
+      "step": 630
+    },
+    {
+      "epoch": 0.012308816670753578,
+      "grad_norm": 10154.603630411342,
+      "learning_rate": 1.9999731813081616e-05,
+      "loss": 4271.5375,
+      "step": 640
+    },
+    {
+      "epoch": 0.012501141931234102,
+      "grad_norm": 10881.758102529358,
+      "learning_rate": 1.9999685253096975e-05,
+      "loss": 4143.4367,
+      "step": 650
+    },
+    {
+      "epoch": 0.012693467191714628,
+      "grad_norm": 10802.768943722904,
+      "learning_rate": 1.999963496839473e-05,
+      "loss": 4074.3352,
+      "step": 660
+    },
+    {
+      "epoch": 0.012885792452195153,
+      "grad_norm": 10995.061114305476,
+      "learning_rate": 1.9999580958993606e-05,
+      "loss": 4057.2422,
+      "step": 670
+    },
+    {
+      "epoch": 0.013078117712675677,
+      "grad_norm": 10730.355493407538,
+      "learning_rate": 1.9999523224913722e-05,
+      "loss": 4027.3887,
+      "step": 680
+    },
+    {
+      "epoch": 0.013270442973156202,
+      "grad_norm": 10705.574888253108,
+      "learning_rate": 1.999946176617659e-05,
+      "loss": 3954.6359,
+      "step": 690
+    },
+    {
+      "epoch": 0.013462768233636726,
+      "grad_norm": 8168.077792098991,
+      "learning_rate": 1.999939658280509e-05,
+      "loss": 3871.0703,
+      "step": 700
+    },
+    {
+      "epoch": 0.01365509349411725,
+      "grad_norm": 10264.234778721111,
+      "learning_rate": 1.9999327674823513e-05,
+      "loss": 3839.8055,
+      "step": 710
+    },
+    {
+      "epoch": 0.013847418754597775,
+      "grad_norm": 8626.566883187743,
+      "learning_rate": 1.9999255042257522e-05,
+      "loss": 3799.2852,
+      "step": 720
+    },
+    {
+      "epoch": 0.0140397440150783,
+      "grad_norm": 8828.928266683495,
+      "learning_rate": 1.999917868513417e-05,
+      "loss": 3700.2406,
+      "step": 730
+    },
+    {
+      "epoch": 0.014232069275558826,
+      "grad_norm": 9660.58301557004,
+      "learning_rate": 1.9999098603481895e-05,
+      "loss": 3615.8523,
+      "step": 740
+    },
+    {
+      "epoch": 0.01442439453603935,
+      "grad_norm": 7058.506989793753,
+      "learning_rate": 1.9999014797330536e-05,
+      "loss": 3648.4711,
+      "step": 750
+    },
+    {
+      "epoch": 0.014616719796519875,
+      "grad_norm": 8612.382081255484,
+      "learning_rate": 1.99989272667113e-05,
+      "loss": 3604.9625,
+      "step": 760
+    },
+    {
+      "epoch": 0.014809045057000399,
+      "grad_norm": 8236.17495506975,
+      "learning_rate": 1.99988360116568e-05,
+      "loss": 3549.2973,
+      "step": 770
+    },
+    {
+      "epoch": 0.015001370317480923,
+      "grad_norm": 7530.487558770446,
+      "learning_rate": 1.999874103220102e-05,
+      "loss": 3491.4262,
+      "step": 780
+    },
+    {
+      "epoch": 0.015193695577961448,
+      "grad_norm": 9216.484832898406,
+      "learning_rate": 1.999864232837934e-05,
+      "loss": 3497.0539,
+      "step": 790
+    },
+    {
+      "epoch": 0.015386020838441972,
+      "grad_norm": 7153.93457818025,
+      "learning_rate": 1.9998539900228526e-05,
+      "loss": 3461.8457,
+      "step": 800
+    },
+    {
+      "epoch": 0.015578346098922499,
+      "grad_norm": 7611.245340816313,
+      "learning_rate": 1.999843374778673e-05,
+      "loss": 3439.5852,
+      "step": 810
+    },
+    {
+      "epoch": 0.01577067135940302,
+      "grad_norm": 8264.323535675565,
+      "learning_rate": 1.999832387109349e-05,
+      "loss": 3387.3648,
+      "step": 820
+    },
+    {
+      "epoch": 0.015962996619883547,
+      "grad_norm": 6642.2333530382075,
+      "learning_rate": 1.9998210270189736e-05,
+      "loss": 3305.0938,
+      "step": 830
+    },
+    {
+      "epoch": 0.01615532188036407,
+      "grad_norm": 8684.38186788829,
+      "learning_rate": 1.9998092945117786e-05,
+      "loss": 3290.0352,
+      "step": 840
+    },
+    {
+      "epoch": 0.016347647140844596,
+      "grad_norm": 6756.283326498615,
+      "learning_rate": 1.999797189592134e-05,
+      "loss": 3261.4875,
+      "step": 850
+    },
+    {
+      "epoch": 0.016539972401325122,
+      "grad_norm": 26631.43185342854,
+      "learning_rate": 1.999784712264548e-05,
+      "loss": 3285.9105,
+      "step": 860
+    },
+    {
+      "epoch": 0.016732297661805645,
+      "grad_norm": 7161.421118835917,
+      "learning_rate": 1.9997718625336686e-05,
+      "loss": 3235.0674,
+      "step": 870
+    },
+    {
+      "epoch": 0.01692462292228617,
+      "grad_norm": 6496.011346595953,
+      "learning_rate": 1.9997586404042825e-05,
+      "loss": 3199.4945,
+      "step": 880
+    },
+    {
+      "epoch": 0.017116948182766694,
+      "grad_norm": 6761.59679428012,
+      "learning_rate": 1.9997450458813142e-05,
+      "loss": 3123.632,
+      "step": 890
+    },
+    {
+      "epoch": 0.01730927344324722,
+      "grad_norm": 6648.806635648011,
+      "learning_rate": 1.9997310789698276e-05,
+      "loss": 3094.5512,
+      "step": 900
+    },
+    {
+      "epoch": 0.017501598703727743,
+      "grad_norm": 7004.756665221598,
+      "learning_rate": 1.999716739675025e-05,
+      "loss": 3095.1094,
+      "step": 910
+    },
+    {
+      "epoch": 0.01769392396420827,
+      "grad_norm": 6858.459203676452,
+      "learning_rate": 1.9997020280022482e-05,
+      "loss": 3063.0963,
+      "step": 920
+    },
+    {
+      "epoch": 0.017886249224688795,
+      "grad_norm": 6661.144629432038,
+      "learning_rate": 1.9996869439569767e-05,
+      "loss": 3016.9559,
+      "step": 930
+    },
+    {
+      "epoch": 0.018078574485169318,
+      "grad_norm": 7214.218460140278,
+      "learning_rate": 1.9996714875448286e-05,
+      "loss": 2984.1994,
+      "step": 940
+    },
+    {
+      "epoch": 0.018270899745649844,
+      "grad_norm": 6001.130120344852,
+      "learning_rate": 1.9996556587715617e-05,
+      "loss": 2966.515,
+      "step": 950
+    },
+    {
+      "epoch": 0.018463225006130367,
+      "grad_norm": 6732.214693785003,
+      "learning_rate": 1.9996394576430716e-05,
+      "loss": 2952.8459,
+      "step": 960
+    },
+    {
+      "epoch": 0.018655550266610893,
+      "grad_norm": 6264.564572575251,
+      "learning_rate": 1.9996228841653932e-05,
+      "loss": 2917.0217,
+      "step": 970
+    },
+    {
+      "epoch": 0.018847875527091416,
+      "grad_norm": 6697.862055110569,
+      "learning_rate": 1.9996059383447e-05,
+      "loss": 2868.5787,
+      "step": 980
+    },
+    {
+      "epoch": 0.019040200787571942,
+      "grad_norm": 8004.141853268536,
+      "learning_rate": 1.999588620187303e-05,
+      "loss": 2838.8414,
+      "step": 990
+    },
+    {
+      "epoch": 0.019232526048052468,
+      "grad_norm": 8137.132667056839,
+      "learning_rate": 1.9995709296996545e-05,
+      "loss": 2848.5451,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01942485130853299,
+      "grad_norm": 5590.029919016462,
+      "learning_rate": 1.999552866888343e-05,
+      "loss": 2827.4316,
+      "step": 1010
+    },
+    {
+      "epoch": 0.019617176569013517,
+      "grad_norm": 7665.639769622731,
+      "learning_rate": 1.9995344317600965e-05,
+      "loss": 2824.509,
+      "step": 1020
+    },
+    {
+      "epoch": 0.01980950182949404,
+      "grad_norm": 14313.87942231357,
+      "learning_rate": 1.9995156243217824e-05,
+      "loss": 2779.2203,
+      "step": 1030
+    },
+    {
+      "epoch": 0.020001827089974566,
+      "grad_norm": 7240.42830736018,
+      "learning_rate": 1.9994964445804054e-05,
+      "loss": 2794.9771,
+      "step": 1040
+    },
+    {
+      "epoch": 0.02019415235045509,
+      "grad_norm": 6362.123548045883,
+      "learning_rate": 1.9994768925431104e-05,
+      "loss": 2734.7777,
+      "step": 1050
+    },
+    {
+      "epoch": 0.020386477610935615,
+      "grad_norm": 6605.1139500342815,
+      "learning_rate": 1.9994569682171796e-05,
+      "loss": 2697.7977,
+      "step": 1060
+    },
+    {
+      "epoch": 0.020578802871416137,
+      "grad_norm": 5802.405497477909,
+      "learning_rate": 1.9994366716100346e-05,
+      "loss": 2666.858,
+      "step": 1070
+    },
+    {
+      "epoch": 0.020771128131896664,
+      "grad_norm": 7937.578426959362,
+      "learning_rate": 1.999416002729236e-05,
+      "loss": 2643.2906,
+      "step": 1080
+    },
+    {
+      "epoch": 0.02096345339237719,
+      "grad_norm": 7943.748330913041,
+      "learning_rate": 1.999394961582482e-05,
+      "loss": 2671.0062,
+      "step": 1090
+    },
+    {
+      "epoch": 0.021155778652857712,
+      "grad_norm": 7739.035172683172,
+      "learning_rate": 1.9993735481776105e-05,
+      "loss": 2579.4945,
+      "step": 1100
+    },
+    {
+      "epoch": 0.02134810391333824,
+      "grad_norm": 6482.674900184367,
+      "learning_rate": 1.9993517625225976e-05,
+      "loss": 2550.8812,
+      "step": 1110
+    },
+    {
+      "epoch": 0.02154042917381876,
+      "grad_norm": 7039.248626270822,
+      "learning_rate": 1.9993296046255578e-05,
+      "loss": 2507.1734,
+      "step": 1120
+    },
+    {
+      "epoch": 0.021732754434299287,
+      "grad_norm": 8778.22134636785,
+      "learning_rate": 1.9993070744947447e-05,
+      "loss": 2565.0559,
+      "step": 1130
+    },
+    {
+      "epoch": 0.02192507969477981,
+      "grad_norm": 6279.14543835544,
+      "learning_rate": 1.9992841721385508e-05,
+      "loss": 2475.0553,
+      "step": 1140
+    },
+    {
+      "epoch": 0.022117404955260336,
+      "grad_norm": 11392.292370308016,
+      "learning_rate": 1.999260897565506e-05,
+      "loss": 2484.5826,
+      "step": 1150
+    },
+    {
+      "epoch": 0.022309730215740863,
+      "grad_norm": 7052.888046527636,
+      "learning_rate": 1.9992372507842807e-05,
+      "loss": 2504.975,
+      "step": 1160
+    },
+    {
+      "epoch": 0.022502055476221385,
+      "grad_norm": 8786.673756747148,
+      "learning_rate": 1.9992132318036825e-05,
+      "loss": 2463.2688,
+      "step": 1170
+    },
+    {
+      "epoch": 0.02269438073670191,
+      "grad_norm": 6774.941164287866,
+      "learning_rate": 1.9991888406326575e-05,
+      "loss": 2469.0289,
+      "step": 1180
+    },
+    {
+      "epoch": 0.022886705997182434,
+      "grad_norm": 6413.011054953301,
+      "learning_rate": 1.9991640772802916e-05,
+      "loss": 2389.6076,
+      "step": 1190
+    },
+    {
+      "epoch": 0.02307903125766296,
+      "grad_norm": 6644.7145028389905,
+      "learning_rate": 1.9991389417558088e-05,
+      "loss": 2368.9789,
+      "step": 1200
+    },
+    {
+      "epoch": 0.023271356518143483,
+      "grad_norm": 8362.347446753292,
+      "learning_rate": 1.9991134340685713e-05,
+      "loss": 2351.3848,
+      "step": 1210
+    },
+    {
+      "epoch": 0.02346368177862401,
+      "grad_norm": 6096.356291539516,
+      "learning_rate": 1.999087554228081e-05,
+      "loss": 2306.2426,
+      "step": 1220
+    },
+    {
+      "epoch": 0.023656007039104535,
+      "grad_norm": 6804.562440191596,
+      "learning_rate": 1.999061302243977e-05,
+      "loss": 2304.1328,
+      "step": 1230
+    },
+    {
+      "epoch": 0.023848332299585058,
+      "grad_norm": 6952.243204119128,
+      "learning_rate": 1.9990346781260378e-05,
+      "loss": 2294.5363,
+      "step": 1240
+    },
+    {
+      "epoch": 0.024040657560065584,
+      "grad_norm": 8363.314403806107,
+      "learning_rate": 1.9990076818841805e-05,
+      "loss": 2245.8201,
+      "step": 1250
+    },
+    {
+      "epoch": 0.024232982820546107,
+      "grad_norm": 5794.89742816201,
+      "learning_rate": 1.998980313528461e-05,
+      "loss": 2252.8531,
+      "step": 1260
+    },
+    {
+      "epoch": 0.024425308081026633,
+      "grad_norm": 7386.9934412876055,
+      "learning_rate": 1.9989525730690736e-05,
+      "loss": 2239.523,
+      "step": 1270
+    },
+    {
+      "epoch": 0.024617633341507156,
+      "grad_norm": 7705.58379475732,
+      "learning_rate": 1.998924460516351e-05,
+      "loss": 2246.0797,
+      "step": 1280
+    },
+    {
+      "epoch": 0.024809958601987682,
+      "grad_norm": 7680.672029723623,
+      "learning_rate": 1.9988959758807645e-05,
+      "loss": 2201.1941,
+      "step": 1290
+    },
+    {
+      "epoch": 0.025002283862468205,
+      "grad_norm": 6815.1772251894035,
+      "learning_rate": 1.9988671191729243e-05,
+      "loss": 2208.9932,
+      "step": 1300
+    },
+    {
+      "epoch": 0.02519460912294873,
+      "grad_norm": 7069.861677780535,
+      "learning_rate": 1.998837890403579e-05,
+      "loss": 2164.1402,
+      "step": 1310
+    },
+    {
+      "epoch": 0.025386934383429257,
+      "grad_norm": 6750.107580515478,
+      "learning_rate": 1.998808289583616e-05,
+      "loss": 2113.6455,
+      "step": 1320
+    },
+    {
+      "epoch": 0.02557925964390978,
+      "grad_norm": 6892.477862219737,
+      "learning_rate": 1.998778316724061e-05,
+      "loss": 2104.1031,
+      "step": 1330
+    },
+    {
+      "epoch": 0.025771584904390306,
+      "grad_norm": 7111.939884524118,
+      "learning_rate": 1.998747971836078e-05,
+      "loss": 2054.852,
+      "step": 1340
+    },
+    {
+      "epoch": 0.02596391016487083,
+      "grad_norm": 7266.3859092421535,
+      "learning_rate": 1.9987172549309707e-05,
+      "loss": 2080.4945,
+      "step": 1350
+    },
+    {
+      "epoch": 0.026156235425351355,
+      "grad_norm": 7013.185786113792,
+      "learning_rate": 1.9986861660201802e-05,
+      "loss": 2046.0879,
+      "step": 1360
+    },
+    {
+      "epoch": 0.026348560685831877,
+      "grad_norm": 7269.11236414404,
+      "learning_rate": 1.998654705115287e-05,
+      "loss": 2090.7902,
+      "step": 1370
+    },
+    {
+      "epoch": 0.026540885946312404,
+      "grad_norm": 7047.400297873044,
+      "learning_rate": 1.9986228722280093e-05,
+      "loss": 2039.8742,
+      "step": 1380
+    },
+    {
+      "epoch": 0.02673321120679293,
+      "grad_norm": 6332.782863580958,
+      "learning_rate": 1.998590667370204e-05,
+      "loss": 2013.5613,
+      "step": 1390
+    },
+    {
+      "epoch": 0.026925536467273452,
+      "grad_norm": 7449.685427230845,
+      "learning_rate": 1.998558090553868e-05,
+      "loss": 2008.0461,
+      "step": 1400
+    },
+    {
+      "epoch": 0.02711786172775398,
+      "grad_norm": 7413.579686598059,
+      "learning_rate": 1.9985251417911347e-05,
+      "loss": 1975.0203,
+      "step": 1410
+    },
+    {
+      "epoch": 0.0273101869882345,
+      "grad_norm": 7033.093276733983,
+      "learning_rate": 1.9984918210942776e-05,
+      "loss": 1940.2811,
+      "step": 1420
+    },
+    {
+      "epoch": 0.027502512248715028,
+      "grad_norm": 5859.880891146991,
+      "learning_rate": 1.9984581284757074e-05,
+      "loss": 1938.102,
+      "step": 1430
+    },
+    {
+      "epoch": 0.02769483750919555,
+      "grad_norm": 6208.344325728506,
+      "learning_rate": 1.9984240639479745e-05,
+      "loss": 1916.9785,
+      "step": 1440
+    },
+    {
+      "epoch": 0.027887162769676076,
+      "grad_norm": 6477.74340167659,
+      "learning_rate": 1.9983896275237677e-05,
+      "loss": 1895.2777,
+      "step": 1450
+    },
+    {
+      "epoch": 0.0280794880301566,
+      "grad_norm": 7475.619950653028,
+      "learning_rate": 1.9983548192159132e-05,
+      "loss": 1879.7949,
+      "step": 1460
+    },
+    {
+      "epoch": 0.028271813290637125,
+      "grad_norm": 6911.270172786403,
+      "learning_rate": 1.998319639037377e-05,
+      "loss": 1878.4859,
+      "step": 1470
+    },
+    {
+      "epoch": 0.02846413855111765,
+      "grad_norm": 7034.868758955199,
+      "learning_rate": 1.9982840870012626e-05,
+      "loss": 1882.0582,
+      "step": 1480
+    },
+    {
+      "epoch": 0.028656463811598174,
+      "grad_norm": 7342.8147967755285,
+      "learning_rate": 1.9982481631208138e-05,
+      "loss": 1831.302,
+      "step": 1490
+    },
+    {
+      "epoch": 0.0288487890720787,
+      "grad_norm": 5374.418078115465,
+      "learning_rate": 1.9982118674094104e-05,
+      "loss": 1820.3521,
+      "step": 1500
+    },
+    {
+      "epoch": 0.029041114332559223,
+      "grad_norm": 9918.535696965788,
+      "learning_rate": 1.9981751998805725e-05,
+      "loss": 1807.5156,
+      "step": 1510
+    },
+    {
+      "epoch": 0.02923343959303975,
+      "grad_norm": 6994.301939045192,
+      "learning_rate": 1.998138160547958e-05,
+      "loss": 1822.6168,
+      "step": 1520
+    },
+    {
+      "epoch": 0.029425764853520272,
+      "grad_norm": 6988.347234586641,
+      "learning_rate": 1.9981007494253638e-05,
+      "loss": 1794.4318,
+      "step": 1530
+    },
+    {
+      "epoch": 0.029618090114000798,
+      "grad_norm": 7182.174762468267,
+      "learning_rate": 1.998062966526724e-05,
+      "loss": 1803.4854,
+      "step": 1540
+    },
+    {
+      "epoch": 0.029810415374481324,
+      "grad_norm": 6032.49938713834,
+      "learning_rate": 1.998024811866113e-05,
+      "loss": 1776.6539,
+      "step": 1550
+    },
+    {
+      "epoch": 0.030002740634961847,
+      "grad_norm": 8451.537205034321,
+      "learning_rate": 1.9979862854577427e-05,
+      "loss": 1767.198,
+      "step": 1560
+    },
+    {
+      "epoch": 0.030195065895442373,
+      "grad_norm": 6324.559095250308,
+      "learning_rate": 1.9979473873159635e-05,
+      "loss": 1739.0559,
+      "step": 1570
+    },
+    {
+      "epoch": 0.030387391155922896,
+      "grad_norm": 6122.708535706743,
+      "learning_rate": 1.9979081174552638e-05,
+      "loss": 1736.2916,
+      "step": 1580
+    },
+    {
+      "epoch": 0.030579716416403422,
+      "grad_norm": 6163.254308596572,
+      "learning_rate": 1.997868475890271e-05,
+      "loss": 1710.7768,
+      "step": 1590
+    },
+    {
+      "epoch": 0.030772041676883945,
+      "grad_norm": 7482.85673113,
+      "learning_rate": 1.997828462635752e-05,
+      "loss": 1675.4146,
+      "step": 1600
+    },
+    {
+      "epoch": 0.03096436693736447,
+      "grad_norm": 7073.578921549694,
+      "learning_rate": 1.99778807770661e-05,
+      "loss": 1677.7027,
+      "step": 1610
+    },
+    {
+      "epoch": 0.031156692197844997,
+      "grad_norm": 6893.3497557177625,
+      "learning_rate": 1.9977473211178882e-05,
+      "loss": 1744.8992,
+      "step": 1620
+    },
+    {
+      "epoch": 0.03134901745832552,
+      "grad_norm": 6140.95848024039,
+      "learning_rate": 1.9977061928847676e-05,
+      "loss": 1692.118,
+      "step": 1630
+    },
+    {
+      "epoch": 0.03154134271880604,
+      "grad_norm": 6336.2767988987025,
+      "learning_rate": 1.9976646930225678e-05,
+      "loss": 1646.1111,
+      "step": 1640
+    },
+    {
+      "epoch": 0.03173366797928657,
+      "grad_norm": 7557.425027234783,
+      "learning_rate": 1.997622821546747e-05,
+      "loss": 1655.5611,
+      "step": 1650
+    },
+    {
+      "epoch": 0.031925993239767095,
+      "grad_norm": 7583.169303374387,
+      "learning_rate": 1.9975805784729008e-05,
+      "loss": 1628.3082,
+      "step": 1660
+    },
+    {
+      "epoch": 0.03211831850024762,
+      "grad_norm": 7173.653913504125,
+      "learning_rate": 1.9975379638167654e-05,
+      "loss": 1637.8015,
+      "step": 1670
+    },
+    {
+      "epoch": 0.03231064376072814,
+      "grad_norm": 5630.425784082243,
+      "learning_rate": 1.9974949775942134e-05,
+      "loss": 1587.5666,
+      "step": 1680
+    },
+    {
+      "epoch": 0.032502969021208666,
+      "grad_norm": 6960.428709379124,
+      "learning_rate": 1.997451619821256e-05,
+      "loss": 1622.5503,
+      "step": 1690
+    },
+    {
+      "epoch": 0.03269529428168919,
+      "grad_norm": 5666.063717228243,
+      "learning_rate": 1.997407890514044e-05,
+      "loss": 1593.3805,
+      "step": 1700
+    },
+    {
+      "epoch": 0.03288761954216972,
+      "grad_norm": 6681.8675048555015,
+      "learning_rate": 1.9973637896888652e-05,
+      "loss": 1558.9164,
+      "step": 1710
+    },
+    {
+      "epoch": 0.033079944802650245,
+      "grad_norm": 6584.118263819194,
+      "learning_rate": 1.997319317362147e-05,
+      "loss": 1554.373,
+      "step": 1720
+    },
+    {
+      "epoch": 0.033272270063130764,
+      "grad_norm": 5627.724155760312,
+      "learning_rate": 1.9972744735504542e-05,
+      "loss": 1531.2031,
+      "step": 1730
+    },
+    {
+      "epoch": 0.03346459532361129,
+      "grad_norm": 7141.653634460501,
+      "learning_rate": 1.9972292582704905e-05,
+      "loss": 1561.4683,
+      "step": 1740
+    },
+    {
+      "epoch": 0.033656920584091816,
+      "grad_norm": 6324.730794149539,
+      "learning_rate": 1.997183671539098e-05,
+      "loss": 1538.1083,
+      "step": 1750
+    },
+    {
+      "epoch": 0.03384924584457234,
+      "grad_norm": 6152.067748579414,
+      "learning_rate": 1.9971377133732567e-05,
+      "loss": 1537.7105,
+      "step": 1760
+    },
+    {
+      "epoch": 0.03404157110505287,
+      "grad_norm": 6727.948648162338,
+      "learning_rate": 1.997091383790086e-05,
+      "loss": 1563.193,
+      "step": 1770
+    },
+    {
+      "epoch": 0.03423389636553339,
+      "grad_norm": 6505.031008665096,
+      "learning_rate": 1.997044682806842e-05,
+      "loss": 1499.8426,
+      "step": 1780
+    },
+    {
+      "epoch": 0.034426221626013914,
+      "grad_norm": 5950.971205507093,
+      "learning_rate": 1.9969976104409202e-05,
+      "loss": 1514.9335,
+      "step": 1790
+    },
+    {
+      "epoch": 0.03461854688649444,
+      "grad_norm": 6142.555502411284,
+      "learning_rate": 1.9969501667098547e-05,
+      "loss": 1519.249,
+      "step": 1800
+    },
+    {
+      "epoch": 0.03481087214697497,
+      "grad_norm": 6715.306709245462,
+      "learning_rate": 1.996902351631317e-05,
+      "loss": 1469.8909,
+      "step": 1810
+    },
+    {
+      "epoch": 0.035003197407455486,
+      "grad_norm": 5243.438368734838,
+      "learning_rate": 1.996854165223118e-05,
+      "loss": 1453.6711,
+      "step": 1820
+    },
+    {
+      "epoch": 0.03519552266793601,
+      "grad_norm": 5982.293698199709,
+      "learning_rate": 1.9968056075032058e-05,
+      "loss": 1455.9143,
+      "step": 1830
+    },
+    {
+      "epoch": 0.03538784792841654,
+      "grad_norm": 5866.6628889997055,
+      "learning_rate": 1.9967566784896676e-05,
+      "loss": 1443.7178,
+      "step": 1840
+    },
+    {
+      "epoch": 0.035580173188897064,
+      "grad_norm": 6016.273160404695,
+      "learning_rate": 1.996707378200729e-05,
+      "loss": 1414.5693,
+      "step": 1850
+    },
+    {
+      "epoch": 0.03577249844937759,
+      "grad_norm": 6048.548916992061,
+      "learning_rate": 1.9966577066547526e-05,
+      "loss": 1422.5311,
+      "step": 1860
+    },
+    {
+      "epoch": 0.03596482370985811,
+      "grad_norm": 6847.995198951612,
+      "learning_rate": 1.9966076638702412e-05,
+      "loss": 1432.3163,
+      "step": 1870
+    },
+    {
+      "epoch": 0.036157148970338636,
+      "grad_norm": 6846.3147885279,
+      "learning_rate": 1.9965572498658346e-05,
+      "loss": 1428.632,
+      "step": 1880
+    },
+    {
+      "epoch": 0.03634947423081916,
+      "grad_norm": 6367.2467448438965,
+      "learning_rate": 1.996506464660311e-05,
+      "loss": 1404.1271,
+      "step": 1890
+    },
+    {
+      "epoch": 0.03654179949129969,
+      "grad_norm": 6149.422226476783,
+      "learning_rate": 1.9964553082725873e-05,
+      "loss": 1391.2855,
+      "step": 1900
+    },
+    {
+      "epoch": 0.03673412475178021,
+      "grad_norm": 5679.545651188765,
+      "learning_rate": 1.996403780721718e-05,
+      "loss": 1393.1797,
+      "step": 1910
+    },
+    {
+      "epoch": 0.036926450012260734,
+      "grad_norm": 6726.248738159026,
+      "learning_rate": 1.9963518820268968e-05,
+      "loss": 1463.1839,
+      "step": 1920
+    },
+    {
+      "epoch": 0.03711877527274126,
+      "grad_norm": 5812.251798998977,
+      "learning_rate": 1.996299612207455e-05,
+      "loss": 1385.4611,
+      "step": 1930
+    },
+    {
+      "epoch": 0.037311100533221786,
+      "grad_norm": 6481.40575070707,
+      "learning_rate": 1.9962469712828613e-05,
+      "loss": 1408.033,
+      "step": 1940
+    },
+    {
+      "epoch": 0.03750342579370231,
+      "grad_norm": 6217.993583939255,
+      "learning_rate": 1.996193959272725e-05,
+      "loss": 1349.4258,
+      "step": 1950
+    },
+    {
+      "epoch": 0.03769575105418283,
+      "grad_norm": 6332.18771920136,
+      "learning_rate": 1.9961405761967914e-05,
+      "loss": 1368.4553,
+      "step": 1960
+    },
+    {
+      "epoch": 0.03788807631466336,
+      "grad_norm": 6952.58839303712,
+      "learning_rate": 1.996086822074945e-05,
+      "loss": 1348.7169,
+      "step": 1970
+    },
+    {
+      "epoch": 0.038080401575143884,
+      "grad_norm": 6009.994644836221,
+      "learning_rate": 1.996032696927208e-05,
+      "loss": 1351.4033,
+      "step": 1980
+    },
+    {
+      "epoch": 0.03827272683562441,
+      "grad_norm": 6277.628853007842,
+      "learning_rate": 1.9959782007737418e-05,
+      "loss": 1340.3785,
+      "step": 1990
+    },
+    {
+      "epoch": 0.038465052096104936,
+      "grad_norm": 6283.430305222497,
+      "learning_rate": 1.9959233336348452e-05,
+      "loss": 1318.3738,
+      "step": 2000
+    },
+    {
+      "epoch": 0.038657377356585455,
+      "grad_norm": 5268.659543280948,
+      "learning_rate": 1.9958680955309546e-05,
+      "loss": 1345.0908,
+      "step": 2010
+    },
+    {
+      "epoch": 0.03884970261706598,
+      "grad_norm": 6018.533082613178,
+      "learning_rate": 1.9958124864826457e-05,
+      "loss": 1335.4022,
+      "step": 2020
+    },
+    {
+      "epoch": 0.03904202787754651,
+      "grad_norm": 6248.182353178989,
+      "learning_rate": 1.9957565065106318e-05,
+      "loss": 1309.4732,
+      "step": 2030
+    },
+    {
+      "epoch": 0.039234353138027034,
+      "grad_norm": 6065.4802712354995,
+      "learning_rate": 1.9957001556357652e-05,
+      "loss": 1308.841,
+      "step": 2040
+    },
+    {
+      "epoch": 0.03942667839850755,
+      "grad_norm": 5488.531257988972,
+      "learning_rate": 1.995643433879035e-05,
+      "loss": 1284.2648,
+      "step": 2050
+    },
+    {
+      "epoch": 0.03961900365898808,
+      "grad_norm": 6036.815153189847,
+      "learning_rate": 1.9955863412615693e-05,
+      "loss": 1301.248,
+      "step": 2060
+    },
+    {
+      "epoch": 0.039811328919468605,
+      "grad_norm": 6303.194316067486,
+      "learning_rate": 1.9955288778046338e-05,
+      "loss": 1294.3275,
+      "step": 2070
+    },
+    {
+      "epoch": 0.04000365417994913,
+      "grad_norm": 6023.2953845852735,
+      "learning_rate": 1.995471043529633e-05,
+      "loss": 1285.6352,
+      "step": 2080
+    },
+    {
+      "epoch": 0.04019597944042966,
+      "grad_norm": 5384.551949622103,
+      "learning_rate": 1.99541283845811e-05,
+      "loss": 1280.3516,
+      "step": 2090
+    },
+    {
+      "epoch": 0.04038830470091018,
+      "grad_norm": 6758.828042617035,
+      "learning_rate": 1.9953542626117437e-05,
+      "loss": 1282.7562,
+      "step": 2100
+    },
+    {
+      "epoch": 0.0405806299613907,
+      "grad_norm": 5895.0203225542455,
+      "learning_rate": 1.9952953160123537e-05,
+      "loss": 1281.4125,
+      "step": 2110
+    },
+    {
+      "epoch": 0.04077295522187123,
+      "grad_norm": 6065.273403921973,
+      "learning_rate": 1.995235998681896e-05,
+      "loss": 1260.6578,
+      "step": 2120
+    },
+    {
+      "epoch": 0.040965280482351756,
+      "grad_norm": 6293.559800161821,
+      "learning_rate": 1.9951763106424658e-05,
+      "loss": 1255.2279,
+      "step": 2130
+    },
+    {
+      "epoch": 0.041157605742832275,
+      "grad_norm": 4700.336619432639,
+      "learning_rate": 1.9951162519162962e-05,
+      "loss": 1239.8465,
+      "step": 2140
+    },
+    {
+      "epoch": 0.0413499310033128,
+      "grad_norm": 5221.876270859536,
+      "learning_rate": 1.9950558225257574e-05,
+      "loss": 1238.9635,
+      "step": 2150
+    },
+    {
+      "epoch": 0.04154225626379333,
+      "grad_norm": 5565.829122934509,
+      "learning_rate": 1.9949950224933584e-05,
+      "loss": 1236.6825,
+      "step": 2160
+    },
+    {
+      "epoch": 0.04173458152427385,
+      "grad_norm": 6147.523806240359,
+      "learning_rate": 1.994933851841747e-05,
+      "loss": 1227.3937,
+      "step": 2170
+    },
+    {
+      "epoch": 0.04192690678475438,
+      "grad_norm": 5974.770554144452,
+      "learning_rate": 1.994872310593707e-05,
+      "loss": 1230.965,
+      "step": 2180
+    },
+    {
+      "epoch": 0.0421192320452349,
+      "grad_norm": 5330.434141430119,
+      "learning_rate": 1.994810398772162e-05,
+      "loss": 1226.8635,
+      "step": 2190
+    },
+    {
+      "epoch": 0.042311557305715425,
+      "grad_norm": 5436.87738586885,
+      "learning_rate": 1.994748116400174e-05,
+      "loss": 1205.6514,
+      "step": 2200
+    },
+    {
+      "epoch": 0.04250388256619595,
+      "grad_norm": 5126.282475196194,
+      "learning_rate": 1.994685463500941e-05,
+      "loss": 1221.4822,
+      "step": 2210
+    },
+    {
+      "epoch": 0.04269620782667648,
+      "grad_norm": 4848.022230952555,
+      "learning_rate": 1.9946224400978006e-05,
+      "loss": 1201.4771,
+      "step": 2220
+    },
+    {
+      "epoch": 0.042888533087157,
+      "grad_norm": 5211.129666115587,
+      "learning_rate": 1.994559046214228e-05,
+      "loss": 1192.5682,
+      "step": 2230
+    },
+    {
+      "epoch": 0.04308085834763752,
+      "grad_norm": 6304.665333599148,
+      "learning_rate": 1.9944952818738366e-05,
+      "loss": 1209.3821,
+      "step": 2240
+    },
+    {
+      "epoch": 0.04327318360811805,
+      "grad_norm": 4778.190236574797,
+      "learning_rate": 1.9944311471003775e-05,
+      "loss": 1210.8347,
+      "step": 2250
+    },
+    {
+      "epoch": 0.043465508868598575,
+      "grad_norm": 5923.088858141527,
+      "learning_rate": 1.9943666419177392e-05,
+      "loss": 1158.5988,
+      "step": 2260
+    },
+    {
+      "epoch": 0.0436578341290791,
+      "grad_norm": 5677.082358004745,
+      "learning_rate": 1.9943017663499497e-05,
+      "loss": 1181.834,
+      "step": 2270
+    },
+    {
+      "epoch": 0.04385015938955962,
+      "grad_norm": 5018.811106626207,
+      "learning_rate": 1.9942365204211734e-05,
+      "loss": 1161.9672,
+      "step": 2280
+    },
+    {
+      "epoch": 0.044042484650040147,
+      "grad_norm": 5519.412534913629,
+      "learning_rate": 1.9941709041557134e-05,
+      "loss": 1165.3775,
+      "step": 2290
+    },
+    {
+      "epoch": 0.04423480991052067,
+      "grad_norm": 4758.708930344798,
+      "learning_rate": 1.994104917578011e-05,
+      "loss": 1189.6666,
+      "step": 2300
+    },
+    {
+      "epoch": 0.0444271351710012,
+      "grad_norm": 5797.702850920498,
+      "learning_rate": 1.9940385607126456e-05,
+      "loss": 1151.2826,
+      "step": 2310
+    },
+    {
+      "epoch": 0.044619460431481725,
+      "grad_norm": 5412.241982469557,
+      "learning_rate": 1.9939718335843326e-05,
+      "loss": 1167.319,
+      "step": 2320
+    },
+    {
+      "epoch": 0.044811785691962244,
+      "grad_norm": 5565.101511080268,
+      "learning_rate": 1.9939047362179283e-05,
+      "loss": 1161.4689,
+      "step": 2330
+    },
+    {
+      "epoch": 0.04500411095244277,
+      "grad_norm": 5556.710865953506,
+      "learning_rate": 1.9938372686384244e-05,
+      "loss": 1127.1202,
+      "step": 2340
+    },
+    {
+      "epoch": 0.0451964362129233,
+      "grad_norm": 5601.832752266584,
+      "learning_rate": 1.9937694308709514e-05,
+      "loss": 1141.9598,
+      "step": 2350
+    },
+    {
+      "epoch": 0.04538876147340382,
+      "grad_norm": 8889.048734758144,
+      "learning_rate": 1.9937012229407783e-05,
+      "loss": 1169.4885,
+      "step": 2360
+    },
+    {
+      "epoch": 0.04558108673388434,
+      "grad_norm": 6129.912153090955,
+      "learning_rate": 1.9936326448733107e-05,
+      "loss": 1172.7059,
+      "step": 2370
+    },
+    {
+      "epoch": 0.04577341199436487,
+      "grad_norm": 4443.626207141793,
+      "learning_rate": 1.9935636966940938e-05,
+      "loss": 1129.7623,
+      "step": 2380
+    },
+    {
+      "epoch": 0.045965737254845394,
+      "grad_norm": 6621.077972481085,
+      "learning_rate": 1.993494378428809e-05,
+      "loss": 1156.3591,
+      "step": 2390
+    },
+    {
+      "epoch": 0.04615806251532592,
+      "grad_norm": 6061.474638019706,
+      "learning_rate": 1.9934246901032764e-05,
+      "loss": 1108.5379,
+      "step": 2400
+    },
+    {
+      "epoch": 0.04635038777580645,
+      "grad_norm": 6122.716279468376,
+      "learning_rate": 1.9933546317434536e-05,
+      "loss": 1131.7629,
+      "step": 2410
+    },
+    {
+      "epoch": 0.046542713036286966,
+      "grad_norm": 4523.510602373938,
+      "learning_rate": 1.9932842033754363e-05,
+      "loss": 1122.6425,
+      "step": 2420
+    },
+    {
+      "epoch": 0.04673503829676749,
+      "grad_norm": 5025.599416057065,
+      "learning_rate": 1.9932134050254578e-05,
+      "loss": 1114.3494,
+      "step": 2430
+    },
+    {
+      "epoch": 0.04692736355724802,
+      "grad_norm": 4899.485765055181,
+      "learning_rate": 1.9931422367198893e-05,
+      "loss": 1099.6018,
+      "step": 2440
+    },
+    {
+      "epoch": 0.047119688817728544,
+      "grad_norm": 5090.665618145162,
+      "learning_rate": 1.9930706984852403e-05,
+      "loss": 1104.125,
+      "step": 2450
+    },
+    {
+      "epoch": 0.04731201407820907,
+      "grad_norm": 4940.992602008076,
+      "learning_rate": 1.992998790348157e-05,
+      "loss": 1115.1102,
+      "step": 2460
+    },
+    {
+      "epoch": 0.04750433933868959,
+      "grad_norm": 4814.86897503154,
+      "learning_rate": 1.992926512335424e-05,
+      "loss": 1116.3245,
+      "step": 2470
+    },
+    {
+      "epoch": 0.047696664599170116,
+      "grad_norm": 4436.093340243908,
+      "learning_rate": 1.992853864473964e-05,
+      "loss": 1084.0676,
+      "step": 2480
+    },
+    {
+      "epoch": 0.04788898985965064,
+      "grad_norm": 5318.538620492033,
+      "learning_rate": 1.992780846790837e-05,
+      "loss": 1099.2612,
+      "step": 2490
+    },
+    {
+      "epoch": 0.04808131512013117,
+      "grad_norm": 5271.639572188547,
+      "learning_rate": 1.9927074593132412e-05,
+      "loss": 1086.9515,
+      "step": 2500
+    },
+    {
+      "epoch": 0.04827364038061169,
+      "grad_norm": 5691.1346926333035,
+      "learning_rate": 1.9926337020685114e-05,
+      "loss": 1096.943,
+      "step": 2510
+    },
+    {
+      "epoch": 0.048465965641092214,
+      "grad_norm": 4964.253910595249,
+      "learning_rate": 1.992559575084122e-05,
+      "loss": 1088.6112,
+      "step": 2520
+    },
+    {
+      "epoch": 0.04865829090157274,
+      "grad_norm": 5474.318377203567,
+      "learning_rate": 1.9924850783876832e-05,
+      "loss": 1078.7441,
+      "step": 2530
+    },
+    {
+      "epoch": 0.048850616162053266,
+      "grad_norm": 4877.686610194673,
+      "learning_rate": 1.992410212006944e-05,
+      "loss": 1075.9105,
+      "step": 2540
+    },
+    {
+      "epoch": 0.04904294142253379,
+      "grad_norm": 4911.033729811151,
+      "learning_rate": 1.9923349759697915e-05,
+      "loss": 1064.0662,
+      "step": 2550
+    },
+    {
+      "epoch": 0.04923526668301431,
+      "grad_norm": 5749.793109960041,
+      "learning_rate": 1.992259370304249e-05,
+      "loss": 1056.484,
+      "step": 2560
+    },
+    {
+      "epoch": 0.04942759194349484,
+      "grad_norm": 5726.810088024783,
+      "learning_rate": 1.992183395038479e-05,
+      "loss": 1068.5598,
+      "step": 2570
+    },
+    {
+      "epoch": 0.049619917203975364,
+      "grad_norm": 4464.140815954304,
+      "learning_rate": 1.9921070502007808e-05,
+      "loss": 1049.9455,
+      "step": 2580
+    },
+    {
+      "epoch": 0.04981224246445589,
+      "grad_norm": 4412.999068295054,
+      "learning_rate": 1.9920303358195916e-05,
+      "loss": 1039.6811,
+      "step": 2590
+    },
+    {
+      "epoch": 0.05000456772493641,
+      "grad_norm": 5669.913135174593,
+      "learning_rate": 1.9919532519234858e-05,
+      "loss": 1058.2753,
+      "step": 2600
+    },
+    {
+      "epoch": 0.050196892985416935,
+      "grad_norm": 5512.474040181994,
+      "learning_rate": 1.9918757985411767e-05,
+      "loss": 1038.1103,
+      "step": 2610
+    },
+    {
+      "epoch": 0.05038921824589746,
+      "grad_norm": 5568.707613091602,
+      "learning_rate": 1.991797975701514e-05,
+      "loss": 1040.8706,
+      "step": 2620
+    },
+    {
+      "epoch": 0.05058154350637799,
+      "grad_norm": 56860.19354775827,
+      "learning_rate": 1.9917197834334858e-05,
+      "loss": 1098.1378,
+      "step": 2630
+    },
+    {
+      "epoch": 0.050773868766858514,
+      "grad_norm": 4838.380630173328,
+      "learning_rate": 1.9916412217662162e-05,
+      "loss": 1032.9431,
+      "step": 2640
+    },
+    {
+      "epoch": 0.05096619402733903,
+      "grad_norm": 4664.195992308137,
+      "learning_rate": 1.9915622907289695e-05,
+      "loss": 1021.9714,
+      "step": 2650
+    },
+    {
+      "epoch": 0.05115851928781956,
+      "grad_norm": 4568.3497516757625,
+      "learning_rate": 1.9914829903511458e-05,
+      "loss": 1014.9593,
+      "step": 2660
+    },
+    {
+      "epoch": 0.051350844548300086,
+      "grad_norm": 5011.072462941165,
+      "learning_rate": 1.9914033206622828e-05,
+      "loss": 1041.5867,
+      "step": 2670
+    },
+    {
+      "epoch": 0.05154316980878061,
+      "grad_norm": 5292.580046561721,
+      "learning_rate": 1.9913232816920565e-05,
+      "loss": 1031.7426,
+      "step": 2680
+    },
+    {
+      "epoch": 0.05173549506926114,
+      "grad_norm": 4823.289178985006,
+      "learning_rate": 1.99124287347028e-05,
+      "loss": 1026.7715,
+      "step": 2690
+    },
+    {
+      "epoch": 0.05192782032974166,
+      "grad_norm": 4752.049768811952,
+      "learning_rate": 1.991162096026904e-05,
+      "loss": 1022.8786,
+      "step": 2700
+    },
+    {
+      "epoch": 0.05212014559022218,
+      "grad_norm": 4705.213430937199,
+      "learning_rate": 1.9910809493920172e-05,
+      "loss": 1017.1159,
+      "step": 2710
+    },
+    {
+      "epoch": 0.05231247085070271,
+      "grad_norm": 4793.854247136831,
+      "learning_rate": 1.990999433595845e-05,
+      "loss": 1019.4643,
+      "step": 2720
+    },
+    {
+      "epoch": 0.052504796111183236,
+      "grad_norm": 4920.728030054162,
+      "learning_rate": 1.99091754866875e-05,
+      "loss": 1007.5629,
+      "step": 2730
+    },
+    {
+      "epoch": 0.052697121371663755,
+      "grad_norm": 4434.570504555348,
+      "learning_rate": 1.990835294641234e-05,
+      "loss": 996.0966,
+      "step": 2740
+    },
+    {
+      "epoch": 0.05288944663214428,
+      "grad_norm": 4893.77566822937,
+      "learning_rate": 1.990752671543935e-05,
+      "loss": 1002.8721,
+      "step": 2750
+    },
+    {
+      "epoch": 0.05308177189262481,
+      "grad_norm": 5102.66497285508,
+      "learning_rate": 1.9906696794076284e-05,
+      "loss": 997.6461,
+      "step": 2760
+    },
+    {
+      "epoch": 0.05327409715310533,
+      "grad_norm": 4818.975735209405,
+      "learning_rate": 1.9905863182632285e-05,
+      "loss": 1007.0695,
+      "step": 2770
+    },
+    {
+      "epoch": 0.05346642241358586,
+      "grad_norm": 4901.244624847641,
+      "learning_rate": 1.990502588141784e-05,
+      "loss": 983.7172,
+      "step": 2780
+    },
+    {
+      "epoch": 0.05365874767406638,
+      "grad_norm": 4937.250155063634,
+      "learning_rate": 1.990418489074485e-05,
+      "loss": 964.584,
+      "step": 2790
+    },
+    {
+      "epoch": 0.053851072934546905,
+      "grad_norm": 4551.201756040109,
+      "learning_rate": 1.9903340210926555e-05,
+      "loss": 959.6879,
+      "step": 2800
+    },
+    {
+      "epoch": 0.05404339819502743,
+      "grad_norm": 5112.548170122441,
+      "learning_rate": 1.9902491842277592e-05,
+      "loss": 1009.366,
+      "step": 2810
+    },
+    {
+      "epoch": 0.05423572345550796,
+      "grad_norm": 4547.733471475647,
+      "learning_rate": 1.9901639785113967e-05,
+      "loss": 989.1688,
+      "step": 2820
+    },
+    {
+      "epoch": 0.05442804871598848,
+      "grad_norm": 4727.651956402866,
+      "learning_rate": 1.990078403975305e-05,
+      "loss": 996.4672,
+      "step": 2830
+    },
+    {
+      "epoch": 0.054620373976469,
+      "grad_norm": 5037.541830739576,
+      "learning_rate": 1.9899924606513593e-05,
+      "loss": 978.5039,
+      "step": 2840
+    },
+    {
+      "epoch": 0.05481269923694953,
+      "grad_norm": 5393.778198327005,
+      "learning_rate": 1.9899061485715726e-05,
+      "loss": 954.7977,
+      "step": 2850
+    },
+    {
+      "epoch": 0.055005024497430055,
+      "grad_norm": 4468.238799938698,
+      "learning_rate": 1.9898194677680943e-05,
+      "loss": 973.9543,
+      "step": 2860
+    },
+    {
+      "epoch": 0.05519734975791058,
+      "grad_norm": 4833.6617348280215,
+      "learning_rate": 1.9897324182732118e-05,
+      "loss": 959.073,
+      "step": 2870
+    },
+    {
+      "epoch": 0.0553896750183911,
+      "grad_norm": 4360.012760415991,
+      "learning_rate": 1.989645000119349e-05,
+      "loss": 950.7967,
+      "step": 2880
+    },
+    {
+      "epoch": 0.05558200027887163,
+      "grad_norm": 4779.795875742137,
+      "learning_rate": 1.9895572133390687e-05,
+      "loss": 951.4395,
+      "step": 2890
+    },
+    {
+      "epoch": 0.05577432553935215,
+      "grad_norm": 4595.241636924597,
+      "learning_rate": 1.9894690579650694e-05,
+      "loss": 953.3818,
+      "step": 2900
+    },
+    {
+      "epoch": 0.05596665079983268,
+      "grad_norm": 4750.078277224996,
+      "learning_rate": 1.9893805340301876e-05,
+      "loss": 972.3777,
+      "step": 2910
+    },
+    {
+      "epoch": 0.0561589760603132,
+      "grad_norm": 4946.332658418301,
+      "learning_rate": 1.989291641567397e-05,
+      "loss": 940.0307,
+      "step": 2920
+    },
+    {
+      "epoch": 0.056351301320793724,
+      "grad_norm": 4820.671381216139,
+      "learning_rate": 1.9892023806098083e-05,
+      "loss": 919.107,
+      "step": 2930
+    },
+    {
+      "epoch": 0.05654362658127425,
+      "grad_norm": 9364.311838278925,
+      "learning_rate": 1.9891127511906703e-05,
+      "loss": 931.4717,
+      "step": 2940
+    },
+    {
+      "epoch": 0.05673595184175478,
+      "grad_norm": 5521.496200012186,
+      "learning_rate": 1.9890227533433685e-05,
+      "loss": 952.6798,
+      "step": 2950
+    },
+    {
+      "epoch": 0.0569282771022353,
+      "grad_norm": 4134.00927852241,
+      "learning_rate": 1.988932387101425e-05,
+      "loss": 953.5514,
+      "step": 2960
+    },
+    {
+      "epoch": 0.05712060236271582,
+      "grad_norm": 5084.677399940139,
+      "learning_rate": 1.9888416524985e-05,
+      "loss": 928.6297,
+      "step": 2970
+    },
+    {
+      "epoch": 0.05731292762319635,
+      "grad_norm": 4406.793507705554,
+      "learning_rate": 1.988750549568391e-05,
+      "loss": 943.6451,
+      "step": 2980
+    },
+    {
+      "epoch": 0.057505252883676874,
+      "grad_norm": 4417.295293118958,
+      "learning_rate": 1.9886590783450317e-05,
+      "loss": 904.7558,
+      "step": 2990
+    },
+    {
+      "epoch": 0.0576975781441574,
+      "grad_norm": 4043.4875527218483,
+      "learning_rate": 1.9885672388624942e-05,
+      "loss": 913.576,
+      "step": 3000
+    },
+    {
+      "epoch": 0.05788990340463793,
+      "grad_norm": 6053.236380815296,
+      "learning_rate": 1.9884750311549868e-05,
+      "loss": 937.1524,
+      "step": 3010
+    },
+    {
+      "epoch": 0.058082228665118446,
+      "grad_norm": 4789.6869984331615,
+      "learning_rate": 1.9883824552568557e-05,
+      "loss": 937.9734,
+      "step": 3020
+    },
+    {
+      "epoch": 0.05827455392559897,
+      "grad_norm": 4965.655181870662,
+      "learning_rate": 1.9882895112025835e-05,
+      "loss": 936.3613,
+      "step": 3030
+    },
+    {
+      "epoch": 0.0584668791860795,
+      "grad_norm": 4977.478694437648,
+      "learning_rate": 1.9881961990267906e-05,
+      "loss": 901.9449,
+      "step": 3040
+    },
+    {
+      "epoch": 0.058659204446560025,
+      "grad_norm": 4910.648766244268,
+      "learning_rate": 1.988102518764234e-05,
+      "loss": 921.7038,
+      "step": 3050
+    },
+    {
+      "epoch": 0.058851529707040544,
+      "grad_norm": 5225.812550012057,
+      "learning_rate": 1.9880084704498084e-05,
+      "loss": 918.5408,
+      "step": 3060
+    },
+    {
+      "epoch": 0.05904385496752107,
+      "grad_norm": 4083.5240466170785,
+      "learning_rate": 1.987914054118545e-05,
+      "loss": 924.0464,
+      "step": 3070
+    },
+    {
+      "epoch": 0.059236180228001596,
+      "grad_norm": 4274.373196959295,
+      "learning_rate": 1.9878192698056125e-05,
+      "loss": 910.9401,
+      "step": 3080
+    },
+    {
+      "epoch": 0.05942850548848212,
+      "grad_norm": 4719.276107575339,
+      "learning_rate": 1.9877241175463165e-05,
+      "loss": 911.0809,
+      "step": 3090
+    },
+    {
+      "epoch": 0.05962083074896265,
+      "grad_norm": 4557.959688711481,
+      "learning_rate": 1.9876285973760993e-05,
+      "loss": 916.6955,
+      "step": 3100
+    },
+    {
+      "epoch": 0.05981315600944317,
+      "grad_norm": 4377.690465567877,
+      "learning_rate": 1.9875327093305405e-05,
+      "loss": 906.8008,
+      "step": 3110
+    },
+    {
+      "epoch": 0.060005481269923694,
+      "grad_norm": 4464.047345330293,
+      "learning_rate": 1.9874364534453577e-05,
+      "loss": 896.7488,
+      "step": 3120
+    },
+    {
+      "epoch": 0.06019780653040422,
+      "grad_norm": 4934.762156507667,
+      "learning_rate": 1.9873398297564036e-05,
+      "loss": 901.9243,
+      "step": 3130
+    },
+    {
+      "epoch": 0.060390131790884746,
+      "grad_norm": 4174.133922786432,
+      "learning_rate": 1.9872428382996697e-05,
+      "loss": 906.8617,
+      "step": 3140
+    },
+    {
+      "epoch": 0.060582457051365265,
+      "grad_norm": 4555.277081853879,
+      "learning_rate": 1.987145479111283e-05,
+      "loss": 872.9479,
+      "step": 3150
+    },
+    {
+      "epoch": 0.06077478231184579,
+      "grad_norm": 5080.292901313894,
+      "learning_rate": 1.987047752227509e-05,
+      "loss": 887.3372,
+      "step": 3160
+    },
+    {
+      "epoch": 0.06096710757232632,
+      "grad_norm": 4435.566792129146,
+      "learning_rate": 1.9869496576847488e-05,
+      "loss": 863.0025,
+      "step": 3170
+    },
+    {
+      "epoch": 0.061159432832806844,
+      "grad_norm": 4418.891159481613,
+      "learning_rate": 1.9868511955195407e-05,
+      "loss": 888.5924,
+      "step": 3180
+    },
+    {
+      "epoch": 0.06135175809328737,
+      "grad_norm": 6415.926300105252,
+      "learning_rate": 1.986752365768561e-05,
+      "loss": 848.7134,
+      "step": 3190
+    },
+    {
+      "epoch": 0.06154408335376789,
+      "grad_norm": 4401.204702087581,
+      "learning_rate": 1.986653168468622e-05,
+      "loss": 879.5176,
+      "step": 3200
+    },
+    {
+      "epoch": 0.061736408614248416,
+      "grad_norm": 5100.704679489723,
+      "learning_rate": 1.9865536036566727e-05,
+      "loss": 873.8359,
+      "step": 3210
+    },
+    {
+      "epoch": 0.06192873387472894,
+      "grad_norm": 4561.358661768131,
+      "learning_rate": 1.9864536713697992e-05,
+      "loss": 905.102,
+      "step": 3220
+    },
+    {
+      "epoch": 0.06212105913520947,
+      "grad_norm": 4042.3756414071713,
+      "learning_rate": 1.986353371645225e-05,
+      "loss": 890.4045,
+      "step": 3230
+    },
+    {
+      "epoch": 0.062313384395689994,
+      "grad_norm": 4432.64429811606,
+      "learning_rate": 1.9862527045203105e-05,
+      "loss": 879.4265,
+      "step": 3240
+    },
+    {
+      "epoch": 0.06250570965617051,
+      "grad_norm": 4304.964417453636,
+      "learning_rate": 1.986151670032552e-05,
+      "loss": 878.1913,
+      "step": 3250
+    },
+    {
+      "epoch": 0.06269803491665105,
+      "grad_norm": 4929.846602020889,
+      "learning_rate": 1.986050268219583e-05,
+      "loss": 856.8271,
+      "step": 3260
+    },
+    {
+      "epoch": 0.06289036017713157,
+      "grad_norm": 4182.158653745027,
+      "learning_rate": 1.9859484991191742e-05,
+      "loss": 874.8608,
+      "step": 3270
+    },
+    {
+      "epoch": 0.06308268543761208,
+      "grad_norm": 4495.4207670603255,
+      "learning_rate": 1.985846362769233e-05,
+      "loss": 880.4451,
+      "step": 3280
+    },
+    {
+      "epoch": 0.06327501069809262,
+      "grad_norm": 4413.667875279064,
+      "learning_rate": 1.9857438592078034e-05,
+      "loss": 856.9678,
+      "step": 3290
+    },
+    {
+      "epoch": 0.06346733595857314,
+      "grad_norm": 4140.851677182371,
+      "learning_rate": 1.9856409884730667e-05,
+      "loss": 859.9239,
+      "step": 3300
+    },
+    {
+      "epoch": 0.06365966121905367,
+      "grad_norm": 4368.5020687747,
+      "learning_rate": 1.98553775060334e-05,
+      "loss": 837.9691,
+      "step": 3310
+    },
+    {
+      "epoch": 0.06385198647953419,
+      "grad_norm": 4160.2297715238465,
+      "learning_rate": 1.9854341456370777e-05,
+      "loss": 823.9158,
+      "step": 3320
+    },
+    {
+      "epoch": 0.06404431174001471,
+      "grad_norm": 4858.730488652083,
+      "learning_rate": 1.9853301736128712e-05,
+      "loss": 846.968,
+      "step": 3330
+    },
+    {
+      "epoch": 0.06423663700049524,
+      "grad_norm": 4403.117774583476,
+      "learning_rate": 1.9852258345694486e-05,
+      "loss": 871.0086,
+      "step": 3340
+    },
+    {
+      "epoch": 0.06442896226097576,
+      "grad_norm": 5190.530620906509,
+      "learning_rate": 1.9851211285456738e-05,
+      "loss": 848.1436,
+      "step": 3350
+    },
+    {
+      "epoch": 0.06462128752145628,
+      "grad_norm": 5214.395170983605,
+      "learning_rate": 1.9850160555805485e-05,
+      "loss": 859.4229,
+      "step": 3360
+    },
+    {
+      "epoch": 0.06481361278193681,
+      "grad_norm": 5113.537918633996,
+      "learning_rate": 1.9849106157132105e-05,
+      "loss": 861.5078,
+      "step": 3370
+    },
+    {
+      "epoch": 0.06500593804241733,
+      "grad_norm": 4078.461363069817,
+      "learning_rate": 1.9848048089829347e-05,
+      "loss": 850.4208,
+      "step": 3380
+    },
+    {
+      "epoch": 0.06519826330289787,
+      "grad_norm": 3898.4248478860036,
+      "learning_rate": 1.9846986354291324e-05,
+      "loss": 831.2916,
+      "step": 3390
+    },
+    {
+      "epoch": 0.06539058856337839,
+      "grad_norm": 5468.591054006592,
+      "learning_rate": 1.9845920950913506e-05,
+      "loss": 853.7027,
+      "step": 3400
+    },
+    {
+      "epoch": 0.0655829138238589,
+      "grad_norm": 4798.514728157639,
+      "learning_rate": 1.9844851880092748e-05,
+      "loss": 817.3862,
+      "step": 3410
+    },
+    {
+      "epoch": 0.06577523908433944,
+      "grad_norm": 4643.5886323774275,
+      "learning_rate": 1.9843779142227258e-05,
+      "loss": 827.4217,
+      "step": 3420
+    },
+    {
+      "epoch": 0.06596756434481996,
+      "grad_norm": 4856.374126046193,
+      "learning_rate": 1.984270273771661e-05,
+      "loss": 825.7255,
+      "step": 3430
+    },
+    {
+      "epoch": 0.06615988960530049,
+      "grad_norm": 4969.591662398133,
+      "learning_rate": 1.9841622666961756e-05,
+      "loss": 836.7778,
+      "step": 3440
+    },
+    {
+      "epoch": 0.06635221486578101,
+      "grad_norm": 4017.7999361640086,
+      "learning_rate": 1.9840538930364992e-05,
+      "loss": 838.4433,
+      "step": 3450
+    },
+    {
+      "epoch": 0.06654454012626153,
+      "grad_norm": 3960.744107457203,
+      "learning_rate": 1.983945152833e-05,
+      "loss": 801.9833,
+      "step": 3460
+    },
+    {
+      "epoch": 0.06673686538674206,
+      "grad_norm": 4703.169822673688,
+      "learning_rate": 1.9838360461261817e-05,
+      "loss": 851.1238,
+      "step": 3470
+    },
+    {
+      "epoch": 0.06692919064722258,
+      "grad_norm": 4395.211216082489,
+      "learning_rate": 1.9837265729566853e-05,
+      "loss": 817.4702,
+      "step": 3480
+    },
+    {
+      "epoch": 0.06712151590770311,
+      "grad_norm": 4481.923669874384,
+      "learning_rate": 1.9836167333652866e-05,
+      "loss": 842.1456,
+      "step": 3490
+    },
+    {
+      "epoch": 0.06731384116818363,
+      "grad_norm": 4237.612272263026,
+      "learning_rate": 1.9835065273929002e-05,
+      "loss": 811.6774,
+      "step": 3500
+    },
+    {
+      "epoch": 0.06750616642866415,
+      "grad_norm": 4336.09376535591,
+      "learning_rate": 1.9833959550805754e-05,
+      "loss": 812.5192,
+      "step": 3510
+    },
+    {
+      "epoch": 0.06769849168914469,
+      "grad_norm": 4534.430920824958,
+      "learning_rate": 1.9832850164694983e-05,
+      "loss": 806.0679,
+      "step": 3520
+    },
+    {
+      "epoch": 0.0678908169496252,
+      "grad_norm": 4737.744932299016,
+      "learning_rate": 1.9831737116009924e-05,
+      "loss": 804.2666,
+      "step": 3530
+    },
+    {
+      "epoch": 0.06808314221010574,
+      "grad_norm": 4205.116332782815,
+      "learning_rate": 1.9830620405165164e-05,
+      "loss": 796.6594,
+      "step": 3540
+    },
+    {
+      "epoch": 0.06827546747058626,
+      "grad_norm": 4999.150392223975,
+      "learning_rate": 1.982950003257666e-05,
+      "loss": 827.5804,
+      "step": 3550
+    },
+    {
+      "epoch": 0.06846779273106678,
+      "grad_norm": 4407.410357341057,
+      "learning_rate": 1.9828375998661738e-05,
+      "loss": 817.3569,
+      "step": 3560
+    },
+    {
+      "epoch": 0.06866011799154731,
+      "grad_norm": 4218.685212880607,
+      "learning_rate": 1.9827248303839073e-05,
+      "loss": 765.9263,
+      "step": 3570
+    },
+    {
+      "epoch": 0.06885244325202783,
+      "grad_norm": 4192.665753801111,
+      "learning_rate": 1.982611694852872e-05,
+      "loss": 792.1777,
+      "step": 3580
+    },
+    {
+      "epoch": 0.06904476851250835,
+      "grad_norm": 4538.948349765603,
+      "learning_rate": 1.9824981933152087e-05,
+      "loss": 777.4994,
+      "step": 3590
+    },
+    {
+      "epoch": 0.06923709377298888,
+      "grad_norm": 4238.988511745942,
+      "learning_rate": 1.9823843258131945e-05,
+      "loss": 796.3801,
+      "step": 3600
+    },
+    {
+      "epoch": 0.0694294190334694,
+      "grad_norm": 4980.483201880425,
+      "learning_rate": 1.9822700923892438e-05,
+      "loss": 818.8832,
+      "step": 3610
+    },
+    {
+      "epoch": 0.06962174429394993,
+      "grad_norm": 4087.560163038672,
+      "learning_rate": 1.9821554930859066e-05,
+      "loss": 794.1205,
+      "step": 3620
+    },
+    {
+      "epoch": 0.06981406955443045,
+      "grad_norm": 4221.2069398196,
+      "learning_rate": 1.982040527945869e-05,
+      "loss": 799.7377,
+      "step": 3630
+    },
+    {
+      "epoch": 0.07000639481491097,
+      "grad_norm": 4103.723339502749,
+      "learning_rate": 1.9819251970119534e-05,
+      "loss": 808.7371,
+      "step": 3640
+    },
+    {
+      "epoch": 0.0701987200753915,
+      "grad_norm": 4073.157123448541,
+      "learning_rate": 1.9818095003271193e-05,
+      "loss": 781.1663,
+      "step": 3650
+    },
+    {
+      "epoch": 0.07039104533587202,
+      "grad_norm": 4201.542638359091,
+      "learning_rate": 1.9816934379344613e-05,
+      "loss": 781.4498,
+      "step": 3660
+    },
+    {
+      "epoch": 0.07058337059635256,
+      "grad_norm": 4369.081933228246,
+      "learning_rate": 1.9815770098772108e-05,
+      "loss": 784.0995,
+      "step": 3670
+    },
+    {
+      "epoch": 0.07077569585683308,
+      "grad_norm": 3845.900065729217,
+      "learning_rate": 1.9814602161987354e-05,
+      "loss": 773.1455,
+      "step": 3680
+    },
+    {
+      "epoch": 0.0709680211173136,
+      "grad_norm": 3738.811478727106,
+      "learning_rate": 1.981343056942539e-05,
+      "loss": 766.453,
+      "step": 3690
+    },
+    {
+      "epoch": 0.07116034637779413,
+      "grad_norm": 5235.3272466201515,
+      "learning_rate": 1.9812255321522614e-05,
+      "loss": 798.5602,
+      "step": 3700
+    },
+    {
+      "epoch": 0.07135267163827465,
+      "grad_norm": 3979.6886324575103,
+      "learning_rate": 1.981107641871678e-05,
+      "loss": 780.4679,
+      "step": 3710
+    },
+    {
+      "epoch": 0.07154499689875518,
+      "grad_norm": 4022.041531127809,
+      "learning_rate": 1.980989386144702e-05,
+      "loss": 774.6857,
+      "step": 3720
+    },
+    {
+      "epoch": 0.0717373221592357,
+      "grad_norm": 4472.923210364339,
+      "learning_rate": 1.980870765015381e-05,
+      "loss": 780.6932,
+      "step": 3730
+    },
+    {
+      "epoch": 0.07192964741971622,
+      "grad_norm": 3689.7253421859155,
+      "learning_rate": 1.9807517785278997e-05,
+      "loss": 772.0474,
+      "step": 3740
+    },
+    {
+      "epoch": 0.07212197268019675,
+      "grad_norm": 4027.0369553735572,
+      "learning_rate": 1.9806324267265786e-05,
+      "loss": 772.4933,
+      "step": 3750
+    },
+    {
+      "epoch": 0.07231429794067727,
+      "grad_norm": 4300.084726674544,
+      "learning_rate": 1.9805127096558742e-05,
+      "loss": 772.58,
+      "step": 3760
+    },
+    {
+      "epoch": 0.0725066232011578,
+      "grad_norm": 4260.686057345299,
+      "learning_rate": 1.980392627360379e-05,
+      "loss": 768.9246,
+      "step": 3770
+    },
+    {
+      "epoch": 0.07269894846163832,
+      "grad_norm": 4182.971510822694,
+      "learning_rate": 1.9802721798848225e-05,
+      "loss": 779.4888,
+      "step": 3780
+    },
+    {
+      "epoch": 0.07289127372211884,
+      "grad_norm": 4067.8402051621556,
+      "learning_rate": 1.980151367274068e-05,
+      "loss": 757.6914,
+      "step": 3790
+    },
+    {
+      "epoch": 0.07308359898259938,
+      "grad_norm": 4435.119921254233,
+      "learning_rate": 1.9800301895731172e-05,
+      "loss": 775.5353,
+      "step": 3800
+    },
+    {
+      "epoch": 0.0732759242430799,
+      "grad_norm": 4169.998600865379,
+      "learning_rate": 1.9799086468271065e-05,
+      "loss": 761.8761,
+      "step": 3810
+    },
+    {
+      "epoch": 0.07346824950356041,
+      "grad_norm": 4250.334802552328,
+      "learning_rate": 1.9797867390813086e-05,
+      "loss": 751.6994,
+      "step": 3820
+    },
+    {
+      "epoch": 0.07366057476404095,
+      "grad_norm": 4161.694459374591,
+      "learning_rate": 1.9796644663811318e-05,
+      "loss": 760.986,
+      "step": 3830
+    },
+    {
+      "epoch": 0.07385290002452147,
+      "grad_norm": 4071.005938035517,
+      "learning_rate": 1.9795418287721215e-05,
+      "loss": 752.5014,
+      "step": 3840
+    },
+    {
+      "epoch": 0.074045225285002,
+      "grad_norm": 3983.5485763971133,
+      "learning_rate": 1.9794188262999574e-05,
+      "loss": 767.3445,
+      "step": 3850
+    },
+    {
+      "epoch": 0.07423755054548252,
+      "grad_norm": 4351.927589412771,
+      "learning_rate": 1.979295459010456e-05,
+      "loss": 752.1044,
+      "step": 3860
+    },
+    {
+      "epoch": 0.07442987580596304,
+      "grad_norm": 4003.554081765813,
+      "learning_rate": 1.9791717269495698e-05,
+      "loss": 735.3016,
+      "step": 3870
+    },
+    {
+      "epoch": 0.07462220106644357,
+      "grad_norm": 4080.9933243213095,
+      "learning_rate": 1.979047630163387e-05,
+      "loss": 740.3427,
+      "step": 3880
+    },
+    {
+      "epoch": 0.07481452632692409,
+      "grad_norm": 4056.4364237507216,
+      "learning_rate": 1.9789231686981313e-05,
+      "loss": 762.9863,
+      "step": 3890
+    },
+    {
+      "epoch": 0.07500685158740462,
+      "grad_norm": 3779.458008967918,
+      "learning_rate": 1.978798342600163e-05,
+      "loss": 739.0848,
+      "step": 3900
+    },
+    {
+      "epoch": 0.07519917684788514,
+      "grad_norm": 3966.5516719775974,
+      "learning_rate": 1.978673151915977e-05,
+      "loss": 748.0446,
+      "step": 3910
+    },
+    {
+      "epoch": 0.07539150210836566,
+      "grad_norm": 3864.9589702649387,
+      "learning_rate": 1.9785475966922055e-05,
+      "loss": 752.2604,
+      "step": 3920
+    },
+    {
+      "epoch": 0.0755838273688462,
+      "grad_norm": 4668.415651232959,
+      "learning_rate": 1.9784216769756156e-05,
+      "loss": 737.1826,
+      "step": 3930
+    },
+    {
+      "epoch": 0.07577615262932672,
+      "grad_norm": 4384.60448412883,
+      "learning_rate": 1.97829539281311e-05,
+      "loss": 731.0745,
+      "step": 3940
+    },
+    {
+      "epoch": 0.07596847788980725,
+      "grad_norm": 3601.994023223988,
+      "learning_rate": 1.9781687442517278e-05,
+      "loss": 736.3513,
+      "step": 3950
+    },
+    {
+      "epoch": 0.07616080315028777,
+      "grad_norm": 3951.141271493915,
+      "learning_rate": 1.9780417313386433e-05,
+      "loss": 750.8261,
+      "step": 3960
+    },
+    {
+      "epoch": 0.07635312841076829,
+      "grad_norm": 4020.710239625638,
+      "learning_rate": 1.9779143541211664e-05,
+      "loss": 733.3558,
+      "step": 3970
+    },
+    {
+      "epoch": 0.07654545367124882,
+      "grad_norm": 3711.5298841098925,
+      "learning_rate": 1.9777866126467436e-05,
+      "loss": 711.2791,
+      "step": 3980
+    },
+    {
+      "epoch": 0.07673777893172934,
+      "grad_norm": 3600.6900243194355,
+      "learning_rate": 1.9776585069629566e-05,
+      "loss": 725.6805,
+      "step": 3990
+    },
+    {
+      "epoch": 0.07693010419220987,
+      "grad_norm": 3791.0140196145912,
+      "learning_rate": 1.9775300371175225e-05,
+      "loss": 731.4292,
+      "step": 4000
+    },
+    {
+      "epoch": 0.07712242945269039,
+      "grad_norm": 6905.455573319279,
+      "learning_rate": 1.9774012031582935e-05,
+      "loss": 717.5897,
+      "step": 4010
+    },
+    {
+      "epoch": 0.07731475471317091,
+      "grad_norm": 4442.5228710944375,
+      "learning_rate": 1.9772720051332585e-05,
+      "loss": 735.1908,
+      "step": 4020
+    },
+    {
+      "epoch": 0.07750707997365144,
+      "grad_norm": 4432.83144236267,
+      "learning_rate": 1.977142443090542e-05,
+      "loss": 727.4589,
+      "step": 4030
+    },
+    {
+      "epoch": 0.07769940523413196,
+      "grad_norm": 4019.6817616779726,
+      "learning_rate": 1.9770125170784035e-05,
+      "loss": 729.0105,
+      "step": 4040
+    },
+    {
+      "epoch": 0.07789173049461248,
+      "grad_norm": 3970.844433192576,
+      "learning_rate": 1.9768822271452385e-05,
+      "loss": 736.296,
+      "step": 4050
+    },
+    {
+      "epoch": 0.07808405575509302,
+      "grad_norm": 4043.051322501368,
+      "learning_rate": 1.9767515733395774e-05,
+      "loss": 717.5287,
+      "step": 4060
+    },
+    {
+      "epoch": 0.07827638101557353,
+      "grad_norm": 6770.272247577486,
+      "learning_rate": 1.976620555710087e-05,
+      "loss": 725.2103,
+      "step": 4070
+    },
+    {
+      "epoch": 0.07846870627605407,
+      "grad_norm": 4234.052491760217,
+      "learning_rate": 1.976489174305569e-05,
+      "loss": 728.0525,
+      "step": 4080
+    },
+    {
+      "epoch": 0.07866103153653459,
+      "grad_norm": 3533.448407372867,
+      "learning_rate": 1.9763574291749603e-05,
+      "loss": 702.9978,
+      "step": 4090
+    },
+    {
+      "epoch": 0.0788533567970151,
+      "grad_norm": 3589.3971111893,
+      "learning_rate": 1.9762253203673348e-05,
+      "loss": 707.1979,
+      "step": 4100
+    },
+    {
+      "epoch": 0.07904568205749564,
+      "grad_norm": 3689.474090623924,
+      "learning_rate": 1.9760928479319003e-05,
+      "loss": 722.2437,
+      "step": 4110
+    },
+    {
+      "epoch": 0.07923800731797616,
+      "grad_norm": 4424.700623564766,
+      "learning_rate": 1.9759600119180005e-05,
+      "loss": 718.8557,
+      "step": 4120
+    },
+    {
+      "epoch": 0.07943033257845669,
+      "grad_norm": 3669.9009921334696,
+      "learning_rate": 1.975826812375115e-05,
+      "loss": 704.1591,
+      "step": 4130
+    },
+    {
+      "epoch": 0.07962265783893721,
+      "grad_norm": 4337.477909996418,
+      "learning_rate": 1.9756932493528583e-05,
+      "loss": 683.2335,
+      "step": 4140
+    },
+    {
+      "epoch": 0.07981498309941773,
+      "grad_norm": 5126.869874135973,
+      "learning_rate": 1.97555932290098e-05,
+      "loss": 764.6667,
+      "step": 4150
+    },
+    {
+      "epoch": 0.08000730835989826,
+      "grad_norm": 3561.896465127024,
+      "learning_rate": 1.9754250330693658e-05,
+      "loss": 708.6442,
+      "step": 4160
+    },
+    {
+      "epoch": 0.08019963362037878,
+      "grad_norm": 4053.5462185031943,
+      "learning_rate": 1.9752903799080366e-05,
+      "loss": 719.9713,
+      "step": 4170
+    },
+    {
+      "epoch": 0.08039195888085932,
+      "grad_norm": 3528.8884049513335,
+      "learning_rate": 1.9751553634671485e-05,
+      "loss": 683.4802,
+      "step": 4180
+    },
+    {
+      "epoch": 0.08058428414133983,
+      "grad_norm": 3373.3511327501215,
+      "learning_rate": 1.9750199837969922e-05,
+      "loss": 690.1696,
+      "step": 4190
+    },
+    {
+      "epoch": 0.08077660940182035,
+      "grad_norm": 3377.279025351652,
+      "learning_rate": 1.9748842409479953e-05,
+      "loss": 698.2177,
+      "step": 4200
+    },
+    {
+      "epoch": 0.08096893466230089,
+      "grad_norm": 3937.853000288384,
+      "learning_rate": 1.9747481349707197e-05,
+      "loss": 708.9531,
+      "step": 4210
+    },
+    {
+      "epoch": 0.0811612599227814,
+      "grad_norm": 3774.9521491888127,
+      "learning_rate": 1.9746116659158618e-05,
+      "loss": 695.156,
+      "step": 4220
+    },
+    {
+      "epoch": 0.08135358518326194,
+      "grad_norm": 3936.348895644883,
+      "learning_rate": 1.9744748338342546e-05,
+      "loss": 692.2083,
+      "step": 4230
+    },
+    {
+      "epoch": 0.08154591044374246,
+      "grad_norm": 4652.120795122975,
+      "learning_rate": 1.974337638776866e-05,
+      "loss": 697.4078,
+      "step": 4240
+    },
+    {
+      "epoch": 0.08173823570422298,
+      "grad_norm": 3915.7909486183266,
+      "learning_rate": 1.9742000807947986e-05,
+      "loss": 684.809,
+      "step": 4250
+    },
+    {
+      "epoch": 0.08193056096470351,
+      "grad_norm": 3965.057614153028,
+      "learning_rate": 1.9740621599392907e-05,
+      "loss": 683.939,
+      "step": 4260
+    },
+    {
+      "epoch": 0.08212288622518403,
+      "grad_norm": 3550.8897728467455,
+      "learning_rate": 1.9739238762617155e-05,
+      "loss": 682.1476,
+      "step": 4270
+    },
+    {
+      "epoch": 0.08231521148566455,
+      "grad_norm": 3286.498018166622,
+      "learning_rate": 1.973785229813581e-05,
+      "loss": 696.0196,
+      "step": 4280
+    },
+    {
+      "epoch": 0.08250753674614508,
+      "grad_norm": 3933.2518693470192,
+      "learning_rate": 1.973646220646531e-05,
+      "loss": 700.2589,
+      "step": 4290
+    },
+    {
+      "epoch": 0.0826998620066256,
+      "grad_norm": 3547.496296922659,
+      "learning_rate": 1.973506848812344e-05,
+      "loss": 682.1439,
+      "step": 4300
+    },
+    {
+      "epoch": 0.08289218726710613,
+      "grad_norm": 3681.139683691825,
+      "learning_rate": 1.9733671143629342e-05,
+      "loss": 715.1272,
+      "step": 4310
+    },
+    {
+      "epoch": 0.08308451252758665,
+      "grad_norm": 3406.1751920767338,
+      "learning_rate": 1.9732270173503493e-05,
+      "loss": 687.8295,
+      "step": 4320
+    },
+    {
+      "epoch": 0.08327683778806717,
+      "grad_norm": 4216.349904376469,
+      "learning_rate": 1.9730865578267745e-05,
+      "loss": 677.5665,
+      "step": 4330
+    },
+    {
+      "epoch": 0.0834691630485477,
+      "grad_norm": 3599.608471801573,
+      "learning_rate": 1.972945735844528e-05,
+      "loss": 688.3549,
+      "step": 4340
+    },
+    {
+      "epoch": 0.08366148830902823,
+      "grad_norm": 3780.9374100271048,
+      "learning_rate": 1.972804551456063e-05,
+      "loss": 690.7379,
+      "step": 4350
+    },
+    {
+      "epoch": 0.08385381356950876,
+      "grad_norm": 3928.9846954940863,
+      "learning_rate": 1.9726630047139695e-05,
+      "loss": 689.8347,
+      "step": 4360
+    },
+    {
+      "epoch": 0.08404613882998928,
+      "grad_norm": 3504.6145494707266,
+      "learning_rate": 1.9725210956709707e-05,
+      "loss": 667.2354,
+      "step": 4370
+    },
+    {
+      "epoch": 0.0842384640904698,
+      "grad_norm": 3819.37645549846,
+      "learning_rate": 1.9723788243799253e-05,
+      "loss": 695.1894,
+      "step": 4380
+    },
+    {
+      "epoch": 0.08443078935095033,
+      "grad_norm": 7974.730780145998,
+      "learning_rate": 1.972236190893827e-05,
+      "loss": 662.5851,
+      "step": 4390
+    },
+    {
+      "epoch": 0.08462311461143085,
+      "grad_norm": 4913.461322493489,
+      "learning_rate": 1.972093195265805e-05,
+      "loss": 696.4384,
+      "step": 4400
+    },
+    {
+      "epoch": 0.08481543987191138,
+      "grad_norm": 5453.674250298887,
+      "learning_rate": 1.9719498375491224e-05,
+      "loss": 678.2224,
+      "step": 4410
+    },
+    {
+      "epoch": 0.0850077651323919,
+      "grad_norm": 4527.548679595546,
+      "learning_rate": 1.9718061177971777e-05,
+      "loss": 681.2826,
+      "step": 4420
+    },
+    {
+      "epoch": 0.08520009039287242,
+      "grad_norm": 3514.229150281252,
+      "learning_rate": 1.9716620360635036e-05,
+      "loss": 684.7965,
+      "step": 4430
+    },
+    {
+      "epoch": 0.08539241565335295,
+      "grad_norm": 3813.8794069158303,
+      "learning_rate": 1.971517592401769e-05,
+      "loss": 689.1514,
+      "step": 4440
+    },
+    {
+      "epoch": 0.08558474091383347,
+      "grad_norm": 3379.7322737514255,
+      "learning_rate": 1.9713727868657764e-05,
+      "loss": 660.6949,
+      "step": 4450
+    },
+    {
+      "epoch": 0.085777066174314,
+      "grad_norm": 3888.111140438278,
+      "learning_rate": 1.971227619509463e-05,
+      "loss": 656.5462,
+      "step": 4460
+    },
+    {
+      "epoch": 0.08596939143479453,
+      "grad_norm": 3449.1515907494745,
+      "learning_rate": 1.971082090386902e-05,
+      "loss": 686.3935,
+      "step": 4470
+    },
+    {
+      "epoch": 0.08616171669527505,
+      "grad_norm": 4084.203049728152,
+      "learning_rate": 1.9709361995523e-05,
+      "loss": 668.6304,
+      "step": 4480
+    },
+    {
+      "epoch": 0.08635404195575558,
+      "grad_norm": 4052.6042528137004,
+      "learning_rate": 1.9707899470599998e-05,
+      "loss": 696.8607,
+      "step": 4490
+    },
+    {
+      "epoch": 0.0865463672162361,
+      "grad_norm": 3839.1314575799543,
+      "learning_rate": 1.970643332964477e-05,
+      "loss": 663.3869,
+      "step": 4500
+    },
+    {
+      "epoch": 0.08673869247671662,
+      "grad_norm": 3777.0372160906877,
+      "learning_rate": 1.9704963573203435e-05,
+      "loss": 662.3137,
+      "step": 4510
+    },
+    {
+      "epoch": 0.08693101773719715,
+      "grad_norm": 4876.160137903215,
+      "learning_rate": 1.970349020182345e-05,
+      "loss": 688.069,
+      "step": 4520
+    },
+    {
+      "epoch": 0.08712334299767767,
+      "grad_norm": 4395.015874331804,
+      "learning_rate": 1.9702013216053623e-05,
+      "loss": 653.012,
+      "step": 4530
+    },
+    {
+      "epoch": 0.0873156682581582,
+      "grad_norm": 3832.486878076765,
+      "learning_rate": 1.9700532616444114e-05,
+      "loss": 676.1902,
+      "step": 4540
+    },
+    {
+      "epoch": 0.08750799351863872,
+      "grad_norm": 4540.329496125654,
+      "learning_rate": 1.969904840354641e-05,
+      "loss": 681.3762,
+      "step": 4550
+    },
+    {
+      "epoch": 0.08770031877911924,
+      "grad_norm": 5575.001699947538,
+      "learning_rate": 1.9697560577913358e-05,
+      "loss": 713.7773,
+      "step": 4560
+    },
+    {
+      "epoch": 0.08789264403959977,
+      "grad_norm": 4208.441348510918,
+      "learning_rate": 1.9696069140099152e-05,
+      "loss": 670.8771,
+      "step": 4570
+    },
+    {
+      "epoch": 0.08808496930008029,
+      "grad_norm": 3979.707409534953,
+      "learning_rate": 1.969457409065933e-05,
+      "loss": 661.7551,
+      "step": 4580
+    },
+    {
+      "epoch": 0.08827729456056083,
+      "grad_norm": 3554.7321912639827,
+      "learning_rate": 1.969307543015077e-05,
+      "loss": 654.7043,
+      "step": 4590
+    },
+    {
+      "epoch": 0.08846961982104135,
+      "grad_norm": 3829.398406063814,
+      "learning_rate": 1.9691573159131696e-05,
+      "loss": 662.6666,
+      "step": 4600
+    },
+    {
+      "epoch": 0.08866194508152186,
+      "grad_norm": 3498.147132863079,
+      "learning_rate": 1.9690067278161686e-05,
+      "loss": 663.5632,
+      "step": 4610
+    },
+    {
+      "epoch": 0.0888542703420024,
+      "grad_norm": 3414.3920898290144,
+      "learning_rate": 1.9688557787801647e-05,
+      "loss": 665.1435,
+      "step": 4620
+    },
+    {
+      "epoch": 0.08904659560248292,
+      "grad_norm": 3687.0786155832902,
+      "learning_rate": 1.968704468861385e-05,
+      "loss": 668.0722,
+      "step": 4630
+    },
+    {
+      "epoch": 0.08923892086296345,
+      "grad_norm": 3221.0008705613354,
+      "learning_rate": 1.968552798116189e-05,
+      "loss": 640.526,
+      "step": 4640
+    },
+    {
+      "epoch": 0.08943124612344397,
+      "grad_norm": 3774.429813675378,
+      "learning_rate": 1.9684007666010716e-05,
+      "loss": 649.5702,
+      "step": 4650
+    },
+    {
+      "epoch": 0.08962357138392449,
+      "grad_norm": 3967.5309119278195,
+      "learning_rate": 1.9682483743726624e-05,
+      "loss": 679.9959,
+      "step": 4660
+    },
+    {
+      "epoch": 0.08981589664440502,
+      "grad_norm": 3632.2186772085734,
+      "learning_rate": 1.968095621487725e-05,
+      "loss": 658.3854,
+      "step": 4670
+    },
+    {
+      "epoch": 0.09000822190488554,
+      "grad_norm": 3596.7121090116234,
+      "learning_rate": 1.9679425080031574e-05,
+      "loss": 643.3561,
+      "step": 4680
+    },
+    {
+      "epoch": 0.09020054716536607,
+      "grad_norm": 3532.2609192610566,
+      "learning_rate": 1.9677890339759914e-05,
+      "loss": 664.0748,
+      "step": 4690
+    },
+    {
+      "epoch": 0.0903928724258466,
+      "grad_norm": 3681.385507188368,
+      "learning_rate": 1.967635199463394e-05,
+      "loss": 639.0781,
+      "step": 4700
+    },
+    {
+      "epoch": 0.09058519768632711,
+      "grad_norm": 3528.3784190038627,
+      "learning_rate": 1.9674810045226658e-05,
+      "loss": 647.9231,
+      "step": 4710
+    },
+    {
+      "epoch": 0.09077752294680765,
+      "grad_norm": 5344.490306861034,
+      "learning_rate": 1.967326449211242e-05,
+      "loss": 650.9944,
+      "step": 4720
+    },
+    {
+      "epoch": 0.09096984820728816,
+      "grad_norm": 3538.6462529986425,
+      "learning_rate": 1.9671715335866915e-05,
+      "loss": 647.9663,
+      "step": 4730
+    },
+    {
+      "epoch": 0.09116217346776868,
+      "grad_norm": 4344.12153977402,
+      "learning_rate": 1.9670162577067182e-05,
+      "loss": 672.18,
+      "step": 4740
+    },
+    {
+      "epoch": 0.09135449872824922,
+      "grad_norm": 4036.565719761374,
+      "learning_rate": 1.9668606216291598e-05,
+      "loss": 647.3205,
+      "step": 4750
+    },
+    {
+      "epoch": 0.09154682398872974,
+      "grad_norm": 3893.713175141919,
+      "learning_rate": 1.9667046254119878e-05,
+      "loss": 630.2145,
+      "step": 4760
+    },
+    {
+      "epoch": 0.09173914924921027,
+      "grad_norm": 3463.037515331363,
+      "learning_rate": 1.966548269113309e-05,
+      "loss": 644.2297,
+      "step": 4770
+    },
+    {
+      "epoch": 0.09193147450969079,
+      "grad_norm": 3854.619786972281,
+      "learning_rate": 1.9663915527913628e-05,
+      "loss": 646.2749,
+      "step": 4780
+    },
+    {
+      "epoch": 0.09212379977017131,
+      "grad_norm": 3486.178830596917,
+      "learning_rate": 1.9662344765045237e-05,
+      "loss": 652.4989,
+      "step": 4790
+    },
+    {
+      "epoch": 0.09231612503065184,
+      "grad_norm": 3496.403710982168,
+      "learning_rate": 1.9660770403112996e-05,
+      "loss": 638.6127,
+      "step": 4800
+    },
+    {
+      "epoch": 0.09250845029113236,
+      "grad_norm": 3473.9121597119747,
+      "learning_rate": 1.9659192442703336e-05,
+      "loss": 642.948,
+      "step": 4810
+    },
+    {
+      "epoch": 0.0927007755516129,
+      "grad_norm": 5673.242669785963,
+      "learning_rate": 1.965761088440402e-05,
+      "loss": 653.6242,
+      "step": 4820
+    },
+    {
+      "epoch": 0.09289310081209341,
+      "grad_norm": 3597.424668915692,
+      "learning_rate": 1.9656025728804147e-05,
+      "loss": 648.2441,
+      "step": 4830
+    },
+    {
+      "epoch": 0.09308542607257393,
+      "grad_norm": 3240.3352994047696,
+      "learning_rate": 1.9654436976494165e-05,
+      "loss": 630.7824,
+      "step": 4840
+    },
+    {
+      "epoch": 0.09327775133305446,
+      "grad_norm": 3953.519280020802,
+      "learning_rate": 1.9652844628065857e-05,
+      "loss": 624.2006,
+      "step": 4850
+    },
+    {
+      "epoch": 0.09347007659353498,
+      "grad_norm": 3325.9017233701265,
+      "learning_rate": 1.965124868411235e-05,
+      "loss": 643.9571,
+      "step": 4860
+    },
+    {
+      "epoch": 0.09366240185401552,
+      "grad_norm": 3811.625234994872,
+      "learning_rate": 1.96496491452281e-05,
+      "loss": 625.1706,
+      "step": 4870
+    },
+    {
+      "epoch": 0.09385472711449604,
+      "grad_norm": 3540.3198944758788,
+      "learning_rate": 1.9648046012008916e-05,
+      "loss": 618.9432,
+      "step": 4880
+    },
+    {
+      "epoch": 0.09404705237497656,
+      "grad_norm": 3787.62609735396,
+      "learning_rate": 1.9646439285051936e-05,
+      "loss": 634.823,
+      "step": 4890
+    },
+    {
+      "epoch": 0.09423937763545709,
+      "grad_norm": 8037.970807135463,
+      "learning_rate": 1.9644828964955633e-05,
+      "loss": 638.2597,
+      "step": 4900
+    },
+    {
+      "epoch": 0.09443170289593761,
+      "grad_norm": 3236.2324896471923,
+      "learning_rate": 1.9643215052319836e-05,
+      "loss": 645.7353,
+      "step": 4910
+    },
+    {
+      "epoch": 0.09462402815641814,
+      "grad_norm": 4459.467303436796,
+      "learning_rate": 1.9641597547745694e-05,
+      "loss": 655.9531,
+      "step": 4920
+    },
+    {
+      "epoch": 0.09481635341689866,
+      "grad_norm": 3528.380784037965,
+      "learning_rate": 1.9639976451835698e-05,
+      "loss": 643.4895,
+      "step": 4930
+    },
+    {
+      "epoch": 0.09500867867737918,
+      "grad_norm": 3544.20401318225,
+      "learning_rate": 1.9638351765193685e-05,
+      "loss": 632.3099,
+      "step": 4940
+    },
+    {
+      "epoch": 0.09520100393785971,
+      "grad_norm": 3482.587986622822,
+      "learning_rate": 1.9636723488424823e-05,
+      "loss": 605.5317,
+      "step": 4950
+    },
+    {
+      "epoch": 0.09539332919834023,
+      "grad_norm": 3616.758129257622,
+      "learning_rate": 1.9635091622135616e-05,
+      "loss": 628.818,
+      "step": 4960
+    },
+    {
+      "epoch": 0.09558565445882075,
+      "grad_norm": 3540.071237234814,
+      "learning_rate": 1.963345616693391e-05,
+      "loss": 617.1661,
+      "step": 4970
+    },
+    {
+      "epoch": 0.09577797971930128,
+      "grad_norm": 3479.171243388493,
+      "learning_rate": 1.9631817123428883e-05,
+      "loss": 628.7106,
+      "step": 4980
+    },
+    {
+      "epoch": 0.0959703049797818,
+      "grad_norm": 3748.258658698097,
+      "learning_rate": 1.9630174492231052e-05,
+      "loss": 619.0021,
+      "step": 4990
+    },
+    {
+      "epoch": 0.09616263024026234,
+      "grad_norm": 3693.4855798077833,
+      "learning_rate": 1.962852827395227e-05,
+      "loss": 643.9448,
+      "step": 5000
+    },
+    {
+      "epoch": 0.09635495550074286,
+      "grad_norm": 3382.9156635765858,
+      "learning_rate": 1.962687846920573e-05,
+      "loss": 616.662,
+      "step": 5010
+    },
+    {
+      "epoch": 0.09654728076122338,
+      "grad_norm": 3460.71630735885,
+      "learning_rate": 1.9625225078605946e-05,
+      "loss": 616.0104,
+      "step": 5020
+    },
+    {
+      "epoch": 0.09673960602170391,
+      "grad_norm": 3282.016150396431,
+      "learning_rate": 1.9623568102768792e-05,
+      "loss": 618.4474,
+      "step": 5030
+    },
+    {
+      "epoch": 0.09693193128218443,
+      "grad_norm": 3798.3562679617376,
+      "learning_rate": 1.9621907542311457e-05,
+      "loss": 623.6288,
+      "step": 5040
+    },
+    {
+      "epoch": 0.09712425654266496,
+      "grad_norm": 3589.7586622669473,
+      "learning_rate": 1.9620243397852473e-05,
+      "loss": 595.6706,
+      "step": 5050
+    },
+    {
+      "epoch": 0.09731658180314548,
+      "grad_norm": 3335.156327211705,
+      "learning_rate": 1.9618575670011705e-05,
+      "loss": 597.4271,
+      "step": 5060
+    },
+    {
+      "epoch": 0.097508907063626,
+      "grad_norm": 4023.356685481532,
+      "learning_rate": 1.9616904359410357e-05,
+      "loss": 605.0412,
+      "step": 5070
+    },
+    {
+      "epoch": 0.09770123232410653,
+      "grad_norm": 3004.0876038850347,
+      "learning_rate": 1.9615229466670963e-05,
+      "loss": 615.485,
+      "step": 5080
+    },
+    {
+      "epoch": 0.09789355758458705,
+      "grad_norm": 3945.3817207057436,
+      "learning_rate": 1.9613550992417396e-05,
+      "loss": 610.2364,
+      "step": 5090
+    },
+    {
+      "epoch": 0.09808588284506758,
+      "grad_norm": 3604.836739141922,
+      "learning_rate": 1.961186893727486e-05,
+      "loss": 612.9978,
+      "step": 5100
+    },
+    {
+      "epoch": 0.0982782081055481,
+      "grad_norm": 3899.040526298082,
+      "learning_rate": 1.9610183301869882e-05,
+      "loss": 593.7491,
+      "step": 5110
+    },
+    {
+      "epoch": 0.09847053336602862,
+      "grad_norm": 3715.6778759239937,
+      "learning_rate": 1.9608494086830348e-05,
+      "loss": 599.9813,
+      "step": 5120
+    },
+    {
+      "epoch": 0.09866285862650916,
+      "grad_norm": 3599.5731398702455,
+      "learning_rate": 1.9606801292785452e-05,
+      "loss": 608.794,
+      "step": 5130
+    },
+    {
+      "epoch": 0.09885518388698968,
+      "grad_norm": 4871.124936102754,
+      "learning_rate": 1.960510492036574e-05,
+      "loss": 610.1908,
+      "step": 5140
+    },
+    {
+      "epoch": 0.09904750914747021,
+      "grad_norm": 3476.5757955014783,
+      "learning_rate": 1.9603404970203078e-05,
+      "loss": 618.4548,
+      "step": 5150
+    },
+    {
+      "epoch": 0.09923983440795073,
+      "grad_norm": 3663.9674310657874,
+      "learning_rate": 1.9601701442930667e-05,
+      "loss": 627.1068,
+      "step": 5160
+    },
+    {
+      "epoch": 0.09943215966843125,
+      "grad_norm": 3325.530184562212,
+      "learning_rate": 1.9599994339183047e-05,
+      "loss": 615.4588,
+      "step": 5170
+    },
+    {
+      "epoch": 0.09962448492891178,
+      "grad_norm": 3582.445627097957,
+      "learning_rate": 1.9598283659596084e-05,
+      "loss": 598.4591,
+      "step": 5180
+    },
+    {
+      "epoch": 0.0998168101893923,
+      "grad_norm": 3783.664326256545,
+      "learning_rate": 1.9596569404806983e-05,
+      "loss": 597.5333,
+      "step": 5190
+    },
+    {
+      "epoch": 0.10000913544987282,
+      "grad_norm": 3347.1457913067775,
+      "learning_rate": 1.9594851575454266e-05,
+      "loss": 596.7927,
+      "step": 5200
+    },
+    {
+      "epoch": 0.10020146071035335,
+      "grad_norm": 3355.4591492034247,
+      "learning_rate": 1.9593130172177806e-05,
+      "loss": 622.0267,
+      "step": 5210
+    },
+    {
+      "epoch": 0.10039378597083387,
+      "grad_norm": 3391.3074568265647,
+      "learning_rate": 1.959140519561879e-05,
+      "loss": 610.5841,
+      "step": 5220
+    },
+    {
+      "epoch": 0.1005861112313144,
+      "grad_norm": 3483.8305438194234,
+      "learning_rate": 1.9589676646419744e-05,
+      "loss": 606.7154,
+      "step": 5230
+    },
+    {
+      "epoch": 0.10077843649179492,
+      "grad_norm": 3343.385466805086,
+      "learning_rate": 1.958794452522453e-05,
+      "loss": 611.2106,
+      "step": 5240
+    },
+    {
+      "epoch": 0.10097076175227544,
+      "grad_norm": 3245.0048993151086,
+      "learning_rate": 1.9586208832678328e-05,
+      "loss": 592.585,
+      "step": 5250
+    },
+    {
+      "epoch": 0.10116308701275598,
+      "grad_norm": 3345.2030475594847,
+      "learning_rate": 1.958446956942766e-05,
+      "loss": 605.1469,
+      "step": 5260
+    },
+    {
+      "epoch": 0.1013554122732365,
+      "grad_norm": 3533.133322503176,
+      "learning_rate": 1.9582726736120365e-05,
+      "loss": 596.0807,
+      "step": 5270
+    },
+    {
+      "epoch": 0.10154773753371703,
+      "grad_norm": 3489.0660001743527,
+      "learning_rate": 1.958098033340563e-05,
+      "loss": 598.1188,
+      "step": 5280
+    },
+    {
+      "epoch": 0.10174006279419755,
+      "grad_norm": 4186.174151433306,
+      "learning_rate": 1.9579230361933952e-05,
+      "loss": 618.7304,
+      "step": 5290
+    },
+    {
+      "epoch": 0.10193238805467807,
+      "grad_norm": 3583.333919258522,
+      "learning_rate": 1.9577476822357174e-05,
+      "loss": 587.8064,
+      "step": 5300
+    },
+    {
+      "epoch": 0.1021247133151586,
+      "grad_norm": 3086.966666664932,
+      "learning_rate": 1.9575719715328457e-05,
+      "loss": 609.2442,
+      "step": 5310
+    },
+    {
+      "epoch": 0.10231703857563912,
+      "grad_norm": 3363.5775793498015,
+      "learning_rate": 1.957395904150229e-05,
+      "loss": 586.4506,
+      "step": 5320
+    },
+    {
+      "epoch": 0.10250936383611965,
+      "grad_norm": 3694.9590168318045,
+      "learning_rate": 1.9572194801534504e-05,
+      "loss": 605.8757,
+      "step": 5330
+    },
+    {
+      "epoch": 0.10270168909660017,
+      "grad_norm": 3332.6618107292793,
+      "learning_rate": 1.9570426996082238e-05,
+      "loss": 599.57,
+      "step": 5340
+    },
+    {
+      "epoch": 0.10289401435708069,
+      "grad_norm": 3317.555584286339,
+      "learning_rate": 1.9568655625803982e-05,
+      "loss": 584.7927,
+      "step": 5350
+    },
+    {
+      "epoch": 0.10308633961756122,
+      "grad_norm": 3586.203961575145,
+      "learning_rate": 1.956688069135954e-05,
+      "loss": 595.3776,
+      "step": 5360
+    },
+    {
+      "epoch": 0.10327866487804174,
+      "grad_norm": 3101.3052556150315,
+      "learning_rate": 1.9565102193410035e-05,
+      "loss": 585.8616,
+      "step": 5370
+    },
+    {
+      "epoch": 0.10347099013852228,
+      "grad_norm": 3142.8410306050705,
+      "learning_rate": 1.956332013261794e-05,
+      "loss": 590.7093,
+      "step": 5380
+    },
+    {
+      "epoch": 0.1036633153990028,
+      "grad_norm": 3237.828944361127,
+      "learning_rate": 1.9561534509647038e-05,
+      "loss": 574.7191,
+      "step": 5390
+    },
+    {
+      "epoch": 0.10385564065948331,
+      "grad_norm": 3487.212973888483,
+      "learning_rate": 1.9559745325162445e-05,
+      "loss": 599.8927,
+      "step": 5400
+    },
+    {
+      "epoch": 0.10404796591996385,
+      "grad_norm": 3320.812445371658,
+      "learning_rate": 1.9557952579830604e-05,
+      "loss": 591.8003,
+      "step": 5410
+    },
+    {
+      "epoch": 0.10424029118044437,
+      "grad_norm": 3088.4325268442735,
+      "learning_rate": 1.955615627431928e-05,
+      "loss": 589.8753,
+      "step": 5420
+    },
+    {
+      "epoch": 0.10443261644092489,
+      "grad_norm": 3059.9743898672727,
+      "learning_rate": 1.955435640929757e-05,
+      "loss": 574.5125,
+      "step": 5430
+    },
+    {
+      "epoch": 0.10462494170140542,
+      "grad_norm": 3230.902278524589,
+      "learning_rate": 1.9552552985435893e-05,
+      "loss": 574.9331,
+      "step": 5440
+    },
+    {
+      "epoch": 0.10481726696188594,
+      "grad_norm": 3112.9865505029543,
+      "learning_rate": 1.9550746003405996e-05,
+      "loss": 581.6879,
+      "step": 5450
+    },
+    {
+      "epoch": 0.10500959222236647,
+      "grad_norm": 3604.150746681098,
+      "learning_rate": 1.9548935463880945e-05,
+      "loss": 590.3337,
+      "step": 5460
+    },
+    {
+      "epoch": 0.10520191748284699,
+      "grad_norm": 3835.5960064615783,
+      "learning_rate": 1.9547121367535143e-05,
+      "loss": 599.1407,
+      "step": 5470
+    },
+    {
+      "epoch": 0.10539424274332751,
+      "grad_norm": 3017.553777298891,
+      "learning_rate": 1.9545303715044305e-05,
+      "loss": 592.7393,
+      "step": 5480
+    },
+    {
+      "epoch": 0.10558656800380804,
+      "grad_norm": 3633.178854381708,
+      "learning_rate": 1.9543482507085484e-05,
+      "loss": 587.8829,
+      "step": 5490
+    },
+    {
+      "epoch": 0.10577889326428856,
+      "grad_norm": 3410.959353056924,
+      "learning_rate": 1.9541657744337038e-05,
+      "loss": 577.7798,
+      "step": 5500
+    },
+    {
+      "epoch": 0.1059712185247691,
+      "grad_norm": 3288.9087004370126,
+      "learning_rate": 1.9539829427478675e-05,
+      "loss": 593.8099,
+      "step": 5510
+    },
+    {
+      "epoch": 0.10616354378524961,
+      "grad_norm": 3389.1495245639303,
+      "learning_rate": 1.95379975571914e-05,
+      "loss": 599.0841,
+      "step": 5520
+    },
+    {
+      "epoch": 0.10635586904573013,
+      "grad_norm": 3408.571500066902,
+      "learning_rate": 1.953616213415756e-05,
+      "loss": 579.6134,
+      "step": 5530
+    },
+    {
+      "epoch": 0.10654819430621067,
+      "grad_norm": 3330.4242393144054,
+      "learning_rate": 1.9534323159060824e-05,
+      "loss": 597.6208,
+      "step": 5540
+    },
+    {
+      "epoch": 0.10674051956669119,
+      "grad_norm": 3452.7173047969713,
+      "learning_rate": 1.9532480632586175e-05,
+      "loss": 587.3513,
+      "step": 5550
+    },
+    {
+      "epoch": 0.10693284482717172,
+      "grad_norm": 3222.0340313059646,
+      "learning_rate": 1.953063455541992e-05,
+      "loss": 593.5667,
+      "step": 5560
+    },
+    {
+      "epoch": 0.10712517008765224,
+      "grad_norm": 3427.3239416696665,
+      "learning_rate": 1.9528784928249703e-05,
+      "loss": 577.5851,
+      "step": 5570
+    },
+    {
+      "epoch": 0.10731749534813276,
+      "grad_norm": 3464.050210020691,
+      "learning_rate": 1.9526931751764467e-05,
+      "loss": 568.4066,
+      "step": 5580
+    },
+    {
+      "epoch": 0.10750982060861329,
+      "grad_norm": 3356.0561712492536,
+      "learning_rate": 1.95250750266545e-05,
+      "loss": 577.8563,
+      "step": 5590
+    },
+    {
+      "epoch": 0.10770214586909381,
+      "grad_norm": 3256.3685934272603,
+      "learning_rate": 1.9523214753611398e-05,
+      "loss": 565.334,
+      "step": 5600
+    },
+    {
+      "epoch": 0.10789447112957434,
+      "grad_norm": 3156.9367460709013,
+      "learning_rate": 1.952135093332808e-05,
+      "loss": 590.5764,
+      "step": 5610
+    },
+    {
+      "epoch": 0.10808679639005486,
+      "grad_norm": 2982.937853922636,
+      "learning_rate": 1.9519483566498788e-05,
+      "loss": 574.5607,
+      "step": 5620
+    },
+    {
+      "epoch": 0.10827912165053538,
+      "grad_norm": 3318.9312930539973,
+      "learning_rate": 1.9517612653819088e-05,
+      "loss": 576.3303,
+      "step": 5630
+    },
+    {
+      "epoch": 0.10847144691101591,
+      "grad_norm": 3292.9260569229145,
+      "learning_rate": 1.9515738195985868e-05,
+      "loss": 597.8799,
+      "step": 5640
+    },
+    {
+      "epoch": 0.10866377217149643,
+      "grad_norm": 3249.3652138353023,
+      "learning_rate": 1.951386019369732e-05,
+      "loss": 566.7925,
+      "step": 5650
+    },
+    {
+      "epoch": 0.10885609743197695,
+      "grad_norm": 2919.023618232392,
+      "learning_rate": 1.9511978647652984e-05,
+      "loss": 566.1437,
+      "step": 5660
+    },
+    {
+      "epoch": 0.10904842269245749,
+      "grad_norm": 3276.7094214119365,
+      "learning_rate": 1.9510093558553687e-05,
+      "loss": 574.3124,
+      "step": 5670
+    },
+    {
+      "epoch": 0.109240747952938,
+      "grad_norm": 2857.4942428475388,
+      "learning_rate": 1.950820492710161e-05,
+      "loss": 574.6187,
+      "step": 5680
+    },
+    {
+      "epoch": 0.10943307321341854,
+      "grad_norm": 3223.1164687825885,
+      "learning_rate": 1.9506312754000235e-05,
+      "loss": 572.9128,
+      "step": 5690
+    },
+    {
+      "epoch": 0.10962539847389906,
+      "grad_norm": 3911.5760853073984,
+      "learning_rate": 1.9504417039954357e-05,
+      "loss": 564.8175,
+      "step": 5700
+    },
+    {
+      "epoch": 0.10981772373437958,
+      "grad_norm": 3277.1211744842003,
+      "learning_rate": 1.9502517785670098e-05,
+      "loss": 583.1017,
+      "step": 5710
+    },
+    {
+      "epoch": 0.11001004899486011,
+      "grad_norm": 3365.049538501484,
+      "learning_rate": 1.950061499185491e-05,
+      "loss": 582.307,
+      "step": 5720
+    },
+    {
+      "epoch": 0.11020237425534063,
+      "grad_norm": 2752.615269421733,
+      "learning_rate": 1.9498708659217542e-05,
+      "loss": 567.6096,
+      "step": 5730
+    },
+    {
+      "epoch": 0.11039469951582116,
+      "grad_norm": 3272.294026166464,
+      "learning_rate": 1.9496798788468077e-05,
+      "loss": 556.4454,
+      "step": 5740
+    },
+    {
+      "epoch": 0.11058702477630168,
+      "grad_norm": 3162.9779601410396,
+      "learning_rate": 1.9494885380317906e-05,
+      "loss": 561.9178,
+      "step": 5750
+    },
+    {
+      "epoch": 0.1107793500367822,
+      "grad_norm": 3087.8642492405775,
+      "learning_rate": 1.9492968435479744e-05,
+      "loss": 558.3578,
+      "step": 5760
+    },
+    {
+      "epoch": 0.11097167529726273,
+      "grad_norm": 3005.3395772840045,
+      "learning_rate": 1.949104795466762e-05,
+      "loss": 560.4387,
+      "step": 5770
+    },
+    {
+      "epoch": 0.11116400055774325,
+      "grad_norm": 3141.6960210526568,
+      "learning_rate": 1.9489123938596886e-05,
+      "loss": 547.4087,
+      "step": 5780
+    },
+    {
+      "epoch": 0.11135632581822379,
+      "grad_norm": 3268.057652349369,
+      "learning_rate": 1.94871963879842e-05,
+      "loss": 566.6952,
+      "step": 5790
+    },
+    {
+      "epoch": 0.1115486510787043,
+      "grad_norm": 3083.8145921229366,
+      "learning_rate": 1.9485265303547547e-05,
+      "loss": 555.4364,
+      "step": 5800
+    },
+    {
+      "epoch": 0.11174097633918482,
+      "grad_norm": 3448.086082615801,
+      "learning_rate": 1.9483330686006223e-05,
+      "loss": 566.9604,
+      "step": 5810
+    },
+    {
+      "epoch": 0.11193330159966536,
+      "grad_norm": 3247.335309053214,
+      "learning_rate": 1.948139253608084e-05,
+      "loss": 594.0994,
+      "step": 5820
+    },
+    {
+      "epoch": 0.11212562686014588,
+      "grad_norm": 3568.8222684517814,
+      "learning_rate": 1.9479450854493327e-05,
+      "loss": 558.641,
+      "step": 5830
+    },
+    {
+      "epoch": 0.1123179521206264,
+      "grad_norm": 3145.708312082184,
+      "learning_rate": 1.9477505641966933e-05,
+      "loss": 552.1804,
+      "step": 5840
+    },
+    {
+      "epoch": 0.11251027738110693,
+      "grad_norm": 3129.4430725024577,
+      "learning_rate": 1.9475556899226213e-05,
+      "loss": 558.4812,
+      "step": 5850
+    },
+    {
+      "epoch": 0.11270260264158745,
+      "grad_norm": 3558.355236191525,
+      "learning_rate": 1.9473604626997037e-05,
+      "loss": 550.7903,
+      "step": 5860
+    },
+    {
+      "epoch": 0.11289492790206798,
+      "grad_norm": 3372.746032867043,
+      "learning_rate": 1.94716488260066e-05,
+      "loss": 564.3837,
+      "step": 5870
+    },
+    {
+      "epoch": 0.1130872531625485,
+      "grad_norm": 2857.0565136248692,
+      "learning_rate": 1.946968949698341e-05,
+      "loss": 556.8066,
+      "step": 5880
+    },
+    {
+      "epoch": 0.11327957842302902,
+      "grad_norm": 3096.655666199947,
+      "learning_rate": 1.9467726640657277e-05,
+      "loss": 547.5038,
+      "step": 5890
+    },
+    {
+      "epoch": 0.11347190368350955,
+      "grad_norm": 3851.2534685833766,
+      "learning_rate": 1.9465760257759336e-05,
+      "loss": 553.2814,
+      "step": 5900
+    },
+    {
+      "epoch": 0.11366422894399007,
+      "grad_norm": 2864.5611039769974,
+      "learning_rate": 1.9463790349022027e-05,
+      "loss": 541.3181,
+      "step": 5910
+    },
+    {
+      "epoch": 0.1138565542044706,
+      "grad_norm": 3540.314022208594,
+      "learning_rate": 1.9461816915179117e-05,
+      "loss": 555.9016,
+      "step": 5920
+    },
+    {
+      "epoch": 0.11404887946495113,
+      "grad_norm": 2959.323534297358,
+      "learning_rate": 1.945983995696567e-05,
+      "loss": 560.4891,
+      "step": 5930
+    },
+    {
+      "epoch": 0.11424120472543164,
+      "grad_norm": 3245.220000408806,
+      "learning_rate": 1.9457859475118077e-05,
+      "loss": 551.865,
+      "step": 5940
+    },
+    {
+      "epoch": 0.11443352998591218,
+      "grad_norm": 3259.892769071125,
+      "learning_rate": 1.9455875470374027e-05,
+      "loss": 572.4262,
+      "step": 5950
+    },
+    {
+      "epoch": 0.1146258552463927,
+      "grad_norm": 3186.270876043077,
+      "learning_rate": 1.9453887943472532e-05,
+      "loss": 533.9896,
+      "step": 5960
+    },
+    {
+      "epoch": 0.11481818050687323,
+      "grad_norm": 3405.081323223352,
+      "learning_rate": 1.945189689515392e-05,
+      "loss": 577.4116,
+      "step": 5970
+    },
+    {
+      "epoch": 0.11501050576735375,
+      "grad_norm": 3057.157431474911,
+      "learning_rate": 1.9449902326159815e-05,
+      "loss": 539.4978,
+      "step": 5980
+    },
+    {
+      "epoch": 0.11520283102783427,
+      "grad_norm": 3090.1708643033476,
+      "learning_rate": 1.9447904237233164e-05,
+      "loss": 552.6425,
+      "step": 5990
+    },
+    {
+      "epoch": 0.1153951562883148,
+      "grad_norm": 3370.0202219698854,
+      "learning_rate": 1.9445902629118223e-05,
+      "loss": 543.0637,
+      "step": 6000
+    },
+    {
+      "epoch": 0.11558748154879532,
+      "grad_norm": 3353.0985821305353,
+      "learning_rate": 1.9443897502560555e-05,
+      "loss": 569.4459,
+      "step": 6010
+    },
+    {
+      "epoch": 0.11577980680927585,
+      "grad_norm": 3128.586725412783,
+      "learning_rate": 1.9441888858307042e-05,
+      "loss": 577.1365,
+      "step": 6020
+    },
+    {
+      "epoch": 0.11597213206975637,
+      "grad_norm": 3093.328840960538,
+      "learning_rate": 1.943987669710586e-05,
+      "loss": 540.1195,
+      "step": 6030
+    },
+    {
+      "epoch": 0.11616445733023689,
+      "grad_norm": 2811.620026745984,
+      "learning_rate": 1.9437861019706522e-05,
+      "loss": 541.466,
+      "step": 6040
+    },
+    {
+      "epoch": 0.11635678259071743,
+      "grad_norm": 3218.77476644157,
+      "learning_rate": 1.943584182685982e-05,
+      "loss": 561.2151,
+      "step": 6050
+    },
+    {
+      "epoch": 0.11654910785119794,
+      "grad_norm": 4422.75107123467,
+      "learning_rate": 1.9433819119317878e-05,
+      "loss": 556.3842,
+      "step": 6060
+    },
+    {
+      "epoch": 0.11674143311167846,
+      "grad_norm": 3247.898932738754,
+      "learning_rate": 1.9431792897834115e-05,
+      "loss": 541.4021,
+      "step": 6070
+    },
+    {
+      "epoch": 0.116933758372159,
+      "grad_norm": 3405.2837613700676,
+      "learning_rate": 1.9429763163163273e-05,
+      "loss": 559.3579,
+      "step": 6080
+    },
+    {
+      "epoch": 0.11712608363263952,
+      "grad_norm": 3133.3314475673415,
+      "learning_rate": 1.942772991606139e-05,
+      "loss": 534.2391,
+      "step": 6090
+    },
+    {
+      "epoch": 0.11731840889312005,
+      "grad_norm": 3360.2424027132392,
+      "learning_rate": 1.9425693157285816e-05,
+      "loss": 545.0727,
+      "step": 6100
+    },
+    {
+      "epoch": 0.11751073415360057,
+      "grad_norm": 3263.1690456164156,
+      "learning_rate": 1.942365288759521e-05,
+      "loss": 535.2984,
+      "step": 6110
+    },
+    {
+      "epoch": 0.11770305941408109,
+      "grad_norm": 3334.7830618196663,
+      "learning_rate": 1.9421609107749542e-05,
+      "loss": 560.9981,
+      "step": 6120
+    },
+    {
+      "epoch": 0.11789538467456162,
+      "grad_norm": 3400.2054641704676,
+      "learning_rate": 1.9419561818510085e-05,
+      "loss": 546.2139,
+      "step": 6130
+    },
+    {
+      "epoch": 0.11808770993504214,
+      "grad_norm": 2909.570257548634,
+      "learning_rate": 1.9417511020639416e-05,
+      "loss": 540.974,
+      "step": 6140
+    },
+    {
+      "epoch": 0.11828003519552267,
+      "grad_norm": 3034.877627466283,
+      "learning_rate": 1.9415456714901432e-05,
+      "loss": 554.5541,
+      "step": 6150
+    },
+    {
+      "epoch": 0.11847236045600319,
+      "grad_norm": 2997.168419008984,
+      "learning_rate": 1.941339890206132e-05,
+      "loss": 554.274,
+      "step": 6160
+    },
+    {
+      "epoch": 0.11866468571648371,
+      "grad_norm": 3087.6845287312244,
+      "learning_rate": 1.9411337582885587e-05,
+      "loss": 543.0096,
+      "step": 6170
+    },
+    {
+      "epoch": 0.11885701097696424,
+      "grad_norm": 3144.6303973691224,
+      "learning_rate": 1.9409272758142034e-05,
+      "loss": 531.5663,
+      "step": 6180
+    },
+    {
+      "epoch": 0.11904933623744476,
+      "grad_norm": 3103.4789362058136,
+      "learning_rate": 1.940720442859978e-05,
+      "loss": 546.0656,
+      "step": 6190
+    },
+    {
+      "epoch": 0.1192416614979253,
+      "grad_norm": 3025.615510900712,
+      "learning_rate": 1.940513259502924e-05,
+      "loss": 545.3059,
+      "step": 6200
+    },
+    {
+      "epoch": 0.11943398675840582,
+      "grad_norm": 3334.150702693289,
+      "learning_rate": 1.9403057258202144e-05,
+      "loss": 538.4965,
+      "step": 6210
+    },
+    {
+      "epoch": 0.11962631201888634,
+      "grad_norm": 3347.3229811869683,
+      "learning_rate": 1.940097841889151e-05,
+      "loss": 536.4763,
+      "step": 6220
+    },
+    {
+      "epoch": 0.11981863727936687,
+      "grad_norm": 2998.9760673185115,
+      "learning_rate": 1.939889607787168e-05,
+      "loss": 548.1066,
+      "step": 6230
+    },
+    {
+      "epoch": 0.12001096253984739,
+      "grad_norm": 3109.4795743976388,
+      "learning_rate": 1.9396810235918287e-05,
+      "loss": 543.2923,
+      "step": 6240
+    },
+    {
+      "epoch": 0.12020328780032792,
+      "grad_norm": 3232.870774546269,
+      "learning_rate": 1.939472089380828e-05,
+      "loss": 534.96,
+      "step": 6250
+    },
+    {
+      "epoch": 0.12039561306080844,
+      "grad_norm": 2978.4002590683954,
+      "learning_rate": 1.9392628052319895e-05,
+      "loss": 543.118,
+      "step": 6260
+    },
+    {
+      "epoch": 0.12058793832128896,
+      "grad_norm": 2919.3960325806106,
+      "learning_rate": 1.9390531712232687e-05,
+      "loss": 542.5694,
+      "step": 6270
+    },
+    {
+      "epoch": 0.12078026358176949,
+      "grad_norm": 2927.486233081042,
+      "learning_rate": 1.9388431874327504e-05,
+      "loss": 564.2301,
+      "step": 6280
+    },
+    {
+      "epoch": 0.12097258884225001,
+      "grad_norm": 3158.569297084666,
+      "learning_rate": 1.9386328539386502e-05,
+      "loss": 549.4159,
+      "step": 6290
+    },
+    {
+      "epoch": 0.12116491410273053,
+      "grad_norm": 2810.2562785000573,
+      "learning_rate": 1.938422170819314e-05,
+      "loss": 543.4707,
+      "step": 6300
+    },
+    {
+      "epoch": 0.12135723936321106,
+      "grad_norm": 3278.6095574682113,
+      "learning_rate": 1.938211138153218e-05,
+      "loss": 530.399,
+      "step": 6310
+    },
+    {
+      "epoch": 0.12154956462369158,
+      "grad_norm": 3300.947726198494,
+      "learning_rate": 1.9379997560189677e-05,
+      "loss": 533.7925,
+      "step": 6320
+    },
+    {
+      "epoch": 0.12174188988417212,
+      "grad_norm": 2896.6190321511403,
+      "learning_rate": 1.9377880244953e-05,
+      "loss": 530.0928,
+      "step": 6330
+    },
+    {
+      "epoch": 0.12193421514465264,
+      "grad_norm": 3128.138150461743,
+      "learning_rate": 1.9375759436610813e-05,
+      "loss": 539.4151,
+      "step": 6340
+    },
+    {
+      "epoch": 0.12212654040513315,
+      "grad_norm": 3188.0923666919643,
+      "learning_rate": 1.937363513595308e-05,
+      "loss": 536.5022,
+      "step": 6350
+    },
+    {
+      "epoch": 0.12231886566561369,
+      "grad_norm": 3007.3963405043382,
+      "learning_rate": 1.937150734377107e-05,
+      "loss": 520.4052,
+      "step": 6360
+    },
+    {
+      "epoch": 0.12251119092609421,
+      "grad_norm": 3084.954357271496,
+      "learning_rate": 1.9369376060857354e-05,
+      "loss": 520.9371,
+      "step": 6370
+    },
+    {
+      "epoch": 0.12270351618657474,
+      "grad_norm": 3083.7171664235193,
+      "learning_rate": 1.936724128800579e-05,
+      "loss": 543.9652,
+      "step": 6380
+    },
+    {
+      "epoch": 0.12289584144705526,
+      "grad_norm": 2909.9761982180416,
+      "learning_rate": 1.9365103026011555e-05,
+      "loss": 535.6667,
+      "step": 6390
+    },
+    {
+      "epoch": 0.12308816670753578,
+      "grad_norm": 4861.786739150577,
+      "learning_rate": 1.9362961275671112e-05,
+      "loss": 545.8405,
+      "step": 6400
+    },
+    {
+      "epoch": 0.12328049196801631,
+      "grad_norm": 3752.910423446671,
+      "learning_rate": 1.936081603778223e-05,
+      "loss": 555.2303,
+      "step": 6410
+    },
+    {
+      "epoch": 0.12347281722849683,
+      "grad_norm": 3161.1223359526,
+      "learning_rate": 1.9358667313143972e-05,
+      "loss": 533.0413,
+      "step": 6420
+    },
+    {
+      "epoch": 0.12366514248897736,
+      "grad_norm": 3084.112356224589,
+      "learning_rate": 1.93565151025567e-05,
+      "loss": 533.5978,
+      "step": 6430
+    },
+    {
+      "epoch": 0.12385746774945788,
+      "grad_norm": 2753.7211396992348,
+      "learning_rate": 1.9354359406822084e-05,
+      "loss": 538.9962,
+      "step": 6440
+    },
+    {
+      "epoch": 0.1240497930099384,
+      "grad_norm": 2830.5405856034,
+      "learning_rate": 1.935220022674308e-05,
+      "loss": 524.1744,
+      "step": 6450
+    },
+    {
+      "epoch": 0.12424211827041894,
+      "grad_norm": 2651.3192158256934,
+      "learning_rate": 1.9350037563123947e-05,
+      "loss": 510.2089,
+      "step": 6460
+    },
+    {
+      "epoch": 0.12443444353089946,
+      "grad_norm": 3168.1823016936264,
+      "learning_rate": 1.9347871416770245e-05,
+      "loss": 536.1155,
+      "step": 6470
+    },
+    {
+      "epoch": 0.12462676879137999,
+      "grad_norm": 3202.03309123853,
+      "learning_rate": 1.9345701788488825e-05,
+      "loss": 520.9955,
+      "step": 6480
+    },
+    {
+      "epoch": 0.12481909405186051,
+      "grad_norm": 2840.598505873904,
+      "learning_rate": 1.9343528679087837e-05,
+      "loss": 527.8342,
+      "step": 6490
+    },
+    {
+      "epoch": 0.12501141931234103,
+      "grad_norm": 3211.6566188190445,
+      "learning_rate": 1.934135208937673e-05,
+      "loss": 530.155,
+      "step": 6500
+    },
+    {
+      "epoch": 0.12520374457282155,
+      "grad_norm": 2955.8310508464415,
+      "learning_rate": 1.9339172020166245e-05,
+      "loss": 539.0327,
+      "step": 6510
+    },
+    {
+      "epoch": 0.1253960698333021,
+      "grad_norm": 2971.960397576802,
+      "learning_rate": 1.933698847226843e-05,
+      "loss": 531.7926,
+      "step": 6520
+    },
+    {
+      "epoch": 0.1255883950937826,
+      "grad_norm": 2881.037520428358,
+      "learning_rate": 1.9334801446496606e-05,
+      "loss": 518.636,
+      "step": 6530
+    },
+    {
+      "epoch": 0.12578072035426313,
+      "grad_norm": 2676.6782659120877,
+      "learning_rate": 1.933261094366542e-05,
+      "loss": 527.1354,
+      "step": 6540
+    },
+    {
+      "epoch": 0.12597304561474365,
+      "grad_norm": 2821.330151632555,
+      "learning_rate": 1.9330416964590787e-05,
+      "loss": 506.6257,
+      "step": 6550
+    },
+    {
+      "epoch": 0.12616537087522417,
+      "grad_norm": 3257.4503429609927,
+      "learning_rate": 1.932821951008993e-05,
+      "loss": 538.6153,
+      "step": 6560
+    },
+    {
+      "epoch": 0.12635769613570472,
+      "grad_norm": 2939.282591331584,
+      "learning_rate": 1.9326018580981373e-05,
+      "loss": 523.2369,
+      "step": 6570
+    },
+    {
+      "epoch": 0.12655002139618524,
+      "grad_norm": 2841.0672894326,
+      "learning_rate": 1.9323814178084914e-05,
+      "loss": 516.4375,
+      "step": 6580
+    },
+    {
+      "epoch": 0.12674234665666576,
+      "grad_norm": 3673.70106053849,
+      "learning_rate": 1.9321606302221662e-05,
+      "loss": 531.7859,
+      "step": 6590
+    },
+    {
+      "epoch": 0.12693467191714627,
+      "grad_norm": 2939.5291681456047,
+      "learning_rate": 1.9319394954214013e-05,
+      "loss": 529.4131,
+      "step": 6600
+    },
+    {
+      "epoch": 0.1271269971776268,
+      "grad_norm": 3374.2767657512536,
+      "learning_rate": 1.931718013488566e-05,
+      "loss": 525.3339,
+      "step": 6610
+    },
+    {
+      "epoch": 0.12731932243810734,
+      "grad_norm": 3028.5653610600807,
+      "learning_rate": 1.9314961845061584e-05,
+      "loss": 529.326,
+      "step": 6620
+    },
+    {
+      "epoch": 0.12751164769858786,
+      "grad_norm": 4369.605625187626,
+      "learning_rate": 1.9312740085568063e-05,
+      "loss": 537.9184,
+      "step": 6630
+    },
+    {
+      "epoch": 0.12770397295906838,
+      "grad_norm": 3241.5922469882184,
+      "learning_rate": 1.9310514857232666e-05,
+      "loss": 513.9442,
+      "step": 6640
+    },
+    {
+      "epoch": 0.1278962982195489,
+      "grad_norm": 2876.8227695115215,
+      "learning_rate": 1.930828616088425e-05,
+      "loss": 504.4727,
+      "step": 6650
+    },
+    {
+      "epoch": 0.12808862348002942,
+      "grad_norm": 2964.3805792671037,
+      "learning_rate": 1.9306053997352973e-05,
+      "loss": 513.8322,
+      "step": 6660
+    },
+    {
+      "epoch": 0.12828094874050994,
+      "grad_norm": 3096.749657798393,
+      "learning_rate": 1.9303818367470274e-05,
+      "loss": 511.4317,
+      "step": 6670
+    },
+    {
+      "epoch": 0.12847327400099048,
+      "grad_norm": 2890.5227620763667,
+      "learning_rate": 1.9301579272068894e-05,
+      "loss": 521.1064,
+      "step": 6680
+    },
+    {
+      "epoch": 0.128665599261471,
+      "grad_norm": 3158.701263857356,
+      "learning_rate": 1.9299336711982853e-05,
+      "loss": 518.0244,
+      "step": 6690
+    },
+    {
+      "epoch": 0.12885792452195152,
+      "grad_norm": 2940.1574261384253,
+      "learning_rate": 1.9297090688047473e-05,
+      "loss": 520.0789,
+      "step": 6700
+    },
+    {
+      "epoch": 0.12905024978243204,
+      "grad_norm": 2992.618765154942,
+      "learning_rate": 1.929484120109936e-05,
+      "loss": 514.3528,
+      "step": 6710
+    },
+    {
+      "epoch": 0.12924257504291256,
+      "grad_norm": 3020.590448166519,
+      "learning_rate": 1.9292588251976404e-05,
+      "loss": 508.5652,
+      "step": 6720
+    },
+    {
+      "epoch": 0.1294349003033931,
+      "grad_norm": 2856.572759086012,
+      "learning_rate": 1.92903318415178e-05,
+      "loss": 519.9399,
+      "step": 6730
+    },
+    {
+      "epoch": 0.12962722556387363,
+      "grad_norm": 2926.8951867528344,
+      "learning_rate": 1.9288071970564015e-05,
+      "loss": 532.937,
+      "step": 6740
+    },
+    {
+      "epoch": 0.12981955082435415,
+      "grad_norm": 3132.1509036620314,
+      "learning_rate": 1.9285808639956823e-05,
+      "loss": 522.9871,
+      "step": 6750
+    },
+    {
+      "epoch": 0.13001187608483467,
+      "grad_norm": 3174.114298022321,
+      "learning_rate": 1.9283541850539272e-05,
+      "loss": 515.1804,
+      "step": 6760
+    },
+    {
+      "epoch": 0.13020420134531518,
+      "grad_norm": 2742.0318887407143,
+      "learning_rate": 1.92812716031557e-05,
+      "loss": 509.1974,
+      "step": 6770
+    },
+    {
+      "epoch": 0.13039652660579573,
+      "grad_norm": 3301.8920060888313,
+      "learning_rate": 1.9278997898651746e-05,
+      "loss": 517.4712,
+      "step": 6780
+    },
+    {
+      "epoch": 0.13058885186627625,
+      "grad_norm": 2984.211272567207,
+      "learning_rate": 1.9276720737874327e-05,
+      "loss": 515.5563,
+      "step": 6790
+    },
+    {
+      "epoch": 0.13078117712675677,
+      "grad_norm": 2876.1911518086067,
+      "learning_rate": 1.9274440121671637e-05,
+      "loss": 503.4981,
+      "step": 6800
+    },
+    {
+      "epoch": 0.1309735023872373,
+      "grad_norm": 3045.9755222607732,
+      "learning_rate": 1.9272156050893173e-05,
+      "loss": 513.2415,
+      "step": 6810
+    },
+    {
+      "epoch": 0.1311658276477178,
+      "grad_norm": 3285.03843223816,
+      "learning_rate": 1.926986852638972e-05,
+      "loss": 513.935,
+      "step": 6820
+    },
+    {
+      "epoch": 0.13135815290819836,
+      "grad_norm": 3552.897797910797,
+      "learning_rate": 1.926757754901333e-05,
+      "loss": 506.6312,
+      "step": 6830
+    },
+    {
+      "epoch": 0.13155047816867887,
+      "grad_norm": 2892.9720519584675,
+      "learning_rate": 1.926528311961737e-05,
+      "loss": 511.6736,
+      "step": 6840
+    },
+    {
+      "epoch": 0.1317428034291594,
+      "grad_norm": 3018.4668902078124,
+      "learning_rate": 1.9262985239056463e-05,
+      "loss": 516.399,
+      "step": 6850
+    },
+    {
+      "epoch": 0.1319351286896399,
+      "grad_norm": 3492.7703250436643,
+      "learning_rate": 1.9260683908186544e-05,
+      "loss": 508.8771,
+      "step": 6860
+    },
+    {
+      "epoch": 0.13212745395012043,
+      "grad_norm": 2737.099406740913,
+      "learning_rate": 1.9258379127864808e-05,
+      "loss": 500.0156,
+      "step": 6870
+    },
+    {
+      "epoch": 0.13231977921060098,
+      "grad_norm": 2758.72034557184,
+      "learning_rate": 1.925607089894976e-05,
+      "loss": 502.6882,
+      "step": 6880
+    },
+    {
+      "epoch": 0.1325121044710815,
+      "grad_norm": 2729.260977484528,
+      "learning_rate": 1.9253759222301168e-05,
+      "loss": 504.7563,
+      "step": 6890
+    },
+    {
+      "epoch": 0.13270442973156202,
+      "grad_norm": 2790.5658413151054,
+      "learning_rate": 1.9251444098780095e-05,
+      "loss": 505.7114,
+      "step": 6900
+    },
+    {
+      "epoch": 0.13289675499204254,
+      "grad_norm": 2753.152180961606,
+      "learning_rate": 1.924912552924889e-05,
+      "loss": 506.7776,
+      "step": 6910
+    },
+    {
+      "epoch": 0.13308908025252306,
+      "grad_norm": 2947.5010065928886,
+      "learning_rate": 1.924680351457118e-05,
+      "loss": 512.5181,
+      "step": 6920
+    },
+    {
+      "epoch": 0.1332814055130036,
+      "grad_norm": 2686.4308328155576,
+      "learning_rate": 1.9244478055611875e-05,
+      "loss": 496.111,
+      "step": 6930
+    },
+    {
+      "epoch": 0.13347373077348412,
+      "grad_norm": 3210.7156628482157,
+      "learning_rate": 1.9242149153237175e-05,
+      "loss": 517.4425,
+      "step": 6940
+    },
+    {
+      "epoch": 0.13366605603396464,
+      "grad_norm": 3073.847999215363,
+      "learning_rate": 1.923981680831455e-05,
+      "loss": 494.8016,
+      "step": 6950
+    },
+    {
+      "epoch": 0.13385838129444516,
+      "grad_norm": 2991.491196704255,
+      "learning_rate": 1.923748102171277e-05,
+      "loss": 512.0045,
+      "step": 6960
+    },
+    {
+      "epoch": 0.13405070655492568,
+      "grad_norm": 3021.09219279553,
+      "learning_rate": 1.9235141794301867e-05,
+      "loss": 507.9525,
+      "step": 6970
+    },
+    {
+      "epoch": 0.13424303181540623,
+      "grad_norm": 3291.2066943333184,
+      "learning_rate": 1.9232799126953173e-05,
+      "loss": 508.618,
+      "step": 6980
+    },
+    {
+      "epoch": 0.13443535707588675,
+      "grad_norm": 3267.4846526856877,
+      "learning_rate": 1.9230453020539285e-05,
+      "loss": 510.1736,
+      "step": 6990
+    },
+    {
+      "epoch": 0.13462768233636727,
+      "grad_norm": 2889.519349382989,
+      "learning_rate": 1.9228103475934096e-05,
+      "loss": 493.2681,
+      "step": 7000
+    },
+    {
+      "epoch": 0.13482000759684779,
+      "grad_norm": 3029.048574747814,
+      "learning_rate": 1.9225750494012767e-05,
+      "loss": 514.1222,
+      "step": 7010
+    },
+    {
+      "epoch": 0.1350123328573283,
+      "grad_norm": 2980.365910026588,
+      "learning_rate": 1.9223394075651748e-05,
+      "loss": 513.1449,
+      "step": 7020
+    },
+    {
+      "epoch": 0.13520465811780885,
+      "grad_norm": 3209.5849732677316,
+      "learning_rate": 1.9221034221728764e-05,
+      "loss": 501.2741,
+      "step": 7030
+    },
+    {
+      "epoch": 0.13539698337828937,
+      "grad_norm": 2934.9755109061402,
+      "learning_rate": 1.9218670933122826e-05,
+      "loss": 508.1023,
+      "step": 7040
+    },
+    {
+      "epoch": 0.1355893086387699,
+      "grad_norm": 2669.4887685364515,
+      "learning_rate": 1.9216304210714213e-05,
+      "loss": 494.1592,
+      "step": 7050
+    },
+    {
+      "epoch": 0.1357816338992504,
+      "grad_norm": 3165.3816885313468,
+      "learning_rate": 1.9213934055384498e-05,
+      "loss": 527.2488,
+      "step": 7060
+    },
+    {
+      "epoch": 0.13597395915973093,
+      "grad_norm": 2931.5462485311364,
+      "learning_rate": 1.9211560468016516e-05,
+      "loss": 512.6491,
+      "step": 7070
+    },
+    {
+      "epoch": 0.13616628442021148,
+      "grad_norm": 3087.9201699188498,
+      "learning_rate": 1.9209183449494397e-05,
+      "loss": 498.9521,
+      "step": 7080
+    },
+    {
+      "epoch": 0.136358609680692,
+      "grad_norm": 2972.71779101732,
+      "learning_rate": 1.9206803000703534e-05,
+      "loss": 505.3806,
+      "step": 7090
+    },
+    {
+      "epoch": 0.1365509349411725,
+      "grad_norm": 3005.4594268142446,
+      "learning_rate": 1.9204419122530614e-05,
+      "loss": 502.5931,
+      "step": 7100
+    },
+    {
+      "epoch": 0.13674326020165303,
+      "grad_norm": 2924.4183496973938,
+      "learning_rate": 1.9202031815863583e-05,
+      "loss": 501.4909,
+      "step": 7110
+    },
+    {
+      "epoch": 0.13693558546213355,
+      "grad_norm": 2793.8027483759365,
+      "learning_rate": 1.919964108159168e-05,
+      "loss": 496.083,
+      "step": 7120
+    },
+    {
+      "epoch": 0.13712791072261407,
+      "grad_norm": 2695.7998871082573,
+      "learning_rate": 1.9197246920605408e-05,
+      "loss": 485.6467,
+      "step": 7130
+    },
+    {
+      "epoch": 0.13732023598309462,
+      "grad_norm": 2966.9919464025984,
+      "learning_rate": 1.9194849333796557e-05,
+      "loss": 496.4655,
+      "step": 7140
+    },
+    {
+      "epoch": 0.13751256124357514,
+      "grad_norm": 2909.544915209876,
+      "learning_rate": 1.9192448322058187e-05,
+      "loss": 505.7636,
+      "step": 7150
+    },
+    {
+      "epoch": 0.13770488650405566,
+      "grad_norm": 2969.974367324485,
+      "learning_rate": 1.9190043886284635e-05,
+      "loss": 495.1523,
+      "step": 7160
+    },
+    {
+      "epoch": 0.13789721176453618,
+      "grad_norm": 3199.5117549484075,
+      "learning_rate": 1.9187636027371518e-05,
+      "loss": 486.4546,
+      "step": 7170
+    },
+    {
+      "epoch": 0.1380895370250167,
+      "grad_norm": 3568.3250452673274,
+      "learning_rate": 1.9185224746215714e-05,
+      "loss": 513.3837,
+      "step": 7180
+    },
+    {
+      "epoch": 0.13828186228549724,
+      "grad_norm": 3190.2351081798324,
+      "learning_rate": 1.9182810043715394e-05,
+      "loss": 492.3429,
+      "step": 7190
+    },
+    {
+      "epoch": 0.13847418754597776,
+      "grad_norm": 3010.424687603582,
+      "learning_rate": 1.9180391920769993e-05,
+      "loss": 491.3337,
+      "step": 7200
+    },
+    {
+      "epoch": 0.13866651280645828,
+      "grad_norm": 2856.1887310591183,
+      "learning_rate": 1.9177970378280215e-05,
+      "loss": 498.1321,
+      "step": 7210
+    },
+    {
+      "epoch": 0.1388588380669388,
+      "grad_norm": 3170.294558194702,
+      "learning_rate": 1.9175545417148056e-05,
+      "loss": 508.6399,
+      "step": 7220
+    },
+    {
+      "epoch": 0.13905116332741932,
+      "grad_norm": 2831.0199704246793,
+      "learning_rate": 1.9173117038276766e-05,
+      "loss": 501.2726,
+      "step": 7230
+    },
+    {
+      "epoch": 0.13924348858789987,
+      "grad_norm": 3381.9769536278372,
+      "learning_rate": 1.9170685242570878e-05,
+      "loss": 508.0261,
+      "step": 7240
+    },
+    {
+      "epoch": 0.13943581384838039,
+      "grad_norm": 2629.6630742994976,
+      "learning_rate": 1.9168250030936195e-05,
+      "loss": 489.3669,
+      "step": 7250
+    },
+    {
+      "epoch": 0.1396281391088609,
+      "grad_norm": 3515.715591808194,
+      "learning_rate": 1.91658114042798e-05,
+      "loss": 497.8372,
+      "step": 7260
+    },
+    {
+      "epoch": 0.13982046436934142,
+      "grad_norm": 3297.661932599785,
+      "learning_rate": 1.9163369363510026e-05,
+      "loss": 493.3461,
+      "step": 7270
+    },
+    {
+      "epoch": 0.14001278962982194,
+      "grad_norm": 2840.2594851903527,
+      "learning_rate": 1.916092390953651e-05,
+      "loss": 498.9034,
+      "step": 7280
+    },
+    {
+      "epoch": 0.1402051148903025,
+      "grad_norm": 2999.0256055345612,
+      "learning_rate": 1.915847504327013e-05,
+      "loss": 493.2006,
+      "step": 7290
+    },
+    {
+      "epoch": 0.140397440150783,
+      "grad_norm": 2751.6678060885597,
+      "learning_rate": 1.9156022765623057e-05,
+      "loss": 490.7306,
+      "step": 7300
+    },
+    {
+      "epoch": 0.14058976541126353,
+      "grad_norm": 3201.114508962532,
+      "learning_rate": 1.9153567077508718e-05,
+      "loss": 483.6938,
+      "step": 7310
+    },
+    {
+      "epoch": 0.14078209067174405,
+      "grad_norm": 2792.437226101636,
+      "learning_rate": 1.9151107979841824e-05,
+      "loss": 489.7729,
+      "step": 7320
+    },
+    {
+      "epoch": 0.14097441593222457,
+      "grad_norm": 2962.013811565237,
+      "learning_rate": 1.9148645473538338e-05,
+      "loss": 478.1479,
+      "step": 7330
+    },
+    {
+      "epoch": 0.14116674119270511,
+      "grad_norm": 2908.3890528948004,
+      "learning_rate": 1.9146179559515507e-05,
+      "loss": 484.4061,
+      "step": 7340
+    },
+    {
+      "epoch": 0.14135906645318563,
+      "grad_norm": 2962.249440808847,
+      "learning_rate": 1.9143710238691847e-05,
+      "loss": 485.9969,
+      "step": 7350
+    },
+    {
+      "epoch": 0.14155139171366615,
+      "grad_norm": 2513.2790256078592,
+      "learning_rate": 1.9141237511987137e-05,
+      "loss": 477.8858,
+      "step": 7360
+    },
+    {
+      "epoch": 0.14174371697414667,
+      "grad_norm": 2706.841797558357,
+      "learning_rate": 1.9138761380322425e-05,
+      "loss": 497.3392,
+      "step": 7370
+    },
+    {
+      "epoch": 0.1419360422346272,
+      "grad_norm": 2925.6397387295415,
+      "learning_rate": 1.913628184462003e-05,
+      "loss": 489.3872,
+      "step": 7380
+    },
+    {
+      "epoch": 0.14212836749510774,
+      "grad_norm": 4446.849487432294,
+      "learning_rate": 1.913379890580354e-05,
+      "loss": 495.8265,
+      "step": 7390
+    },
+    {
+      "epoch": 0.14232069275558826,
+      "grad_norm": 2791.091878429826,
+      "learning_rate": 1.9131312564797805e-05,
+      "loss": 489.4853,
+      "step": 7400
+    },
+    {
+      "epoch": 0.14251301801606878,
+      "grad_norm": 2626.5949527078046,
+      "learning_rate": 1.912882282252895e-05,
+      "loss": 475.7432,
+      "step": 7410
+    },
+    {
+      "epoch": 0.1427053432765493,
+      "grad_norm": 3383.8341351294794,
+      "learning_rate": 1.9126329679924364e-05,
+      "loss": 496.1304,
+      "step": 7420
+    },
+    {
+      "epoch": 0.14289766853702981,
+      "grad_norm": 2978.863005353464,
+      "learning_rate": 1.9123833137912693e-05,
+      "loss": 489.6156,
+      "step": 7430
+    },
+    {
+      "epoch": 0.14308999379751036,
+      "grad_norm": 2849.684134046931,
+      "learning_rate": 1.9121333197423867e-05,
+      "loss": 512.0472,
+      "step": 7440
+    },
+    {
+      "epoch": 0.14328231905799088,
+      "grad_norm": 3088.67894949292,
+      "learning_rate": 1.9118829859389067e-05,
+      "loss": 502.16,
+      "step": 7450
+    },
+    {
+      "epoch": 0.1434746443184714,
+      "grad_norm": 2819.8772951772203,
+      "learning_rate": 1.9116323124740748e-05,
+      "loss": 485.5934,
+      "step": 7460
+    },
+    {
+      "epoch": 0.14366696957895192,
+      "grad_norm": 2773.713141370265,
+      "learning_rate": 1.9113812994412627e-05,
+      "loss": 490.0072,
+      "step": 7470
+    },
+    {
+      "epoch": 0.14385929483943244,
+      "grad_norm": 2833.560775422032,
+      "learning_rate": 1.911129946933968e-05,
+      "loss": 492.3943,
+      "step": 7480
+    },
+    {
+      "epoch": 0.14405162009991299,
+      "grad_norm": 2598.159765312937,
+      "learning_rate": 1.9108782550458164e-05,
+      "loss": 501.1888,
+      "step": 7490
+    },
+    {
+      "epoch": 0.1442439453603935,
+      "grad_norm": 2947.8731284449723,
+      "learning_rate": 1.9106262238705583e-05,
+      "loss": 495.5496,
+      "step": 7500
+    },
+    {
+      "epoch": 0.14443627062087402,
+      "grad_norm": 2674.6330467886787,
+      "learning_rate": 1.9103738535020713e-05,
+      "loss": 494.459,
+      "step": 7510
+    },
+    {
+      "epoch": 0.14462859588135454,
+      "grad_norm": 2855.64514682562,
+      "learning_rate": 1.910121144034359e-05,
+      "loss": 478.4898,
+      "step": 7520
+    },
+    {
+      "epoch": 0.14482092114183506,
+      "grad_norm": 3014.0201415361366,
+      "learning_rate": 1.909868095561552e-05,
+      "loss": 481.9886,
+      "step": 7530
+    },
+    {
+      "epoch": 0.1450132464023156,
+      "grad_norm": 2834.7522007383313,
+      "learning_rate": 1.9096147081779063e-05,
+      "loss": 494.806,
+      "step": 7540
+    },
+    {
+      "epoch": 0.14520557166279613,
+      "grad_norm": 2843.4863006339915,
+      "learning_rate": 1.9093609819778044e-05,
+      "loss": 482.2569,
+      "step": 7550
+    },
+    {
+      "epoch": 0.14539789692327665,
+      "grad_norm": 2734.9134302578213,
+      "learning_rate": 1.9091069170557554e-05,
+      "loss": 489.1965,
+      "step": 7560
+    },
+    {
+      "epoch": 0.14559022218375717,
+      "grad_norm": 2970.862474311478,
+      "learning_rate": 1.9088525135063944e-05,
+      "loss": 484.0156,
+      "step": 7570
+    },
+    {
+      "epoch": 0.1457825474442377,
+      "grad_norm": 2527.0690158392335,
+      "learning_rate": 1.9085977714244822e-05,
+      "loss": 482.1195,
+      "step": 7580
+    },
+    {
+      "epoch": 0.1459748727047182,
+      "grad_norm": 3162.881517394372,
+      "learning_rate": 1.908342690904906e-05,
+      "loss": 492.1695,
+      "step": 7590
+    },
+    {
+      "epoch": 0.14616719796519875,
+      "grad_norm": 3077.277460070853,
+      "learning_rate": 1.9080872720426793e-05,
+      "loss": 483.6169,
+      "step": 7600
+    },
+    {
+      "epoch": 0.14635952322567927,
+      "grad_norm": 2742.044614693097,
+      "learning_rate": 1.9078315149329413e-05,
+      "loss": 461.6744,
+      "step": 7610
+    },
+    {
+      "epoch": 0.1465518484861598,
+      "grad_norm": 2681.675491050549,
+      "learning_rate": 1.9075754196709574e-05,
+      "loss": 490.2057,
+      "step": 7620
+    },
+    {
+      "epoch": 0.1467441737466403,
+      "grad_norm": 2626.59863424605,
+      "learning_rate": 1.9073189863521184e-05,
+      "loss": 477.6793,
+      "step": 7630
+    },
+    {
+      "epoch": 0.14693649900712083,
+      "grad_norm": 2643.2531040384692,
+      "learning_rate": 1.9070622150719423e-05,
+      "loss": 482.2477,
+      "step": 7640
+    },
+    {
+      "epoch": 0.14712882426760138,
+      "grad_norm": 3172.2098393406486,
+      "learning_rate": 1.9068051059260716e-05,
+      "loss": 481.8371,
+      "step": 7650
+    },
+    {
+      "epoch": 0.1473211495280819,
+      "grad_norm": 3256.479733327193,
+      "learning_rate": 1.9065476590102752e-05,
+      "loss": 491.9813,
+      "step": 7660
+    },
+    {
+      "epoch": 0.14751347478856242,
+      "grad_norm": 2630.5854410247307,
+      "learning_rate": 1.906289874420448e-05,
+      "loss": 498.6755,
+      "step": 7670
+    },
+    {
+      "epoch": 0.14770580004904293,
+      "grad_norm": 2748.258611251968,
+      "learning_rate": 1.9060317522526105e-05,
+      "loss": 470.4862,
+      "step": 7680
+    },
+    {
+      "epoch": 0.14789812530952345,
+      "grad_norm": 2775.263570459274,
+      "learning_rate": 1.905773292602909e-05,
+      "loss": 485.3623,
+      "step": 7690
+    },
+    {
+      "epoch": 0.148090450570004,
+      "grad_norm": 2886.085140627227,
+      "learning_rate": 1.905514495567615e-05,
+      "loss": 476.224,
+      "step": 7700
+    },
+    {
+      "epoch": 0.14828277583048452,
+      "grad_norm": 2702.3882118156494,
+      "learning_rate": 1.9052553612431268e-05,
+      "loss": 496.2374,
+      "step": 7710
+    },
+    {
+      "epoch": 0.14847510109096504,
+      "grad_norm": 2755.746818929295,
+      "learning_rate": 1.9049958897259674e-05,
+      "loss": 473.6089,
+      "step": 7720
+    },
+    {
+      "epoch": 0.14866742635144556,
+      "grad_norm": 2893.7558399190953,
+      "learning_rate": 1.904736081112785e-05,
+      "loss": 482.383,
+      "step": 7730
+    },
+    {
+      "epoch": 0.14885975161192608,
+      "grad_norm": 2613.4543216423335,
+      "learning_rate": 1.9044759355003552e-05,
+      "loss": 487.0586,
+      "step": 7740
+    },
+    {
+      "epoch": 0.14905207687240662,
+      "grad_norm": 2880.935110610246,
+      "learning_rate": 1.904215452985577e-05,
+      "loss": 473.3734,
+      "step": 7750
+    },
+    {
+      "epoch": 0.14924440213288714,
+      "grad_norm": 3768.0153704315408,
+      "learning_rate": 1.9039546336654765e-05,
+      "loss": 503.639,
+      "step": 7760
+    },
+    {
+      "epoch": 0.14943672739336766,
+      "grad_norm": 3132.526013928591,
+      "learning_rate": 1.903693477637204e-05,
+      "loss": 470.5541,
+      "step": 7770
+    },
+    {
+      "epoch": 0.14962905265384818,
+      "grad_norm": 2774.32343686497,
+      "learning_rate": 1.903431984998036e-05,
+      "loss": 474.7524,
+      "step": 7780
+    },
+    {
+      "epoch": 0.1498213779143287,
+      "grad_norm": 2938.971764833828,
+      "learning_rate": 1.9031701558453747e-05,
+      "loss": 498.0345,
+      "step": 7790
+    },
+    {
+      "epoch": 0.15001370317480925,
+      "grad_norm": 3693.358094860785,
+      "learning_rate": 1.902907990276746e-05,
+      "loss": 458.7667,
+      "step": 7800
+    },
+    {
+      "epoch": 0.15020602843528977,
+      "grad_norm": 3306.4578035535965,
+      "learning_rate": 1.9026454883898036e-05,
+      "loss": 486.2284,
+      "step": 7810
+    },
+    {
+      "epoch": 0.1503983536957703,
+      "grad_norm": 3100.144077778144,
+      "learning_rate": 1.902382650282324e-05,
+      "loss": 467.8271,
+      "step": 7820
+    },
+    {
+      "epoch": 0.1505906789562508,
+      "grad_norm": 2732.474309499422,
+      "learning_rate": 1.902119476052211e-05,
+      "loss": 461.508,
+      "step": 7830
+    },
+    {
+      "epoch": 0.15078300421673133,
+      "grad_norm": 2770.421004144993,
+      "learning_rate": 1.9018559657974918e-05,
+      "loss": 476.7328,
+      "step": 7840
+    },
+    {
+      "epoch": 0.15097532947721187,
+      "grad_norm": 2909.490670250935,
+      "learning_rate": 1.90159211961632e-05,
+      "loss": 530.7372,
+      "step": 7850
+    },
+    {
+      "epoch": 0.1511676547376924,
+      "grad_norm": 3188.832267172363,
+      "learning_rate": 1.901327937606974e-05,
+      "loss": 502.5419,
+      "step": 7860
+    },
+    {
+      "epoch": 0.1513599799981729,
+      "grad_norm": 2612.9993956686194,
+      "learning_rate": 1.901063419867857e-05,
+      "loss": 486.2078,
+      "step": 7870
+    },
+    {
+      "epoch": 0.15155230525865343,
+      "grad_norm": 2747.7614659293913,
+      "learning_rate": 1.900798566497498e-05,
+      "loss": 474.4527,
+      "step": 7880
+    },
+    {
+      "epoch": 0.15174463051913395,
+      "grad_norm": 2509.4355263152893,
+      "learning_rate": 1.9005333775945496e-05,
+      "loss": 479.4521,
+      "step": 7890
+    },
+    {
+      "epoch": 0.1519369557796145,
+      "grad_norm": 2732.3112735518544,
+      "learning_rate": 1.9002678532577915e-05,
+      "loss": 469.8596,
+      "step": 7900
+    },
+    {
+      "epoch": 0.15212928104009502,
+      "grad_norm": 2649.1751809974876,
+      "learning_rate": 1.900001993586126e-05,
+      "loss": 481.8939,
+      "step": 7910
+    },
+    {
+      "epoch": 0.15232160630057553,
+      "grad_norm": 2677.0014270735664,
+      "learning_rate": 1.8997357986785822e-05,
+      "loss": 489.0299,
+      "step": 7920
+    },
+    {
+      "epoch": 0.15251393156105605,
+      "grad_norm": 2700.5217249604525,
+      "learning_rate": 1.899469268634313e-05,
+      "loss": 481.1306,
+      "step": 7930
+    },
+    {
+      "epoch": 0.15270625682153657,
+      "grad_norm": 2508.298814733503,
+      "learning_rate": 1.8992024035525964e-05,
+      "loss": 463.1267,
+      "step": 7940
+    },
+    {
+      "epoch": 0.15289858208201712,
+      "grad_norm": 2981.2789852308065,
+      "learning_rate": 1.8989352035328352e-05,
+      "loss": 477.5609,
+      "step": 7950
+    },
+    {
+      "epoch": 0.15309090734249764,
+      "grad_norm": 2891.7242881516418,
+      "learning_rate": 1.8986676686745572e-05,
+      "loss": 473.6297,
+      "step": 7960
+    },
+    {
+      "epoch": 0.15328323260297816,
+      "grad_norm": 2879.0509933380317,
+      "learning_rate": 1.8983997990774145e-05,
+      "loss": 494.6183,
+      "step": 7970
+    },
+    {
+      "epoch": 0.15347555786345868,
+      "grad_norm": 2763.657335141747,
+      "learning_rate": 1.8981315948411842e-05,
+      "loss": 486.7721,
+      "step": 7980
+    },
+    {
+      "epoch": 0.1536678831239392,
+      "grad_norm": 3176.949500486039,
+      "learning_rate": 1.897863056065768e-05,
+      "loss": 490.7333,
+      "step": 7990
+    },
+    {
+      "epoch": 0.15386020838441974,
+      "grad_norm": 2831.83018369395,
+      "learning_rate": 1.8975941828511923e-05,
+      "loss": 474.9769,
+      "step": 8000
+    },
+    {
+      "epoch": 0.15405253364490026,
+      "grad_norm": 2898.392383273504,
+      "learning_rate": 1.8973249752976075e-05,
+      "loss": 474.2976,
+      "step": 8010
+    },
+    {
+      "epoch": 0.15424485890538078,
+      "grad_norm": 2836.3775964522815,
+      "learning_rate": 1.8970554335052897e-05,
+      "loss": 464.9511,
+      "step": 8020
+    },
+    {
+      "epoch": 0.1544371841658613,
+      "grad_norm": 2788.359892750998,
+      "learning_rate": 1.8967855575746375e-05,
+      "loss": 467.5968,
+      "step": 8030
+    },
+    {
+      "epoch": 0.15462950942634182,
+      "grad_norm": 2674.441048869643,
+      "learning_rate": 1.8965153476061763e-05,
+      "loss": 475.4356,
+      "step": 8040
+    },
+    {
+      "epoch": 0.15482183468682234,
+      "grad_norm": 2820.9973278343778,
+      "learning_rate": 1.896244803700555e-05,
+      "loss": 470.7019,
+      "step": 8050
+    },
+    {
+      "epoch": 0.1550141599473029,
+      "grad_norm": 2833.5184081204247,
+      "learning_rate": 1.8959739259585458e-05,
+      "loss": 467.0874,
+      "step": 8060
+    },
+    {
+      "epoch": 0.1552064852077834,
+      "grad_norm": 2670.554800576522,
+      "learning_rate": 1.895702714481047e-05,
+      "loss": 454.6197,
+      "step": 8070
+    },
+    {
+      "epoch": 0.15539881046826393,
+      "grad_norm": 2654.2861147762346,
+      "learning_rate": 1.8954311693690798e-05,
+      "loss": 450.6652,
+      "step": 8080
+    },
+    {
+      "epoch": 0.15559113572874445,
+      "grad_norm": 2885.771301584405,
+      "learning_rate": 1.8951592907237906e-05,
+      "loss": 493.7055,
+      "step": 8090
+    },
+    {
+      "epoch": 0.15578346098922496,
+      "grad_norm": 3661.354185149759,
+      "learning_rate": 1.8948870786464496e-05,
+      "loss": 486.4298,
+      "step": 8100
+    },
+    {
+      "epoch": 0.1559757862497055,
+      "grad_norm": 2794.957543952636,
+      "learning_rate": 1.8946145332384515e-05,
+      "loss": 455.7606,
+      "step": 8110
+    },
+    {
+      "epoch": 0.15616811151018603,
+      "grad_norm": 2789.9608572044913,
+      "learning_rate": 1.8943416546013148e-05,
+      "loss": 482.6221,
+      "step": 8120
+    },
+    {
+      "epoch": 0.15636043677066655,
+      "grad_norm": 2402.1141955166977,
+      "learning_rate": 1.894068442836682e-05,
+      "loss": 466.4746,
+      "step": 8130
+    },
+    {
+      "epoch": 0.15655276203114707,
+      "grad_norm": 3050.247166384424,
+      "learning_rate": 1.8937948980463207e-05,
+      "loss": 463.4685,
+      "step": 8140
+    },
+    {
+      "epoch": 0.1567450872916276,
+      "grad_norm": 2976.425797495626,
+      "learning_rate": 1.893521020332121e-05,
+      "loss": 481.4321,
+      "step": 8150
+    },
+    {
+      "epoch": 0.15693741255210814,
+      "grad_norm": 3450.0250925142286,
+      "learning_rate": 1.8932468097960988e-05,
+      "loss": 462.0546,
+      "step": 8160
+    },
+    {
+      "epoch": 0.15712973781258865,
+      "grad_norm": 2702.7549440163066,
+      "learning_rate": 1.892972266540392e-05,
+      "loss": 448.9424,
+      "step": 8170
+    },
+    {
+      "epoch": 0.15732206307306917,
+      "grad_norm": 2596.001292073025,
+      "learning_rate": 1.8926973906672635e-05,
+      "loss": 454.9815,
+      "step": 8180
+    },
+    {
+      "epoch": 0.1575143883335497,
+      "grad_norm": 3009.8509523185962,
+      "learning_rate": 1.892422182279101e-05,
+      "loss": 471.6159,
+      "step": 8190
+    },
+    {
+      "epoch": 0.1577067135940302,
+      "grad_norm": 2652.4358538229417,
+      "learning_rate": 1.892146641478414e-05,
+      "loss": 467.1265,
+      "step": 8200
+    },
+    {
+      "epoch": 0.15789903885451076,
+      "grad_norm": 2495.2359093695,
+      "learning_rate": 1.8918707683678376e-05,
+      "loss": 452.5336,
+      "step": 8210
+    },
+    {
+      "epoch": 0.15809136411499128,
+      "grad_norm": 3007.1314531523863,
+      "learning_rate": 1.8915945630501296e-05,
+      "loss": 462.3906,
+      "step": 8220
+    },
+    {
+      "epoch": 0.1582836893754718,
+      "grad_norm": 2798.396673862514,
+      "learning_rate": 1.8913180256281723e-05,
+      "loss": 463.8158,
+      "step": 8230
+    },
+    {
+      "epoch": 0.15847601463595232,
+      "grad_norm": 2544.692471398825,
+      "learning_rate": 1.8910411562049706e-05,
+      "loss": 454.6521,
+      "step": 8240
+    },
+    {
+      "epoch": 0.15866833989643284,
+      "grad_norm": 2527.1761584761894,
+      "learning_rate": 1.8907639548836548e-05,
+      "loss": 459.9999,
+      "step": 8250
+    },
+    {
+      "epoch": 0.15886066515691338,
+      "grad_norm": 2723.1644502431964,
+      "learning_rate": 1.8904864217674766e-05,
+      "loss": 484.4615,
+      "step": 8260
+    },
+    {
+      "epoch": 0.1590529904173939,
+      "grad_norm": 2496.344877980351,
+      "learning_rate": 1.8902085569598136e-05,
+      "loss": 452.6649,
+      "step": 8270
+    },
+    {
+      "epoch": 0.15924531567787442,
+      "grad_norm": 2496.2472071614447,
+      "learning_rate": 1.889930360564165e-05,
+      "loss": 454.2822,
+      "step": 8280
+    },
+    {
+      "epoch": 0.15943764093835494,
+      "grad_norm": 2593.1348017621613,
+      "learning_rate": 1.8896518326841554e-05,
+      "loss": 447.8645,
+      "step": 8290
+    },
+    {
+      "epoch": 0.15962996619883546,
+      "grad_norm": 2661.784519983517,
+      "learning_rate": 1.889372973423531e-05,
+      "loss": 460.3713,
+      "step": 8300
+    },
+    {
+      "epoch": 0.159822291459316,
+      "grad_norm": 2438.8354895739203,
+      "learning_rate": 1.889093782886162e-05,
+      "loss": 457.8413,
+      "step": 8310
+    },
+    {
+      "epoch": 0.16001461671979653,
+      "grad_norm": 2714.4516476368362,
+      "learning_rate": 1.8888142611760433e-05,
+      "loss": 453.8086,
+      "step": 8320
+    },
+    {
+      "epoch": 0.16020694198027705,
+      "grad_norm": 3906.6661312224155,
+      "learning_rate": 1.8885344083972912e-05,
+      "loss": 453.6644,
+      "step": 8330
+    },
+    {
+      "epoch": 0.16039926724075756,
+      "grad_norm": 3294.581965773663,
+      "learning_rate": 1.8882542246541468e-05,
+      "loss": 487.4707,
+      "step": 8340
+    },
+    {
+      "epoch": 0.16059159250123808,
+      "grad_norm": 2935.270245095748,
+      "learning_rate": 1.887973710050974e-05,
+      "loss": 477.0995,
+      "step": 8350
+    },
+    {
+      "epoch": 0.16078391776171863,
+      "grad_norm": 2575.6822947435367,
+      "learning_rate": 1.887692864692259e-05,
+      "loss": 451.8498,
+      "step": 8360
+    },
+    {
+      "epoch": 0.16097624302219915,
+      "grad_norm": 2546.8471243887843,
+      "learning_rate": 1.887411688682613e-05,
+      "loss": 471.3553,
+      "step": 8370
+    },
+    {
+      "epoch": 0.16116856828267967,
+      "grad_norm": 2804.982756489293,
+      "learning_rate": 1.887130182126769e-05,
+      "loss": 469.4985,
+      "step": 8380
+    },
+    {
+      "epoch": 0.1613608935431602,
+      "grad_norm": 2934.456138274911,
+      "learning_rate": 1.8868483451295835e-05,
+      "loss": 463.9745,
+      "step": 8390
+    },
+    {
+      "epoch": 0.1615532188036407,
+      "grad_norm": 2794.9079840955537,
+      "learning_rate": 1.8865661777960366e-05,
+      "loss": 455.6153,
+      "step": 8400
+    },
+    {
+      "epoch": 0.16174554406412125,
+      "grad_norm": 2621.9513638770195,
+      "learning_rate": 1.88628368023123e-05,
+      "loss": 463.6387,
+      "step": 8410
+    },
+    {
+      "epoch": 0.16193786932460177,
+      "grad_norm": 7409.282526206722,
+      "learning_rate": 1.8860008525403903e-05,
+      "loss": 488.0506,
+      "step": 8420
+    },
+    {
+      "epoch": 0.1621301945850823,
+      "grad_norm": 2552.0709512542307,
+      "learning_rate": 1.885717694828866e-05,
+      "loss": 460.1391,
+      "step": 8430
+    },
+    {
+      "epoch": 0.1623225198455628,
+      "grad_norm": 2642.468928233332,
+      "learning_rate": 1.8854342072021282e-05,
+      "loss": 444.1377,
+      "step": 8440
+    },
+    {
+      "epoch": 0.16251484510604333,
+      "grad_norm": 2516.648950037554,
+      "learning_rate": 1.8851503897657717e-05,
+      "loss": 459.1028,
+      "step": 8450
+    },
+    {
+      "epoch": 0.16270717036652388,
+      "grad_norm": 2536.2265636656502,
+      "learning_rate": 1.8848662426255135e-05,
+      "loss": 468.7705,
+      "step": 8460
+    },
+    {
+      "epoch": 0.1628994956270044,
+      "grad_norm": 2614.4965010157607,
+      "learning_rate": 1.8845817658871942e-05,
+      "loss": 461.9733,
+      "step": 8470
+    },
+    {
+      "epoch": 0.16309182088748492,
+      "grad_norm": 2486.652627734411,
+      "learning_rate": 1.8842969596567765e-05,
+      "loss": 470.0281,
+      "step": 8480
+    },
+    {
+      "epoch": 0.16328414614796544,
+      "grad_norm": 2544.035398439454,
+      "learning_rate": 1.884011824040346e-05,
+      "loss": 479.6681,
+      "step": 8490
+    },
+    {
+      "epoch": 0.16347647140844596,
+      "grad_norm": 3201.5775268815228,
+      "learning_rate": 1.883726359144111e-05,
+      "loss": 464.7123,
+      "step": 8500
+    },
+    {
+      "epoch": 0.16366879666892648,
+      "grad_norm": 2701.606297904168,
+      "learning_rate": 1.8834405650744023e-05,
+      "loss": 454.8366,
+      "step": 8510
+    },
+    {
+      "epoch": 0.16386112192940702,
+      "grad_norm": 2836.1064810811786,
+      "learning_rate": 1.883154441937674e-05,
+      "loss": 465.3839,
+      "step": 8520
+    },
+    {
+      "epoch": 0.16405344718988754,
+      "grad_norm": 2525.191898053846,
+      "learning_rate": 1.8828679898405015e-05,
+      "loss": 444.021,
+      "step": 8530
+    },
+    {
+      "epoch": 0.16424577245036806,
+      "grad_norm": 2684.247548099907,
+      "learning_rate": 1.8825812088895835e-05,
+      "loss": 463.5473,
+      "step": 8540
+    },
+    {
+      "epoch": 0.16443809771084858,
+      "grad_norm": 2726.5514179263896,
+      "learning_rate": 1.882294099191742e-05,
+      "loss": 463.1741,
+      "step": 8550
+    },
+    {
+      "epoch": 0.1646304229713291,
+      "grad_norm": 2651.7242239650836,
+      "learning_rate": 1.88200666085392e-05,
+      "loss": 452.2353,
+      "step": 8560
+    },
+    {
+      "epoch": 0.16482274823180965,
+      "grad_norm": 2604.6174059849163,
+      "learning_rate": 1.8817188939831838e-05,
+      "loss": 462.9798,
+      "step": 8570
+    },
+    {
+      "epoch": 0.16501507349229017,
+      "grad_norm": 3034.7929991319465,
+      "learning_rate": 1.8814307986867214e-05,
+      "loss": 442.6848,
+      "step": 8580
+    },
+    {
+      "epoch": 0.16520739875277068,
+      "grad_norm": 2748.234644255372,
+      "learning_rate": 1.881142375071844e-05,
+      "loss": 435.3604,
+      "step": 8590
+    },
+    {
+      "epoch": 0.1653997240132512,
+      "grad_norm": 2939.0726535785348,
+      "learning_rate": 1.8808536232459844e-05,
+      "loss": 463.7852,
+      "step": 8600
+    },
+    {
+      "epoch": 0.16559204927373172,
+      "grad_norm": 2461.040963357909,
+      "learning_rate": 1.8805645433166976e-05,
+      "loss": 467.5215,
+      "step": 8610
+    },
+    {
+      "epoch": 0.16578437453421227,
+      "grad_norm": 2889.624399623839,
+      "learning_rate": 1.8802751353916618e-05,
+      "loss": 462.004,
+      "step": 8620
+    },
+    {
+      "epoch": 0.1659766997946928,
+      "grad_norm": 2518.8337305606265,
+      "learning_rate": 1.8799853995786763e-05,
+      "loss": 454.456,
+      "step": 8630
+    },
+    {
+      "epoch": 0.1661690250551733,
+      "grad_norm": 2468.759031477647,
+      "learning_rate": 1.8796953359856626e-05,
+      "loss": 433.6745,
+      "step": 8640
+    },
+    {
+      "epoch": 0.16636135031565383,
+      "grad_norm": 2723.029927952221,
+      "learning_rate": 1.879404944720665e-05,
+      "loss": 451.1816,
+      "step": 8650
+    },
+    {
+      "epoch": 0.16655367557613435,
+      "grad_norm": 2688.333294514795,
+      "learning_rate": 1.8791142258918496e-05,
+      "loss": 452.7185,
+      "step": 8660
+    },
+    {
+      "epoch": 0.1667460008366149,
+      "grad_norm": 2525.239445850532,
+      "learning_rate": 1.8788231796075037e-05,
+      "loss": 468.3976,
+      "step": 8670
+    },
+    {
+      "epoch": 0.1669383260970954,
+      "grad_norm": 2394.840633479949,
+      "learning_rate": 1.8785318059760384e-05,
+      "loss": 440.3838,
+      "step": 8680
+    },
+    {
+      "epoch": 0.16713065135757593,
+      "grad_norm": 2657.3516753061485,
+      "learning_rate": 1.8782401051059838e-05,
+      "loss": 446.2088,
+      "step": 8690
+    },
+    {
+      "epoch": 0.16732297661805645,
+      "grad_norm": 2572.4322494026887,
+      "learning_rate": 1.8779480771059954e-05,
+      "loss": 433.9772,
+      "step": 8700
+    },
+    {
+      "epoch": 0.16751530187853697,
+      "grad_norm": 2532.5718843841155,
+      "learning_rate": 1.8776557220848477e-05,
+      "loss": 445.3365,
+      "step": 8710
+    },
+    {
+      "epoch": 0.16770762713901752,
+      "grad_norm": 22990.43305437463,
+      "learning_rate": 1.8773630401514388e-05,
+      "loss": 457.1438,
+      "step": 8720
+    },
+    {
+      "epoch": 0.16789995239949804,
+      "grad_norm": 3069.2404349042918,
+      "learning_rate": 1.877070031414787e-05,
+      "loss": 459.5575,
+      "step": 8730
+    },
+    {
+      "epoch": 0.16809227765997856,
+      "grad_norm": 3377.1212825848525,
+      "learning_rate": 1.876776695984034e-05,
+      "loss": 444.4089,
+      "step": 8740
+    },
+    {
+      "epoch": 0.16828460292045908,
+      "grad_norm": 2746.360379939743,
+      "learning_rate": 1.8764830339684426e-05,
+      "loss": 441.5095,
+      "step": 8750
+    },
+    {
+      "epoch": 0.1684769281809396,
+      "grad_norm": 2821.7195242780617,
+      "learning_rate": 1.8761890454773965e-05,
+      "loss": 449.9007,
+      "step": 8760
+    },
+    {
+      "epoch": 0.16866925344142014,
+      "grad_norm": 2504.713284221407,
+      "learning_rate": 1.8758947306204012e-05,
+      "loss": 457.6672,
+      "step": 8770
+    },
+    {
+      "epoch": 0.16886157870190066,
+      "grad_norm": 2618.0904141710457,
+      "learning_rate": 1.8756000895070854e-05,
+      "loss": 431.5223,
+      "step": 8780
+    },
+    {
+      "epoch": 0.16905390396238118,
+      "grad_norm": 2377.2910977937076,
+      "learning_rate": 1.8753051222471968e-05,
+      "loss": 454.6217,
+      "step": 8790
+    },
+    {
+      "epoch": 0.1692462292228617,
+      "grad_norm": 2761.4493946217513,
+      "learning_rate": 1.8750098289506066e-05,
+      "loss": 442.4448,
+      "step": 8800
+    },
+    {
+      "epoch": 0.16943855448334222,
+      "grad_norm": 2633.4902607595054,
+      "learning_rate": 1.8747142097273057e-05,
+      "loss": 439.2786,
+      "step": 8810
+    },
+    {
+      "epoch": 0.16963087974382277,
+      "grad_norm": 2495.2262974002697,
+      "learning_rate": 1.8744182646874085e-05,
+      "loss": 448.5426,
+      "step": 8820
+    },
+    {
+      "epoch": 0.16982320500430328,
+      "grad_norm": 2628.0239127628934,
+      "learning_rate": 1.8741219939411494e-05,
+      "loss": 461.3018,
+      "step": 8830
+    },
+    {
+      "epoch": 0.1700155302647838,
+      "grad_norm": 2860.2357493755394,
+      "learning_rate": 1.873825397598884e-05,
+      "loss": 457.4539,
+      "step": 8840
+    },
+    {
+      "epoch": 0.17020785552526432,
+      "grad_norm": 2594.084372111208,
+      "learning_rate": 1.8735284757710897e-05,
+      "loss": 452.0995,
+      "step": 8850
+    },
+    {
+      "epoch": 0.17040018078574484,
+      "grad_norm": 2710.2131400228295,
+      "learning_rate": 1.873231228568365e-05,
+      "loss": 455.7155,
+      "step": 8860
+    },
+    {
+      "epoch": 0.1705925060462254,
+      "grad_norm": 2665.492809143777,
+      "learning_rate": 1.8729336561014294e-05,
+      "loss": 455.2856,
+      "step": 8870
+    },
+    {
+      "epoch": 0.1707848313067059,
+      "grad_norm": 2561.649795156198,
+      "learning_rate": 1.8726357584811242e-05,
+      "loss": 441.5276,
+      "step": 8880
+    },
+    {
+      "epoch": 0.17097715656718643,
+      "grad_norm": 2415.065192258531,
+      "learning_rate": 1.8723375358184107e-05,
+      "loss": 454.0223,
+      "step": 8890
+    },
+    {
+      "epoch": 0.17116948182766695,
+      "grad_norm": 2746.989536334842,
+      "learning_rate": 1.8720389882243722e-05,
+      "loss": 441.6445,
+      "step": 8900
+    },
+    {
+      "epoch": 0.17136180708814747,
+      "grad_norm": 2780.5009929539547,
+      "learning_rate": 1.871740115810213e-05,
+      "loss": 468.2937,
+      "step": 8910
+    },
+    {
+      "epoch": 0.171554132348628,
+      "grad_norm": 2632.9240695650924,
+      "learning_rate": 1.8714409186872578e-05,
+      "loss": 453.4708,
+      "step": 8920
+    },
+    {
+      "epoch": 0.17174645760910853,
+      "grad_norm": 18478.258419009937,
+      "learning_rate": 1.871141396966953e-05,
+      "loss": 458.452,
+      "step": 8930
+    },
+    {
+      "epoch": 0.17193878286958905,
+      "grad_norm": 2573.099238197207,
+      "learning_rate": 1.8708415507608647e-05,
+      "loss": 443.7543,
+      "step": 8940
+    },
+    {
+      "epoch": 0.17213110813006957,
+      "grad_norm": 2470.2442302466566,
+      "learning_rate": 1.8705413801806817e-05,
+      "loss": 446.0874,
+      "step": 8950
+    },
+    {
+      "epoch": 0.1723234333905501,
+      "grad_norm": 2944.5258247958905,
+      "learning_rate": 1.8702408853382114e-05,
+      "loss": 436.7016,
+      "step": 8960
+    },
+    {
+      "epoch": 0.1725157586510306,
+      "grad_norm": 2428.471223091907,
+      "learning_rate": 1.8699400663453842e-05,
+      "loss": 439.9705,
+      "step": 8970
+    },
+    {
+      "epoch": 0.17270808391151116,
+      "grad_norm": 2779.5787718497054,
+      "learning_rate": 1.8696389233142498e-05,
+      "loss": 436.2561,
+      "step": 8980
+    },
+    {
+      "epoch": 0.17290040917199168,
+      "grad_norm": 2801.7844262610247,
+      "learning_rate": 1.8693374563569792e-05,
+      "loss": 440.0648,
+      "step": 8990
+    },
+    {
+      "epoch": 0.1730927344324722,
+      "grad_norm": 2468.3211507447713,
+      "learning_rate": 1.8690356655858634e-05,
+      "loss": 444.2379,
+      "step": 9000
+    },
+    {
+      "epoch": 0.17328505969295271,
+      "grad_norm": 2950.9049283145578,
+      "learning_rate": 1.868733551113315e-05,
+      "loss": 457.8609,
+      "step": 9010
+    },
+    {
+      "epoch": 0.17347738495343323,
+      "grad_norm": 2623.2794604038363,
+      "learning_rate": 1.8684311130518663e-05,
+      "loss": 440.1009,
+      "step": 9020
+    },
+    {
+      "epoch": 0.17366971021391378,
+      "grad_norm": 2555.9618966907633,
+      "learning_rate": 1.8681283515141705e-05,
+      "loss": 437.3212,
+      "step": 9030
+    },
+    {
+      "epoch": 0.1738620354743943,
+      "grad_norm": 2927.0106727794473,
+      "learning_rate": 1.8678252666130016e-05,
+      "loss": 445.1681,
+      "step": 9040
+    },
+    {
+      "epoch": 0.17405436073487482,
+      "grad_norm": 2550.7522383571295,
+      "learning_rate": 1.8675218584612534e-05,
+      "loss": 437.728,
+      "step": 9050
+    },
+    {
+      "epoch": 0.17424668599535534,
+      "grad_norm": 2649.529870631337,
+      "learning_rate": 1.8672181271719406e-05,
+      "loss": 446.4435,
+      "step": 9060
+    },
+    {
+      "epoch": 0.17443901125583586,
+      "grad_norm": 2720.653259222087,
+      "learning_rate": 1.866914072858198e-05,
+      "loss": 455.3857,
+      "step": 9070
+    },
+    {
+      "epoch": 0.1746313365163164,
+      "grad_norm": 2804.8181920843176,
+      "learning_rate": 1.8666096956332805e-05,
+      "loss": 443.0305,
+      "step": 9080
+    },
+    {
+      "epoch": 0.17482366177679692,
+      "grad_norm": 2729.2129717363414,
+      "learning_rate": 1.8663049956105642e-05,
+      "loss": 440.492,
+      "step": 9090
+    },
+    {
+      "epoch": 0.17501598703727744,
+      "grad_norm": 2609.3562559386232,
+      "learning_rate": 1.8659999729035445e-05,
+      "loss": 437.4828,
+      "step": 9100
+    },
+    {
+      "epoch": 0.17520831229775796,
+      "grad_norm": 2451.155422918878,
+      "learning_rate": 1.8656946276258373e-05,
+      "loss": 434.4389,
+      "step": 9110
+    },
+    {
+      "epoch": 0.17540063755823848,
+      "grad_norm": 2512.828164974454,
+      "learning_rate": 1.8653889598911787e-05,
+      "loss": 446.8909,
+      "step": 9120
+    },
+    {
+      "epoch": 0.17559296281871903,
+      "grad_norm": 2806.0533918149017,
+      "learning_rate": 1.865082969813425e-05,
+      "loss": 435.8528,
+      "step": 9130
+    },
+    {
+      "epoch": 0.17578528807919955,
+      "grad_norm": 2548.2490140445034,
+      "learning_rate": 1.8647766575065523e-05,
+      "loss": 442.6416,
+      "step": 9140
+    },
+    {
+      "epoch": 0.17597761333968007,
+      "grad_norm": 2803.8978902199083,
+      "learning_rate": 1.864470023084657e-05,
+      "loss": 462.2274,
+      "step": 9150
+    },
+    {
+      "epoch": 0.17616993860016059,
+      "grad_norm": 2623.634212275952,
+      "learning_rate": 1.864163066661955e-05,
+      "loss": 442.0449,
+      "step": 9160
+    },
+    {
+      "epoch": 0.1763622638606411,
+      "grad_norm": 2470.2873050666863,
+      "learning_rate": 1.8638557883527833e-05,
+      "loss": 435.2117,
+      "step": 9170
+    },
+    {
+      "epoch": 0.17655458912112165,
+      "grad_norm": 2408.3928470769733,
+      "learning_rate": 1.8635481882715975e-05,
+      "loss": 426.3458,
+      "step": 9180
+    },
+    {
+      "epoch": 0.17674691438160217,
+      "grad_norm": 2335.9253478763662,
+      "learning_rate": 1.863240266532973e-05,
+      "loss": 426.5856,
+      "step": 9190
+    },
+    {
+      "epoch": 0.1769392396420827,
+      "grad_norm": 2516.6826056669256,
+      "learning_rate": 1.8629320232516063e-05,
+      "loss": 444.586,
+      "step": 9200
+    },
+    {
+      "epoch": 0.1771315649025632,
+      "grad_norm": 2295.24317998568,
+      "learning_rate": 1.862623458542313e-05,
+      "loss": 440.7661,
+      "step": 9210
+    },
+    {
+      "epoch": 0.17732389016304373,
+      "grad_norm": 2402.3695694836456,
+      "learning_rate": 1.862314572520028e-05,
+      "loss": 434.8042,
+      "step": 9220
+    },
+    {
+      "epoch": 0.17751621542352428,
+      "grad_norm": 2393.47139003244,
+      "learning_rate": 1.862005365299806e-05,
+      "loss": 428.6981,
+      "step": 9230
+    },
+    {
+      "epoch": 0.1777085406840048,
+      "grad_norm": 2601.4400613546713,
+      "learning_rate": 1.8616958369968223e-05,
+      "loss": 457.2905,
+      "step": 9240
+    },
+    {
+      "epoch": 0.17790086594448531,
+      "grad_norm": 2598.4376595452686,
+      "learning_rate": 1.8613859877263708e-05,
+      "loss": 437.9435,
+      "step": 9250
+    },
+    {
+      "epoch": 0.17809319120496583,
+      "grad_norm": 2480.4261551646964,
+      "learning_rate": 1.8610758176038647e-05,
+      "loss": 424.9386,
+      "step": 9260
+    },
+    {
+      "epoch": 0.17828551646544635,
+      "grad_norm": 2725.104556430378,
+      "learning_rate": 1.860765326744838e-05,
+      "loss": 431.2261,
+      "step": 9270
+    },
+    {
+      "epoch": 0.1784778417259269,
+      "grad_norm": 2712.507553065823,
+      "learning_rate": 1.8604545152649426e-05,
+      "loss": 443.1223,
+      "step": 9280
+    },
+    {
+      "epoch": 0.17867016698640742,
+      "grad_norm": 2372.062633725075,
+      "learning_rate": 1.860143383279952e-05,
+      "loss": 443.9025,
+      "step": 9290
+    },
+    {
+      "epoch": 0.17886249224688794,
+      "grad_norm": 2683.398947528759,
+      "learning_rate": 1.859831930905756e-05,
+      "loss": 430.8124,
+      "step": 9300
+    },
+    {
+      "epoch": 0.17905481750736846,
+      "grad_norm": 2387.22497436682,
+      "learning_rate": 1.8595201582583668e-05,
+      "loss": 442.3059,
+      "step": 9310
+    },
+    {
+      "epoch": 0.17924714276784898,
+      "grad_norm": 2537.502536146372,
+      "learning_rate": 1.859208065453914e-05,
+      "loss": 428.5348,
+      "step": 9320
+    },
+    {
+      "epoch": 0.17943946802832952,
+      "grad_norm": 2647.7503285995936,
+      "learning_rate": 1.8588956526086472e-05,
+      "loss": 460.5251,
+      "step": 9330
+    },
+    {
+      "epoch": 0.17963179328881004,
+      "grad_norm": 2761.5945195069867,
+      "learning_rate": 1.8585829198389347e-05,
+      "loss": 430.1305,
+      "step": 9340
+    },
+    {
+      "epoch": 0.17982411854929056,
+      "grad_norm": 2527.638340963992,
+      "learning_rate": 1.8582698672612646e-05,
+      "loss": 427.3626,
+      "step": 9350
+    },
+    {
+      "epoch": 0.18001644380977108,
+      "grad_norm": 2681.904507082835,
+      "learning_rate": 1.8579564949922438e-05,
+      "loss": 427.4606,
+      "step": 9360
+    },
+    {
+      "epoch": 0.1802087690702516,
+      "grad_norm": 2520.4384581669665,
+      "learning_rate": 1.8576428031485984e-05,
+      "loss": 422.3001,
+      "step": 9370
+    },
+    {
+      "epoch": 0.18040109433073215,
+      "grad_norm": 2282.314596930165,
+      "learning_rate": 1.8573287918471728e-05,
+      "loss": 432.4351,
+      "step": 9380
+    },
+    {
+      "epoch": 0.18059341959121267,
+      "grad_norm": 2695.8164683961345,
+      "learning_rate": 1.8570144612049322e-05,
+      "loss": 430.8169,
+      "step": 9390
+    },
+    {
+      "epoch": 0.1807857448516932,
+      "grad_norm": 3000.4497862681146,
+      "learning_rate": 1.856699811338958e-05,
+      "loss": 430.5135,
+      "step": 9400
+    },
+    {
+      "epoch": 0.1809780701121737,
+      "grad_norm": 2584.238550307117,
+      "learning_rate": 1.8563848423664536e-05,
+      "loss": 442.1215,
+      "step": 9410
+    },
+    {
+      "epoch": 0.18117039537265422,
+      "grad_norm": 2200.2629717432324,
+      "learning_rate": 1.8560695544047388e-05,
+      "loss": 452.2628,
+      "step": 9420
+    },
+    {
+      "epoch": 0.18136272063313474,
+      "grad_norm": 2675.380971957908,
+      "learning_rate": 1.8557539475712538e-05,
+      "loss": 453.2036,
+      "step": 9430
+    },
+    {
+      "epoch": 0.1815550458936153,
+      "grad_norm": 2785.1580728918234,
+      "learning_rate": 1.855438021983556e-05,
+      "loss": 430.1964,
+      "step": 9440
+    },
+    {
+      "epoch": 0.1817473711540958,
+      "grad_norm": 2756.895002991828,
+      "learning_rate": 1.8551217777593233e-05,
+      "loss": 430.8823,
+      "step": 9450
+    },
+    {
+      "epoch": 0.18193969641457633,
+      "grad_norm": 2614.708846604072,
+      "learning_rate": 1.8548052150163514e-05,
+      "loss": 427.9357,
+      "step": 9460
+    },
+    {
+      "epoch": 0.18213202167505685,
+      "grad_norm": 2599.5556389036838,
+      "learning_rate": 1.8544883338725544e-05,
+      "loss": 416.8458,
+      "step": 9470
+    },
+    {
+      "epoch": 0.18232434693553737,
+      "grad_norm": 3151.279742287895,
+      "learning_rate": 1.8541711344459652e-05,
+      "loss": 430.7875,
+      "step": 9480
+    },
+    {
+      "epoch": 0.18251667219601792,
+      "grad_norm": 3042.373648887261,
+      "learning_rate": 1.8538536168547353e-05,
+      "loss": 443.7286,
+      "step": 9490
+    },
+    {
+      "epoch": 0.18270899745649843,
+      "grad_norm": 2903.0010357803503,
+      "learning_rate": 1.8535357812171356e-05,
+      "loss": 446.3522,
+      "step": 9500
+    },
+    {
+      "epoch": 0.18290132271697895,
+      "grad_norm": 2767.1010436271704,
+      "learning_rate": 1.8532176276515538e-05,
+      "loss": 429.4558,
+      "step": 9510
+    },
+    {
+      "epoch": 0.18309364797745947,
+      "grad_norm": 2603.6546380581676,
+      "learning_rate": 1.8528991562764967e-05,
+      "loss": 431.4535,
+      "step": 9520
+    },
+    {
+      "epoch": 0.18328597323794,
+      "grad_norm": 2559.83977407247,
+      "learning_rate": 1.85258036721059e-05,
+      "loss": 425.6351,
+      "step": 9530
+    },
+    {
+      "epoch": 0.18347829849842054,
+      "grad_norm": 2771.59820476445,
+      "learning_rate": 1.8522612605725777e-05,
+      "loss": 436.8646,
+      "step": 9540
+    },
+    {
+      "epoch": 0.18367062375890106,
+      "grad_norm": 2550.3542997155123,
+      "learning_rate": 1.8519418364813215e-05,
+      "loss": 429.8599,
+      "step": 9550
+    },
+    {
+      "epoch": 0.18386294901938158,
+      "grad_norm": 2349.4890345142303,
+      "learning_rate": 1.851622095055801e-05,
+      "loss": 425.3041,
+      "step": 9560
+    },
+    {
+      "epoch": 0.1840552742798621,
+      "grad_norm": 2356.380351723672,
+      "learning_rate": 1.8513020364151155e-05,
+      "loss": 428.1639,
+      "step": 9570
+    },
+    {
+      "epoch": 0.18424759954034262,
+      "grad_norm": 3117.0554903200723,
+      "learning_rate": 1.850981660678481e-05,
+      "loss": 427.749,
+      "step": 9580
+    },
+    {
+      "epoch": 0.18443992480082316,
+      "grad_norm": 2663.287913875028,
+      "learning_rate": 1.8506609679652323e-05,
+      "loss": 426.2586,
+      "step": 9590
+    },
+    {
+      "epoch": 0.18463225006130368,
+      "grad_norm": 2538.139707007356,
+      "learning_rate": 1.8503399583948224e-05,
+      "loss": 438.3786,
+      "step": 9600
+    },
+    {
+      "epoch": 0.1848245753217842,
+      "grad_norm": 2473.0098406620523,
+      "learning_rate": 1.8500186320868215e-05,
+      "loss": 419.1407,
+      "step": 9610
+    },
+    {
+      "epoch": 0.18501690058226472,
+      "grad_norm": 2508.558120856946,
+      "learning_rate": 1.8496969891609186e-05,
+      "loss": 444.3701,
+      "step": 9620
+    },
+    {
+      "epoch": 0.18520922584274524,
+      "grad_norm": 2503.30195361973,
+      "learning_rate": 1.8493750297369208e-05,
+      "loss": 442.5193,
+      "step": 9630
+    },
+    {
+      "epoch": 0.1854015511032258,
+      "grad_norm": 2408.15371260309,
+      "learning_rate": 1.849052753934752e-05,
+      "loss": 438.0247,
+      "step": 9640
+    },
+    {
+      "epoch": 0.1855938763637063,
+      "grad_norm": 2994.015716679271,
+      "learning_rate": 1.8487301618744552e-05,
+      "loss": 440.4727,
+      "step": 9650
+    },
+    {
+      "epoch": 0.18578620162418683,
+      "grad_norm": 2383.8264213177185,
+      "learning_rate": 1.84840725367619e-05,
+      "loss": 415.5592,
+      "step": 9660
+    },
+    {
+      "epoch": 0.18597852688466734,
+      "grad_norm": 2437.0946812610778,
+      "learning_rate": 1.8480840294602352e-05,
+      "loss": 421.3766,
+      "step": 9670
+    },
+    {
+      "epoch": 0.18617085214514786,
+      "grad_norm": 2591.1241802281756,
+      "learning_rate": 1.8477604893469857e-05,
+      "loss": 438.2731,
+      "step": 9680
+    },
+    {
+      "epoch": 0.1863631774056284,
+      "grad_norm": 2426.706370070809,
+      "learning_rate": 1.847436633456955e-05,
+      "loss": 436.0906,
+      "step": 9690
+    },
+    {
+      "epoch": 0.18655550266610893,
+      "grad_norm": 2249.9550569722387,
+      "learning_rate": 1.8471124619107744e-05,
+      "loss": 416.2272,
+      "step": 9700
+    },
+    {
+      "epoch": 0.18674782792658945,
+      "grad_norm": 2218.356740664979,
+      "learning_rate": 1.846787974829192e-05,
+      "loss": 422.8797,
+      "step": 9710
+    },
+    {
+      "epoch": 0.18694015318706997,
+      "grad_norm": 2675.610068113475,
+      "learning_rate": 1.8464631723330745e-05,
+      "loss": 432.1363,
+      "step": 9720
+    },
+    {
+      "epoch": 0.1871324784475505,
+      "grad_norm": 2308.209685452142,
+      "learning_rate": 1.8461380545434054e-05,
+      "loss": 415.4586,
+      "step": 9730
+    },
+    {
+      "epoch": 0.18732480370803103,
+      "grad_norm": 2387.1713189371644,
+      "learning_rate": 1.8458126215812848e-05,
+      "loss": 415.1146,
+      "step": 9740
+    },
+    {
+      "epoch": 0.18751712896851155,
+      "grad_norm": 2676.7830843197266,
+      "learning_rate": 1.845486873567932e-05,
+      "loss": 424.1718,
+      "step": 9750
+    },
+    {
+      "epoch": 0.18770945422899207,
+      "grad_norm": 2472.5215166530497,
+      "learning_rate": 1.8451608106246822e-05,
+      "loss": 414.0548,
+      "step": 9760
+    },
+    {
+      "epoch": 0.1879017794894726,
+      "grad_norm": 10329.26524510405,
+      "learning_rate": 1.8448344328729893e-05,
+      "loss": 433.7043,
+      "step": 9770
+    },
+    {
+      "epoch": 0.1880941047499531,
+      "grad_norm": 2483.168622703879,
+      "learning_rate": 1.8445077404344226e-05,
+      "loss": 413.1561,
+      "step": 9780
+    },
+    {
+      "epoch": 0.18828643001043366,
+      "grad_norm": 2367.901590900312,
+      "learning_rate": 1.8441807334306702e-05,
+      "loss": 419.4771,
+      "step": 9790
+    },
+    {
+      "epoch": 0.18847875527091418,
+      "grad_norm": 2407.46881799832,
+      "learning_rate": 1.8438534119835365e-05,
+      "loss": 409.0065,
+      "step": 9800
+    },
+    {
+      "epoch": 0.1886710805313947,
+      "grad_norm": 2634.4653171152536,
+      "learning_rate": 1.8435257762149436e-05,
+      "loss": 426.1817,
+      "step": 9810
+    },
+    {
+      "epoch": 0.18886340579187522,
+      "grad_norm": 2282.70870343972,
+      "learning_rate": 1.8431978262469305e-05,
+      "loss": 431.1925,
+      "step": 9820
+    },
+    {
+      "epoch": 0.18905573105235574,
+      "grad_norm": 2429.9620623316027,
+      "learning_rate": 1.8428695622016532e-05,
+      "loss": 414.264,
+      "step": 9830
+    },
+    {
+      "epoch": 0.18924805631283628,
+      "grad_norm": 2594.6612237535387,
+      "learning_rate": 1.8425409842013843e-05,
+      "loss": 428.9275,
+      "step": 9840
+    },
+    {
+      "epoch": 0.1894403815733168,
+      "grad_norm": 2650.7523909612446,
+      "learning_rate": 1.8422120923685135e-05,
+      "loss": 410.5828,
+      "step": 9850
+    },
+    {
+      "epoch": 0.18963270683379732,
+      "grad_norm": 2444.278485671017,
+      "learning_rate": 1.8418828868255484e-05,
+      "loss": 418.427,
+      "step": 9860
+    },
+    {
+      "epoch": 0.18982503209427784,
+      "grad_norm": 2456.8782042718963,
+      "learning_rate": 1.8415533676951117e-05,
+      "loss": 426.9792,
+      "step": 9870
+    },
+    {
+      "epoch": 0.19001735735475836,
+      "grad_norm": 2491.4483740263368,
+      "learning_rate": 1.8412235350999444e-05,
+      "loss": 432.1644,
+      "step": 9880
+    },
+    {
+      "epoch": 0.19020968261523888,
+      "grad_norm": 2476.0588561727045,
+      "learning_rate": 1.840893389162903e-05,
+      "loss": 462.9826,
+      "step": 9890
+    },
+    {
+      "epoch": 0.19040200787571943,
+      "grad_norm": 2732.8547628771053,
+      "learning_rate": 1.8405629300069626e-05,
+      "loss": 427.2529,
+      "step": 9900
+    },
+    {
+      "epoch": 0.19059433313619994,
+      "grad_norm": 2595.2345543319143,
+      "learning_rate": 1.840232157755213e-05,
+      "loss": 426.8976,
+      "step": 9910
+    },
+    {
+      "epoch": 0.19078665839668046,
+      "grad_norm": 2884.291239899689,
+      "learning_rate": 1.8399010725308616e-05,
+      "loss": 415.4808,
+      "step": 9920
+    },
+    {
+      "epoch": 0.19097898365716098,
+      "grad_norm": 2383.5731911539187,
+      "learning_rate": 1.839569674457232e-05,
+      "loss": 424.0588,
+      "step": 9930
+    },
+    {
+      "epoch": 0.1911713089176415,
+      "grad_norm": 2524.5105683796714,
+      "learning_rate": 1.8392379636577647e-05,
+      "loss": 439.3013,
+      "step": 9940
+    },
+    {
+      "epoch": 0.19136363417812205,
+      "grad_norm": 2566.1891981854415,
+      "learning_rate": 1.8389059402560165e-05,
+      "loss": 427.5253,
+      "step": 9950
+    },
+    {
+      "epoch": 0.19155595943860257,
+      "grad_norm": 2357.3355139382893,
+      "learning_rate": 1.8385736043756605e-05,
+      "loss": 421.2647,
+      "step": 9960
+    },
+    {
+      "epoch": 0.1917482846990831,
+      "grad_norm": 2578.7377395645394,
+      "learning_rate": 1.838240956140486e-05,
+      "loss": 427.3051,
+      "step": 9970
+    },
+    {
+      "epoch": 0.1919406099595636,
+      "grad_norm": 2396.6699178415674,
+      "learning_rate": 1.8379079956743996e-05,
+      "loss": 441.4921,
+      "step": 9980
+    },
+    {
+      "epoch": 0.19213293522004413,
+      "grad_norm": 28306.667678786922,
+      "learning_rate": 1.8375747231014233e-05,
+      "loss": 425.5845,
+      "step": 9990
+    },
+    {
+      "epoch": 0.19232526048052467,
+      "grad_norm": 2911.719873512512,
+      "learning_rate": 1.8372411385456956e-05,
+      "loss": 427.8772,
+      "step": 10000
+    },
+    {
+      "epoch": 0.1925175857410052,
+      "grad_norm": 2397.3024696914417,
+      "learning_rate": 1.8369072421314717e-05,
+      "loss": 415.5873,
+      "step": 10010
+    },
+    {
+      "epoch": 0.1927099110014857,
+      "grad_norm": 2596.583132820523,
+      "learning_rate": 1.8365730339831212e-05,
+      "loss": 412.709,
+      "step": 10020
+    },
+    {
+      "epoch": 0.19290223626196623,
+      "grad_norm": 2447.7325883524313,
+      "learning_rate": 1.8362385142251328e-05,
+      "loss": 424.9756,
+      "step": 10030
+    },
+    {
+      "epoch": 0.19309456152244675,
+      "grad_norm": 2610.1759248125973,
+      "learning_rate": 1.8359036829821085e-05,
+      "loss": 415.7236,
+      "step": 10040
+    },
+    {
+      "epoch": 0.1932868867829273,
+      "grad_norm": 2568.3642489728477,
+      "learning_rate": 1.8355685403787677e-05,
+      "loss": 410.082,
+      "step": 10050
+    },
+    {
+      "epoch": 0.19347921204340782,
+      "grad_norm": 2297.950766179293,
+      "learning_rate": 1.8352330865399457e-05,
+      "loss": 407.8644,
+      "step": 10060
+    },
+    {
+      "epoch": 0.19367153730388834,
+      "grad_norm": 2746.3368471131,
+      "learning_rate": 1.834897321590593e-05,
+      "loss": 415.3305,
+      "step": 10070
+    },
+    {
+      "epoch": 0.19386386256436886,
+      "grad_norm": 2174.4847725716004,
+      "learning_rate": 1.8345612456557767e-05,
+      "loss": 404.8042,
+      "step": 10080
+    },
+    {
+      "epoch": 0.19405618782484937,
+      "grad_norm": 2282.3197924485125,
+      "learning_rate": 1.83422485886068e-05,
+      "loss": 408.8012,
+      "step": 10090
+    },
+    {
+      "epoch": 0.19424851308532992,
+      "grad_norm": 2823.9083708204867,
+      "learning_rate": 1.833888161330601e-05,
+      "loss": 435.0374,
+      "step": 10100
+    },
+    {
+      "epoch": 0.19444083834581044,
+      "grad_norm": 2285.562841291326,
+      "learning_rate": 1.833551153190954e-05,
+      "loss": 410.8455,
+      "step": 10110
+    },
+    {
+      "epoch": 0.19463316360629096,
+      "grad_norm": 2394.9424523734256,
+      "learning_rate": 1.8332138345672686e-05,
+      "loss": 408.7798,
+      "step": 10120
+    },
+    {
+      "epoch": 0.19482548886677148,
+      "grad_norm": 2483.776180779221,
+      "learning_rate": 1.832876205585191e-05,
+      "loss": 419.2998,
+      "step": 10130
+    },
+    {
+      "epoch": 0.195017814127252,
+      "grad_norm": 2348.916678173444,
+      "learning_rate": 1.8325382663704826e-05,
+      "loss": 414.339,
+      "step": 10140
+    },
+    {
+      "epoch": 0.19521013938773255,
+      "grad_norm": 2257.972693658309,
+      "learning_rate": 1.8322000170490194e-05,
+      "loss": 408.5225,
+      "step": 10150
+    },
+    {
+      "epoch": 0.19540246464821306,
+      "grad_norm": 2546.3267920016447,
+      "learning_rate": 1.831861457746794e-05,
+      "loss": 417.0976,
+      "step": 10160
+    },
+    {
+      "epoch": 0.19559478990869358,
+      "grad_norm": 2321.8695956707966,
+      "learning_rate": 1.8315225885899144e-05,
+      "loss": 422.5677,
+      "step": 10170
+    },
+    {
+      "epoch": 0.1957871151691741,
+      "grad_norm": 2483.177157201622,
+      "learning_rate": 1.8311834097046038e-05,
+      "loss": 407.5825,
+      "step": 10180
+    },
+    {
+      "epoch": 0.19597944042965462,
+      "grad_norm": 3199.911756595318,
+      "learning_rate": 1.8308439212172e-05,
+      "loss": 420.1181,
+      "step": 10190
+    },
+    {
+      "epoch": 0.19617176569013517,
+      "grad_norm": 2405.0842741328424,
+      "learning_rate": 1.830504123254158e-05,
+      "loss": 428.5215,
+      "step": 10200
+    },
+    {
+      "epoch": 0.1963640909506157,
+      "grad_norm": 2853.6141960433793,
+      "learning_rate": 1.830164015942046e-05,
+      "loss": 416.0092,
+      "step": 10210
+    },
+    {
+      "epoch": 0.1965564162110962,
+      "grad_norm": 6150.725113057045,
+      "learning_rate": 1.8298235994075488e-05,
+      "loss": 407.6205,
+      "step": 10220
+    },
+    {
+      "epoch": 0.19674874147157673,
+      "grad_norm": 2941.968311887401,
+      "learning_rate": 1.829482873777466e-05,
+      "loss": 425.1357,
+      "step": 10230
+    },
+    {
+      "epoch": 0.19694106673205725,
+      "grad_norm": 13541.236660979861,
+      "learning_rate": 1.8291418391787116e-05,
+      "loss": 430.1037,
+      "step": 10240
+    },
+    {
+      "epoch": 0.1971333919925378,
+      "grad_norm": 2687.732781345709,
+      "learning_rate": 1.8288004957383162e-05,
+      "loss": 416.6701,
+      "step": 10250
+    },
+    {
+      "epoch": 0.1973257172530183,
+      "grad_norm": 2488.094560088087,
+      "learning_rate": 1.8284588435834242e-05,
+      "loss": 415.4613,
+      "step": 10260
+    },
+    {
+      "epoch": 0.19751804251349883,
+      "grad_norm": 2585.5638282418886,
+      "learning_rate": 1.828116882841295e-05,
+      "loss": 416.3445,
+      "step": 10270
+    },
+    {
+      "epoch": 0.19771036777397935,
+      "grad_norm": 2391.2602293948394,
+      "learning_rate": 1.8277746136393042e-05,
+      "loss": 428.4682,
+      "step": 10280
+    },
+    {
+      "epoch": 0.19790269303445987,
+      "grad_norm": 2629.0570084707033,
+      "learning_rate": 1.827432036104941e-05,
+      "loss": 411.7794,
+      "step": 10290
+    },
+    {
+      "epoch": 0.19809501829494042,
+      "grad_norm": 2444.479259928908,
+      "learning_rate": 1.8270891503658096e-05,
+      "loss": 411.8743,
+      "step": 10300
+    },
+    {
+      "epoch": 0.19828734355542094,
+      "grad_norm": 2421.4364067604934,
+      "learning_rate": 1.8267459565496298e-05,
+      "loss": 417.5099,
+      "step": 10310
+    },
+    {
+      "epoch": 0.19847966881590146,
+      "grad_norm": 2475.220447059186,
+      "learning_rate": 1.8264024547842346e-05,
+      "loss": 412.6634,
+      "step": 10320
+    },
+    {
+      "epoch": 0.19867199407638197,
+      "grad_norm": 2148.9450390888987,
+      "learning_rate": 1.8260586451975745e-05,
+      "loss": 406.1828,
+      "step": 10330
+    },
+    {
+      "epoch": 0.1988643193368625,
+      "grad_norm": 2668.7152817319225,
+      "learning_rate": 1.825714527917711e-05,
+      "loss": 422.5888,
+      "step": 10340
+    },
+    {
+      "epoch": 0.199056644597343,
+      "grad_norm": 2846.3524448522394,
+      "learning_rate": 1.8253701030728235e-05,
+      "loss": 417.4071,
+      "step": 10350
+    },
+    {
+      "epoch": 0.19924896985782356,
+      "grad_norm": 2663.3106333984706,
+      "learning_rate": 1.8250253707912036e-05,
+      "loss": 421.4405,
+      "step": 10360
+    },
+    {
+      "epoch": 0.19944129511830408,
+      "grad_norm": 2278.902298079059,
+      "learning_rate": 1.8246803312012593e-05,
+      "loss": 399.4237,
+      "step": 10370
+    },
+    {
+      "epoch": 0.1996336203787846,
+      "grad_norm": 2258.4201981721167,
+      "learning_rate": 1.8243349844315116e-05,
+      "loss": 404.1829,
+      "step": 10380
+    },
+    {
+      "epoch": 0.19982594563926512,
+      "grad_norm": 2696.685765780273,
+      "learning_rate": 1.8239893306105966e-05,
+      "loss": 404.6022,
+      "step": 10390
+    },
+    {
+      "epoch": 0.20001827089974564,
+      "grad_norm": 2250.8881015934917,
+      "learning_rate": 1.823643369867264e-05,
+      "loss": 405.7177,
+      "step": 10400
+    },
+    {
+      "epoch": 0.20021059616022618,
+      "grad_norm": 2173.0644028037655,
+      "learning_rate": 1.8232971023303798e-05,
+      "loss": 396.2673,
+      "step": 10410
+    },
+    {
+      "epoch": 0.2004029214207067,
+      "grad_norm": 2752.758966792303,
+      "learning_rate": 1.8229505281289216e-05,
+      "loss": 419.2991,
+      "step": 10420
+    },
+    {
+      "epoch": 0.20059524668118722,
+      "grad_norm": 2492.381749055844,
+      "learning_rate": 1.8226036473919836e-05,
+      "loss": 398.8003,
+      "step": 10430
+    },
+    {
+      "epoch": 0.20078757194166774,
+      "grad_norm": 2327.00730066168,
+      "learning_rate": 1.8222564602487724e-05,
+      "loss": 408.9353,
+      "step": 10440
+    },
+    {
+      "epoch": 0.20097989720214826,
+      "grad_norm": 2359.3698107997043,
+      "learning_rate": 1.82190896682861e-05,
+      "loss": 406.7283,
+      "step": 10450
+    },
+    {
+      "epoch": 0.2011722224626288,
+      "grad_norm": 2594.1865103408004,
+      "learning_rate": 1.8215611672609316e-05,
+      "loss": 392.2045,
+      "step": 10460
+    },
+    {
+      "epoch": 0.20136454772310933,
+      "grad_norm": 2727.18681436104,
+      "learning_rate": 1.821213061675287e-05,
+      "loss": 406.7791,
+      "step": 10470
+    },
+    {
+      "epoch": 0.20155687298358985,
+      "grad_norm": 2739.1610554666677,
+      "learning_rate": 1.8208646502013395e-05,
+      "loss": 407.5338,
+      "step": 10480
+    },
+    {
+      "epoch": 0.20174919824407037,
+      "grad_norm": 2392.15464023485,
+      "learning_rate": 1.820515932968867e-05,
+      "loss": 406.5358,
+      "step": 10490
+    },
+    {
+      "epoch": 0.20194152350455088,
+      "grad_norm": 2452.371794680503,
+      "learning_rate": 1.8201669101077608e-05,
+      "loss": 409.1662,
+      "step": 10500
+    },
+    {
+      "epoch": 0.20213384876503143,
+      "grad_norm": 2394.1563961392794,
+      "learning_rate": 1.819817581748026e-05,
+      "loss": 426.1443,
+      "step": 10510
+    },
+    {
+      "epoch": 0.20232617402551195,
+      "grad_norm": 2527.0573621370754,
+      "learning_rate": 1.8194679480197817e-05,
+      "loss": 408.445,
+      "step": 10520
+    },
+    {
+      "epoch": 0.20251849928599247,
+      "grad_norm": 2493.0292909377595,
+      "learning_rate": 1.8191180090532608e-05,
+      "loss": 411.8142,
+      "step": 10530
+    },
+    {
+      "epoch": 0.202710824546473,
+      "grad_norm": 2706.38906962919,
+      "learning_rate": 1.8187677649788097e-05,
+      "loss": 401.5184,
+      "step": 10540
+    },
+    {
+      "epoch": 0.2029031498069535,
+      "grad_norm": 2332.4821868204363,
+      "learning_rate": 1.8184172159268884e-05,
+      "loss": 417.2245,
+      "step": 10550
+    },
+    {
+      "epoch": 0.20309547506743406,
+      "grad_norm": 2431.6873975187805,
+      "learning_rate": 1.818066362028071e-05,
+      "loss": 408.8991,
+      "step": 10560
+    },
+    {
+      "epoch": 0.20328780032791458,
+      "grad_norm": 2241.9349140611075,
+      "learning_rate": 1.8177152034130442e-05,
+      "loss": 420.855,
+      "step": 10570
+    },
+    {
+      "epoch": 0.2034801255883951,
+      "grad_norm": 2713.146169055902,
+      "learning_rate": 1.8173637402126093e-05,
+      "loss": 421.2833,
+      "step": 10580
+    },
+    {
+      "epoch": 0.2036724508488756,
+      "grad_norm": 2450.1724149154816,
+      "learning_rate": 1.8170119725576808e-05,
+      "loss": 411.6564,
+      "step": 10590
+    },
+    {
+      "epoch": 0.20386477610935613,
+      "grad_norm": 2394.9678777535923,
+      "learning_rate": 1.8166599005792852e-05,
+      "loss": 405.5297,
+      "step": 10600
+    },
+    {
+      "epoch": 0.20405710136983668,
+      "grad_norm": 2388.8301954499243,
+      "learning_rate": 1.816307524408565e-05,
+      "loss": 398.6396,
+      "step": 10610
+    },
+    {
+      "epoch": 0.2042494266303172,
+      "grad_norm": 2616.6871594197464,
+      "learning_rate": 1.8159548441767732e-05,
+      "loss": 403.034,
+      "step": 10620
+    },
+    {
+      "epoch": 0.20444175189079772,
+      "grad_norm": 2345.9857639268953,
+      "learning_rate": 1.8156018600152777e-05,
+      "loss": 408.3324,
+      "step": 10630
+    },
+    {
+      "epoch": 0.20463407715127824,
+      "grad_norm": 2600.5443813452302,
+      "learning_rate": 1.81524857205556e-05,
+      "loss": 410.9663,
+      "step": 10640
+    },
+    {
+      "epoch": 0.20482640241175876,
+      "grad_norm": 2491.259876868476,
+      "learning_rate": 1.814894980429213e-05,
+      "loss": 428.4407,
+      "step": 10650
+    },
+    {
+      "epoch": 0.2050187276722393,
+      "grad_norm": 2227.1327928258866,
+      "learning_rate": 1.8145410852679447e-05,
+      "loss": 409.756,
+      "step": 10660
+    },
+    {
+      "epoch": 0.20521105293271982,
+      "grad_norm": 2400.673689900987,
+      "learning_rate": 1.8141868867035745e-05,
+      "loss": 400.873,
+      "step": 10670
+    },
+    {
+      "epoch": 0.20540337819320034,
+      "grad_norm": 2463.62094591511,
+      "learning_rate": 1.8138323848680354e-05,
+      "loss": 399.8795,
+      "step": 10680
+    },
+    {
+      "epoch": 0.20559570345368086,
+      "grad_norm": 2262.0422015220383,
+      "learning_rate": 1.813477579893374e-05,
+      "loss": 402.7038,
+      "step": 10690
+    },
+    {
+      "epoch": 0.20578802871416138,
+      "grad_norm": 2747.5668526902605,
+      "learning_rate": 1.8131224719117497e-05,
+      "loss": 403.4552,
+      "step": 10700
+    },
+    {
+      "epoch": 0.20598035397464193,
+      "grad_norm": 2403.8496516176165,
+      "learning_rate": 1.8127670610554332e-05,
+      "loss": 406.5367,
+      "step": 10710
+    },
+    {
+      "epoch": 0.20617267923512245,
+      "grad_norm": 2206.889261105498,
+      "learning_rate": 1.81241134745681e-05,
+      "loss": 399.0284,
+      "step": 10720
+    },
+    {
+      "epoch": 0.20636500449560297,
+      "grad_norm": 2559.568540180695,
+      "learning_rate": 1.812055331248377e-05,
+      "loss": 411.964,
+      "step": 10730
+    },
+    {
+      "epoch": 0.20655732975608349,
+      "grad_norm": 2581.72622460684,
+      "learning_rate": 1.811699012562745e-05,
+      "loss": 410.2328,
+      "step": 10740
+    },
+    {
+      "epoch": 0.206749655016564,
+      "grad_norm": 2680.00126818257,
+      "learning_rate": 1.8113423915326362e-05,
+      "loss": 419.1748,
+      "step": 10750
+    },
+    {
+      "epoch": 0.20694198027704455,
+      "grad_norm": 2513.56967677966,
+      "learning_rate": 1.8109854682908864e-05,
+      "loss": 418.2968,
+      "step": 10760
+    },
+    {
+      "epoch": 0.20713430553752507,
+      "grad_norm": 2623.4003108539373,
+      "learning_rate": 1.8106282429704436e-05,
+      "loss": 410.6109,
+      "step": 10770
+    },
+    {
+      "epoch": 0.2073266307980056,
+      "grad_norm": 2405.1786942492386,
+      "learning_rate": 1.810270715704368e-05,
+      "loss": 420.2175,
+      "step": 10780
+    },
+    {
+      "epoch": 0.2075189560584861,
+      "grad_norm": 2570.270819036147,
+      "learning_rate": 1.809912886625833e-05,
+      "loss": 404.7516,
+      "step": 10790
+    },
+    {
+      "epoch": 0.20771128131896663,
+      "grad_norm": 2297.8982421671208,
+      "learning_rate": 1.8095547558681243e-05,
+      "loss": 420.8726,
+      "step": 10800
+    },
+    {
+      "epoch": 0.20790360657944715,
+      "grad_norm": 2471.2642815329295,
+      "learning_rate": 1.809196323564639e-05,
+      "loss": 416.5768,
+      "step": 10810
+    },
+    {
+      "epoch": 0.2080959318399277,
+      "grad_norm": 2485.2638635002304,
+      "learning_rate": 1.8088375898488873e-05,
+      "loss": 397.1862,
+      "step": 10820
+    },
+    {
+      "epoch": 0.20828825710040821,
+      "grad_norm": 2240.332621458524,
+      "learning_rate": 1.808478554854492e-05,
+      "loss": 409.2945,
+      "step": 10830
+    },
+    {
+      "epoch": 0.20848058236088873,
+      "grad_norm": 2310.0913790318946,
+      "learning_rate": 1.8081192187151873e-05,
+      "loss": 403.7675,
+      "step": 10840
+    },
+    {
+      "epoch": 0.20867290762136925,
+      "grad_norm": 2197.1117063416928,
+      "learning_rate": 1.8077595815648202e-05,
+      "loss": 397.342,
+      "step": 10850
+    },
+    {
+      "epoch": 0.20886523288184977,
+      "grad_norm": 2195.730145545457,
+      "learning_rate": 1.8073996435373494e-05,
+      "loss": 407.2368,
+      "step": 10860
+    },
+    {
+      "epoch": 0.20905755814233032,
+      "grad_norm": 2427.4795901523935,
+      "learning_rate": 1.8070394047668466e-05,
+      "loss": 403.1966,
+      "step": 10870
+    },
+    {
+      "epoch": 0.20924988340281084,
+      "grad_norm": 2226.296377914443,
+      "learning_rate": 1.8066788653874936e-05,
+      "loss": 390.1516,
+      "step": 10880
+    },
+    {
+      "epoch": 0.20944220866329136,
+      "grad_norm": 2542.001444994491,
+      "learning_rate": 1.806318025533586e-05,
+      "loss": 403.8007,
+      "step": 10890
+    },
+    {
+      "epoch": 0.20963453392377188,
+      "grad_norm": 2509.4441559480447,
+      "learning_rate": 1.805956885339531e-05,
+      "loss": 413.4315,
+      "step": 10900
+    },
+    {
+      "epoch": 0.2098268591842524,
+      "grad_norm": 2531.563606008494,
+      "learning_rate": 1.8055954449398472e-05,
+      "loss": 419.1451,
+      "step": 10910
+    },
+    {
+      "epoch": 0.21001918444473294,
+      "grad_norm": 3092.553750298103,
+      "learning_rate": 1.8052337044691648e-05,
+      "loss": 400.9566,
+      "step": 10920
+    },
+    {
+      "epoch": 0.21021150970521346,
+      "grad_norm": 2382.377838914781,
+      "learning_rate": 1.8048716640622262e-05,
+      "loss": 408.889,
+      "step": 10930
+    },
+    {
+      "epoch": 0.21040383496569398,
+      "grad_norm": 2490.8763516619233,
+      "learning_rate": 1.8045093238538856e-05,
+      "loss": 413.3729,
+      "step": 10940
+    },
+    {
+      "epoch": 0.2105961602261745,
+      "grad_norm": 2660.505139688422,
+      "learning_rate": 1.8041466839791087e-05,
+      "loss": 398.592,
+      "step": 10950
+    },
+    {
+      "epoch": 0.21078848548665502,
+      "grad_norm": 2164.5340715385973,
+      "learning_rate": 1.8037837445729733e-05,
+      "loss": 398.3774,
+      "step": 10960
+    },
+    {
+      "epoch": 0.21098081074713557,
+      "grad_norm": 2349.449614351497,
+      "learning_rate": 1.803420505770668e-05,
+      "loss": 395.1659,
+      "step": 10970
+    },
+    {
+      "epoch": 0.21117313600761609,
+      "grad_norm": 2435.3497667190113,
+      "learning_rate": 1.803056967707493e-05,
+      "loss": 396.0531,
+      "step": 10980
+    },
+    {
+      "epoch": 0.2113654612680966,
+      "grad_norm": 2246.816610446081,
+      "learning_rate": 1.8026931305188603e-05,
+      "loss": 410.702,
+      "step": 10990
+    },
+    {
+      "epoch": 0.21155778652857712,
+      "grad_norm": 2356.254917816634,
+      "learning_rate": 1.802328994340294e-05,
+      "loss": 404.5293,
+      "step": 11000
+    },
+    {
+      "epoch": 0.21175011178905764,
+      "grad_norm": 3024.863226923607,
+      "learning_rate": 1.8019645593074275e-05,
+      "loss": 401.4539,
+      "step": 11010
+    },
+    {
+      "epoch": 0.2119424370495382,
+      "grad_norm": 2458.6315749528885,
+      "learning_rate": 1.8015998255560082e-05,
+      "loss": 383.7941,
+      "step": 11020
+    },
+    {
+      "epoch": 0.2121347623100187,
+      "grad_norm": 2482.937050752312,
+      "learning_rate": 1.801234793221892e-05,
+      "loss": 390.7949,
+      "step": 11030
+    },
+    {
+      "epoch": 0.21232708757049923,
+      "grad_norm": 2130.659983516312,
+      "learning_rate": 1.800869462441049e-05,
+      "loss": 374.8555,
+      "step": 11040
+    },
+    {
+      "epoch": 0.21251941283097975,
+      "grad_norm": 3436.892560409446,
+      "learning_rate": 1.8005038333495572e-05,
+      "loss": 398.9813,
+      "step": 11050
+    },
+    {
+      "epoch": 0.21271173809146027,
+      "grad_norm": 2549.1782406530206,
+      "learning_rate": 1.8001379060836088e-05,
+      "loss": 403.375,
+      "step": 11060
+    },
+    {
+      "epoch": 0.21290406335194081,
+      "grad_norm": 2266.0715382189323,
+      "learning_rate": 1.7997716807795046e-05,
+      "loss": 404.9078,
+      "step": 11070
+    },
+    {
+      "epoch": 0.21309638861242133,
+      "grad_norm": 2373.7615169944665,
+      "learning_rate": 1.7994051575736585e-05,
+      "loss": 395.2838,
+      "step": 11080
+    },
+    {
+      "epoch": 0.21328871387290185,
+      "grad_norm": 2478.8311099791295,
+      "learning_rate": 1.7990383366025935e-05,
+      "loss": 405.3711,
+      "step": 11090
+    },
+    {
+      "epoch": 0.21348103913338237,
+      "grad_norm": 2577.515646094947,
+      "learning_rate": 1.7986712180029448e-05,
+      "loss": 393.1514,
+      "step": 11100
+    },
+    {
+      "epoch": 0.2136733643938629,
+      "grad_norm": 2362.7725035869066,
+      "learning_rate": 1.798303801911458e-05,
+      "loss": 398.3662,
+      "step": 11110
+    },
+    {
+      "epoch": 0.21386568965434344,
+      "grad_norm": 2163.3205259141787,
+      "learning_rate": 1.7979360884649894e-05,
+      "loss": 395.3951,
+      "step": 11120
+    },
+    {
+      "epoch": 0.21405801491482396,
+      "grad_norm": 2235.8150793368454,
+      "learning_rate": 1.7975680778005058e-05,
+      "loss": 392.3371,
+      "step": 11130
+    },
+    {
+      "epoch": 0.21425034017530448,
+      "grad_norm": 2299.195976343412,
+      "learning_rate": 1.7971997700550856e-05,
+      "loss": 402.759,
+      "step": 11140
+    },
+    {
+      "epoch": 0.214442665435785,
+      "grad_norm": 2592.0103304619306,
+      "learning_rate": 1.7968311653659177e-05,
+      "loss": 403.5507,
+      "step": 11150
+    },
+    {
+      "epoch": 0.21463499069626552,
+      "grad_norm": 2377.799379865453,
+      "learning_rate": 1.7964622638703003e-05,
+      "loss": 407.2532,
+      "step": 11160
+    },
+    {
+      "epoch": 0.21482731595674606,
+      "grad_norm": 3131.9266768005905,
+      "learning_rate": 1.796093065705644e-05,
+      "loss": 394.6042,
+      "step": 11170
+    },
+    {
+      "epoch": 0.21501964121722658,
+      "grad_norm": 2221.285884736338,
+      "learning_rate": 1.7957235710094686e-05,
+      "loss": 381.3053,
+      "step": 11180
+    },
+    {
+      "epoch": 0.2152119664777071,
+      "grad_norm": 2451.7127486980853,
+      "learning_rate": 1.7953537799194042e-05,
+      "loss": 397.5707,
+      "step": 11190
+    },
+    {
+      "epoch": 0.21540429173818762,
+      "grad_norm": 2261.694226947071,
+      "learning_rate": 1.7949836925731934e-05,
+      "loss": 411.3794,
+      "step": 11200
+    },
+    {
+      "epoch": 0.21559661699866814,
+      "grad_norm": 2344.328939293039,
+      "learning_rate": 1.7946133091086858e-05,
+      "loss": 405.38,
+      "step": 11210
+    },
+    {
+      "epoch": 0.21578894225914869,
+      "grad_norm": 2744.2452677672222,
+      "learning_rate": 1.7942426296638447e-05,
+      "loss": 400.7316,
+      "step": 11220
+    },
+    {
+      "epoch": 0.2159812675196292,
+      "grad_norm": 2179.5410545432596,
+      "learning_rate": 1.7938716543767412e-05,
+      "loss": 395.5047,
+      "step": 11230
+    },
+    {
+      "epoch": 0.21617359278010972,
+      "grad_norm": 2932.745050424784,
+      "learning_rate": 1.7935003833855576e-05,
+      "loss": 403.2667,
+      "step": 11240
+    },
+    {
+      "epoch": 0.21636591804059024,
+      "grad_norm": 2328.6822430694315,
+      "learning_rate": 1.7931288168285863e-05,
+      "loss": 400.5162,
+      "step": 11250
+    },
+    {
+      "epoch": 0.21655824330107076,
+      "grad_norm": 2404.7254552353556,
+      "learning_rate": 1.79275695484423e-05,
+      "loss": 394.3482,
+      "step": 11260
+    },
+    {
+      "epoch": 0.21675056856155128,
+      "grad_norm": 2724.9999609740257,
+      "learning_rate": 1.7923847975710003e-05,
+      "loss": 400.9097,
+      "step": 11270
+    },
+    {
+      "epoch": 0.21694289382203183,
+      "grad_norm": 2676.729401971507,
+      "learning_rate": 1.7920123451475203e-05,
+      "loss": 408.8387,
+      "step": 11280
+    },
+    {
+      "epoch": 0.21713521908251235,
+      "grad_norm": 2608.657703456628,
+      "learning_rate": 1.7916395977125227e-05,
+      "loss": 388.8734,
+      "step": 11290
+    },
+    {
+      "epoch": 0.21732754434299287,
+      "grad_norm": 2554.1094913876814,
+      "learning_rate": 1.7912665554048486e-05,
+      "loss": 401.6312,
+      "step": 11300
+    },
+    {
+      "epoch": 0.2175198696034734,
+      "grad_norm": 2526.7331695023895,
+      "learning_rate": 1.7908932183634515e-05,
+      "loss": 393.5017,
+      "step": 11310
+    },
+    {
+      "epoch": 0.2177121948639539,
+      "grad_norm": 2248.5706584848404,
+      "learning_rate": 1.790519586727392e-05,
+      "loss": 402.7919,
+      "step": 11320
+    },
+    {
+      "epoch": 0.21790452012443445,
+      "grad_norm": 2479.6053722356187,
+      "learning_rate": 1.790145660635843e-05,
+      "loss": 395.3233,
+      "step": 11330
+    },
+    {
+      "epoch": 0.21809684538491497,
+      "grad_norm": 2431.351989522283,
+      "learning_rate": 1.7897714402280844e-05,
+      "loss": 399.1128,
+      "step": 11340
+    },
+    {
+      "epoch": 0.2182891706453955,
+      "grad_norm": 2321.3161115556927,
+      "learning_rate": 1.789396925643508e-05,
+      "loss": 414.4136,
+      "step": 11350
+    },
+    {
+      "epoch": 0.218481495905876,
+      "grad_norm": 2433.942334044243,
+      "learning_rate": 1.7890221170216144e-05,
+      "loss": 388.3796,
+      "step": 11360
+    },
+    {
+      "epoch": 0.21867382116635653,
+      "grad_norm": 2242.8210732726634,
+      "learning_rate": 1.788647014502013e-05,
+      "loss": 387.7581,
+      "step": 11370
+    },
+    {
+      "epoch": 0.21886614642683708,
+      "grad_norm": 2310.3153801960866,
+      "learning_rate": 1.7882716182244242e-05,
+      "loss": 386.1181,
+      "step": 11380
+    },
+    {
+      "epoch": 0.2190584716873176,
+      "grad_norm": 2381.915282265231,
+      "learning_rate": 1.7878959283286758e-05,
+      "loss": 388.257,
+      "step": 11390
+    },
+    {
+      "epoch": 0.21925079694779812,
+      "grad_norm": 2586.260447381824,
+      "learning_rate": 1.787519944954707e-05,
+      "loss": 388.8421,
+      "step": 11400
+    },
+    {
+      "epoch": 0.21944312220827863,
+      "grad_norm": 2199.8917309435546,
+      "learning_rate": 1.7871436682425645e-05,
+      "loss": 391.1387,
+      "step": 11410
+    },
+    {
+      "epoch": 0.21963544746875915,
+      "grad_norm": 4909.373677771122,
+      "learning_rate": 1.786767098332406e-05,
+      "loss": 404.8643,
+      "step": 11420
+    },
+    {
+      "epoch": 0.2198277727292397,
+      "grad_norm": 2265.699700181102,
+      "learning_rate": 1.7863902353644972e-05,
+      "loss": 380.425,
+      "step": 11430
+    },
+    {
+      "epoch": 0.22002009798972022,
+      "grad_norm": 2264.2999697652035,
+      "learning_rate": 1.7860130794792137e-05,
+      "loss": 389.3438,
+      "step": 11440
+    },
+    {
+      "epoch": 0.22021242325020074,
+      "grad_norm": 2600.7568210842096,
+      "learning_rate": 1.7856356308170394e-05,
+      "loss": 404.331,
+      "step": 11450
+    },
+    {
+      "epoch": 0.22040474851068126,
+      "grad_norm": 2253.3199406105978,
+      "learning_rate": 1.7852578895185675e-05,
+      "loss": 409.1266,
+      "step": 11460
+    },
+    {
+      "epoch": 0.22059707377116178,
+      "grad_norm": 2961.2445056928277,
+      "learning_rate": 1.7848798557245008e-05,
+      "loss": 398.4247,
+      "step": 11470
+    },
+    {
+      "epoch": 0.22078939903164232,
+      "grad_norm": 2488.328905763648,
+      "learning_rate": 1.7845015295756506e-05,
+      "loss": 402.7198,
+      "step": 11480
+    },
+    {
+      "epoch": 0.22098172429212284,
+      "grad_norm": 2115.781398309855,
+      "learning_rate": 1.784122911212937e-05,
+      "loss": 388.8867,
+      "step": 11490
+    },
+    {
+      "epoch": 0.22117404955260336,
+      "grad_norm": 2136.812712503691,
+      "learning_rate": 1.7837440007773895e-05,
+      "loss": 386.4281,
+      "step": 11500
+    },
+    {
+      "epoch": 0.22136637481308388,
+      "grad_norm": 2347.8785529560864,
+      "learning_rate": 1.783364798410146e-05,
+      "loss": 387.8281,
+      "step": 11510
+    },
+    {
+      "epoch": 0.2215587000735644,
+      "grad_norm": 2228.3889903200293,
+      "learning_rate": 1.782985304252452e-05,
+      "loss": 390.3264,
+      "step": 11520
+    },
+    {
+      "epoch": 0.22175102533404495,
+      "grad_norm": 2508.0085433196264,
+      "learning_rate": 1.7826055184456643e-05,
+      "loss": 400.6156,
+      "step": 11530
+    },
+    {
+      "epoch": 0.22194335059452547,
+      "grad_norm": 2729.6083957527267,
+      "learning_rate": 1.7822254411312455e-05,
+      "loss": 397.1274,
+      "step": 11540
+    },
+    {
+      "epoch": 0.222135675855006,
+      "grad_norm": 2183.676311189381,
+      "learning_rate": 1.781845072450769e-05,
+      "loss": 385.762,
+      "step": 11550
+    },
+    {
+      "epoch": 0.2223280011154865,
+      "grad_norm": 2987.6666025107056,
+      "learning_rate": 1.7814644125459157e-05,
+      "loss": 399.8271,
+      "step": 11560
+    },
+    {
+      "epoch": 0.22252032637596703,
+      "grad_norm": 2403.9241230398407,
+      "learning_rate": 1.781083461558475e-05,
+      "loss": 390.9242,
+      "step": 11570
+    },
+    {
+      "epoch": 0.22271265163644757,
+      "grad_norm": 2532.620143806683,
+      "learning_rate": 1.7807022196303447e-05,
+      "loss": 395.1546,
+      "step": 11580
+    },
+    {
+      "epoch": 0.2229049768969281,
+      "grad_norm": 2304.3143606905132,
+      "learning_rate": 1.7803206869035318e-05,
+      "loss": 390.9656,
+      "step": 11590
+    },
+    {
+      "epoch": 0.2230973021574086,
+      "grad_norm": 2497.0977539681685,
+      "learning_rate": 1.7799388635201498e-05,
+      "loss": 396.0081,
+      "step": 11600
+    },
+    {
+      "epoch": 0.22328962741788913,
+      "grad_norm": 2137.3276684673474,
+      "learning_rate": 1.7795567496224226e-05,
+      "loss": 390.5375,
+      "step": 11610
+    },
+    {
+      "epoch": 0.22348195267836965,
+      "grad_norm": 2475.2489591013623,
+      "learning_rate": 1.7791743453526812e-05,
+      "loss": 399.8304,
+      "step": 11620
+    },
+    {
+      "epoch": 0.2236742779388502,
+      "grad_norm": 2302.8248429456294,
+      "learning_rate": 1.778791650853364e-05,
+      "loss": 388.7546,
+      "step": 11630
+    },
+    {
+      "epoch": 0.22386660319933072,
+      "grad_norm": 2541.3533131710237,
+      "learning_rate": 1.77840866626702e-05,
+      "loss": 385.4423,
+      "step": 11640
+    },
+    {
+      "epoch": 0.22405892845981124,
+      "grad_norm": 2256.7805083032003,
+      "learning_rate": 1.7780253917363026e-05,
+      "loss": 388.9178,
+      "step": 11650
+    },
+    {
+      "epoch": 0.22425125372029175,
+      "grad_norm": 2453.797965663399,
+      "learning_rate": 1.7776418274039767e-05,
+      "loss": 387.315,
+      "step": 11660
+    },
+    {
+      "epoch": 0.22444357898077227,
+      "grad_norm": 2676.6923687295252,
+      "learning_rate": 1.7772579734129136e-05,
+      "loss": 396.885,
+      "step": 11670
+    },
+    {
+      "epoch": 0.2246359042412528,
+      "grad_norm": 3086.89156146947,
+      "learning_rate": 1.776873829906092e-05,
+      "loss": 389.7317,
+      "step": 11680
+    },
+    {
+      "epoch": 0.22482822950173334,
+      "grad_norm": 2424.621620486519,
+      "learning_rate": 1.776489397026599e-05,
+      "loss": 406.4202,
+      "step": 11690
+    },
+    {
+      "epoch": 0.22502055476221386,
+      "grad_norm": 2276.7448812268194,
+      "learning_rate": 1.7761046749176302e-05,
+      "loss": 383.7106,
+      "step": 11700
+    },
+    {
+      "epoch": 0.22521288002269438,
+      "grad_norm": 2313.6311991836037,
+      "learning_rate": 1.7757196637224874e-05,
+      "loss": 391.2429,
+      "step": 11710
+    },
+    {
+      "epoch": 0.2254052052831749,
+      "grad_norm": 2341.3697884828307,
+      "learning_rate": 1.7753343635845817e-05,
+      "loss": 386.1084,
+      "step": 11720
+    },
+    {
+      "epoch": 0.22559753054365542,
+      "grad_norm": 2376.0300132140237,
+      "learning_rate": 1.7749487746474305e-05,
+      "loss": 385.0678,
+      "step": 11730
+    },
+    {
+      "epoch": 0.22578985580413596,
+      "grad_norm": 2347.218519597106,
+      "learning_rate": 1.7745628970546592e-05,
+      "loss": 389.4445,
+      "step": 11740
+    },
+    {
+      "epoch": 0.22598218106461648,
+      "grad_norm": 2687.338690227089,
+      "learning_rate": 1.774176730950001e-05,
+      "loss": 373.979,
+      "step": 11750
+    },
+    {
+      "epoch": 0.226174506325097,
+      "grad_norm": 2307.7770863294495,
+      "learning_rate": 1.7737902764772967e-05,
+      "loss": 386.6169,
+      "step": 11760
+    },
+    {
+      "epoch": 0.22636683158557752,
+      "grad_norm": 3200.131284017193,
+      "learning_rate": 1.773403533780494e-05,
+      "loss": 387.5548,
+      "step": 11770
+    },
+    {
+      "epoch": 0.22655915684605804,
+      "grad_norm": 2325.677583423592,
+      "learning_rate": 1.7730165030036482e-05,
+      "loss": 383.3013,
+      "step": 11780
+    },
+    {
+      "epoch": 0.2267514821065386,
+      "grad_norm": 2438.4287045483115,
+      "learning_rate": 1.772629184290922e-05,
+      "loss": 385.3212,
+      "step": 11790
+    },
+    {
+      "epoch": 0.2269438073670191,
+      "grad_norm": 2324.824326296251,
+      "learning_rate": 1.7722415777865845e-05,
+      "loss": 375.5994,
+      "step": 11800
+    },
+    {
+      "epoch": 0.22713613262749963,
+      "grad_norm": 2443.6408456885065,
+      "learning_rate": 1.771853683635014e-05,
+      "loss": 386.032,
+      "step": 11810
+    },
+    {
+      "epoch": 0.22732845788798015,
+      "grad_norm": 2335.801762020265,
+      "learning_rate": 1.7714655019806932e-05,
+      "loss": 393.7155,
+      "step": 11820
+    },
+    {
+      "epoch": 0.22752078314846066,
+      "grad_norm": 2395.537989034479,
+      "learning_rate": 1.7710770329682145e-05,
+      "loss": 400.2536,
+      "step": 11830
+    },
+    {
+      "epoch": 0.2277131084089412,
+      "grad_norm": 2787.2395348230343,
+      "learning_rate": 1.770688276742276e-05,
+      "loss": 390.3201,
+      "step": 11840
+    },
+    {
+      "epoch": 0.22790543366942173,
+      "grad_norm": 2710.4724119666625,
+      "learning_rate": 1.770299233447682e-05,
+      "loss": 387.9175,
+      "step": 11850
+    },
+    {
+      "epoch": 0.22809775892990225,
+      "grad_norm": 2323.435954751655,
+      "learning_rate": 1.769909903229346e-05,
+      "loss": 389.8935,
+      "step": 11860
+    },
+    {
+      "epoch": 0.22829008419038277,
+      "grad_norm": 2622.2302563890967,
+      "learning_rate": 1.7695202862322863e-05,
+      "loss": 386.7145,
+      "step": 11870
+    },
+    {
+      "epoch": 0.2284824094508633,
+      "grad_norm": 2343.815966484445,
+      "learning_rate": 1.769130382601629e-05,
+      "loss": 394.5964,
+      "step": 11880
+    },
+    {
+      "epoch": 0.22867473471134384,
+      "grad_norm": 2363.7595061816423,
+      "learning_rate": 1.768740192482607e-05,
+      "loss": 388.6387,
+      "step": 11890
+    },
+    {
+      "epoch": 0.22886705997182435,
+      "grad_norm": 2269.353120654822,
+      "learning_rate": 1.7683497160205595e-05,
+      "loss": 393.3567,
+      "step": 11900
+    },
+    {
+      "epoch": 0.22905938523230487,
+      "grad_norm": 2472.494570261814,
+      "learning_rate": 1.7679589533609323e-05,
+      "loss": 392.5933,
+      "step": 11910
+    },
+    {
+      "epoch": 0.2292517104927854,
+      "grad_norm": 2223.128949708475,
+      "learning_rate": 1.767567904649278e-05,
+      "loss": 400.179,
+      "step": 11920
+    },
+    {
+      "epoch": 0.2294440357532659,
+      "grad_norm": 2589.5018549123547,
+      "learning_rate": 1.7671765700312567e-05,
+      "loss": 392.6122,
+      "step": 11930
+    },
+    {
+      "epoch": 0.22963636101374646,
+      "grad_norm": 2111.446795298321,
+      "learning_rate": 1.7667849496526327e-05,
+      "loss": 379.3064,
+      "step": 11940
+    },
+    {
+      "epoch": 0.22982868627422698,
+      "grad_norm": 2378.717511793757,
+      "learning_rate": 1.7663930436592793e-05,
+      "loss": 389.1514,
+      "step": 11950
+    },
+    {
+      "epoch": 0.2300210115347075,
+      "grad_norm": 2346.2808680197413,
+      "learning_rate": 1.7660008521971744e-05,
+      "loss": 386.1958,
+      "step": 11960
+    },
+    {
+      "epoch": 0.23021333679518802,
+      "grad_norm": 2541.89480853922,
+      "learning_rate": 1.765608375412403e-05,
+      "loss": 383.4571,
+      "step": 11970
+    },
+    {
+      "epoch": 0.23040566205566854,
+      "grad_norm": 2160.62311131137,
+      "learning_rate": 1.7652156134511563e-05,
+      "loss": 387.4947,
+      "step": 11980
+    },
+    {
+      "epoch": 0.23059798731614908,
+      "grad_norm": 2233.5666004308027,
+      "learning_rate": 1.7648225664597314e-05,
+      "loss": 390.8314,
+      "step": 11990
+    },
+    {
+      "epoch": 0.2307903125766296,
+      "grad_norm": 2263.0223621099176,
+      "learning_rate": 1.764429234584532e-05,
+      "loss": 385.8524,
+      "step": 12000
+    },
+    {
+      "epoch": 0.23098263783711012,
+      "grad_norm": 2330.289995225945,
+      "learning_rate": 1.7640356179720674e-05,
+      "loss": 365.5703,
+      "step": 12010
+    },
+    {
+      "epoch": 0.23117496309759064,
+      "grad_norm": 2282.907061035763,
+      "learning_rate": 1.7636417167689538e-05,
+      "loss": 392.4571,
+      "step": 12020
+    },
+    {
+      "epoch": 0.23136728835807116,
+      "grad_norm": 2230.4093858725714,
+      "learning_rate": 1.7632475311219125e-05,
+      "loss": 377.4121,
+      "step": 12030
+    },
+    {
+      "epoch": 0.2315596136185517,
+      "grad_norm": 2317.6649136602114,
+      "learning_rate": 1.7628530611777716e-05,
+      "loss": 384.7892,
+      "step": 12040
+    },
+    {
+      "epoch": 0.23175193887903223,
+      "grad_norm": 2202.679921921386,
+      "learning_rate": 1.7624583070834646e-05,
+      "loss": 389.4707,
+      "step": 12050
+    },
+    {
+      "epoch": 0.23194426413951275,
+      "grad_norm": 2417.2271221250494,
+      "learning_rate": 1.7620632689860298e-05,
+      "loss": 393.0368,
+      "step": 12060
+    },
+    {
+      "epoch": 0.23213658939999327,
+      "grad_norm": 2560.741612554429,
+      "learning_rate": 1.761667947032614e-05,
+      "loss": 397.0752,
+      "step": 12070
+    },
+    {
+      "epoch": 0.23232891466047378,
+      "grad_norm": 2994.689721313683,
+      "learning_rate": 1.761272341370467e-05,
+      "loss": 382.2997,
+      "step": 12080
+    },
+    {
+      "epoch": 0.23252123992095433,
+      "grad_norm": 2260.7791320488623,
+      "learning_rate": 1.7608764521469456e-05,
+      "loss": 373.9792,
+      "step": 12090
+    },
+    {
+      "epoch": 0.23271356518143485,
+      "grad_norm": 2302.393193266844,
+      "learning_rate": 1.760480279509512e-05,
+      "loss": 382.4867,
+      "step": 12100
+    },
+    {
+      "epoch": 0.23290589044191537,
+      "grad_norm": 2327.0769193689284,
+      "learning_rate": 1.7600838236057342e-05,
+      "loss": 384.5145,
+      "step": 12110
+    },
+    {
+      "epoch": 0.2330982157023959,
+      "grad_norm": 2150.98676300647,
+      "learning_rate": 1.759687084583285e-05,
+      "loss": 378.755,
+      "step": 12120
+    },
+    {
+      "epoch": 0.2332905409628764,
+      "grad_norm": 2256.504732823715,
+      "learning_rate": 1.7592900625899437e-05,
+      "loss": 389.5562,
+      "step": 12130
+    },
+    {
+      "epoch": 0.23348286622335693,
+      "grad_norm": 2274.3014291975965,
+      "learning_rate": 1.7588927577735943e-05,
+      "loss": 375.5534,
+      "step": 12140
+    },
+    {
+      "epoch": 0.23367519148383747,
+      "grad_norm": 2114.0374629396065,
+      "learning_rate": 1.758495170282226e-05,
+      "loss": 380.2994,
+      "step": 12150
+    },
+    {
+      "epoch": 0.233867516744318,
+      "grad_norm": 2357.7964137669774,
+      "learning_rate": 1.7580973002639337e-05,
+      "loss": 388.1062,
+      "step": 12160
+    },
+    {
+      "epoch": 0.2340598420047985,
+      "grad_norm": 2272.161426274605,
+      "learning_rate": 1.7576991478669174e-05,
+      "loss": 384.5954,
+      "step": 12170
+    },
+    {
+      "epoch": 0.23425216726527903,
+      "grad_norm": 2127.431549340675,
+      "learning_rate": 1.7573007132394823e-05,
+      "loss": 381.7495,
+      "step": 12180
+    },
+    {
+      "epoch": 0.23444449252575955,
+      "grad_norm": 2494.721838509276,
+      "learning_rate": 1.756901996530039e-05,
+      "loss": 384.8348,
+      "step": 12190
+    },
+    {
+      "epoch": 0.2346368177862401,
+      "grad_norm": 2877.4404041265866,
+      "learning_rate": 1.7565029978871025e-05,
+      "loss": 400.2325,
+      "step": 12200
+    },
+    {
+      "epoch": 0.23482914304672062,
+      "grad_norm": 2557.969871364385,
+      "learning_rate": 1.7561037174592933e-05,
+      "loss": 381.2346,
+      "step": 12210
+    },
+    {
+      "epoch": 0.23502146830720114,
+      "grad_norm": 2565.029485905372,
+      "learning_rate": 1.7557041553953368e-05,
+      "loss": 374.5849,
+      "step": 12220
+    },
+    {
+      "epoch": 0.23521379356768166,
+      "grad_norm": 2581.207375088013,
+      "learning_rate": 1.7553043118440634e-05,
+      "loss": 375.8843,
+      "step": 12230
+    },
+    {
+      "epoch": 0.23540611882816218,
+      "grad_norm": 2579.923372208383,
+      "learning_rate": 1.7549041869544077e-05,
+      "loss": 378.0457,
+      "step": 12240
+    },
+    {
+      "epoch": 0.23559844408864272,
+      "grad_norm": 2827.048997734147,
+      "learning_rate": 1.7545037808754105e-05,
+      "loss": 375.2411,
+      "step": 12250
+    },
+    {
+      "epoch": 0.23579076934912324,
+      "grad_norm": 2208.4241563716687,
+      "learning_rate": 1.754103093756216e-05,
+      "loss": 364.0572,
+      "step": 12260
+    },
+    {
+      "epoch": 0.23598309460960376,
+      "grad_norm": 2012.2013360667677,
+      "learning_rate": 1.7537021257460732e-05,
+      "loss": 373.4937,
+      "step": 12270
+    },
+    {
+      "epoch": 0.23617541987008428,
+      "grad_norm": 2337.7945293318235,
+      "learning_rate": 1.7533008769943366e-05,
+      "loss": 371.4375,
+      "step": 12280
+    },
+    {
+      "epoch": 0.2363677451305648,
+      "grad_norm": 2017.197630179005,
+      "learning_rate": 1.7528993476504644e-05,
+      "loss": 375.1286,
+      "step": 12290
+    },
+    {
+      "epoch": 0.23656007039104535,
+      "grad_norm": 2208.9437121421356,
+      "learning_rate": 1.7524975378640198e-05,
+      "loss": 377.4877,
+      "step": 12300
+    },
+    {
+      "epoch": 0.23675239565152587,
+      "grad_norm": 2021.4147609550564,
+      "learning_rate": 1.75209544778467e-05,
+      "loss": 375.5848,
+      "step": 12310
+    },
+    {
+      "epoch": 0.23694472091200638,
+      "grad_norm": 2605.063531196968,
+      "learning_rate": 1.7516930775621873e-05,
+      "loss": 395.9945,
+      "step": 12320
+    },
+    {
+      "epoch": 0.2371370461724869,
+      "grad_norm": 2604.6354111838214,
+      "learning_rate": 1.751290427346448e-05,
+      "loss": 380.5949,
+      "step": 12330
+    },
+    {
+      "epoch": 0.23732937143296742,
+      "grad_norm": 2134.523800512569,
+      "learning_rate": 1.7508874972874325e-05,
+      "loss": 385.1871,
+      "step": 12340
+    },
+    {
+      "epoch": 0.23752169669344797,
+      "grad_norm": 2557.6205213871312,
+      "learning_rate": 1.7504842875352254e-05,
+      "loss": 380.2885,
+      "step": 12350
+    },
+    {
+      "epoch": 0.2377140219539285,
+      "grad_norm": 2402.018914592033,
+      "learning_rate": 1.7500807982400155e-05,
+      "loss": 376.2682,
+      "step": 12360
+    },
+    {
+      "epoch": 0.237906347214409,
+      "grad_norm": 2165.764987759702,
+      "learning_rate": 1.749677029552097e-05,
+      "loss": 371.1269,
+      "step": 12370
+    },
+    {
+      "epoch": 0.23809867247488953,
+      "grad_norm": 2072.7914361726002,
+      "learning_rate": 1.7492729816218656e-05,
+      "loss": 378.911,
+      "step": 12380
+    },
+    {
+      "epoch": 0.23829099773537005,
+      "grad_norm": 2397.20874623808,
+      "learning_rate": 1.7488686545998237e-05,
+      "loss": 369.1646,
+      "step": 12390
+    },
+    {
+      "epoch": 0.2384833229958506,
+      "grad_norm": 2393.3178214051827,
+      "learning_rate": 1.7484640486365757e-05,
+      "loss": 377.2179,
+      "step": 12400
+    },
+    {
+      "epoch": 0.2386756482563311,
+      "grad_norm": 2623.916253511651,
+      "learning_rate": 1.748059163882831e-05,
+      "loss": 383.2641,
+      "step": 12410
+    },
+    {
+      "epoch": 0.23886797351681163,
+      "grad_norm": 2152.0668589588495,
+      "learning_rate": 1.747654000489402e-05,
+      "loss": 373.3912,
+      "step": 12420
+    },
+    {
+      "epoch": 0.23906029877729215,
+      "grad_norm": 2530.02619281479,
+      "learning_rate": 1.7472485586072062e-05,
+      "loss": 373.3627,
+      "step": 12430
+    },
+    {
+      "epoch": 0.23925262403777267,
+      "grad_norm": 2508.9927435530876,
+      "learning_rate": 1.746842838387264e-05,
+      "loss": 369.6259,
+      "step": 12440
+    },
+    {
+      "epoch": 0.23944494929825322,
+      "grad_norm": 3179.2708015154562,
+      "learning_rate": 1.746436839980698e-05,
+      "loss": 384.0702,
+      "step": 12450
+    },
+    {
+      "epoch": 0.23963727455873374,
+      "grad_norm": 2410.888904472343,
+      "learning_rate": 1.7460305635387372e-05,
+      "loss": 382.8646,
+      "step": 12460
+    },
+    {
+      "epoch": 0.23982959981921426,
+      "grad_norm": 2469.982232789963,
+      "learning_rate": 1.745624009212713e-05,
+      "loss": 381.0362,
+      "step": 12470
+    },
+    {
+      "epoch": 0.24002192507969478,
+      "grad_norm": 2569.9909175390508,
+      "learning_rate": 1.7452171771540593e-05,
+      "loss": 383.5919,
+      "step": 12480
+    },
+    {
+      "epoch": 0.2402142503401753,
+      "grad_norm": 2249.847929516714,
+      "learning_rate": 1.744810067514315e-05,
+      "loss": 376.9287,
+      "step": 12490
+    },
+    {
+      "epoch": 0.24040657560065584,
+      "grad_norm": 2038.242923459335,
+      "learning_rate": 1.7444026804451214e-05,
+      "loss": 373.5053,
+      "step": 12500
+    },
+    {
+      "epoch": 0.24059890086113636,
+      "grad_norm": 2250.1997887785706,
+      "learning_rate": 1.7439950160982236e-05,
+      "loss": 375.1043,
+      "step": 12510
+    },
+    {
+      "epoch": 0.24079122612161688,
+      "grad_norm": 2219.2837718383635,
+      "learning_rate": 1.7435870746254697e-05,
+      "loss": 376.5237,
+      "step": 12520
+    },
+    {
+      "epoch": 0.2409835513820974,
+      "grad_norm": 2281.0283481339156,
+      "learning_rate": 1.7431788561788116e-05,
+      "loss": 372.6687,
+      "step": 12530
+    },
+    {
+      "epoch": 0.24117587664257792,
+      "grad_norm": 2078.6147727217976,
+      "learning_rate": 1.742770360910303e-05,
+      "loss": 384.8385,
+      "step": 12540
+    },
+    {
+      "epoch": 0.24136820190305847,
+      "grad_norm": 2162.7392171748047,
+      "learning_rate": 1.7423615889721027e-05,
+      "loss": 372.5577,
+      "step": 12550
+    },
+    {
+      "epoch": 0.24156052716353899,
+      "grad_norm": 2386.920379603306,
+      "learning_rate": 1.7419525405164705e-05,
+      "loss": 389.3319,
+      "step": 12560
+    },
+    {
+      "epoch": 0.2417528524240195,
+      "grad_norm": 2756.8727323904436,
+      "learning_rate": 1.7415432156957713e-05,
+      "loss": 379.5574,
+      "step": 12570
+    },
+    {
+      "epoch": 0.24194517768450002,
+      "grad_norm": 2099.2502349035385,
+      "learning_rate": 1.741133614662471e-05,
+      "loss": 378.9946,
+      "step": 12580
+    },
+    {
+      "epoch": 0.24213750294498054,
+      "grad_norm": 2380.329571350227,
+      "learning_rate": 1.7407237375691394e-05,
+      "loss": 390.7249,
+      "step": 12590
+    },
+    {
+      "epoch": 0.24232982820546106,
+      "grad_norm": 2095.2206685250503,
+      "learning_rate": 1.740313584568449e-05,
+      "loss": 372.3252,
+      "step": 12600
+    },
+    {
+      "epoch": 0.2425221534659416,
+      "grad_norm": 2217.826751681227,
+      "learning_rate": 1.739903155813175e-05,
+      "loss": 385.4555,
+      "step": 12610
+    },
+    {
+      "epoch": 0.24271447872642213,
+      "grad_norm": 2386.987243566276,
+      "learning_rate": 1.7394924514561955e-05,
+      "loss": 377.4909,
+      "step": 12620
+    },
+    {
+      "epoch": 0.24290680398690265,
+      "grad_norm": 2204.0162900759146,
+      "learning_rate": 1.739081471650491e-05,
+      "loss": 369.1897,
+      "step": 12630
+    },
+    {
+      "epoch": 0.24309912924738317,
+      "grad_norm": 2073.5013741947837,
+      "learning_rate": 1.7386702165491443e-05,
+      "loss": 383.1417,
+      "step": 12640
+    },
+    {
+      "epoch": 0.24329145450786369,
+      "grad_norm": 2362.364142856857,
+      "learning_rate": 1.738258686305342e-05,
+      "loss": 371.7277,
+      "step": 12650
+    },
+    {
+      "epoch": 0.24348377976834423,
+      "grad_norm": 2643.743376662587,
+      "learning_rate": 1.7378468810723713e-05,
+      "loss": 385.6666,
+      "step": 12660
+    },
+    {
+      "epoch": 0.24367610502882475,
+      "grad_norm": 2286.0604513845683,
+      "learning_rate": 1.7374348010036235e-05,
+      "loss": 378.0156,
+      "step": 12670
+    },
+    {
+      "epoch": 0.24386843028930527,
+      "grad_norm": 2273.380518491826,
+      "learning_rate": 1.7370224462525916e-05,
+      "loss": 376.7685,
+      "step": 12680
+    },
+    {
+      "epoch": 0.2440607555497858,
+      "grad_norm": 2583.2678746445936,
+      "learning_rate": 1.736609816972871e-05,
+      "loss": 381.597,
+      "step": 12690
+    },
+    {
+      "epoch": 0.2442530808102663,
+      "grad_norm": 2175.324993896793,
+      "learning_rate": 1.7361969133181585e-05,
+      "loss": 361.4785,
+      "step": 12700
+    },
+    {
+      "epoch": 0.24444540607074686,
+      "grad_norm": 2113.7376989641184,
+      "learning_rate": 1.735783735442255e-05,
+      "loss": 377.5192,
+      "step": 12710
+    },
+    {
+      "epoch": 0.24463773133122738,
+      "grad_norm": 2136.1028577925226,
+      "learning_rate": 1.7353702834990617e-05,
+      "loss": 397.3817,
+      "step": 12720
+    },
+    {
+      "epoch": 0.2448300565917079,
+      "grad_norm": 2353.5458717950337,
+      "learning_rate": 1.7349565576425828e-05,
+      "loss": 360.6521,
+      "step": 12730
+    },
+    {
+      "epoch": 0.24502238185218841,
+      "grad_norm": 2289.400301786185,
+      "learning_rate": 1.7345425580269245e-05,
+      "loss": 378.4829,
+      "step": 12740
+    },
+    {
+      "epoch": 0.24521470711266893,
+      "grad_norm": 2301.400052972841,
+      "learning_rate": 1.734128284806295e-05,
+      "loss": 372.5589,
+      "step": 12750
+    },
+    {
+      "epoch": 0.24540703237314948,
+      "grad_norm": 2392.7181586747943,
+      "learning_rate": 1.7337137381350033e-05,
+      "loss": 373.6964,
+      "step": 12760
+    },
+    {
+      "epoch": 0.24559935763363,
+      "grad_norm": 2587.9513884335347,
+      "learning_rate": 1.7332989181674623e-05,
+      "loss": 363.099,
+      "step": 12770
+    },
+    {
+      "epoch": 0.24579168289411052,
+      "grad_norm": 2304.8713947992815,
+      "learning_rate": 1.7328838250581846e-05,
+      "loss": 370.9114,
+      "step": 12780
+    },
+    {
+      "epoch": 0.24598400815459104,
+      "grad_norm": 2353.5580929841517,
+      "learning_rate": 1.7324684589617862e-05,
+      "loss": 367.49,
+      "step": 12790
+    },
+    {
+      "epoch": 0.24617633341507156,
+      "grad_norm": 2552.8687113219935,
+      "learning_rate": 1.7320528200329846e-05,
+      "loss": 377.0836,
+      "step": 12800
+    },
+    {
+      "epoch": 0.2463686586755521,
+      "grad_norm": 2277.3914450397547,
+      "learning_rate": 1.7316369084265973e-05,
+      "loss": 363.0377,
+      "step": 12810
+    },
+    {
+      "epoch": 0.24656098393603262,
+      "grad_norm": 2229.9469580267078,
+      "learning_rate": 1.731220724297545e-05,
+      "loss": 361.0781,
+      "step": 12820
+    },
+    {
+      "epoch": 0.24675330919651314,
+      "grad_norm": 2290.8165063255447,
+      "learning_rate": 1.730804267800849e-05,
+      "loss": 373.4446,
+      "step": 12830
+    },
+    {
+      "epoch": 0.24694563445699366,
+      "grad_norm": 2501.9933224803535,
+      "learning_rate": 1.7303875390916338e-05,
+      "loss": 362.5407,
+      "step": 12840
+    },
+    {
+      "epoch": 0.24713795971747418,
+      "grad_norm": 2295.3147239736013,
+      "learning_rate": 1.729970538325122e-05,
+      "loss": 362.7664,
+      "step": 12850
+    },
+    {
+      "epoch": 0.24733028497795473,
+      "grad_norm": 2400.941128732704,
+      "learning_rate": 1.7295532656566413e-05,
+      "loss": 378.3463,
+      "step": 12860
+    },
+    {
+      "epoch": 0.24752261023843525,
+      "grad_norm": 2241.701808611237,
+      "learning_rate": 1.729135721241618e-05,
+      "loss": 366.0954,
+      "step": 12870
+    },
+    {
+      "epoch": 0.24771493549891577,
+      "grad_norm": 2342.4637468850588,
+      "learning_rate": 1.7287179052355803e-05,
+      "loss": 378.5448,
+      "step": 12880
+    },
+    {
+      "epoch": 0.2479072607593963,
+      "grad_norm": 2288.097396796908,
+      "learning_rate": 1.7282998177941586e-05,
+      "loss": 383.0611,
+      "step": 12890
+    },
+    {
+      "epoch": 0.2480995860198768,
+      "grad_norm": 2432.4276696859324,
+      "learning_rate": 1.7278814590730826e-05,
+      "loss": 365.5417,
+      "step": 12900
+    },
+    {
+      "epoch": 0.24829191128035735,
+      "grad_norm": 2339.524064330787,
+      "learning_rate": 1.7274628292281846e-05,
+      "loss": 363.7857,
+      "step": 12910
+    },
+    {
+      "epoch": 0.24848423654083787,
+      "grad_norm": 2257.424850624551,
+      "learning_rate": 1.727043928415397e-05,
+      "loss": 367.3295,
+      "step": 12920
+    },
+    {
+      "epoch": 0.2486765618013184,
+      "grad_norm": 2297.566011024205,
+      "learning_rate": 1.726624756790754e-05,
+      "loss": 374.8469,
+      "step": 12930
+    },
+    {
+      "epoch": 0.2488688870617989,
+      "grad_norm": 2268.6771212686576,
+      "learning_rate": 1.7262053145103893e-05,
+      "loss": 378.9152,
+      "step": 12940
+    },
+    {
+      "epoch": 0.24906121232227943,
+      "grad_norm": 2326.5117844320807,
+      "learning_rate": 1.7257856017305387e-05,
+      "loss": 373.2381,
+      "step": 12950
+    },
+    {
+      "epoch": 0.24925353758275998,
+      "grad_norm": 2166.3961909250083,
+      "learning_rate": 1.725365618607538e-05,
+      "loss": 363.4617,
+      "step": 12960
+    },
+    {
+      "epoch": 0.2494458628432405,
+      "grad_norm": 2123.8157116986677,
+      "learning_rate": 1.7249453652978244e-05,
+      "loss": 366.4645,
+      "step": 12970
+    },
+    {
+      "epoch": 0.24963818810372101,
+      "grad_norm": 1954.5908539168058,
+      "learning_rate": 1.7245248419579353e-05,
+      "loss": 375.104,
+      "step": 12980
+    },
+    {
+      "epoch": 0.24983051336420153,
+      "grad_norm": 2311.455331979595,
+      "learning_rate": 1.7241040487445082e-05,
+      "loss": 366.1221,
+      "step": 12990
+    },
+    {
+      "epoch": 0.25002283862468205,
+      "grad_norm": 1952.5267212095614,
+      "learning_rate": 1.723682985814282e-05,
+      "loss": 366.4078,
+      "step": 13000
+    },
+    {
+      "epoch": 0.2502151638851626,
+      "grad_norm": 2494.0820832215604,
+      "learning_rate": 1.7232616533240958e-05,
+      "loss": 370.3326,
+      "step": 13010
+    },
+    {
+      "epoch": 0.2504074891456431,
+      "grad_norm": 2318.699171466557,
+      "learning_rate": 1.7228400514308884e-05,
+      "loss": 375.1649,
+      "step": 13020
+    },
+    {
+      "epoch": 0.2505998144061236,
+      "grad_norm": 2923.064596255426,
+      "learning_rate": 1.7224181802917003e-05,
+      "loss": 365.9617,
+      "step": 13030
+    },
+    {
+      "epoch": 0.2507921396666042,
+      "grad_norm": 2235.8162398618665,
+      "learning_rate": 1.721996040063671e-05,
+      "loss": 371.2755,
+      "step": 13040
+    },
+    {
+      "epoch": 0.2509844649270847,
+      "grad_norm": 3402.491641507786,
+      "learning_rate": 1.7215736309040408e-05,
+      "loss": 364.7011,
+      "step": 13050
+    },
+    {
+      "epoch": 0.2511767901875652,
+      "grad_norm": 2190.4111841868817,
+      "learning_rate": 1.7211509529701507e-05,
+      "loss": 364.3978,
+      "step": 13060
+    },
+    {
+      "epoch": 0.25136911544804574,
+      "grad_norm": 2182.905249871719,
+      "learning_rate": 1.7207280064194403e-05,
+      "loss": 370.612,
+      "step": 13070
+    },
+    {
+      "epoch": 0.25156144070852626,
+      "grad_norm": 2412.7129913277554,
+      "learning_rate": 1.7203047914094514e-05,
+      "loss": 363.9965,
+      "step": 13080
+    },
+    {
+      "epoch": 0.2517537659690068,
+      "grad_norm": 2262.078866173598,
+      "learning_rate": 1.7198813080978235e-05,
+      "loss": 368.0875,
+      "step": 13090
+    },
+    {
+      "epoch": 0.2519460912294873,
+      "grad_norm": 2141.2010465395024,
+      "learning_rate": 1.7194575566422975e-05,
+      "loss": 372.1631,
+      "step": 13100
+    },
+    {
+      "epoch": 0.2521384164899678,
+      "grad_norm": 2236.876871069789,
+      "learning_rate": 1.719033537200714e-05,
+      "loss": 357.9208,
+      "step": 13110
+    },
+    {
+      "epoch": 0.25233074175044834,
+      "grad_norm": 2368.896756757804,
+      "learning_rate": 1.7186092499310133e-05,
+      "loss": 370.6848,
+      "step": 13120
+    },
+    {
+      "epoch": 0.25252306701092886,
+      "grad_norm": 2034.295919620777,
+      "learning_rate": 1.7181846949912347e-05,
+      "loss": 365.5694,
+      "step": 13130
+    },
+    {
+      "epoch": 0.25271539227140943,
+      "grad_norm": 2112.365699259346,
+      "learning_rate": 1.717759872539519e-05,
+      "loss": 357.8564,
+      "step": 13140
+    },
+    {
+      "epoch": 0.25290771753188995,
+      "grad_norm": 2236.7844958234014,
+      "learning_rate": 1.7173347827341046e-05,
+      "loss": 363.274,
+      "step": 13150
+    },
+    {
+      "epoch": 0.25310004279237047,
+      "grad_norm": 2131.5027472197603,
+      "learning_rate": 1.7169094257333307e-05,
+      "loss": 377.5089,
+      "step": 13160
+    },
+    {
+      "epoch": 0.253292368052851,
+      "grad_norm": 2201.601387732003,
+      "learning_rate": 1.716483801695636e-05,
+      "loss": 372.2224,
+      "step": 13170
+    },
+    {
+      "epoch": 0.2534846933133315,
+      "grad_norm": 2270.5089754773767,
+      "learning_rate": 1.7160579107795587e-05,
+      "loss": 371.2687,
+      "step": 13180
+    },
+    {
+      "epoch": 0.25367701857381203,
+      "grad_norm": 2087.7274790080764,
+      "learning_rate": 1.715631753143735e-05,
+      "loss": 361.2863,
+      "step": 13190
+    },
+    {
+      "epoch": 0.25386934383429255,
+      "grad_norm": 2203.274867124545,
+      "learning_rate": 1.715205328946903e-05,
+      "loss": 375.0043,
+      "step": 13200
+    },
+    {
+      "epoch": 0.25406166909477307,
+      "grad_norm": 2183.0265198187403,
+      "learning_rate": 1.7147786383478978e-05,
+      "loss": 365.3814,
+      "step": 13210
+    },
+    {
+      "epoch": 0.2542539943552536,
+      "grad_norm": 2071.064381032265,
+      "learning_rate": 1.7143516815056545e-05,
+      "loss": 373.6266,
+      "step": 13220
+    },
+    {
+      "epoch": 0.2544463196157341,
+      "grad_norm": 2120.748161443223,
+      "learning_rate": 1.713924458579208e-05,
+      "loss": 381.5468,
+      "step": 13230
+    },
+    {
+      "epoch": 0.2546386448762147,
+      "grad_norm": 2381.4264809998035,
+      "learning_rate": 1.7134969697276912e-05,
+      "loss": 361.3351,
+      "step": 13240
+    },
+    {
+      "epoch": 0.2548309701366952,
+      "grad_norm": 2204.218987668362,
+      "learning_rate": 1.7130692151103373e-05,
+      "loss": 360.3402,
+      "step": 13250
+    },
+    {
+      "epoch": 0.2550232953971757,
+      "grad_norm": 2287.9938308956316,
+      "learning_rate": 1.7126411948864776e-05,
+      "loss": 361.9453,
+      "step": 13260
+    },
+    {
+      "epoch": 0.25521562065765624,
+      "grad_norm": 2352.708402483653,
+      "learning_rate": 1.7122129092155422e-05,
+      "loss": 372.9736,
+      "step": 13270
+    },
+    {
+      "epoch": 0.25540794591813676,
+      "grad_norm": 2365.0151964818424,
+      "learning_rate": 1.7117843582570608e-05,
+      "loss": 363.5916,
+      "step": 13280
+    },
+    {
+      "epoch": 0.2556002711786173,
+      "grad_norm": 2428.924702841836,
+      "learning_rate": 1.711355542170661e-05,
+      "loss": 354.8736,
+      "step": 13290
+    },
+    {
+      "epoch": 0.2557925964390978,
+      "grad_norm": 2373.2535106694277,
+      "learning_rate": 1.710926461116071e-05,
+      "loss": 354.0335,
+      "step": 13300
+    },
+    {
+      "epoch": 0.2559849216995783,
+      "grad_norm": 2174.4421393852217,
+      "learning_rate": 1.710497115253115e-05,
+      "loss": 367.6362,
+      "step": 13310
+    },
+    {
+      "epoch": 0.25617724696005884,
+      "grad_norm": 2100.664711178483,
+      "learning_rate": 1.7100675047417178e-05,
+      "loss": 362.7153,
+      "step": 13320
+    },
+    {
+      "epoch": 0.25636957222053935,
+      "grad_norm": 2265.959941281571,
+      "learning_rate": 1.7096376297419027e-05,
+      "loss": 364.1679,
+      "step": 13330
+    },
+    {
+      "epoch": 0.2565618974810199,
+      "grad_norm": 2303.4915717391596,
+      "learning_rate": 1.70920749041379e-05,
+      "loss": 368.2602,
+      "step": 13340
+    },
+    {
+      "epoch": 0.25675422274150045,
+      "grad_norm": 2253.541679059977,
+      "learning_rate": 1.7087770869176005e-05,
+      "loss": 363.7776,
+      "step": 13350
+    },
+    {
+      "epoch": 0.25694654800198097,
+      "grad_norm": 2619.359812239953,
+      "learning_rate": 1.7083464194136517e-05,
+      "loss": 364.5733,
+      "step": 13360
+    },
+    {
+      "epoch": 0.2571388732624615,
+      "grad_norm": 2327.6661959592793,
+      "learning_rate": 1.707915488062361e-05,
+      "loss": 367.0563,
+      "step": 13370
+    },
+    {
+      "epoch": 0.257331198522942,
+      "grad_norm": 2210.701600130927,
+      "learning_rate": 1.7074842930242418e-05,
+      "loss": 361.3447,
+      "step": 13380
+    },
+    {
+      "epoch": 0.2575235237834225,
+      "grad_norm": 2255.424614088036,
+      "learning_rate": 1.7070528344599083e-05,
+      "loss": 366.5962,
+      "step": 13390
+    },
+    {
+      "epoch": 0.25771584904390304,
+      "grad_norm": 1939.1075943547062,
+      "learning_rate": 1.7066211125300713e-05,
+      "loss": 355.21,
+      "step": 13400
+    },
+    {
+      "epoch": 0.25790817430438356,
+      "grad_norm": 2218.0890782546844,
+      "learning_rate": 1.70618912739554e-05,
+      "loss": 354.0135,
+      "step": 13410
+    },
+    {
+      "epoch": 0.2581004995648641,
+      "grad_norm": 2228.9430765714455,
+      "learning_rate": 1.705756879217222e-05,
+      "loss": 360.8117,
+      "step": 13420
+    },
+    {
+      "epoch": 0.2582928248253446,
+      "grad_norm": 2339.181835304501,
+      "learning_rate": 1.7053243681561225e-05,
+      "loss": 391.1912,
+      "step": 13430
+    },
+    {
+      "epoch": 0.2584851500858251,
+      "grad_norm": 2942.846542274916,
+      "learning_rate": 1.7048915943733444e-05,
+      "loss": 374.7205,
+      "step": 13440
+    },
+    {
+      "epoch": 0.2586774753463057,
+      "grad_norm": 2108.8569405750263,
+      "learning_rate": 1.704458558030089e-05,
+      "loss": 361.5591,
+      "step": 13450
+    },
+    {
+      "epoch": 0.2588698006067862,
+      "grad_norm": 2173.377418923527,
+      "learning_rate": 1.704025259287656e-05,
+      "loss": 349.6717,
+      "step": 13460
+    },
+    {
+      "epoch": 0.25906212586726673,
+      "grad_norm": 2270.1040677140495,
+      "learning_rate": 1.7035916983074405e-05,
+      "loss": 365.4547,
+      "step": 13470
+    },
+    {
+      "epoch": 0.25925445112774725,
+      "grad_norm": 2191.429983903243,
+      "learning_rate": 1.7031578752509377e-05,
+      "loss": 376.328,
+      "step": 13480
+    },
+    {
+      "epoch": 0.2594467763882278,
+      "grad_norm": 2341.776015951565,
+      "learning_rate": 1.70272379027974e-05,
+      "loss": 353.0333,
+      "step": 13490
+    },
+    {
+      "epoch": 0.2596391016487083,
+      "grad_norm": 2991.992840272398,
+      "learning_rate": 1.7022894435555356e-05,
+      "loss": 364.1902,
+      "step": 13500
+    },
+    {
+      "epoch": 0.2598314269091888,
+      "grad_norm": 2251.311098855147,
+      "learning_rate": 1.7018548352401123e-05,
+      "loss": 360.6715,
+      "step": 13510
+    },
+    {
+      "epoch": 0.26002375216966933,
+      "grad_norm": 2069.2219256224384,
+      "learning_rate": 1.7014199654953543e-05,
+      "loss": 354.686,
+      "step": 13520
+    },
+    {
+      "epoch": 0.26021607743014985,
+      "grad_norm": 2339.4828322112303,
+      "learning_rate": 1.700984834483244e-05,
+      "loss": 357.82,
+      "step": 13530
+    },
+    {
+      "epoch": 0.26040840269063037,
+      "grad_norm": 2275.0942544119675,
+      "learning_rate": 1.7005494423658598e-05,
+      "loss": 361.6596,
+      "step": 13540
+    },
+    {
+      "epoch": 0.26060072795111094,
+      "grad_norm": 2423.1712189884956,
+      "learning_rate": 1.7001137893053782e-05,
+      "loss": 352.1861,
+      "step": 13550
+    },
+    {
+      "epoch": 0.26079305321159146,
+      "grad_norm": 1952.630341986935,
+      "learning_rate": 1.6996778754640727e-05,
+      "loss": 355.9204,
+      "step": 13560
+    },
+    {
+      "epoch": 0.260985378472072,
+      "grad_norm": 2803.1007945938454,
+      "learning_rate": 1.6992417010043144e-05,
+      "loss": 362.2839,
+      "step": 13570
+    },
+    {
+      "epoch": 0.2611777037325525,
+      "grad_norm": 2491.0123601699406,
+      "learning_rate": 1.6988052660885707e-05,
+      "loss": 356.3197,
+      "step": 13580
+    },
+    {
+      "epoch": 0.261370028993033,
+      "grad_norm": 2249.6902256937415,
+      "learning_rate": 1.6983685708794064e-05,
+      "loss": 351.8635,
+      "step": 13590
+    },
+    {
+      "epoch": 0.26156235425351354,
+      "grad_norm": 2154.262615559068,
+      "learning_rate": 1.6979316155394834e-05,
+      "loss": 360.5741,
+      "step": 13600
+    },
+    {
+      "epoch": 0.26175467951399406,
+      "grad_norm": 2290.60966799732,
+      "learning_rate": 1.6974944002315605e-05,
+      "loss": 352.6973,
+      "step": 13610
+    },
+    {
+      "epoch": 0.2619470047744746,
+      "grad_norm": 2141.0607880001735,
+      "learning_rate": 1.697056925118493e-05,
+      "loss": 355.3396,
+      "step": 13620
+    },
+    {
+      "epoch": 0.2621393300349551,
+      "grad_norm": 2161.2153868605674,
+      "learning_rate": 1.696619190363233e-05,
+      "loss": 372.7554,
+      "step": 13630
+    },
+    {
+      "epoch": 0.2623316552954356,
+      "grad_norm": 2306.6074489399607,
+      "learning_rate": 1.69618119612883e-05,
+      "loss": 352.0928,
+      "step": 13640
+    },
+    {
+      "epoch": 0.2625239805559162,
+      "grad_norm": 2372.277682089138,
+      "learning_rate": 1.695742942578429e-05,
+      "loss": 360.1005,
+      "step": 13650
+    },
+    {
+      "epoch": 0.2627163058163967,
+      "grad_norm": 2148.801092905007,
+      "learning_rate": 1.6953044298752724e-05,
+      "loss": 354.6568,
+      "step": 13660
+    },
+    {
+      "epoch": 0.26290863107687723,
+      "grad_norm": 2107.752353383053,
+      "learning_rate": 1.694865658182699e-05,
+      "loss": 376.3746,
+      "step": 13670
+    },
+    {
+      "epoch": 0.26310095633735775,
+      "grad_norm": 2251.879910727671,
+      "learning_rate": 1.6944266276641442e-05,
+      "loss": 359.1013,
+      "step": 13680
+    },
+    {
+      "epoch": 0.26329328159783827,
+      "grad_norm": 2280.915123621908,
+      "learning_rate": 1.6939873384831394e-05,
+      "loss": 352.8126,
+      "step": 13690
+    },
+    {
+      "epoch": 0.2634856068583188,
+      "grad_norm": 2151.763211719724,
+      "learning_rate": 1.6935477908033124e-05,
+      "loss": 365.3831,
+      "step": 13700
+    },
+    {
+      "epoch": 0.2636779321187993,
+      "grad_norm": 2219.681130064892,
+      "learning_rate": 1.6931079847883877e-05,
+      "loss": 360.9801,
+      "step": 13710
+    },
+    {
+      "epoch": 0.2638702573792798,
+      "grad_norm": 2148.437330167889,
+      "learning_rate": 1.692667920602186e-05,
+      "loss": 360.1392,
+      "step": 13720
+    },
+    {
+      "epoch": 0.26406258263976035,
+      "grad_norm": 2311.2112790663277,
+      "learning_rate": 1.6922275984086233e-05,
+      "loss": 368.1768,
+      "step": 13730
+    },
+    {
+      "epoch": 0.26425490790024087,
+      "grad_norm": 3347.3742915926946,
+      "learning_rate": 1.691787018371713e-05,
+      "loss": 359.7262,
+      "step": 13740
+    },
+    {
+      "epoch": 0.2644472331607214,
+      "grad_norm": 1958.7131328256862,
+      "learning_rate": 1.6913461806555635e-05,
+      "loss": 366.6636,
+      "step": 13750
+    },
+    {
+      "epoch": 0.26463955842120196,
+      "grad_norm": 2124.205274840122,
+      "learning_rate": 1.6909050854243797e-05,
+      "loss": 361.5742,
+      "step": 13760
+    },
+    {
+      "epoch": 0.2648318836816825,
+      "grad_norm": 2196.083347899615,
+      "learning_rate": 1.690463732842462e-05,
+      "loss": 358.3661,
+      "step": 13770
+    },
+    {
+      "epoch": 0.265024208942163,
+      "grad_norm": 2099.8495510558323,
+      "learning_rate": 1.6900221230742073e-05,
+      "loss": 363.142,
+      "step": 13780
+    },
+    {
+      "epoch": 0.2652165342026435,
+      "grad_norm": 2287.4135454319503,
+      "learning_rate": 1.689580256284108e-05,
+      "loss": 371.7677,
+      "step": 13790
+    },
+    {
+      "epoch": 0.26540885946312404,
+      "grad_norm": 2159.738428224536,
+      "learning_rate": 1.689138132636752e-05,
+      "loss": 355.8717,
+      "step": 13800
+    },
+    {
+      "epoch": 0.26560118472360456,
+      "grad_norm": 2417.10382081598,
+      "learning_rate": 1.688695752296823e-05,
+      "loss": 353.2436,
+      "step": 13810
+    },
+    {
+      "epoch": 0.2657935099840851,
+      "grad_norm": 2175.4065633322334,
+      "learning_rate": 1.6882531154291007e-05,
+      "loss": 357.5786,
+      "step": 13820
+    },
+    {
+      "epoch": 0.2659858352445656,
+      "grad_norm": 2448.4463448035663,
+      "learning_rate": 1.6878102221984593e-05,
+      "loss": 366.5969,
+      "step": 13830
+    },
+    {
+      "epoch": 0.2661781605050461,
+      "grad_norm": 2311.1519905753667,
+      "learning_rate": 1.6873670727698702e-05,
+      "loss": 370.8688,
+      "step": 13840
+    },
+    {
+      "epoch": 0.26637048576552663,
+      "grad_norm": 1833.2445852137944,
+      "learning_rate": 1.686923667308398e-05,
+      "loss": 359.431,
+      "step": 13850
+    },
+    {
+      "epoch": 0.2665628110260072,
+      "grad_norm": 2311.8042821976733,
+      "learning_rate": 1.6864800059792057e-05,
+      "loss": 358.09,
+      "step": 13860
+    },
+    {
+      "epoch": 0.2667551362864877,
+      "grad_norm": 2290.26105451503,
+      "learning_rate": 1.686036088947548e-05,
+      "loss": 378.6407,
+      "step": 13870
+    },
+    {
+      "epoch": 0.26694746154696825,
+      "grad_norm": 2308.049201900373,
+      "learning_rate": 1.6855919163787777e-05,
+      "loss": 368.1997,
+      "step": 13880
+    },
+    {
+      "epoch": 0.26713978680744876,
+      "grad_norm": 2402.727721477322,
+      "learning_rate": 1.6851474884383416e-05,
+      "loss": 352.0517,
+      "step": 13890
+    },
+    {
+      "epoch": 0.2673321120679293,
+      "grad_norm": 2126.4202684865063,
+      "learning_rate": 1.6847028052917814e-05,
+      "loss": 356.8757,
+      "step": 13900
+    },
+    {
+      "epoch": 0.2675244373284098,
+      "grad_norm": 2219.704431840823,
+      "learning_rate": 1.6842578671047345e-05,
+      "loss": 351.6863,
+      "step": 13910
+    },
+    {
+      "epoch": 0.2677167625888903,
+      "grad_norm": 2165.459820766716,
+      "learning_rate": 1.683812674042933e-05,
+      "loss": 343.2167,
+      "step": 13920
+    },
+    {
+      "epoch": 0.26790908784937084,
+      "grad_norm": 1973.858326095211,
+      "learning_rate": 1.683367226272204e-05,
+      "loss": 354.5079,
+      "step": 13930
+    },
+    {
+      "epoch": 0.26810141310985136,
+      "grad_norm": 2009.8594355693149,
+      "learning_rate": 1.6829215239584695e-05,
+      "loss": 359.1503,
+      "step": 13940
+    },
+    {
+      "epoch": 0.2682937383703319,
+      "grad_norm": 2476.9372914391574,
+      "learning_rate": 1.6824755672677458e-05,
+      "loss": 359.3496,
+      "step": 13950
+    },
+    {
+      "epoch": 0.26848606363081245,
+      "grad_norm": 2533.3646696574024,
+      "learning_rate": 1.682029356366145e-05,
+      "loss": 357.923,
+      "step": 13960
+    },
+    {
+      "epoch": 0.268678388891293,
+      "grad_norm": 1972.045103103336,
+      "learning_rate": 1.6815828914198732e-05,
+      "loss": 343.1023,
+      "step": 13970
+    },
+    {
+      "epoch": 0.2688707141517735,
+      "grad_norm": 2168.6922610639886,
+      "learning_rate": 1.6811361725952308e-05,
+      "loss": 369.1714,
+      "step": 13980
+    },
+    {
+      "epoch": 0.269063039412254,
+      "grad_norm": 2238.709758857822,
+      "learning_rate": 1.6806892000586135e-05,
+      "loss": 356.6833,
+      "step": 13990
+    },
+    {
+      "epoch": 0.26925536467273453,
+      "grad_norm": 2216.9579415444555,
+      "learning_rate": 1.6802419739765114e-05,
+      "loss": 356.4481,
+      "step": 14000
+    },
+    {
+      "epoch": 0.26944768993321505,
+      "grad_norm": 2411.948012072841,
+      "learning_rate": 1.679794494515508e-05,
+      "loss": 357.2391,
+      "step": 14010
+    },
+    {
+      "epoch": 0.26964001519369557,
+      "grad_norm": 2333.3106707951483,
+      "learning_rate": 1.6793467618422828e-05,
+      "loss": 351.3738,
+      "step": 14020
+    },
+    {
+      "epoch": 0.2698323404541761,
+      "grad_norm": 2344.896906114102,
+      "learning_rate": 1.6788987761236088e-05,
+      "loss": 359.5713,
+      "step": 14030
+    },
+    {
+      "epoch": 0.2700246657146566,
+      "grad_norm": 2047.8087364894836,
+      "learning_rate": 1.6784505375263533e-05,
+      "loss": 351.5925,
+      "step": 14040
+    },
+    {
+      "epoch": 0.27021699097513713,
+      "grad_norm": 2289.9871298987428,
+      "learning_rate": 1.678002046217477e-05,
+      "loss": 354.071,
+      "step": 14050
+    },
+    {
+      "epoch": 0.2704093162356177,
+      "grad_norm": 2297.079878289623,
+      "learning_rate": 1.6775533023640363e-05,
+      "loss": 382.4293,
+      "step": 14060
+    },
+    {
+      "epoch": 0.2706016414960982,
+      "grad_norm": 2707.1323362802527,
+      "learning_rate": 1.6771043061331806e-05,
+      "loss": 379.08,
+      "step": 14070
+    },
+    {
+      "epoch": 0.27079396675657874,
+      "grad_norm": 2140.7356612219182,
+      "learning_rate": 1.6766550576921533e-05,
+      "loss": 350.7611,
+      "step": 14080
+    },
+    {
+      "epoch": 0.27098629201705926,
+      "grad_norm": 2148.0099854408904,
+      "learning_rate": 1.676205557208293e-05,
+      "loss": 361.6202,
+      "step": 14090
+    },
+    {
+      "epoch": 0.2711786172775398,
+      "grad_norm": 2315.1241871264983,
+      "learning_rate": 1.67575580484903e-05,
+      "loss": 341.0929,
+      "step": 14100
+    },
+    {
+      "epoch": 0.2713709425380203,
+      "grad_norm": 2316.589255783513,
+      "learning_rate": 1.6753058007818906e-05,
+      "loss": 360.6659,
+      "step": 14110
+    },
+    {
+      "epoch": 0.2715632677985008,
+      "grad_norm": 1899.0355660504529,
+      "learning_rate": 1.674855545174493e-05,
+      "loss": 351.1972,
+      "step": 14120
+    },
+    {
+      "epoch": 0.27175559305898134,
+      "grad_norm": 2222.084557494,
+      "learning_rate": 1.6744050381945507e-05,
+      "loss": 361.3624,
+      "step": 14130
+    },
+    {
+      "epoch": 0.27194791831946186,
+      "grad_norm": 2264.881700045289,
+      "learning_rate": 1.67395428000987e-05,
+      "loss": 352.5764,
+      "step": 14140
+    },
+    {
+      "epoch": 0.2721402435799424,
+      "grad_norm": 2189.680143711977,
+      "learning_rate": 1.6735032707883502e-05,
+      "loss": 355.9011,
+      "step": 14150
+    },
+    {
+      "epoch": 0.27233256884042295,
+      "grad_norm": 1972.6086135796374,
+      "learning_rate": 1.6730520106979855e-05,
+      "loss": 341.7274,
+      "step": 14160
+    },
+    {
+      "epoch": 0.27252489410090347,
+      "grad_norm": 2110.300745632065,
+      "learning_rate": 1.672600499906863e-05,
+      "loss": 353.6648,
+      "step": 14170
+    },
+    {
+      "epoch": 0.272717219361384,
+      "grad_norm": 2204.052763725582,
+      "learning_rate": 1.6721487385831622e-05,
+      "loss": 357.5317,
+      "step": 14180
+    },
+    {
+      "epoch": 0.2729095446218645,
+      "grad_norm": 2041.8102406745647,
+      "learning_rate": 1.6716967268951574e-05,
+      "loss": 358.4835,
+      "step": 14190
+    },
+    {
+      "epoch": 0.273101869882345,
+      "grad_norm": 2005.8114284987464,
+      "learning_rate": 1.6712444650112152e-05,
+      "loss": 362.9404,
+      "step": 14200
+    },
+    {
+      "epoch": 0.27329419514282555,
+      "grad_norm": 2412.752289840888,
+      "learning_rate": 1.6707919530997956e-05,
+      "loss": 363.9804,
+      "step": 14210
+    },
+    {
+      "epoch": 0.27348652040330607,
+      "grad_norm": 2471.0214667614428,
+      "learning_rate": 1.6703391913294524e-05,
+      "loss": 354.7934,
+      "step": 14220
+    },
+    {
+      "epoch": 0.2736788456637866,
+      "grad_norm": 2675.282545326385,
+      "learning_rate": 1.6698861798688312e-05,
+      "loss": 360.2001,
+      "step": 14230
+    },
+    {
+      "epoch": 0.2738711709242671,
+      "grad_norm": 2087.417377825284,
+      "learning_rate": 1.6694329188866717e-05,
+      "loss": 358.0862,
+      "step": 14240
+    },
+    {
+      "epoch": 0.2740634961847476,
+      "grad_norm": 2181.953738189635,
+      "learning_rate": 1.6689794085518057e-05,
+      "loss": 345.6908,
+      "step": 14250
+    },
+    {
+      "epoch": 0.27425582144522814,
+      "grad_norm": 2214.301271829383,
+      "learning_rate": 1.668525649033159e-05,
+      "loss": 360.2512,
+      "step": 14260
+    },
+    {
+      "epoch": 0.2744481467057087,
+      "grad_norm": 2070.4968198587185,
+      "learning_rate": 1.6680716404997482e-05,
+      "loss": 351.3011,
+      "step": 14270
+    },
+    {
+      "epoch": 0.27464047196618924,
+      "grad_norm": 1970.425874835042,
+      "learning_rate": 1.667617383120686e-05,
+      "loss": 359.1019,
+      "step": 14280
+    },
+    {
+      "epoch": 0.27483279722666976,
+      "grad_norm": 1979.2190282830913,
+      "learning_rate": 1.667162877065174e-05,
+      "loss": 352.6497,
+      "step": 14290
+    },
+    {
+      "epoch": 0.2750251224871503,
+      "grad_norm": 2053.884922474694,
+      "learning_rate": 1.6667081225025087e-05,
+      "loss": 347.5574,
+      "step": 14300
+    },
+    {
+      "epoch": 0.2752174477476308,
+      "grad_norm": 2823.2874066492827,
+      "learning_rate": 1.666253119602079e-05,
+      "loss": 361.0676,
+      "step": 14310
+    },
+    {
+      "epoch": 0.2754097730081113,
+      "grad_norm": 2429.729872293584,
+      "learning_rate": 1.665797868533366e-05,
+      "loss": 345.4754,
+      "step": 14320
+    },
+    {
+      "epoch": 0.27560209826859183,
+      "grad_norm": 2564.8506157499537,
+      "learning_rate": 1.6653423694659433e-05,
+      "loss": 366.7636,
+      "step": 14330
+    },
+    {
+      "epoch": 0.27579442352907235,
+      "grad_norm": 2126.2268195384677,
+      "learning_rate": 1.6648866225694757e-05,
+      "loss": 354.3418,
+      "step": 14340
+    },
+    {
+      "epoch": 0.27598674878955287,
+      "grad_norm": 2174.4487955254194,
+      "learning_rate": 1.6644306280137227e-05,
+      "loss": 350.8759,
+      "step": 14350
+    },
+    {
+      "epoch": 0.2761790740500334,
+      "grad_norm": 2121.795515758063,
+      "learning_rate": 1.6639743859685336e-05,
+      "loss": 358.6837,
+      "step": 14360
+    },
+    {
+      "epoch": 0.27637139931051397,
+      "grad_norm": 2235.7630797497222,
+      "learning_rate": 1.663517896603852e-05,
+      "loss": 353.1709,
+      "step": 14370
+    },
+    {
+      "epoch": 0.2765637245709945,
+      "grad_norm": 2119.0977257896884,
+      "learning_rate": 1.6630611600897126e-05,
+      "loss": 352.6597,
+      "step": 14380
+    },
+    {
+      "epoch": 0.276756049831475,
+      "grad_norm": 2501.0019998363905,
+      "learning_rate": 1.6626041765962413e-05,
+      "loss": 364.0342,
+      "step": 14390
+    },
+    {
+      "epoch": 0.2769483750919555,
+      "grad_norm": 2166.2492933080384,
+      "learning_rate": 1.662146946293658e-05,
+      "loss": 354.1955,
+      "step": 14400
+    },
+    {
+      "epoch": 0.27714070035243604,
+      "grad_norm": 2126.9277228427763,
+      "learning_rate": 1.6616894693522727e-05,
+      "loss": 346.3621,
+      "step": 14410
+    },
+    {
+      "epoch": 0.27733302561291656,
+      "grad_norm": 2257.7783824903768,
+      "learning_rate": 1.6612317459424884e-05,
+      "loss": 345.0017,
+      "step": 14420
+    },
+    {
+      "epoch": 0.2775253508733971,
+      "grad_norm": 2179.224609595095,
+      "learning_rate": 1.6607737762347987e-05,
+      "loss": 352.2429,
+      "step": 14430
+    },
+    {
+      "epoch": 0.2777176761338776,
+      "grad_norm": 2067.1268379003523,
+      "learning_rate": 1.6603155603997908e-05,
+      "loss": 350.3075,
+      "step": 14440
+    },
+    {
+      "epoch": 0.2779100013943581,
+      "grad_norm": 2304.950825416585,
+      "learning_rate": 1.6598570986081424e-05,
+      "loss": 353.2081,
+      "step": 14450
+    },
+    {
+      "epoch": 0.27810232665483864,
+      "grad_norm": 14988.110072366713,
+      "learning_rate": 1.6593983910306225e-05,
+      "loss": 361.7833,
+      "step": 14460
+    },
+    {
+      "epoch": 0.2782946519153192,
+      "grad_norm": 2311.952332545013,
+      "learning_rate": 1.658939437838092e-05,
+      "loss": 345.4867,
+      "step": 14470
+    },
+    {
+      "epoch": 0.27848697717579973,
+      "grad_norm": 2068.3808973281393,
+      "learning_rate": 1.658480239201504e-05,
+      "loss": 348.6313,
+      "step": 14480
+    },
+    {
+      "epoch": 0.27867930243628025,
+      "grad_norm": 2170.969441943965,
+      "learning_rate": 1.6580207952919018e-05,
+      "loss": 342.5731,
+      "step": 14490
+    },
+    {
+      "epoch": 0.27887162769676077,
+      "grad_norm": 2021.5175783147224,
+      "learning_rate": 1.657561106280421e-05,
+      "loss": 346.9506,
+      "step": 14500
+    },
+    {
+      "epoch": 0.2790639529572413,
+      "grad_norm": 2028.6282206966307,
+      "learning_rate": 1.6571011723382882e-05,
+      "loss": 346.9975,
+      "step": 14510
+    },
+    {
+      "epoch": 0.2792562782177218,
+      "grad_norm": 2187.477258236756,
+      "learning_rate": 1.6566409936368207e-05,
+      "loss": 356.4622,
+      "step": 14520
+    },
+    {
+      "epoch": 0.27944860347820233,
+      "grad_norm": 2137.1956438430307,
+      "learning_rate": 1.6561805703474285e-05,
+      "loss": 342.7526,
+      "step": 14530
+    },
+    {
+      "epoch": 0.27964092873868285,
+      "grad_norm": 2218.8060348649406,
+      "learning_rate": 1.655719902641611e-05,
+      "loss": 352.7918,
+      "step": 14540
+    },
+    {
+      "epoch": 0.27983325399916337,
+      "grad_norm": 2144.055837550634,
+      "learning_rate": 1.6552589906909586e-05,
+      "loss": 348.8337,
+      "step": 14550
+    },
+    {
+      "epoch": 0.2800255792596439,
+      "grad_norm": 2253.7484428585017,
+      "learning_rate": 1.654797834667155e-05,
+      "loss": 336.3861,
+      "step": 14560
+    },
+    {
+      "epoch": 0.28021790452012446,
+      "grad_norm": 2236.6492524430246,
+      "learning_rate": 1.6543364347419714e-05,
+      "loss": 354.0255,
+      "step": 14570
+    },
+    {
+      "epoch": 0.280410229780605,
+      "grad_norm": 2160.2015474686873,
+      "learning_rate": 1.6538747910872733e-05,
+      "loss": 350.5711,
+      "step": 14580
+    },
+    {
+      "epoch": 0.2806025550410855,
+      "grad_norm": 2135.825692923429,
+      "learning_rate": 1.6534129038750145e-05,
+      "loss": 351.2991,
+      "step": 14590
+    },
+    {
+      "epoch": 0.280794880301566,
+      "grad_norm": 2102.410992561339,
+      "learning_rate": 1.65295077327724e-05,
+      "loss": 347.3173,
+      "step": 14600
+    },
+    {
+      "epoch": 0.28098720556204654,
+      "grad_norm": 1960.491867519005,
+      "learning_rate": 1.652488399466087e-05,
+      "loss": 338.7215,
+      "step": 14610
+    },
+    {
+      "epoch": 0.28117953082252706,
+      "grad_norm": 2216.935971908477,
+      "learning_rate": 1.6520257826137807e-05,
+      "loss": 355.5081,
+      "step": 14620
+    },
+    {
+      "epoch": 0.2813718560830076,
+      "grad_norm": 2545.0622226305786,
+      "learning_rate": 1.6515629228926396e-05,
+      "loss": 357.2568,
+      "step": 14630
+    },
+    {
+      "epoch": 0.2815641813434881,
+      "grad_norm": 2415.123601846704,
+      "learning_rate": 1.6510998204750702e-05,
+      "loss": 364.0826,
+      "step": 14640
+    },
+    {
+      "epoch": 0.2817565066039686,
+      "grad_norm": 2157.644372041305,
+      "learning_rate": 1.650636475533571e-05,
+      "loss": 349.2333,
+      "step": 14650
+    },
+    {
+      "epoch": 0.28194883186444913,
+      "grad_norm": 2045.6376577360497,
+      "learning_rate": 1.6501728882407305e-05,
+      "loss": 350.3113,
+      "step": 14660
+    },
+    {
+      "epoch": 0.28214115712492965,
+      "grad_norm": 2119.425827077217,
+      "learning_rate": 1.649709058769227e-05,
+      "loss": 338.2437,
+      "step": 14670
+    },
+    {
+      "epoch": 0.28233348238541023,
+      "grad_norm": 2559.4017979471387,
+      "learning_rate": 1.6492449872918293e-05,
+      "loss": 337.4597,
+      "step": 14680
+    },
+    {
+      "epoch": 0.28252580764589075,
+      "grad_norm": 2034.6266755131946,
+      "learning_rate": 1.6487806739813966e-05,
+      "loss": 343.1722,
+      "step": 14690
+    },
+    {
+      "epoch": 0.28271813290637127,
+      "grad_norm": 2027.6246653423716,
+      "learning_rate": 1.6483161190108778e-05,
+      "loss": 348.8538,
+      "step": 14700
+    },
+    {
+      "epoch": 0.2829104581668518,
+      "grad_norm": 2391.058467009256,
+      "learning_rate": 1.6478513225533117e-05,
+      "loss": 354.9481,
+      "step": 14710
+    },
+    {
+      "epoch": 0.2831027834273323,
+      "grad_norm": 1935.4540753060578,
+      "learning_rate": 1.647386284781828e-05,
+      "loss": 353.477,
+      "step": 14720
+    },
+    {
+      "epoch": 0.2832951086878128,
+      "grad_norm": 1900.7454901121991,
+      "learning_rate": 1.6469210058696448e-05,
+      "loss": 356.1835,
+      "step": 14730
+    },
+    {
+      "epoch": 0.28348743394829334,
+      "grad_norm": 2178.9793520560766,
+      "learning_rate": 1.646455485990071e-05,
+      "loss": 344.4227,
+      "step": 14740
+    },
+    {
+      "epoch": 0.28367975920877386,
+      "grad_norm": 2317.573864365449,
+      "learning_rate": 1.645989725316506e-05,
+      "loss": 342.7161,
+      "step": 14750
+    },
+    {
+      "epoch": 0.2838720844692544,
+      "grad_norm": 2148.3270520104556,
+      "learning_rate": 1.6455237240224364e-05,
+      "loss": 341.3651,
+      "step": 14760
+    },
+    {
+      "epoch": 0.2840644097297349,
+      "grad_norm": 1890.0661872101743,
+      "learning_rate": 1.6450574822814412e-05,
+      "loss": 348.7713,
+      "step": 14770
+    },
+    {
+      "epoch": 0.2842567349902155,
+      "grad_norm": 2111.166658494511,
+      "learning_rate": 1.6445910002671872e-05,
+      "loss": 346.0004,
+      "step": 14780
+    },
+    {
+      "epoch": 0.284449060250696,
+      "grad_norm": 2182.8152442404066,
+      "learning_rate": 1.644124278153431e-05,
+      "loss": 349.5256,
+      "step": 14790
+    },
+    {
+      "epoch": 0.2846413855111765,
+      "grad_norm": 2174.55236921231,
+      "learning_rate": 1.64365731611402e-05,
+      "loss": 351.512,
+      "step": 14800
+    },
+    {
+      "epoch": 0.28483371077165703,
+      "grad_norm": 1925.2918978791486,
+      "learning_rate": 1.6431901143228888e-05,
+      "loss": 341.6268,
+      "step": 14810
+    },
+    {
+      "epoch": 0.28502603603213755,
+      "grad_norm": 2174.7874966646423,
+      "learning_rate": 1.6427226729540623e-05,
+      "loss": 340.9609,
+      "step": 14820
+    },
+    {
+      "epoch": 0.2852183612926181,
+      "grad_norm": 2134.1172658835803,
+      "learning_rate": 1.6422549921816556e-05,
+      "loss": 351.6038,
+      "step": 14830
+    },
+    {
+      "epoch": 0.2854106865530986,
+      "grad_norm": 2217.3516934749564,
+      "learning_rate": 1.641787072179871e-05,
+      "loss": 341.5353,
+      "step": 14840
+    },
+    {
+      "epoch": 0.2856030118135791,
+      "grad_norm": 2223.8972668867937,
+      "learning_rate": 1.6413189131230022e-05,
+      "loss": 341.5969,
+      "step": 14850
+    },
+    {
+      "epoch": 0.28579533707405963,
+      "grad_norm": 2054.9513926731647,
+      "learning_rate": 1.6408505151854292e-05,
+      "loss": 342.8476,
+      "step": 14860
+    },
+    {
+      "epoch": 0.28598766233454015,
+      "grad_norm": 2009.6768638747901,
+      "learning_rate": 1.6403818785416236e-05,
+      "loss": 339.1113,
+      "step": 14870
+    },
+    {
+      "epoch": 0.2861799875950207,
+      "grad_norm": 2053.080049113835,
+      "learning_rate": 1.6399130033661444e-05,
+      "loss": 357.4447,
+      "step": 14880
+    },
+    {
+      "epoch": 0.28637231285550124,
+      "grad_norm": 2127.5793168709356,
+      "learning_rate": 1.6394438898336402e-05,
+      "loss": 347.4203,
+      "step": 14890
+    },
+    {
+      "epoch": 0.28656463811598176,
+      "grad_norm": 2410.0591550511567,
+      "learning_rate": 1.6389745381188475e-05,
+      "loss": 353.4183,
+      "step": 14900
+    },
+    {
+      "epoch": 0.2867569633764623,
+      "grad_norm": 1984.7275396999785,
+      "learning_rate": 1.6385049483965926e-05,
+      "loss": 339.8347,
+      "step": 14910
+    },
+    {
+      "epoch": 0.2869492886369428,
+      "grad_norm": 2272.169987853921,
+      "learning_rate": 1.6380351208417897e-05,
+      "loss": 345.1773,
+      "step": 14920
+    },
+    {
+      "epoch": 0.2871416138974233,
+      "grad_norm": 2174.804837594622,
+      "learning_rate": 1.6375650556294417e-05,
+      "loss": 341.4035,
+      "step": 14930
+    },
+    {
+      "epoch": 0.28733393915790384,
+      "grad_norm": 2274.165185256845,
+      "learning_rate": 1.6370947529346404e-05,
+      "loss": 346.0246,
+      "step": 14940
+    },
+    {
+      "epoch": 0.28752626441838436,
+      "grad_norm": 2070.155956987362,
+      "learning_rate": 1.6366242129325652e-05,
+      "loss": 347.7661,
+      "step": 14950
+    },
+    {
+      "epoch": 0.2877185896788649,
+      "grad_norm": 1993.8221844616796,
+      "learning_rate": 1.636153435798485e-05,
+      "loss": 352.0007,
+      "step": 14960
+    },
+    {
+      "epoch": 0.2879109149393454,
+      "grad_norm": 2197.798183522224,
+      "learning_rate": 1.6356824217077564e-05,
+      "loss": 333.8231,
+      "step": 14970
+    },
+    {
+      "epoch": 0.28810324019982597,
+      "grad_norm": 2411.3087500087086,
+      "learning_rate": 1.6352111708358243e-05,
+      "loss": 343.7047,
+      "step": 14980
+    },
+    {
+      "epoch": 0.2882955654603065,
+      "grad_norm": 2028.8896675886378,
+      "learning_rate": 1.6347396833582224e-05,
+      "loss": 348.8022,
+      "step": 14990
+    },
+    {
+      "epoch": 0.288487890720787,
+      "grad_norm": 2053.5154042317085,
+      "learning_rate": 1.634267959450571e-05,
+      "loss": 339.8348,
+      "step": 15000
+    },
+    {
+      "epoch": 0.28868021598126753,
+      "grad_norm": 2449.9333994756007,
+      "learning_rate": 1.63379599928858e-05,
+      "loss": 354.8625,
+      "step": 15010
+    },
+    {
+      "epoch": 0.28887254124174805,
+      "grad_norm": 1991.0620960043839,
+      "learning_rate": 1.6333238030480473e-05,
+      "loss": 349.4698,
+      "step": 15020
+    },
+    {
+      "epoch": 0.28906486650222857,
+      "grad_norm": 2156.6215816389918,
+      "learning_rate": 1.6328513709048573e-05,
+      "loss": 336.3499,
+      "step": 15030
+    },
+    {
+      "epoch": 0.2892571917627091,
+      "grad_norm": 2159.9912011033466,
+      "learning_rate": 1.6323787030349833e-05,
+      "loss": 343.4849,
+      "step": 15040
+    },
+    {
+      "epoch": 0.2894495170231896,
+      "grad_norm": 2466.2459075722577,
+      "learning_rate": 1.6319057996144868e-05,
+      "loss": 348.7814,
+      "step": 15050
+    },
+    {
+      "epoch": 0.2896418422836701,
+      "grad_norm": 2103.788500815738,
+      "learning_rate": 1.631432660819516e-05,
+      "loss": 335.7118,
+      "step": 15060
+    },
+    {
+      "epoch": 0.28983416754415064,
+      "grad_norm": 2249.819119280314,
+      "learning_rate": 1.6309592868263075e-05,
+      "loss": 344.7883,
+      "step": 15070
+    },
+    {
+      "epoch": 0.2900264928046312,
+      "grad_norm": 2113.4292951546504,
+      "learning_rate": 1.630485677811185e-05,
+      "loss": 349.2526,
+      "step": 15080
+    },
+    {
+      "epoch": 0.29021881806511174,
+      "grad_norm": 2016.642538768806,
+      "learning_rate": 1.630011833950561e-05,
+      "loss": 343.9141,
+      "step": 15090
+    },
+    {
+      "epoch": 0.29041114332559226,
+      "grad_norm": 2323.251189811375,
+      "learning_rate": 1.6295377554209338e-05,
+      "loss": 357.7049,
+      "step": 15100
+    },
+    {
+      "epoch": 0.2906034685860728,
+      "grad_norm": 2313.839921820489,
+      "learning_rate": 1.6290634423988897e-05,
+      "loss": 347.8679,
+      "step": 15110
+    },
+    {
+      "epoch": 0.2907957938465533,
+      "grad_norm": 1922.0937291669225,
+      "learning_rate": 1.6285888950611023e-05,
+      "loss": 350.3313,
+      "step": 15120
+    },
+    {
+      "epoch": 0.2909881191070338,
+      "grad_norm": 2172.165611825818,
+      "learning_rate": 1.6281141135843334e-05,
+      "loss": 339.3288,
+      "step": 15130
+    },
+    {
+      "epoch": 0.29118044436751434,
+      "grad_norm": 1902.3087888659554,
+      "learning_rate": 1.6276390981454306e-05,
+      "loss": 345.3844,
+      "step": 15140
+    },
+    {
+      "epoch": 0.29137276962799485,
+      "grad_norm": 2484.71928775544,
+      "learning_rate": 1.6271638489213297e-05,
+      "loss": 354.9697,
+      "step": 15150
+    },
+    {
+      "epoch": 0.2915650948884754,
+      "grad_norm": 2314.1098835200887,
+      "learning_rate": 1.6266883660890527e-05,
+      "loss": 357.2189,
+      "step": 15160
+    },
+    {
+      "epoch": 0.2917574201489559,
+      "grad_norm": 2079.488309143478,
+      "learning_rate": 1.6262126498257098e-05,
+      "loss": 342.9789,
+      "step": 15170
+    },
+    {
+      "epoch": 0.2919497454094364,
+      "grad_norm": 2159.5246672841668,
+      "learning_rate": 1.625736700308497e-05,
+      "loss": 345.0607,
+      "step": 15180
+    },
+    {
+      "epoch": 0.292142070669917,
+      "grad_norm": 2039.1375020606483,
+      "learning_rate": 1.6252605177146978e-05,
+      "loss": 343.0139,
+      "step": 15190
+    },
+    {
+      "epoch": 0.2923343959303975,
+      "grad_norm": 2201.209101113219,
+      "learning_rate": 1.624784102221682e-05,
+      "loss": 369.8984,
+      "step": 15200
+    },
+    {
+      "epoch": 0.292526721190878,
+      "grad_norm": 2235.822789944362,
+      "learning_rate": 1.6243074540069067e-05,
+      "loss": 341.1845,
+      "step": 15210
+    },
+    {
+      "epoch": 0.29271904645135854,
+      "grad_norm": 2071.1335185407997,
+      "learning_rate": 1.6238305732479158e-05,
+      "loss": 349.0349,
+      "step": 15220
+    },
+    {
+      "epoch": 0.29291137171183906,
+      "grad_norm": 2392.5512339717075,
+      "learning_rate": 1.6233534601223396e-05,
+      "loss": 345.4016,
+      "step": 15230
+    },
+    {
+      "epoch": 0.2931036969723196,
+      "grad_norm": 2054.6089100408317,
+      "learning_rate": 1.6228761148078943e-05,
+      "loss": 346.1329,
+      "step": 15240
+    },
+    {
+      "epoch": 0.2932960222328001,
+      "grad_norm": 1999.651223568873,
+      "learning_rate": 1.622398537482383e-05,
+      "loss": 350.039,
+      "step": 15250
+    },
+    {
+      "epoch": 0.2934883474932806,
+      "grad_norm": 2452.492930003192,
+      "learning_rate": 1.621920728323696e-05,
+      "loss": 350.0799,
+      "step": 15260
+    },
+    {
+      "epoch": 0.29368067275376114,
+      "grad_norm": 2051.588893775578,
+      "learning_rate": 1.621442687509809e-05,
+      "loss": 341.6577,
+      "step": 15270
+    },
+    {
+      "epoch": 0.29387299801424166,
+      "grad_norm": 2056.5558340694497,
+      "learning_rate": 1.6209644152187848e-05,
+      "loss": 351.3471,
+      "step": 15280
+    },
+    {
+      "epoch": 0.29406532327472223,
+      "grad_norm": 2147.400731848501,
+      "learning_rate": 1.620485911628771e-05,
+      "loss": 350.3942,
+      "step": 15290
+    },
+    {
+      "epoch": 0.29425764853520275,
+      "grad_norm": 2146.700466555457,
+      "learning_rate": 1.6200071769180026e-05,
+      "loss": 328.6214,
+      "step": 15300
+    },
+    {
+      "epoch": 0.2944499737956833,
+      "grad_norm": 2055.8986477886388,
+      "learning_rate": 1.6195282112648007e-05,
+      "loss": 344.675,
+      "step": 15310
+    },
+    {
+      "epoch": 0.2946422990561638,
+      "grad_norm": 2392.27956202332,
+      "learning_rate": 1.6190490148475724e-05,
+      "loss": 337.4436,
+      "step": 15320
+    },
+    {
+      "epoch": 0.2948346243166443,
+      "grad_norm": 2050.8781501790213,
+      "learning_rate": 1.6185695878448094e-05,
+      "loss": 343.9708,
+      "step": 15330
+    },
+    {
+      "epoch": 0.29502694957712483,
+      "grad_norm": 2032.4167013905073,
+      "learning_rate": 1.6180899304350915e-05,
+      "loss": 329.2228,
+      "step": 15340
+    },
+    {
+      "epoch": 0.29521927483760535,
+      "grad_norm": 1959.2517802295768,
+      "learning_rate": 1.6176100427970826e-05,
+      "loss": 335.7417,
+      "step": 15350
+    },
+    {
+      "epoch": 0.29541160009808587,
+      "grad_norm": 2342.0114212060766,
+      "learning_rate": 1.6171299251095324e-05,
+      "loss": 341.7732,
+      "step": 15360
+    },
+    {
+      "epoch": 0.2956039253585664,
+      "grad_norm": 2192.902401093125,
+      "learning_rate": 1.6166495775512777e-05,
+      "loss": 351.5493,
+      "step": 15370
+    },
+    {
+      "epoch": 0.2957962506190469,
+      "grad_norm": 2263.4632326344313,
+      "learning_rate": 1.6161690003012392e-05,
+      "loss": 347.1629,
+      "step": 15380
+    },
+    {
+      "epoch": 0.2959885758795275,
+      "grad_norm": 2219.263283930718,
+      "learning_rate": 1.615688193538425e-05,
+      "loss": 334.7667,
+      "step": 15390
+    },
+    {
+      "epoch": 0.296180901140008,
+      "grad_norm": 2480.6173716009744,
+      "learning_rate": 1.615207157441927e-05,
+      "loss": 346.333,
+      "step": 15400
+    },
+    {
+      "epoch": 0.2963732264004885,
+      "grad_norm": 2028.1183731241908,
+      "learning_rate": 1.6147258921909236e-05,
+      "loss": 335.6766,
+      "step": 15410
+    },
+    {
+      "epoch": 0.29656555166096904,
+      "grad_norm": 2486.808921182866,
+      "learning_rate": 1.6142443979646774e-05,
+      "loss": 353.8469,
+      "step": 15420
+    },
+    {
+      "epoch": 0.29675787692144956,
+      "grad_norm": 2116.3810332652265,
+      "learning_rate": 1.6137626749425377e-05,
+      "loss": 335.7973,
+      "step": 15430
+    },
+    {
+      "epoch": 0.2969502021819301,
+      "grad_norm": 2025.8910088704392,
+      "learning_rate": 1.6132807233039382e-05,
+      "loss": 340.9344,
+      "step": 15440
+    },
+    {
+      "epoch": 0.2971425274424106,
+      "grad_norm": 2012.3189811695688,
+      "learning_rate": 1.612798543228398e-05,
+      "loss": 337.8212,
+      "step": 15450
+    },
+    {
+      "epoch": 0.2973348527028911,
+      "grad_norm": 2030.5678851637938,
+      "learning_rate": 1.612316134895521e-05,
+      "loss": 343.1717,
+      "step": 15460
+    },
+    {
+      "epoch": 0.29752717796337164,
+      "grad_norm": 2066.6076610157006,
+      "learning_rate": 1.611833498484997e-05,
+      "loss": 342.1694,
+      "step": 15470
+    },
+    {
+      "epoch": 0.29771950322385216,
+      "grad_norm": 2040.0285872280083,
+      "learning_rate": 1.611350634176599e-05,
+      "loss": 344.1515,
+      "step": 15480
+    },
+    {
+      "epoch": 0.29791182848433273,
+      "grad_norm": 1885.0376331820423,
+      "learning_rate": 1.6108675421501865e-05,
+      "loss": 346.8648,
+      "step": 15490
+    },
+    {
+      "epoch": 0.29810415374481325,
+      "grad_norm": 2378.5574491418756,
+      "learning_rate": 1.610384222585704e-05,
+      "loss": 348.066,
+      "step": 15500
+    },
+    {
+      "epoch": 0.29829647900529377,
+      "grad_norm": 2142.635413293242,
+      "learning_rate": 1.609900675663179e-05,
+      "loss": 335.4527,
+      "step": 15510
+    },
+    {
+      "epoch": 0.2984888042657743,
+      "grad_norm": 2174.693092624955,
+      "learning_rate": 1.609416901562725e-05,
+      "loss": 329.3521,
+      "step": 15520
+    },
+    {
+      "epoch": 0.2986811295262548,
+      "grad_norm": 2082.419491545453,
+      "learning_rate": 1.60893290046454e-05,
+      "loss": 338.2611,
+      "step": 15530
+    },
+    {
+      "epoch": 0.2988734547867353,
+      "grad_norm": 1990.4989218204114,
+      "learning_rate": 1.608448672548907e-05,
+      "loss": 343.3369,
+      "step": 15540
+    },
+    {
+      "epoch": 0.29906578004721585,
+      "grad_norm": 2058.5567291446737,
+      "learning_rate": 1.6079642179961917e-05,
+      "loss": 328.6461,
+      "step": 15550
+    },
+    {
+      "epoch": 0.29925810530769636,
+      "grad_norm": 2059.5328825023216,
+      "learning_rate": 1.6074795369868463e-05,
+      "loss": 344.672,
+      "step": 15560
+    },
+    {
+      "epoch": 0.2994504305681769,
+      "grad_norm": 2042.0033589316513,
+      "learning_rate": 1.6069946297014064e-05,
+      "loss": 337.5279,
+      "step": 15570
+    },
+    {
+      "epoch": 0.2996427558286574,
+      "grad_norm": 2113.1761985993344,
+      "learning_rate": 1.6065094963204915e-05,
+      "loss": 335.2345,
+      "step": 15580
+    },
+    {
+      "epoch": 0.2998350810891379,
+      "grad_norm": 2028.4594574437606,
+      "learning_rate": 1.6060241370248064e-05,
+      "loss": 340.8168,
+      "step": 15590
+    },
+    {
+      "epoch": 0.3000274063496185,
+      "grad_norm": 2071.6367684556894,
+      "learning_rate": 1.605538551995139e-05,
+      "loss": 347.7569,
+      "step": 15600
+    },
+    {
+      "epoch": 0.300219731610099,
+      "grad_norm": 2022.8654457668224,
+      "learning_rate": 1.6050527414123616e-05,
+      "loss": 337.1365,
+      "step": 15610
+    },
+    {
+      "epoch": 0.30041205687057954,
+      "grad_norm": 1984.6657992908138,
+      "learning_rate": 1.604566705457431e-05,
+      "loss": 326.2922,
+      "step": 15620
+    },
+    {
+      "epoch": 0.30060438213106006,
+      "grad_norm": 2108.2426764165125,
+      "learning_rate": 1.604080444311387e-05,
+      "loss": 332.3005,
+      "step": 15630
+    },
+    {
+      "epoch": 0.3007967073915406,
+      "grad_norm": 2417.668494699437,
+      "learning_rate": 1.6035939581553543e-05,
+      "loss": 343.7115,
+      "step": 15640
+    },
+    {
+      "epoch": 0.3009890326520211,
+      "grad_norm": 2098.9251251981427,
+      "learning_rate": 1.603107247170541e-05,
+      "loss": 350.848,
+      "step": 15650
+    },
+    {
+      "epoch": 0.3011813579125016,
+      "grad_norm": 2187.8632945932964,
+      "learning_rate": 1.6026203115382392e-05,
+      "loss": 350.0126,
+      "step": 15660
+    },
+    {
+      "epoch": 0.30137368317298213,
+      "grad_norm": 2081.12109471647,
+      "learning_rate": 1.6021331514398233e-05,
+      "loss": 331.2589,
+      "step": 15670
+    },
+    {
+      "epoch": 0.30156600843346265,
+      "grad_norm": 2026.1622593795485,
+      "learning_rate": 1.6016457670567535e-05,
+      "loss": 350.4996,
+      "step": 15680
+    },
+    {
+      "epoch": 0.30175833369394317,
+      "grad_norm": 1947.1323797000855,
+      "learning_rate": 1.6011581585705715e-05,
+      "loss": 338.5748,
+      "step": 15690
+    },
+    {
+      "epoch": 0.30195065895442375,
+      "grad_norm": 1978.9177872431721,
+      "learning_rate": 1.600670326162904e-05,
+      "loss": 345.5526,
+      "step": 15700
+    },
+    {
+      "epoch": 0.30214298421490426,
+      "grad_norm": 2065.192048660437,
+      "learning_rate": 1.600182270015461e-05,
+      "loss": 332.2462,
+      "step": 15710
+    },
+    {
+      "epoch": 0.3023353094753848,
+      "grad_norm": 2122.800141350639,
+      "learning_rate": 1.5996939903100338e-05,
+      "loss": 330.6463,
+      "step": 15720
+    },
+    {
+      "epoch": 0.3025276347358653,
+      "grad_norm": 2216.346427216386,
+      "learning_rate": 1.5992054872285005e-05,
+      "loss": 333.9136,
+      "step": 15730
+    },
+    {
+      "epoch": 0.3027199599963458,
+      "grad_norm": 1781.8287120360487,
+      "learning_rate": 1.5987167609528187e-05,
+      "loss": 341.6092,
+      "step": 15740
+    },
+    {
+      "epoch": 0.30291228525682634,
+      "grad_norm": 2240.8663635102666,
+      "learning_rate": 1.598227811665032e-05,
+      "loss": 350.1469,
+      "step": 15750
+    },
+    {
+      "epoch": 0.30310461051730686,
+      "grad_norm": 2311.8158310427875,
+      "learning_rate": 1.597738639547265e-05,
+      "loss": 329.6627,
+      "step": 15760
+    },
+    {
+      "epoch": 0.3032969357777874,
+      "grad_norm": 1823.629312620248,
+      "learning_rate": 1.597249244781727e-05,
+      "loss": 341.2837,
+      "step": 15770
+    },
+    {
+      "epoch": 0.3034892610382679,
+      "grad_norm": 2163.3408733522306,
+      "learning_rate": 1.5967596275507094e-05,
+      "loss": 344.6266,
+      "step": 15780
+    },
+    {
+      "epoch": 0.3036815862987484,
+      "grad_norm": 1933.7887376312376,
+      "learning_rate": 1.5962697880365863e-05,
+      "loss": 330.588,
+      "step": 15790
+    },
+    {
+      "epoch": 0.303873911559229,
+      "grad_norm": 1908.057850733314,
+      "learning_rate": 1.5957797264218145e-05,
+      "loss": 328.8406,
+      "step": 15800
+    },
+    {
+      "epoch": 0.3040662368197095,
+      "grad_norm": 2161.8909836142493,
+      "learning_rate": 1.5952894428889347e-05,
+      "loss": 354.0483,
+      "step": 15810
+    },
+    {
+      "epoch": 0.30425856208019003,
+      "grad_norm": 1892.9751899058763,
+      "learning_rate": 1.594798937620569e-05,
+      "loss": 329.8559,
+      "step": 15820
+    },
+    {
+      "epoch": 0.30445088734067055,
+      "grad_norm": 2106.3950913699387,
+      "learning_rate": 1.594308210799422e-05,
+      "loss": 336.1968,
+      "step": 15830
+    },
+    {
+      "epoch": 0.30464321260115107,
+      "grad_norm": 2247.289635998585,
+      "learning_rate": 1.5938172626082823e-05,
+      "loss": 342.8016,
+      "step": 15840
+    },
+    {
+      "epoch": 0.3048355378616316,
+      "grad_norm": 2085.1920447215507,
+      "learning_rate": 1.5933260932300192e-05,
+      "loss": 337.2578,
+      "step": 15850
+    },
+    {
+      "epoch": 0.3050278631221121,
+      "grad_norm": 2044.9909918871356,
+      "learning_rate": 1.5928347028475855e-05,
+      "loss": 338.7811,
+      "step": 15860
+    },
+    {
+      "epoch": 0.3052201883825926,
+      "grad_norm": 2056.4942800640515,
+      "learning_rate": 1.592343091644016e-05,
+      "loss": 334.6297,
+      "step": 15870
+    },
+    {
+      "epoch": 0.30541251364307315,
+      "grad_norm": 2279.2176455707217,
+      "learning_rate": 1.5918512598024275e-05,
+      "loss": 324.1793,
+      "step": 15880
+    },
+    {
+      "epoch": 0.30560483890355367,
+      "grad_norm": 2421.7687068141227,
+      "learning_rate": 1.59135920750602e-05,
+      "loss": 348.8277,
+      "step": 15890
+    },
+    {
+      "epoch": 0.30579716416403424,
+      "grad_norm": 2122.412821515118,
+      "learning_rate": 1.590866934938074e-05,
+      "loss": 340.6207,
+      "step": 15900
+    },
+    {
+      "epoch": 0.30598948942451476,
+      "grad_norm": 1895.899102182779,
+      "learning_rate": 1.590374442281953e-05,
+      "loss": 330.9981,
+      "step": 15910
+    },
+    {
+      "epoch": 0.3061818146849953,
+      "grad_norm": 2101.911142862094,
+      "learning_rate": 1.5898817297211028e-05,
+      "loss": 331.7635,
+      "step": 15920
+    },
+    {
+      "epoch": 0.3063741399454758,
+      "grad_norm": 2185.9010198046053,
+      "learning_rate": 1.589388797439051e-05,
+      "loss": 343.5025,
+      "step": 15930
+    },
+    {
+      "epoch": 0.3065664652059563,
+      "grad_norm": 2039.0075617015264,
+      "learning_rate": 1.5888956456194056e-05,
+      "loss": 336.8323,
+      "step": 15940
+    },
+    {
+      "epoch": 0.30675879046643684,
+      "grad_norm": 2004.7136092241692,
+      "learning_rate": 1.588402274445858e-05,
+      "loss": 351.7526,
+      "step": 15950
+    },
+    {
+      "epoch": 0.30695111572691736,
+      "grad_norm": 2021.5746304781997,
+      "learning_rate": 1.5879086841021815e-05,
+      "loss": 344.0881,
+      "step": 15960
+    },
+    {
+      "epoch": 0.3071434409873979,
+      "grad_norm": 2081.0173856942433,
+      "learning_rate": 1.5874148747722294e-05,
+      "loss": 339.0723,
+      "step": 15970
+    },
+    {
+      "epoch": 0.3073357662478784,
+      "grad_norm": 2130.4156752434787,
+      "learning_rate": 1.5869208466399382e-05,
+      "loss": 335.951,
+      "step": 15980
+    },
+    {
+      "epoch": 0.3075280915083589,
+      "grad_norm": 2292.130303399673,
+      "learning_rate": 1.586426599889325e-05,
+      "loss": 339.7939,
+      "step": 15990
+    },
+    {
+      "epoch": 0.3077204167688395,
+      "grad_norm": 2493.4941406807498,
+      "learning_rate": 1.5859321347044882e-05,
+      "loss": 347.21,
+      "step": 16000
+    },
+    {
+      "epoch": 0.30791274202932,
+      "grad_norm": 2053.740886710575,
+      "learning_rate": 1.5854374512696084e-05,
+      "loss": 342.1399,
+      "step": 16010
+    },
+    {
+      "epoch": 0.3081050672898005,
+      "grad_norm": 2447.624340403277,
+      "learning_rate": 1.584942549768947e-05,
+      "loss": 336.8734,
+      "step": 16020
+    },
+    {
+      "epoch": 0.30829739255028105,
+      "grad_norm": 2086.5396459126378,
+      "learning_rate": 1.584447430386846e-05,
+      "loss": 349.1976,
+      "step": 16030
+    },
+    {
+      "epoch": 0.30848971781076157,
+      "grad_norm": 2075.0906560218377,
+      "learning_rate": 1.58395209330773e-05,
+      "loss": 322.2554,
+      "step": 16040
+    },
+    {
+      "epoch": 0.3086820430712421,
+      "grad_norm": 2372.653710833521,
+      "learning_rate": 1.5834565387161034e-05,
+      "loss": 341.3229,
+      "step": 16050
+    },
+    {
+      "epoch": 0.3088743683317226,
+      "grad_norm": 2211.589477666567,
+      "learning_rate": 1.5829607667965524e-05,
+      "loss": 333.5114,
+      "step": 16060
+    },
+    {
+      "epoch": 0.3090666935922031,
+      "grad_norm": 2570.1561750505557,
+      "learning_rate": 1.5824647777337433e-05,
+      "loss": 358.9544,
+      "step": 16070
+    },
+    {
+      "epoch": 0.30925901885268364,
+      "grad_norm": 2180.9552364678066,
+      "learning_rate": 1.5819685717124245e-05,
+      "loss": 331.2237,
+      "step": 16080
+    },
+    {
+      "epoch": 0.30945134411316416,
+      "grad_norm": 2085.2442998980287,
+      "learning_rate": 1.5814721489174246e-05,
+      "loss": 333.0478,
+      "step": 16090
+    },
+    {
+      "epoch": 0.3096436693736447,
+      "grad_norm": 2065.1309089259485,
+      "learning_rate": 1.580975509533652e-05,
+      "loss": 339.2547,
+      "step": 16100
+    },
+    {
+      "epoch": 0.30983599463412526,
+      "grad_norm": 2234.1041916736517,
+      "learning_rate": 1.5804786537460972e-05,
+      "loss": 328.4009,
+      "step": 16110
+    },
+    {
+      "epoch": 0.3100283198946058,
+      "grad_norm": 2002.5156719897268,
+      "learning_rate": 1.5799815817398312e-05,
+      "loss": 334.0761,
+      "step": 16120
+    },
+    {
+      "epoch": 0.3102206451550863,
+      "grad_norm": 2989.9680736875594,
+      "learning_rate": 1.579484293700004e-05,
+      "loss": 339.9199,
+      "step": 16130
+    },
+    {
+      "epoch": 0.3104129704155668,
+      "grad_norm": 2140.2577985832277,
+      "learning_rate": 1.578986789811849e-05,
+      "loss": 348.1439,
+      "step": 16140
+    },
+    {
+      "epoch": 0.31060529567604733,
+      "grad_norm": 2037.7938563831885,
+      "learning_rate": 1.5784890702606763e-05,
+      "loss": 330.0218,
+      "step": 16150
+    },
+    {
+      "epoch": 0.31079762093652785,
+      "grad_norm": 2273.0544599516516,
+      "learning_rate": 1.5779911352318792e-05,
+      "loss": 335.1149,
+      "step": 16160
+    },
+    {
+      "epoch": 0.31098994619700837,
+      "grad_norm": 2158.9584996291364,
+      "learning_rate": 1.5774929849109303e-05,
+      "loss": 334.0113,
+      "step": 16170
+    },
+    {
+      "epoch": 0.3111822714574889,
+      "grad_norm": 1986.953982832818,
+      "learning_rate": 1.5769946194833816e-05,
+      "loss": 350.6712,
+      "step": 16180
+    },
+    {
+      "epoch": 0.3113745967179694,
+      "grad_norm": 2680.7694975063,
+      "learning_rate": 1.5764960391348666e-05,
+      "loss": 344.6785,
+      "step": 16190
+    },
+    {
+      "epoch": 0.31156692197844993,
+      "grad_norm": 1836.2954392767801,
+      "learning_rate": 1.5759972440510985e-05,
+      "loss": 323.0379,
+      "step": 16200
+    },
+    {
+      "epoch": 0.3117592472389305,
+      "grad_norm": 1891.9186317164206,
+      "learning_rate": 1.5754982344178697e-05,
+      "loss": 324.2129,
+      "step": 16210
+    },
+    {
+      "epoch": 0.311951572499411,
+      "grad_norm": 1858.3685193919584,
+      "learning_rate": 1.5749990104210534e-05,
+      "loss": 338.1616,
+      "step": 16220
+    },
+    {
+      "epoch": 0.31214389775989154,
+      "grad_norm": 1942.769362270575,
+      "learning_rate": 1.574499572246602e-05,
+      "loss": 331.0952,
+      "step": 16230
+    },
+    {
+      "epoch": 0.31233622302037206,
+      "grad_norm": 2391.8832959326223,
+      "learning_rate": 1.5739999200805483e-05,
+      "loss": 349.1809,
+      "step": 16240
+    },
+    {
+      "epoch": 0.3125285482808526,
+      "grad_norm": 1896.4588090836314,
+      "learning_rate": 1.573500054109004e-05,
+      "loss": 336.1771,
+      "step": 16250
+    },
+    {
+      "epoch": 0.3127208735413331,
+      "grad_norm": 1933.2231685213192,
+      "learning_rate": 1.5729999745181617e-05,
+      "loss": 333.6326,
+      "step": 16260
+    },
+    {
+      "epoch": 0.3129131988018136,
+      "grad_norm": 1957.8009081451296,
+      "learning_rate": 1.572499681494292e-05,
+      "loss": 333.6753,
+      "step": 16270
+    },
+    {
+      "epoch": 0.31310552406229414,
+      "grad_norm": 1930.2970190906028,
+      "learning_rate": 1.571999175223746e-05,
+      "loss": 329.0752,
+      "step": 16280
+    },
+    {
+      "epoch": 0.31329784932277466,
+      "grad_norm": 2199.0292486815474,
+      "learning_rate": 1.571498455892954e-05,
+      "loss": 341.7442,
+      "step": 16290
+    },
+    {
+      "epoch": 0.3134901745832552,
+      "grad_norm": 1986.7851098666854,
+      "learning_rate": 1.570997523688426e-05,
+      "loss": 340.3837,
+      "step": 16300
+    },
+    {
+      "epoch": 0.31368249984373575,
+      "grad_norm": 1964.379709960503,
+      "learning_rate": 1.570496378796751e-05,
+      "loss": 330.2182,
+      "step": 16310
+    },
+    {
+      "epoch": 0.31387482510421627,
+      "grad_norm": 2158.3643877281206,
+      "learning_rate": 1.5699950214045966e-05,
+      "loss": 342.3301,
+      "step": 16320
+    },
+    {
+      "epoch": 0.3140671503646968,
+      "grad_norm": 2159.7701323799,
+      "learning_rate": 1.5694934516987102e-05,
+      "loss": 340.2655,
+      "step": 16330
+    },
+    {
+      "epoch": 0.3142594756251773,
+      "grad_norm": 2126.4845282418337,
+      "learning_rate": 1.5689916698659193e-05,
+      "loss": 342.693,
+      "step": 16340
+    },
+    {
+      "epoch": 0.31445180088565783,
+      "grad_norm": 1911.873268297658,
+      "learning_rate": 1.568489676093128e-05,
+      "loss": 322.9534,
+      "step": 16350
+    },
+    {
+      "epoch": 0.31464412614613835,
+      "grad_norm": 1942.7507027055012,
+      "learning_rate": 1.5679874705673215e-05,
+      "loss": 336.9441,
+      "step": 16360
+    },
+    {
+      "epoch": 0.31483645140661887,
+      "grad_norm": 2114.4984494087657,
+      "learning_rate": 1.5674850534755628e-05,
+      "loss": 340.4918,
+      "step": 16370
+    },
+    {
+      "epoch": 0.3150287766670994,
+      "grad_norm": 2074.3587412918373,
+      "learning_rate": 1.566982425004994e-05,
+      "loss": 334.4541,
+      "step": 16380
+    },
+    {
+      "epoch": 0.3152211019275799,
+      "grad_norm": 2020.809113927804,
+      "learning_rate": 1.5664795853428357e-05,
+      "loss": 333.6528,
+      "step": 16390
+    },
+    {
+      "epoch": 0.3154134271880604,
+      "grad_norm": 1933.6723048394228,
+      "learning_rate": 1.565976534676388e-05,
+      "loss": 330.6611,
+      "step": 16400
+    },
+    {
+      "epoch": 0.315605752448541,
+      "grad_norm": 2222.27062645372,
+      "learning_rate": 1.5654732731930286e-05,
+      "loss": 334.5393,
+      "step": 16410
+    },
+    {
+      "epoch": 0.3157980777090215,
+      "grad_norm": 2212.9341483115013,
+      "learning_rate": 1.5649698010802138e-05,
+      "loss": 352.7397,
+      "step": 16420
+    },
+    {
+      "epoch": 0.31599040296950204,
+      "grad_norm": 1878.5701680525785,
+      "learning_rate": 1.564466118525479e-05,
+      "loss": 330.0373,
+      "step": 16430
+    },
+    {
+      "epoch": 0.31618272822998256,
+      "grad_norm": 2013.2482855645796,
+      "learning_rate": 1.5639622257164372e-05,
+      "loss": 334.0851,
+      "step": 16440
+    },
+    {
+      "epoch": 0.3163750534904631,
+      "grad_norm": 2027.427984102896,
+      "learning_rate": 1.5634581228407807e-05,
+      "loss": 321.8146,
+      "step": 16450
+    },
+    {
+      "epoch": 0.3165673787509436,
+      "grad_norm": 2079.9571648910664,
+      "learning_rate": 1.562953810086279e-05,
+      "loss": 335.0436,
+      "step": 16460
+    },
+    {
+      "epoch": 0.3167597040114241,
+      "grad_norm": 2090.528441509643,
+      "learning_rate": 1.562449287640781e-05,
+      "loss": 326.4796,
+      "step": 16470
+    },
+    {
+      "epoch": 0.31695202927190463,
+      "grad_norm": 1981.8539034822998,
+      "learning_rate": 1.5619445556922118e-05,
+      "loss": 335.8691,
+      "step": 16480
+    },
+    {
+      "epoch": 0.31714435453238515,
+      "grad_norm": 1966.4134138299942,
+      "learning_rate": 1.561439614428577e-05,
+      "loss": 339.6275,
+      "step": 16490
+    },
+    {
+      "epoch": 0.3173366797928657,
+      "grad_norm": 1907.2904507812912,
+      "learning_rate": 1.5609344640379585e-05,
+      "loss": 335.7618,
+      "step": 16500
+    },
+    {
+      "epoch": 0.3175290050533462,
+      "grad_norm": 2095.980158379129,
+      "learning_rate": 1.560429104708516e-05,
+      "loss": 326.9014,
+      "step": 16510
+    },
+    {
+      "epoch": 0.31772133031382677,
+      "grad_norm": 1960.8340623964773,
+      "learning_rate": 1.5599235366284874e-05,
+      "loss": 329.9941,
+      "step": 16520
+    },
+    {
+      "epoch": 0.3179136555743073,
+      "grad_norm": 2122.109266842947,
+      "learning_rate": 1.5594177599861894e-05,
+      "loss": 337.7616,
+      "step": 16530
+    },
+    {
+      "epoch": 0.3181059808347878,
+      "grad_norm": 2447.161113783872,
+      "learning_rate": 1.5589117749700147e-05,
+      "loss": 335.4703,
+      "step": 16540
+    },
+    {
+      "epoch": 0.3182983060952683,
+      "grad_norm": 1930.1834990275436,
+      "learning_rate": 1.5584055817684346e-05,
+      "loss": 325.2849,
+      "step": 16550
+    },
+    {
+      "epoch": 0.31849063135574884,
+      "grad_norm": 2203.4758860547086,
+      "learning_rate": 1.5578991805699975e-05,
+      "loss": 333.9773,
+      "step": 16560
+    },
+    {
+      "epoch": 0.31868295661622936,
+      "grad_norm": 2133.660235062438,
+      "learning_rate": 1.5573925715633297e-05,
+      "loss": 332.5323,
+      "step": 16570
+    },
+    {
+      "epoch": 0.3188752818767099,
+      "grad_norm": 1922.932702918146,
+      "learning_rate": 1.5568857549371348e-05,
+      "loss": 337.5257,
+      "step": 16580
+    },
+    {
+      "epoch": 0.3190676071371904,
+      "grad_norm": 2138.1516124156665,
+      "learning_rate": 1.5563787308801934e-05,
+      "loss": 335.7361,
+      "step": 16590
+    },
+    {
+      "epoch": 0.3192599323976709,
+      "grad_norm": 2088.3037741374383,
+      "learning_rate": 1.5558714995813636e-05,
+      "loss": 330.6313,
+      "step": 16600
+    },
+    {
+      "epoch": 0.31945225765815144,
+      "grad_norm": 1949.1565451246443,
+      "learning_rate": 1.5553640612295807e-05,
+      "loss": 330.558,
+      "step": 16610
+    },
+    {
+      "epoch": 0.319644582918632,
+      "grad_norm": 2117.9288843146896,
+      "learning_rate": 1.5548564160138572e-05,
+      "loss": 338.077,
+      "step": 16620
+    },
+    {
+      "epoch": 0.31983690817911253,
+      "grad_norm": 2041.70053444773,
+      "learning_rate": 1.5543485641232825e-05,
+      "loss": 325.9343,
+      "step": 16630
+    },
+    {
+      "epoch": 0.32002923343959305,
+      "grad_norm": 2192.4520615181723,
+      "learning_rate": 1.553840505747023e-05,
+      "loss": 322.8957,
+      "step": 16640
+    },
+    {
+      "epoch": 0.32022155870007357,
+      "grad_norm": 1936.7697945460645,
+      "learning_rate": 1.5533322410743223e-05,
+      "loss": 332.07,
+      "step": 16650
+    },
+    {
+      "epoch": 0.3204138839605541,
+      "grad_norm": 2078.863803346763,
+      "learning_rate": 1.5528237702945e-05,
+      "loss": 330.5714,
+      "step": 16660
+    },
+    {
+      "epoch": 0.3206062092210346,
+      "grad_norm": 23951.936369128667,
+      "learning_rate": 1.5523150935969534e-05,
+      "loss": 327.5922,
+      "step": 16670
+    },
+    {
+      "epoch": 0.32079853448151513,
+      "grad_norm": 2094.1039223778803,
+      "learning_rate": 1.5518062111711566e-05,
+      "loss": 328.2249,
+      "step": 16680
+    },
+    {
+      "epoch": 0.32099085974199565,
+      "grad_norm": 2203.447765173993,
+      "learning_rate": 1.5512971232066593e-05,
+      "loss": 326.7832,
+      "step": 16690
+    },
+    {
+      "epoch": 0.32118318500247617,
+      "grad_norm": 2267.712583511857,
+      "learning_rate": 1.5507878298930888e-05,
+      "loss": 329.2262,
+      "step": 16700
+    },
+    {
+      "epoch": 0.3213755102629567,
+      "grad_norm": 1825.498688551304,
+      "learning_rate": 1.5502783314201478e-05,
+      "loss": 333.0049,
+      "step": 16710
+    },
+    {
+      "epoch": 0.32156783552343726,
+      "grad_norm": 2131.370407651775,
+      "learning_rate": 1.549768627977617e-05,
+      "loss": 333.8423,
+      "step": 16720
+    },
+    {
+      "epoch": 0.3217601607839178,
+      "grad_norm": 2237.4984306172946,
+      "learning_rate": 1.5492587197553517e-05,
+      "loss": 339.3328,
+      "step": 16730
+    },
+    {
+      "epoch": 0.3219524860443983,
+      "grad_norm": 2015.8269770930942,
+      "learning_rate": 1.5487486069432848e-05,
+      "loss": 328.2787,
+      "step": 16740
+    },
+    {
+      "epoch": 0.3221448113048788,
+      "grad_norm": 1950.4103937897498,
+      "learning_rate": 1.5482382897314243e-05,
+      "loss": 327.7073,
+      "step": 16750
+    },
+    {
+      "epoch": 0.32233713656535934,
+      "grad_norm": 2134.0277555161397,
+      "learning_rate": 1.5477277683098555e-05,
+      "loss": 341.0341,
+      "step": 16760
+    },
+    {
+      "epoch": 0.32252946182583986,
+      "grad_norm": 2112.634876932766,
+      "learning_rate": 1.547217042868739e-05,
+      "loss": 323.5574,
+      "step": 16770
+    },
+    {
+      "epoch": 0.3227217870863204,
+      "grad_norm": 1942.7595504479032,
+      "learning_rate": 1.546706113598312e-05,
+      "loss": 322.569,
+      "step": 16780
+    },
+    {
+      "epoch": 0.3229141123468009,
+      "grad_norm": 2122.0691961567495,
+      "learning_rate": 1.5461949806888867e-05,
+      "loss": 323.3565,
+      "step": 16790
+    },
+    {
+      "epoch": 0.3231064376072814,
+      "grad_norm": 2072.3440021732094,
+      "learning_rate": 1.5456836443308512e-05,
+      "loss": 329.8078,
+      "step": 16800
+    },
+    {
+      "epoch": 0.32329876286776194,
+      "grad_norm": 2163.2838632362523,
+      "learning_rate": 1.545172104714671e-05,
+      "loss": 329.1739,
+      "step": 16810
+    },
+    {
+      "epoch": 0.3234910881282425,
+      "grad_norm": 1971.8664793007088,
+      "learning_rate": 1.544660362030886e-05,
+      "loss": 326.3791,
+      "step": 16820
+    },
+    {
+      "epoch": 0.32368341338872303,
+      "grad_norm": 2179.2026442975875,
+      "learning_rate": 1.544148416470111e-05,
+      "loss": 332.5055,
+      "step": 16830
+    },
+    {
+      "epoch": 0.32387573864920355,
+      "grad_norm": 2206.3075491996556,
+      "learning_rate": 1.5436362682230378e-05,
+      "loss": 336.3999,
+      "step": 16840
+    },
+    {
+      "epoch": 0.32406806390968407,
+      "grad_norm": 1935.5689654090077,
+      "learning_rate": 1.543123917480433e-05,
+      "loss": 332.8678,
+      "step": 16850
+    },
+    {
+      "epoch": 0.3242603891701646,
+      "grad_norm": 2012.1462183908066,
+      "learning_rate": 1.542611364433139e-05,
+      "loss": 335.3134,
+      "step": 16860
+    },
+    {
+      "epoch": 0.3244527144306451,
+      "grad_norm": 2007.0735892387465,
+      "learning_rate": 1.5420986092720735e-05,
+      "loss": 323.2146,
+      "step": 16870
+    },
+    {
+      "epoch": 0.3246450396911256,
+      "grad_norm": 1911.6239401173323,
+      "learning_rate": 1.541585652188229e-05,
+      "loss": 324.8375,
+      "step": 16880
+    },
+    {
+      "epoch": 0.32483736495160614,
+      "grad_norm": 2069.4784281287316,
+      "learning_rate": 1.5410724933726732e-05,
+      "loss": 336.6477,
+      "step": 16890
+    },
+    {
+      "epoch": 0.32502969021208666,
+      "grad_norm": 1926.7697556794687,
+      "learning_rate": 1.5405591330165503e-05,
+      "loss": 318.2225,
+      "step": 16900
+    },
+    {
+      "epoch": 0.3252220154725672,
+      "grad_norm": 2011.9796932076379,
+      "learning_rate": 1.5400455713110777e-05,
+      "loss": 332.4606,
+      "step": 16910
+    },
+    {
+      "epoch": 0.32541434073304776,
+      "grad_norm": 2013.159557686021,
+      "learning_rate": 1.539531808447549e-05,
+      "loss": 336.5361,
+      "step": 16920
+    },
+    {
+      "epoch": 0.3256066659935283,
+      "grad_norm": 2070.4194640024884,
+      "learning_rate": 1.5390178446173325e-05,
+      "loss": 333.021,
+      "step": 16930
+    },
+    {
+      "epoch": 0.3257989912540088,
+      "grad_norm": 2017.7817211832491,
+      "learning_rate": 1.538503680011871e-05,
+      "loss": 328.0961,
+      "step": 16940
+    },
+    {
+      "epoch": 0.3259913165144893,
+      "grad_norm": 1908.0387238264814,
+      "learning_rate": 1.537989314822682e-05,
+      "loss": 326.2898,
+      "step": 16950
+    },
+    {
+      "epoch": 0.32618364177496983,
+      "grad_norm": 2089.0363119673184,
+      "learning_rate": 1.5374747492413587e-05,
+      "loss": 330.2057,
+      "step": 16960
+    },
+    {
+      "epoch": 0.32637596703545035,
+      "grad_norm": 2111.0971296183175,
+      "learning_rate": 1.536959983459568e-05,
+      "loss": 335.5717,
+      "step": 16970
+    },
+    {
+      "epoch": 0.3265682922959309,
+      "grad_norm": 1928.90596373864,
+      "learning_rate": 1.536445017669052e-05,
+      "loss": 324.4423,
+      "step": 16980
+    },
+    {
+      "epoch": 0.3267606175564114,
+      "grad_norm": 1950.6647464743678,
+      "learning_rate": 1.535929852061626e-05,
+      "loss": 320.3136,
+      "step": 16990
+    },
+    {
+      "epoch": 0.3269529428168919,
+      "grad_norm": 2002.1189101190598,
+      "learning_rate": 1.5354144868291817e-05,
+      "loss": 337.2886,
+      "step": 17000
+    },
+    {
+      "epoch": 0.32714526807737243,
+      "grad_norm": 2063.863592766914,
+      "learning_rate": 1.5348989221636835e-05,
+      "loss": 325.4392,
+      "step": 17010
+    },
+    {
+      "epoch": 0.32733759333785295,
+      "grad_norm": 1989.6880282565426,
+      "learning_rate": 1.5343831582571706e-05,
+      "loss": 344.3409,
+      "step": 17020
+    },
+    {
+      "epoch": 0.3275299185983335,
+      "grad_norm": 2126.617739771563,
+      "learning_rate": 1.5338671953017576e-05,
+      "loss": 323.6003,
+      "step": 17030
+    },
+    {
+      "epoch": 0.32772224385881404,
+      "grad_norm": 1980.7580303789032,
+      "learning_rate": 1.5333510334896308e-05,
+      "loss": 323.6241,
+      "step": 17040
+    },
+    {
+      "epoch": 0.32791456911929456,
+      "grad_norm": 7989.5692703841,
+      "learning_rate": 1.532834673013053e-05,
+      "loss": 338.0232,
+      "step": 17050
+    },
+    {
+      "epoch": 0.3281068943797751,
+      "grad_norm": 2105.4461893366692,
+      "learning_rate": 1.5323181140643598e-05,
+      "loss": 324.2593,
+      "step": 17060
+    },
+    {
+      "epoch": 0.3282992196402556,
+      "grad_norm": 2233.1294595468275,
+      "learning_rate": 1.5318013568359603e-05,
+      "loss": 325.8482,
+      "step": 17070
+    },
+    {
+      "epoch": 0.3284915449007361,
+      "grad_norm": 2053.89517412598,
+      "learning_rate": 1.531284401520338e-05,
+      "loss": 320.8966,
+      "step": 17080
+    },
+    {
+      "epoch": 0.32868387016121664,
+      "grad_norm": 1804.1815399037755,
+      "learning_rate": 1.530767248310051e-05,
+      "loss": 324.8647,
+      "step": 17090
+    },
+    {
+      "epoch": 0.32887619542169716,
+      "grad_norm": 1984.4429863739215,
+      "learning_rate": 1.53024989739773e-05,
+      "loss": 326.1241,
+      "step": 17100
+    },
+    {
+      "epoch": 0.3290685206821777,
+      "grad_norm": 2119.3650834321793,
+      "learning_rate": 1.5297323489760792e-05,
+      "loss": 325.1836,
+      "step": 17110
+    },
+    {
+      "epoch": 0.3292608459426582,
+      "grad_norm": 2011.8162230472658,
+      "learning_rate": 1.5292146032378778e-05,
+      "loss": 331.957,
+      "step": 17120
+    },
+    {
+      "epoch": 0.3294531712031388,
+      "grad_norm": 2208.5290778087833,
+      "learning_rate": 1.5286966603759767e-05,
+      "loss": 329.0348,
+      "step": 17130
+    },
+    {
+      "epoch": 0.3296454964636193,
+      "grad_norm": 1983.1830730633092,
+      "learning_rate": 1.5281785205833013e-05,
+      "loss": 324.2075,
+      "step": 17140
+    },
+    {
+      "epoch": 0.3298378217240998,
+      "grad_norm": 2177.5842019402803,
+      "learning_rate": 1.52766018405285e-05,
+      "loss": 321.3308,
+      "step": 17150
+    },
+    {
+      "epoch": 0.33003014698458033,
+      "grad_norm": 2017.432680715218,
+      "learning_rate": 1.5271416509776948e-05,
+      "loss": 321.9001,
+      "step": 17160
+    },
+    {
+      "epoch": 0.33022247224506085,
+      "grad_norm": 1924.177275402273,
+      "learning_rate": 1.5266229215509806e-05,
+      "loss": 323.9805,
+      "step": 17170
+    },
+    {
+      "epoch": 0.33041479750554137,
+      "grad_norm": 2325.5354430363386,
+      "learning_rate": 1.5261039959659257e-05,
+      "loss": 330.6658,
+      "step": 17180
+    },
+    {
+      "epoch": 0.3306071227660219,
+      "grad_norm": 1991.2666160434483,
+      "learning_rate": 1.5255848744158214e-05,
+      "loss": 313.3101,
+      "step": 17190
+    },
+    {
+      "epoch": 0.3307994480265024,
+      "grad_norm": 2004.8450498962638,
+      "learning_rate": 1.5250655570940317e-05,
+      "loss": 339.1153,
+      "step": 17200
+    },
+    {
+      "epoch": 0.3309917732869829,
+      "grad_norm": 2047.0739413470224,
+      "learning_rate": 1.5245460441939934e-05,
+      "loss": 330.8931,
+      "step": 17210
+    },
+    {
+      "epoch": 0.33118409854746345,
+      "grad_norm": 2156.6426925055903,
+      "learning_rate": 1.5240263359092167e-05,
+      "loss": 328.6984,
+      "step": 17220
+    },
+    {
+      "epoch": 0.331376423807944,
+      "grad_norm": 2027.7322002174321,
+      "learning_rate": 1.5235064324332846e-05,
+      "loss": 335.8807,
+      "step": 17230
+    },
+    {
+      "epoch": 0.33156874906842454,
+      "grad_norm": 1998.791675253118,
+      "learning_rate": 1.5229863339598528e-05,
+      "loss": 321.5385,
+      "step": 17240
+    },
+    {
+      "epoch": 0.33176107432890506,
+      "grad_norm": 1957.281613309926,
+      "learning_rate": 1.5224660406826486e-05,
+      "loss": 318.0045,
+      "step": 17250
+    },
+    {
+      "epoch": 0.3319533995893856,
+      "grad_norm": 2137.653876681757,
+      "learning_rate": 1.5219455527954732e-05,
+      "loss": 336.5705,
+      "step": 17260
+    },
+    {
+      "epoch": 0.3321457248498661,
+      "grad_norm": 1958.2701604944473,
+      "learning_rate": 1.5214248704921995e-05,
+      "loss": 323.9953,
+      "step": 17270
+    },
+    {
+      "epoch": 0.3323380501103466,
+      "grad_norm": 2307.58974962407,
+      "learning_rate": 1.5209039939667731e-05,
+      "loss": 325.2405,
+      "step": 17280
+    },
+    {
+      "epoch": 0.33253037537082714,
+      "grad_norm": 2035.1142977745817,
+      "learning_rate": 1.5203829234132118e-05,
+      "loss": 325.5495,
+      "step": 17290
+    },
+    {
+      "epoch": 0.33272270063130766,
+      "grad_norm": 1946.6222467425596,
+      "learning_rate": 1.5198616590256064e-05,
+      "loss": 330.5086,
+      "step": 17300
+    },
+    {
+      "epoch": 0.3329150258917882,
+      "grad_norm": 2005.549179065666,
+      "learning_rate": 1.5193402009981187e-05,
+      "loss": 314.0701,
+      "step": 17310
+    },
+    {
+      "epoch": 0.3331073511522687,
+      "grad_norm": 2482.5043269857088,
+      "learning_rate": 1.5188185495249832e-05,
+      "loss": 330.9168,
+      "step": 17320
+    },
+    {
+      "epoch": 0.33329967641274927,
+      "grad_norm": 1789.6741231107126,
+      "learning_rate": 1.5182967048005069e-05,
+      "loss": 326.3284,
+      "step": 17330
+    },
+    {
+      "epoch": 0.3334920016732298,
+      "grad_norm": 1999.3730211320772,
+      "learning_rate": 1.5177746670190674e-05,
+      "loss": 320.489,
+      "step": 17340
+    },
+    {
+      "epoch": 0.3336843269337103,
+      "grad_norm": 1835.1780236735635,
+      "learning_rate": 1.5172524363751162e-05,
+      "loss": 318.593,
+      "step": 17350
+    },
+    {
+      "epoch": 0.3338766521941908,
+      "grad_norm": 2175.5909436717416,
+      "learning_rate": 1.5167300130631748e-05,
+      "loss": 324.3285,
+      "step": 17360
+    },
+    {
+      "epoch": 0.33406897745467135,
+      "grad_norm": 2004.7006577466032,
+      "learning_rate": 1.516207397277838e-05,
+      "loss": 322.8537,
+      "step": 17370
+    },
+    {
+      "epoch": 0.33426130271515186,
+      "grad_norm": 1998.9144930847553,
+      "learning_rate": 1.5156845892137711e-05,
+      "loss": 311.381,
+      "step": 17380
+    },
+    {
+      "epoch": 0.3344536279756324,
+      "grad_norm": 2198.4087850985256,
+      "learning_rate": 1.5151615890657113e-05,
+      "loss": 328.1361,
+      "step": 17390
+    },
+    {
+      "epoch": 0.3346459532361129,
+      "grad_norm": 2002.820027979471,
+      "learning_rate": 1.5146383970284679e-05,
+      "loss": 324.4853,
+      "step": 17400
+    },
+    {
+      "epoch": 0.3348382784965934,
+      "grad_norm": 1805.7839113516109,
+      "learning_rate": 1.514115013296921e-05,
+      "loss": 323.149,
+      "step": 17410
+    },
+    {
+      "epoch": 0.33503060375707394,
+      "grad_norm": 2183.193993180048,
+      "learning_rate": 1.513591438066023e-05,
+      "loss": 328.4162,
+      "step": 17420
+    },
+    {
+      "epoch": 0.33522292901755446,
+      "grad_norm": 2182.4459137606827,
+      "learning_rate": 1.5130676715307962e-05,
+      "loss": 337.7823,
+      "step": 17430
+    },
+    {
+      "epoch": 0.33541525427803504,
+      "grad_norm": 2017.3594994786833,
+      "learning_rate": 1.5125437138863353e-05,
+      "loss": 336.6125,
+      "step": 17440
+    },
+    {
+      "epoch": 0.33560757953851555,
+      "grad_norm": 1980.9282396432104,
+      "learning_rate": 1.512019565327806e-05,
+      "loss": 329.9132,
+      "step": 17450
+    },
+    {
+      "epoch": 0.3357999047989961,
+      "grad_norm": 2265.1804667663673,
+      "learning_rate": 1.5114952260504448e-05,
+      "loss": 336.1914,
+      "step": 17460
+    },
+    {
+      "epoch": 0.3359922300594766,
+      "grad_norm": 1906.9452157750625,
+      "learning_rate": 1.5109706962495596e-05,
+      "loss": 327.1075,
+      "step": 17470
+    },
+    {
+      "epoch": 0.3361845553199571,
+      "grad_norm": 2493.3960080989086,
+      "learning_rate": 1.510445976120529e-05,
+      "loss": 320.5026,
+      "step": 17480
+    },
+    {
+      "epoch": 0.33637688058043763,
+      "grad_norm": 2204.2267700891744,
+      "learning_rate": 1.5099210658588029e-05,
+      "loss": 326.1179,
+      "step": 17490
+    },
+    {
+      "epoch": 0.33656920584091815,
+      "grad_norm": 2051.901544526736,
+      "learning_rate": 1.5093959656599008e-05,
+      "loss": 327.8618,
+      "step": 17500
+    },
+    {
+      "epoch": 0.33676153110139867,
+      "grad_norm": 1917.0879883303546,
+      "learning_rate": 1.5088706757194147e-05,
+      "loss": 316.0583,
+      "step": 17510
+    },
+    {
+      "epoch": 0.3369538563618792,
+      "grad_norm": 1878.027348792607,
+      "learning_rate": 1.5083451962330062e-05,
+      "loss": 319.3782,
+      "step": 17520
+    },
+    {
+      "epoch": 0.3371461816223597,
+      "grad_norm": 2319.7363964115143,
+      "learning_rate": 1.5078195273964081e-05,
+      "loss": 320.0514,
+      "step": 17530
+    },
+    {
+      "epoch": 0.3373385068828403,
+      "grad_norm": 2373.517712199625,
+      "learning_rate": 1.5072936694054222e-05,
+      "loss": 319.1762,
+      "step": 17540
+    },
+    {
+      "epoch": 0.3375308321433208,
+      "grad_norm": 2096.8316519822724,
+      "learning_rate": 1.5067676224559231e-05,
+      "loss": 324.5782,
+      "step": 17550
+    },
+    {
+      "epoch": 0.3377231574038013,
+      "grad_norm": 1921.033773791416,
+      "learning_rate": 1.506241386743854e-05,
+      "loss": 325.5894,
+      "step": 17560
+    },
+    {
+      "epoch": 0.33791548266428184,
+      "grad_norm": 2054.43677617778,
+      "learning_rate": 1.5057149624652297e-05,
+      "loss": 313.1354,
+      "step": 17570
+    },
+    {
+      "epoch": 0.33810780792476236,
+      "grad_norm": 1862.8960805804957,
+      "learning_rate": 1.5051883498161334e-05,
+      "loss": 320.0346,
+      "step": 17580
+    },
+    {
+      "epoch": 0.3383001331852429,
+      "grad_norm": 1971.0422353913254,
+      "learning_rate": 1.5046615489927206e-05,
+      "loss": 319.7538,
+      "step": 17590
+    },
+    {
+      "epoch": 0.3384924584457234,
+      "grad_norm": 2030.5111586380638,
+      "learning_rate": 1.5041345601912154e-05,
+      "loss": 320.2064,
+      "step": 17600
+    },
+    {
+      "epoch": 0.3386847837062039,
+      "grad_norm": 1959.3174767968349,
+      "learning_rate": 1.5036073836079128e-05,
+      "loss": 328.8023,
+      "step": 17610
+    },
+    {
+      "epoch": 0.33887710896668444,
+      "grad_norm": 1934.1969978271761,
+      "learning_rate": 1.5030800194391773e-05,
+      "loss": 314.1831,
+      "step": 17620
+    },
+    {
+      "epoch": 0.33906943422716496,
+      "grad_norm": 2169.611251447064,
+      "learning_rate": 1.5025524678814428e-05,
+      "loss": 326.154,
+      "step": 17630
+    },
+    {
+      "epoch": 0.33926175948764553,
+      "grad_norm": 1922.8586206121417,
+      "learning_rate": 1.5020247291312138e-05,
+      "loss": 325.2667,
+      "step": 17640
+    },
+    {
+      "epoch": 0.33945408474812605,
+      "grad_norm": 2085.36074181669,
+      "learning_rate": 1.5014968033850647e-05,
+      "loss": 338.1495,
+      "step": 17650
+    },
+    {
+      "epoch": 0.33964641000860657,
+      "grad_norm": 2373.3831354809463,
+      "learning_rate": 1.5009686908396388e-05,
+      "loss": 325.9303,
+      "step": 17660
+    },
+    {
+      "epoch": 0.3398387352690871,
+      "grad_norm": 1884.2010035743306,
+      "learning_rate": 1.5004403916916494e-05,
+      "loss": 325.0139,
+      "step": 17670
+    },
+    {
+      "epoch": 0.3400310605295676,
+      "grad_norm": 2169.9003744987435,
+      "learning_rate": 1.4999119061378791e-05,
+      "loss": 346.9868,
+      "step": 17680
+    },
+    {
+      "epoch": 0.3402233857900481,
+      "grad_norm": 2292.426596440604,
+      "learning_rate": 1.4993832343751802e-05,
+      "loss": 338.4296,
+      "step": 17690
+    },
+    {
+      "epoch": 0.34041571105052865,
+      "grad_norm": 2114.140782796333,
+      "learning_rate": 1.4988543766004737e-05,
+      "loss": 316.1823,
+      "step": 17700
+    },
+    {
+      "epoch": 0.34060803631100917,
+      "grad_norm": 1888.9179425005343,
+      "learning_rate": 1.498325333010751e-05,
+      "loss": 309.8166,
+      "step": 17710
+    },
+    {
+      "epoch": 0.3408003615714897,
+      "grad_norm": 1951.0823971475256,
+      "learning_rate": 1.497796103803072e-05,
+      "loss": 325.782,
+      "step": 17720
+    },
+    {
+      "epoch": 0.3409926868319702,
+      "grad_norm": 2088.9968188153207,
+      "learning_rate": 1.4972666891745655e-05,
+      "loss": 323.7715,
+      "step": 17730
+    },
+    {
+      "epoch": 0.3411850120924508,
+      "grad_norm": 2036.82448764238,
+      "learning_rate": 1.49673708932243e-05,
+      "loss": 310.4458,
+      "step": 17740
+    },
+    {
+      "epoch": 0.3413773373529313,
+      "grad_norm": 1970.2248684614185,
+      "learning_rate": 1.4962073044439328e-05,
+      "loss": 326.2854,
+      "step": 17750
+    },
+    {
+      "epoch": 0.3415696626134118,
+      "grad_norm": 1924.3678663716555,
+      "learning_rate": 1.4956773347364095e-05,
+      "loss": 327.319,
+      "step": 17760
+    },
+    {
+      "epoch": 0.34176198787389234,
+      "grad_norm": 1907.118918461094,
+      "learning_rate": 1.4951471803972657e-05,
+      "loss": 323.0438,
+      "step": 17770
+    },
+    {
+      "epoch": 0.34195431313437286,
+      "grad_norm": 1893.8868023094,
+      "learning_rate": 1.4946168416239746e-05,
+      "loss": 320.827,
+      "step": 17780
+    },
+    {
+      "epoch": 0.3421466383948534,
+      "grad_norm": 2047.7456871641295,
+      "learning_rate": 1.4940863186140788e-05,
+      "loss": 334.5147,
+      "step": 17790
+    },
+    {
+      "epoch": 0.3423389636553339,
+      "grad_norm": 1935.984066118262,
+      "learning_rate": 1.4935556115651898e-05,
+      "loss": 311.2683,
+      "step": 17800
+    },
+    {
+      "epoch": 0.3425312889158144,
+      "grad_norm": 2157.4059313849093,
+      "learning_rate": 1.4930247206749863e-05,
+      "loss": 316.3749,
+      "step": 17810
+    },
+    {
+      "epoch": 0.34272361417629493,
+      "grad_norm": 2154.6794877591947,
+      "learning_rate": 1.4924936461412172e-05,
+      "loss": 327.4048,
+      "step": 17820
+    },
+    {
+      "epoch": 0.34291593943677545,
+      "grad_norm": 2129.092134332189,
+      "learning_rate": 1.4919623881616989e-05,
+      "loss": 315.438,
+      "step": 17830
+    },
+    {
+      "epoch": 0.343108264697256,
+      "grad_norm": 1769.4656200328968,
+      "learning_rate": 1.4914309469343158e-05,
+      "loss": 308.637,
+      "step": 17840
+    },
+    {
+      "epoch": 0.34330058995773655,
+      "grad_norm": 1921.0105116747557,
+      "learning_rate": 1.4908993226570214e-05,
+      "loss": 335.8036,
+      "step": 17850
+    },
+    {
+      "epoch": 0.34349291521821707,
+      "grad_norm": 2489.4731367913478,
+      "learning_rate": 1.4903675155278365e-05,
+      "loss": 316.3473,
+      "step": 17860
+    },
+    {
+      "epoch": 0.3436852404786976,
+      "grad_norm": 1754.5021339822188,
+      "learning_rate": 1.4898355257448508e-05,
+      "loss": 326.5426,
+      "step": 17870
+    },
+    {
+      "epoch": 0.3438775657391781,
+      "grad_norm": 1747.207452600057,
+      "learning_rate": 1.4893033535062219e-05,
+      "loss": 315.3213,
+      "step": 17880
+    },
+    {
+      "epoch": 0.3440698909996586,
+      "grad_norm": 1905.074965064672,
+      "learning_rate": 1.4887709990101748e-05,
+      "loss": 321.391,
+      "step": 17890
+    },
+    {
+      "epoch": 0.34426221626013914,
+      "grad_norm": 2023.926765494719,
+      "learning_rate": 1.4882384624550028e-05,
+      "loss": 326.0344,
+      "step": 17900
+    },
+    {
+      "epoch": 0.34445454152061966,
+      "grad_norm": 2048.243543418818,
+      "learning_rate": 1.4877057440390672e-05,
+      "loss": 312.2275,
+      "step": 17910
+    },
+    {
+      "epoch": 0.3446468667811002,
+      "grad_norm": 2440.3189371072713,
+      "learning_rate": 1.4871728439607967e-05,
+      "loss": 322.354,
+      "step": 17920
+    },
+    {
+      "epoch": 0.3448391920415807,
+      "grad_norm": 2049.14963222684,
+      "learning_rate": 1.4866397624186877e-05,
+      "loss": 319.812,
+      "step": 17930
+    },
+    {
+      "epoch": 0.3450315173020612,
+      "grad_norm": 1764.1111453288904,
+      "learning_rate": 1.4861064996113042e-05,
+      "loss": 314.0869,
+      "step": 17940
+    },
+    {
+      "epoch": 0.3452238425625418,
+      "grad_norm": 2046.198191920155,
+      "learning_rate": 1.485573055737278e-05,
+      "loss": 326.3094,
+      "step": 17950
+    },
+    {
+      "epoch": 0.3454161678230223,
+      "grad_norm": 2385.1610373708513,
+      "learning_rate": 1.485039430995308e-05,
+      "loss": 320.4093,
+      "step": 17960
+    },
+    {
+      "epoch": 0.34560849308350283,
+      "grad_norm": 2310.3930441294906,
+      "learning_rate": 1.4845056255841608e-05,
+      "loss": 324.5415,
+      "step": 17970
+    },
+    {
+      "epoch": 0.34580081834398335,
+      "grad_norm": 2326.4672852586045,
+      "learning_rate": 1.48397163970267e-05,
+      "loss": 332.3383,
+      "step": 17980
+    },
+    {
+      "epoch": 0.34599314360446387,
+      "grad_norm": 2074.804843858858,
+      "learning_rate": 1.4834374735497362e-05,
+      "loss": 319.3209,
+      "step": 17990
+    },
+    {
+      "epoch": 0.3461854688649444,
+      "grad_norm": 2094.009201572699,
+      "learning_rate": 1.4829031273243277e-05,
+      "loss": 327.2733,
+      "step": 18000
+    },
+    {
+      "epoch": 0.3463777941254249,
+      "grad_norm": 1859.8269819007644,
+      "learning_rate": 1.4823686012254798e-05,
+      "loss": 314.172,
+      "step": 18010
+    },
+    {
+      "epoch": 0.34657011938590543,
+      "grad_norm": 1765.6420807287334,
+      "learning_rate": 1.4818338954522943e-05,
+      "loss": 311.1994,
+      "step": 18020
+    },
+    {
+      "epoch": 0.34676244464638595,
+      "grad_norm": 2018.194964187766,
+      "learning_rate": 1.4812990102039411e-05,
+      "loss": 320.7778,
+      "step": 18030
+    },
+    {
+      "epoch": 0.34695476990686647,
+      "grad_norm": 1875.9220214495574,
+      "learning_rate": 1.480763945679655e-05,
+      "loss": 314.9387,
+      "step": 18040
+    },
+    {
+      "epoch": 0.34714709516734704,
+      "grad_norm": 2084.257906742831,
+      "learning_rate": 1.4802287020787396e-05,
+      "loss": 316.8192,
+      "step": 18050
+    },
+    {
+      "epoch": 0.34733942042782756,
+      "grad_norm": 2012.1965208451954,
+      "learning_rate": 1.4796932796005634e-05,
+      "loss": 322.3936,
+      "step": 18060
+    },
+    {
+      "epoch": 0.3475317456883081,
+      "grad_norm": 1865.3618847142252,
+      "learning_rate": 1.4791576784445632e-05,
+      "loss": 313.6889,
+      "step": 18070
+    },
+    {
+      "epoch": 0.3477240709487886,
+      "grad_norm": 1972.297950123333,
+      "learning_rate": 1.4786218988102414e-05,
+      "loss": 316.1889,
+      "step": 18080
+    },
+    {
+      "epoch": 0.3479163962092691,
+      "grad_norm": 1931.513425895944,
+      "learning_rate": 1.4780859408971668e-05,
+      "loss": 309.7523,
+      "step": 18090
+    },
+    {
+      "epoch": 0.34810872146974964,
+      "grad_norm": 2143.1011270207578,
+      "learning_rate": 1.4775498049049754e-05,
+      "loss": 321.8954,
+      "step": 18100
+    },
+    {
+      "epoch": 0.34830104673023016,
+      "grad_norm": 1931.306303255075,
+      "learning_rate": 1.4770134910333684e-05,
+      "loss": 325.3682,
+      "step": 18110
+    },
+    {
+      "epoch": 0.3484933719907107,
+      "grad_norm": 1838.3266126720682,
+      "learning_rate": 1.4764769994821145e-05,
+      "loss": 327.3196,
+      "step": 18120
+    },
+    {
+      "epoch": 0.3486856972511912,
+      "grad_norm": 2205.6414409446684,
+      "learning_rate": 1.4759403304510472e-05,
+      "loss": 325.9669,
+      "step": 18130
+    },
+    {
+      "epoch": 0.3488780225116717,
+      "grad_norm": 2158.1387264897116,
+      "learning_rate": 1.475403484140067e-05,
+      "loss": 328.6126,
+      "step": 18140
+    },
+    {
+      "epoch": 0.3490703477721523,
+      "grad_norm": 1933.6185484531904,
+      "learning_rate": 1.4748664607491408e-05,
+      "loss": 318.6722,
+      "step": 18150
+    },
+    {
+      "epoch": 0.3492626730326328,
+      "grad_norm": 1798.4376151061012,
+      "learning_rate": 1.4743292604783008e-05,
+      "loss": 322.1551,
+      "step": 18160
+    },
+    {
+      "epoch": 0.34945499829311333,
+      "grad_norm": 1880.9531443315814,
+      "learning_rate": 1.4737918835276451e-05,
+      "loss": 317.39,
+      "step": 18170
+    },
+    {
+      "epoch": 0.34964732355359385,
+      "grad_norm": 2041.1887987519235,
+      "learning_rate": 1.4732543300973374e-05,
+      "loss": 317.1812,
+      "step": 18180
+    },
+    {
+      "epoch": 0.34983964881407437,
+      "grad_norm": 1905.7255850185568,
+      "learning_rate": 1.472716600387608e-05,
+      "loss": 317.1819,
+      "step": 18190
+    },
+    {
+      "epoch": 0.3500319740745549,
+      "grad_norm": 1853.7110640473709,
+      "learning_rate": 1.4721786945987519e-05,
+      "loss": 324.7371,
+      "step": 18200
+    },
+    {
+      "epoch": 0.3502242993350354,
+      "grad_norm": 2109.624381698033,
+      "learning_rate": 1.4716406129311307e-05,
+      "loss": 322.3356,
+      "step": 18210
+    },
+    {
+      "epoch": 0.3504166245955159,
+      "grad_norm": 2084.676895145801,
+      "learning_rate": 1.4711023555851702e-05,
+      "loss": 324.9865,
+      "step": 18220
+    },
+    {
+      "epoch": 0.35060894985599644,
+      "grad_norm": 2066.4189840773197,
+      "learning_rate": 1.470563922761363e-05,
+      "loss": 307.3408,
+      "step": 18230
+    },
+    {
+      "epoch": 0.35080127511647696,
+      "grad_norm": 1869.6423144397857,
+      "learning_rate": 1.470025314660266e-05,
+      "loss": 319.2566,
+      "step": 18240
+    },
+    {
+      "epoch": 0.35099360037695754,
+      "grad_norm": 2095.261229177306,
+      "learning_rate": 1.4694865314825024e-05,
+      "loss": 322.8215,
+      "step": 18250
+    },
+    {
+      "epoch": 0.35118592563743806,
+      "grad_norm": 2143.808312555047,
+      "learning_rate": 1.4689475734287596e-05,
+      "loss": 321.3808,
+      "step": 18260
+    },
+    {
+      "epoch": 0.3513782508979186,
+      "grad_norm": 2027.4728901627552,
+      "learning_rate": 1.4684084406997903e-05,
+      "loss": 322.6761,
+      "step": 18270
+    },
+    {
+      "epoch": 0.3515705761583991,
+      "grad_norm": 2067.7195730213252,
+      "learning_rate": 1.467869133496413e-05,
+      "loss": 318.2469,
+      "step": 18280
+    },
+    {
+      "epoch": 0.3517629014188796,
+      "grad_norm": 2030.820218952797,
+      "learning_rate": 1.4673296520195105e-05,
+      "loss": 314.2379,
+      "step": 18290
+    },
+    {
+      "epoch": 0.35195522667936013,
+      "grad_norm": 1892.0949388087715,
+      "learning_rate": 1.4667899964700309e-05,
+      "loss": 316.7988,
+      "step": 18300
+    },
+    {
+      "epoch": 0.35214755193984065,
+      "grad_norm": 1967.7305646277794,
+      "learning_rate": 1.466250167048987e-05,
+      "loss": 317.0118,
+      "step": 18310
+    },
+    {
+      "epoch": 0.35233987720032117,
+      "grad_norm": 1784.1315707035938,
+      "learning_rate": 1.4657101639574563e-05,
+      "loss": 324.7445,
+      "step": 18320
+    },
+    {
+      "epoch": 0.3525322024608017,
+      "grad_norm": 1809.1691018959052,
+      "learning_rate": 1.4651699873965808e-05,
+      "loss": 319.2028,
+      "step": 18330
+    },
+    {
+      "epoch": 0.3527245277212822,
+      "grad_norm": 1938.2220324138657,
+      "learning_rate": 1.4646296375675676e-05,
+      "loss": 326.4758,
+      "step": 18340
+    },
+    {
+      "epoch": 0.35291685298176273,
+      "grad_norm": 1970.6963131640166,
+      "learning_rate": 1.464089114671688e-05,
+      "loss": 328.4086,
+      "step": 18350
+    },
+    {
+      "epoch": 0.3531091782422433,
+      "grad_norm": 1993.257745861049,
+      "learning_rate": 1.4635484189102776e-05,
+      "loss": 316.6254,
+      "step": 18360
+    },
+    {
+      "epoch": 0.3533015035027238,
+      "grad_norm": 2419.8754322232635,
+      "learning_rate": 1.4630075504847373e-05,
+      "loss": 318.0136,
+      "step": 18370
+    },
+    {
+      "epoch": 0.35349382876320434,
+      "grad_norm": 1734.813658359721,
+      "learning_rate": 1.4624665095965311e-05,
+      "loss": 307.9769,
+      "step": 18380
+    },
+    {
+      "epoch": 0.35368615402368486,
+      "grad_norm": 2101.34467235573,
+      "learning_rate": 1.4619252964471881e-05,
+      "loss": 313.932,
+      "step": 18390
+    },
+    {
+      "epoch": 0.3538784792841654,
+      "grad_norm": 2127.022887820255,
+      "learning_rate": 1.461383911238301e-05,
+      "loss": 329.91,
+      "step": 18400
+    },
+    {
+      "epoch": 0.3540708045446459,
+      "grad_norm": 2532.5380742352836,
+      "learning_rate": 1.4608423541715273e-05,
+      "loss": 317.393,
+      "step": 18410
+    },
+    {
+      "epoch": 0.3542631298051264,
+      "grad_norm": 2032.124816421271,
+      "learning_rate": 1.4603006254485874e-05,
+      "loss": 318.3746,
+      "step": 18420
+    },
+    {
+      "epoch": 0.35445545506560694,
+      "grad_norm": 1923.7326716813545,
+      "learning_rate": 1.4597587252712666e-05,
+      "loss": 318.1964,
+      "step": 18430
+    },
+    {
+      "epoch": 0.35464778032608746,
+      "grad_norm": 1959.1941719642653,
+      "learning_rate": 1.4592166538414136e-05,
+      "loss": 308.2598,
+      "step": 18440
+    },
+    {
+      "epoch": 0.354840105586568,
+      "grad_norm": 2058.6767455449412,
+      "learning_rate": 1.4586744113609416e-05,
+      "loss": 319.3535,
+      "step": 18450
+    },
+    {
+      "epoch": 0.35503243084704855,
+      "grad_norm": 1855.3898169160702,
+      "learning_rate": 1.4581319980318266e-05,
+      "loss": 323.4018,
+      "step": 18460
+    },
+    {
+      "epoch": 0.35522475610752907,
+      "grad_norm": 2078.0727114579167,
+      "learning_rate": 1.4575894140561086e-05,
+      "loss": 306.4881,
+      "step": 18470
+    },
+    {
+      "epoch": 0.3554170813680096,
+      "grad_norm": 2669.718554169241,
+      "learning_rate": 1.4570466596358914e-05,
+      "loss": 328.2359,
+      "step": 18480
+    },
+    {
+      "epoch": 0.3556094066284901,
+      "grad_norm": 1828.7174987731157,
+      "learning_rate": 1.4565037349733415e-05,
+      "loss": 313.8513,
+      "step": 18490
+    },
+    {
+      "epoch": 0.35580173188897063,
+      "grad_norm": 2146.5401200268298,
+      "learning_rate": 1.45596064027069e-05,
+      "loss": 316.1958,
+      "step": 18500
+    },
+    {
+      "epoch": 0.35599405714945115,
+      "grad_norm": 2058.8989907633168,
+      "learning_rate": 1.4554173757302303e-05,
+      "loss": 308.0151,
+      "step": 18510
+    },
+    {
+      "epoch": 0.35618638240993167,
+      "grad_norm": 2076.327720638844,
+      "learning_rate": 1.4548739415543197e-05,
+      "loss": 310.2676,
+      "step": 18520
+    },
+    {
+      "epoch": 0.3563787076704122,
+      "grad_norm": 1808.30379456728,
+      "learning_rate": 1.454330337945378e-05,
+      "loss": 311.1484,
+      "step": 18530
+    },
+    {
+      "epoch": 0.3565710329308927,
+      "grad_norm": 1867.0293368215587,
+      "learning_rate": 1.4537865651058893e-05,
+      "loss": 310.1113,
+      "step": 18540
+    },
+    {
+      "epoch": 0.3567633581913732,
+      "grad_norm": 1916.9749913680243,
+      "learning_rate": 1.4532426232383998e-05,
+      "loss": 315.0335,
+      "step": 18550
+    },
+    {
+      "epoch": 0.3569556834518538,
+      "grad_norm": 1888.0848784079424,
+      "learning_rate": 1.4526985125455184e-05,
+      "loss": 318.2653,
+      "step": 18560
+    },
+    {
+      "epoch": 0.3571480087123343,
+      "grad_norm": 2015.6795246459603,
+      "learning_rate": 1.4521542332299177e-05,
+      "loss": 320.7423,
+      "step": 18570
+    },
+    {
+      "epoch": 0.35734033397281484,
+      "grad_norm": 1837.266065593453,
+      "learning_rate": 1.4516097854943325e-05,
+      "loss": 311.4094,
+      "step": 18580
+    },
+    {
+      "epoch": 0.35753265923329536,
+      "grad_norm": 1981.0795432117868,
+      "learning_rate": 1.4510651695415612e-05,
+      "loss": 304.5394,
+      "step": 18590
+    },
+    {
+      "epoch": 0.3577249844937759,
+      "grad_norm": 2795.787894580473,
+      "learning_rate": 1.4505203855744637e-05,
+      "loss": 326.8149,
+      "step": 18600
+    },
+    {
+      "epoch": 0.3579173097542564,
+      "grad_norm": 2086.2310172510206,
+      "learning_rate": 1.4499754337959628e-05,
+      "loss": 317.7597,
+      "step": 18610
+    },
+    {
+      "epoch": 0.3581096350147369,
+      "grad_norm": 1921.11740535252,
+      "learning_rate": 1.449430314409045e-05,
+      "loss": 310.1388,
+      "step": 18620
+    },
+    {
+      "epoch": 0.35830196027521743,
+      "grad_norm": 2069.939321975512,
+      "learning_rate": 1.4488850276167572e-05,
+      "loss": 315.3359,
+      "step": 18630
+    },
+    {
+      "epoch": 0.35849428553569795,
+      "grad_norm": 1969.5201478033393,
+      "learning_rate": 1.44833957362221e-05,
+      "loss": 320.9641,
+      "step": 18640
+    },
+    {
+      "epoch": 0.3586866107961785,
+      "grad_norm": 2108.6232107583496,
+      "learning_rate": 1.4477939526285767e-05,
+      "loss": 322.8485,
+      "step": 18650
+    },
+    {
+      "epoch": 0.35887893605665905,
+      "grad_norm": 2163.322998180366,
+      "learning_rate": 1.4472481648390914e-05,
+      "loss": 329.6273,
+      "step": 18660
+    },
+    {
+      "epoch": 0.35907126131713957,
+      "grad_norm": 2242.9781783935637,
+      "learning_rate": 1.4467022104570514e-05,
+      "loss": 319.2365,
+      "step": 18670
+    },
+    {
+      "epoch": 0.3592635865776201,
+      "grad_norm": 2119.987093377786,
+      "learning_rate": 1.4461560896858156e-05,
+      "loss": 317.7456,
+      "step": 18680
+    },
+    {
+      "epoch": 0.3594559118381006,
+      "grad_norm": 1910.747115614292,
+      "learning_rate": 1.4456098027288046e-05,
+      "loss": 312.7495,
+      "step": 18690
+    },
+    {
+      "epoch": 0.3596482370985811,
+      "grad_norm": 1992.3958919806419,
+      "learning_rate": 1.4450633497895017e-05,
+      "loss": 312.3302,
+      "step": 18700
+    },
+    {
+      "epoch": 0.35984056235906164,
+      "grad_norm": 1867.6226183600268,
+      "learning_rate": 1.4445167310714514e-05,
+      "loss": 322.3002,
+      "step": 18710
+    },
+    {
+      "epoch": 0.36003288761954216,
+      "grad_norm": 2016.9168798952683,
+      "learning_rate": 1.4439699467782602e-05,
+      "loss": 311.2525,
+      "step": 18720
+    },
+    {
+      "epoch": 0.3602252128800227,
+      "grad_norm": 1879.312367571334,
+      "learning_rate": 1.4434229971135965e-05,
+      "loss": 318.1439,
+      "step": 18730
+    },
+    {
+      "epoch": 0.3604175381405032,
+      "grad_norm": 2128.9900564083687,
+      "learning_rate": 1.4428758822811894e-05,
+      "loss": 314.8298,
+      "step": 18740
+    },
+    {
+      "epoch": 0.3606098634009837,
+      "grad_norm": 2124.0174372540587,
+      "learning_rate": 1.442328602484831e-05,
+      "loss": 322.5594,
+      "step": 18750
+    },
+    {
+      "epoch": 0.3608021886614643,
+      "grad_norm": 1844.7012521631195,
+      "learning_rate": 1.441781157928373e-05,
+      "loss": 309.2597,
+      "step": 18760
+    },
+    {
+      "epoch": 0.3609945139219448,
+      "grad_norm": 1891.4001240238235,
+      "learning_rate": 1.44123354881573e-05,
+      "loss": 305.0057,
+      "step": 18770
+    },
+    {
+      "epoch": 0.36118683918242533,
+      "grad_norm": 1984.21550273332,
+      "learning_rate": 1.4406857753508772e-05,
+      "loss": 323.6916,
+      "step": 18780
+    },
+    {
+      "epoch": 0.36137916444290585,
+      "grad_norm": 2282.0088909131864,
+      "learning_rate": 1.4401378377378512e-05,
+      "loss": 316.7283,
+      "step": 18790
+    },
+    {
+      "epoch": 0.3615714897033864,
+      "grad_norm": 2158.795102463002,
+      "learning_rate": 1.43958973618075e-05,
+      "loss": 306.5132,
+      "step": 18800
+    },
+    {
+      "epoch": 0.3617638149638669,
+      "grad_norm": 1711.2221485116238,
+      "learning_rate": 1.4390414708837322e-05,
+      "loss": 308.3174,
+      "step": 18810
+    },
+    {
+      "epoch": 0.3619561402243474,
+      "grad_norm": 5177.026732398226,
+      "learning_rate": 1.4384930420510173e-05,
+      "loss": 309.7688,
+      "step": 18820
+    },
+    {
+      "epoch": 0.36214846548482793,
+      "grad_norm": 2371.9891617859353,
+      "learning_rate": 1.4379444498868864e-05,
+      "loss": 309.2038,
+      "step": 18830
+    },
+    {
+      "epoch": 0.36234079074530845,
+      "grad_norm": 1973.3877647265913,
+      "learning_rate": 1.4373956945956807e-05,
+      "loss": 328.1908,
+      "step": 18840
+    },
+    {
+      "epoch": 0.36253311600578897,
+      "grad_norm": 1985.5771462030311,
+      "learning_rate": 1.4368467763818026e-05,
+      "loss": 307.2351,
+      "step": 18850
+    },
+    {
+      "epoch": 0.3627254412662695,
+      "grad_norm": 1963.4280274222835,
+      "learning_rate": 1.436297695449715e-05,
+      "loss": 320.2213,
+      "step": 18860
+    },
+    {
+      "epoch": 0.36291776652675006,
+      "grad_norm": 1826.2839625130428,
+      "learning_rate": 1.4357484520039412e-05,
+      "loss": 310.4313,
+      "step": 18870
+    },
+    {
+      "epoch": 0.3631100917872306,
+      "grad_norm": 2162.835794138138,
+      "learning_rate": 1.4351990462490662e-05,
+      "loss": 312.1796,
+      "step": 18880
+    },
+    {
+      "epoch": 0.3633024170477111,
+      "grad_norm": 1685.3395633462474,
+      "learning_rate": 1.434649478389734e-05,
+      "loss": 306.479,
+      "step": 18890
+    },
+    {
+      "epoch": 0.3634947423081916,
+      "grad_norm": 1776.8404205917066,
+      "learning_rate": 1.4340997486306491e-05,
+      "loss": 325.0773,
+      "step": 18900
+    },
+    {
+      "epoch": 0.36368706756867214,
+      "grad_norm": 1697.9502374028443,
+      "learning_rate": 1.4335498571765777e-05,
+      "loss": 319.4545,
+      "step": 18910
+    },
+    {
+      "epoch": 0.36387939282915266,
+      "grad_norm": 2193.8686567811064,
+      "learning_rate": 1.4329998042323447e-05,
+      "loss": 303.4668,
+      "step": 18920
+    },
+    {
+      "epoch": 0.3640717180896332,
+      "grad_norm": 2084.9775769730727,
+      "learning_rate": 1.4324495900028358e-05,
+      "loss": 319.325,
+      "step": 18930
+    },
+    {
+      "epoch": 0.3642640433501137,
+      "grad_norm": 2012.3829496433511,
+      "learning_rate": 1.4318992146929967e-05,
+      "loss": 311.6177,
+      "step": 18940
+    },
+    {
+      "epoch": 0.3644563686105942,
+      "grad_norm": 1820.7741656741277,
+      "learning_rate": 1.4313486785078335e-05,
+      "loss": 310.1186,
+      "step": 18950
+    },
+    {
+      "epoch": 0.36464869387107474,
+      "grad_norm": 1762.4376145747967,
+      "learning_rate": 1.4307979816524111e-05,
+      "loss": 315.7654,
+      "step": 18960
+    },
+    {
+      "epoch": 0.3648410191315553,
+      "grad_norm": 2166.390370436035,
+      "learning_rate": 1.4302471243318554e-05,
+      "loss": 318.3813,
+      "step": 18970
+    },
+    {
+      "epoch": 0.36503334439203583,
+      "grad_norm": 1983.5687903411094,
+      "learning_rate": 1.4296961067513519e-05,
+      "loss": 312.6711,
+      "step": 18980
+    },
+    {
+      "epoch": 0.36522566965251635,
+      "grad_norm": 1926.7835146225914,
+      "learning_rate": 1.4291449291161452e-05,
+      "loss": 310.5246,
+      "step": 18990
+    },
+    {
+      "epoch": 0.36541799491299687,
+      "grad_norm": 1927.507952019359,
+      "learning_rate": 1.4285935916315401e-05,
+      "loss": 310.5273,
+      "step": 19000
+    },
+    {
+      "epoch": 0.3656103201734774,
+      "grad_norm": 2016.7856479857473,
+      "learning_rate": 1.4280420945029004e-05,
+      "loss": 321.7089,
+      "step": 19010
+    },
+    {
+      "epoch": 0.3658026454339579,
+      "grad_norm": 1786.1187781237063,
+      "learning_rate": 1.4274904379356498e-05,
+      "loss": 314.2794,
+      "step": 19020
+    },
+    {
+      "epoch": 0.3659949706944384,
+      "grad_norm": 2282.8631987246968,
+      "learning_rate": 1.4269386221352714e-05,
+      "loss": 314.0156,
+      "step": 19030
+    },
+    {
+      "epoch": 0.36618729595491895,
+      "grad_norm": 1928.6441685431853,
+      "learning_rate": 1.4263866473073076e-05,
+      "loss": 310.3671,
+      "step": 19040
+    },
+    {
+      "epoch": 0.36637962121539946,
+      "grad_norm": 1907.3715722787026,
+      "learning_rate": 1.4258345136573595e-05,
+      "loss": 303.5411,
+      "step": 19050
+    },
+    {
+      "epoch": 0.36657194647588,
+      "grad_norm": 2103.9555527095317,
+      "learning_rate": 1.4252822213910878e-05,
+      "loss": 313.5066,
+      "step": 19060
+    },
+    {
+      "epoch": 0.36676427173636056,
+      "grad_norm": 2010.3933330536374,
+      "learning_rate": 1.4247297707142126e-05,
+      "loss": 310.4386,
+      "step": 19070
+    },
+    {
+      "epoch": 0.3669565969968411,
+      "grad_norm": 2063.4211860022233,
+      "learning_rate": 1.4241771618325123e-05,
+      "loss": 309.1785,
+      "step": 19080
+    },
+    {
+      "epoch": 0.3671489222573216,
+      "grad_norm": 1956.2391677728288,
+      "learning_rate": 1.4236243949518249e-05,
+      "loss": 320.1582,
+      "step": 19090
+    },
+    {
+      "epoch": 0.3673412475178021,
+      "grad_norm": 1950.6993736481784,
+      "learning_rate": 1.4230714702780466e-05,
+      "loss": 310.1097,
+      "step": 19100
+    },
+    {
+      "epoch": 0.36753357277828264,
+      "grad_norm": 1904.7005026276433,
+      "learning_rate": 1.422518388017133e-05,
+      "loss": 317.9098,
+      "step": 19110
+    },
+    {
+      "epoch": 0.36772589803876315,
+      "grad_norm": 1915.5051774378703,
+      "learning_rate": 1.4219651483750978e-05,
+      "loss": 308.9524,
+      "step": 19120
+    },
+    {
+      "epoch": 0.3679182232992437,
+      "grad_norm": 1809.2140718054884,
+      "learning_rate": 1.4214117515580139e-05,
+      "loss": 313.6805,
+      "step": 19130
+    },
+    {
+      "epoch": 0.3681105485597242,
+      "grad_norm": 2092.510817558125,
+      "learning_rate": 1.4208581977720124e-05,
+      "loss": 317.2792,
+      "step": 19140
+    },
+    {
+      "epoch": 0.3683028738202047,
+      "grad_norm": 1933.8141245130055,
+      "learning_rate": 1.420304487223283e-05,
+      "loss": 305.4637,
+      "step": 19150
+    },
+    {
+      "epoch": 0.36849519908068523,
+      "grad_norm": 1952.8186864404734,
+      "learning_rate": 1.4197506201180737e-05,
+      "loss": 313.2031,
+      "step": 19160
+    },
+    {
+      "epoch": 0.3686875243411658,
+      "grad_norm": 2378.1138098497067,
+      "learning_rate": 1.4191965966626908e-05,
+      "loss": 301.6322,
+      "step": 19170
+    },
+    {
+      "epoch": 0.3688798496016463,
+      "grad_norm": 1934.6936088973725,
+      "learning_rate": 1.418642417063499e-05,
+      "loss": 303.935,
+      "step": 19180
+    },
+    {
+      "epoch": 0.36907217486212684,
+      "grad_norm": 1952.5898928401564,
+      "learning_rate": 1.4180880815269207e-05,
+      "loss": 301.3802,
+      "step": 19190
+    },
+    {
+      "epoch": 0.36926450012260736,
+      "grad_norm": 2078.106652840737,
+      "learning_rate": 1.4175335902594372e-05,
+      "loss": 307.9113,
+      "step": 19200
+    },
+    {
+      "epoch": 0.3694568253830879,
+      "grad_norm": 1852.4005499181155,
+      "learning_rate": 1.416978943467587e-05,
+      "loss": 314.8365,
+      "step": 19210
+    },
+    {
+      "epoch": 0.3696491506435684,
+      "grad_norm": 2110.3624721328297,
+      "learning_rate": 1.4164241413579669e-05,
+      "loss": 315.4488,
+      "step": 19220
+    },
+    {
+      "epoch": 0.3698414759040489,
+      "grad_norm": 2166.5131944429427,
+      "learning_rate": 1.4158691841372318e-05,
+      "loss": 316.4969,
+      "step": 19230
+    },
+    {
+      "epoch": 0.37003380116452944,
+      "grad_norm": 1998.1800692681916,
+      "learning_rate": 1.4153140720120936e-05,
+      "loss": 307.1049,
+      "step": 19240
+    },
+    {
+      "epoch": 0.37022612642500996,
+      "grad_norm": 2095.3634420311373,
+      "learning_rate": 1.4147588051893233e-05,
+      "loss": 307.5317,
+      "step": 19250
+    },
+    {
+      "epoch": 0.3704184516854905,
+      "grad_norm": 1965.7569767201355,
+      "learning_rate": 1.4142033838757476e-05,
+      "loss": 308.2359,
+      "step": 19260
+    },
+    {
+      "epoch": 0.370610776945971,
+      "grad_norm": 1809.7745368880326,
+      "learning_rate": 1.4136478082782525e-05,
+      "loss": 308.7504,
+      "step": 19270
+    },
+    {
+      "epoch": 0.3708031022064516,
+      "grad_norm": 1730.1898504855028,
+      "learning_rate": 1.4130920786037798e-05,
+      "loss": 309.5616,
+      "step": 19280
+    },
+    {
+      "epoch": 0.3709954274669321,
+      "grad_norm": 1896.750624329191,
+      "learning_rate": 1.412536195059331e-05,
+      "loss": 315.0295,
+      "step": 19290
+    },
+    {
+      "epoch": 0.3711877527274126,
+      "grad_norm": 1807.6157393935891,
+      "learning_rate": 1.4119801578519625e-05,
+      "loss": 316.3897,
+      "step": 19300
+    },
+    {
+      "epoch": 0.37138007798789313,
+      "grad_norm": 1857.1932913980472,
+      "learning_rate": 1.4114239671887892e-05,
+      "loss": 318.6112,
+      "step": 19310
+    },
+    {
+      "epoch": 0.37157240324837365,
+      "grad_norm": 1868.5201341392255,
+      "learning_rate": 1.4108676232769831e-05,
+      "loss": 311.7844,
+      "step": 19320
+    },
+    {
+      "epoch": 0.37176472850885417,
+      "grad_norm": 1856.0649615078016,
+      "learning_rate": 1.410311126323773e-05,
+      "loss": 309.1805,
+      "step": 19330
+    },
+    {
+      "epoch": 0.3719570537693347,
+      "grad_norm": 1996.4311550702541,
+      "learning_rate": 1.409754476536445e-05,
+      "loss": 303.6167,
+      "step": 19340
+    },
+    {
+      "epoch": 0.3721493790298152,
+      "grad_norm": 2106.725941757717,
+      "learning_rate": 1.4091976741223414e-05,
+      "loss": 304.3662,
+      "step": 19350
+    },
+    {
+      "epoch": 0.3723417042902957,
+      "grad_norm": 3005.4569479533657,
+      "learning_rate": 1.408640719288863e-05,
+      "loss": 325.7058,
+      "step": 19360
+    },
+    {
+      "epoch": 0.37253402955077625,
+      "grad_norm": 1718.3012210985255,
+      "learning_rate": 1.408083612243465e-05,
+      "loss": 316.0893,
+      "step": 19370
+    },
+    {
+      "epoch": 0.3727263548112568,
+      "grad_norm": 1824.6290617258912,
+      "learning_rate": 1.4075263531936614e-05,
+      "loss": 313.317,
+      "step": 19380
+    },
+    {
+      "epoch": 0.37291868007173734,
+      "grad_norm": 1905.9594764430149,
+      "learning_rate": 1.4069689423470219e-05,
+      "loss": 304.4165,
+      "step": 19390
+    },
+    {
+      "epoch": 0.37311100533221786,
+      "grad_norm": 1924.1121000284206,
+      "learning_rate": 1.4064113799111725e-05,
+      "loss": 312.3041,
+      "step": 19400
+    },
+    {
+      "epoch": 0.3733033305926984,
+      "grad_norm": 1846.1835277819953,
+      "learning_rate": 1.405853666093796e-05,
+      "loss": 307.0441,
+      "step": 19410
+    },
+    {
+      "epoch": 0.3734956558531789,
+      "grad_norm": 1973.9086806717667,
+      "learning_rate": 1.405295801102632e-05,
+      "loss": 309.8415,
+      "step": 19420
+    },
+    {
+      "epoch": 0.3736879811136594,
+      "grad_norm": 1915.473961806632,
+      "learning_rate": 1.4047377851454758e-05,
+      "loss": 317.7333,
+      "step": 19430
+    },
+    {
+      "epoch": 0.37388030637413994,
+      "grad_norm": 3382.5123701927755,
+      "learning_rate": 1.4041796184301788e-05,
+      "loss": 303.0723,
+      "step": 19440
+    },
+    {
+      "epoch": 0.37407263163462046,
+      "grad_norm": 1971.8609093903099,
+      "learning_rate": 1.4036213011646496e-05,
+      "loss": 316.7415,
+      "step": 19450
+    },
+    {
+      "epoch": 0.374264956895101,
+      "grad_norm": 2101.1370131171916,
+      "learning_rate": 1.4030628335568515e-05,
+      "loss": 317.4839,
+      "step": 19460
+    },
+    {
+      "epoch": 0.3744572821555815,
+      "grad_norm": 1850.3315744382107,
+      "learning_rate": 1.4025042158148048e-05,
+      "loss": 307.7922,
+      "step": 19470
+    },
+    {
+      "epoch": 0.37464960741606207,
+      "grad_norm": 2139.3139482021966,
+      "learning_rate": 1.4019454481465853e-05,
+      "loss": 311.2271,
+      "step": 19480
+    },
+    {
+      "epoch": 0.3748419326765426,
+      "grad_norm": 2051.129950819537,
+      "learning_rate": 1.4013865307603248e-05,
+      "loss": 304.0343,
+      "step": 19490
+    },
+    {
+      "epoch": 0.3750342579370231,
+      "grad_norm": 1986.422851101012,
+      "learning_rate": 1.4008274638642103e-05,
+      "loss": 316.6552,
+      "step": 19500
+    },
+    {
+      "epoch": 0.3752265831975036,
+      "grad_norm": 2141.2965455420413,
+      "learning_rate": 1.4002682476664857e-05,
+      "loss": 307.9586,
+      "step": 19510
+    },
+    {
+      "epoch": 0.37541890845798415,
+      "grad_norm": 1783.2098075083138,
+      "learning_rate": 1.3997088823754494e-05,
+      "loss": 301.2444,
+      "step": 19520
+    },
+    {
+      "epoch": 0.37561123371846467,
+      "grad_norm": 2356.57486767964,
+      "learning_rate": 1.399149368199456e-05,
+      "loss": 314.9871,
+      "step": 19530
+    },
+    {
+      "epoch": 0.3758035589789452,
+      "grad_norm": 1921.3499142703806,
+      "learning_rate": 1.398589705346915e-05,
+      "loss": 306.1402,
+      "step": 19540
+    },
+    {
+      "epoch": 0.3759958842394257,
+      "grad_norm": 1994.8936359819968,
+      "learning_rate": 1.3980298940262918e-05,
+      "loss": 308.3559,
+      "step": 19550
+    },
+    {
+      "epoch": 0.3761882094999062,
+      "grad_norm": 1916.4051343548426,
+      "learning_rate": 1.3974699344461065e-05,
+      "loss": 306.8087,
+      "step": 19560
+    },
+    {
+      "epoch": 0.37638053476038674,
+      "grad_norm": 1862.185972088084,
+      "learning_rate": 1.396909826814935e-05,
+      "loss": 314.2664,
+      "step": 19570
+    },
+    {
+      "epoch": 0.3765728600208673,
+      "grad_norm": 2152.1525009731063,
+      "learning_rate": 1.3963495713414085e-05,
+      "loss": 308.783,
+      "step": 19580
+    },
+    {
+      "epoch": 0.37676518528134784,
+      "grad_norm": 1719.5065081571609,
+      "learning_rate": 1.3957891682342127e-05,
+      "loss": 310.7855,
+      "step": 19590
+    },
+    {
+      "epoch": 0.37695751054182836,
+      "grad_norm": 2097.0431765370918,
+      "learning_rate": 1.3952286177020879e-05,
+      "loss": 309.2558,
+      "step": 19600
+    },
+    {
+      "epoch": 0.3771498358023089,
+      "grad_norm": 1944.0466706544819,
+      "learning_rate": 1.3946679199538308e-05,
+      "loss": 302.2271,
+      "step": 19610
+    },
+    {
+      "epoch": 0.3773421610627894,
+      "grad_norm": 1918.8782355396409,
+      "learning_rate": 1.3941070751982917e-05,
+      "loss": 302.6986,
+      "step": 19620
+    },
+    {
+      "epoch": 0.3775344863232699,
+      "grad_norm": 1861.0998667746328,
+      "learning_rate": 1.3935460836443758e-05,
+      "loss": 308.8521,
+      "step": 19630
+    },
+    {
+      "epoch": 0.37772681158375043,
+      "grad_norm": 1865.3471595427097,
+      "learning_rate": 1.3929849455010433e-05,
+      "loss": 314.5706,
+      "step": 19640
+    },
+    {
+      "epoch": 0.37791913684423095,
+      "grad_norm": 2021.8855646959091,
+      "learning_rate": 1.3924236609773094e-05,
+      "loss": 309.5404,
+      "step": 19650
+    },
+    {
+      "epoch": 0.37811146210471147,
+      "grad_norm": 2017.1147507819485,
+      "learning_rate": 1.3918622302822425e-05,
+      "loss": 306.997,
+      "step": 19660
+    },
+    {
+      "epoch": 0.378303787365192,
+      "grad_norm": 2030.0589372727097,
+      "learning_rate": 1.391300653624967e-05,
+      "loss": 316.3002,
+      "step": 19670
+    },
+    {
+      "epoch": 0.37849611262567256,
+      "grad_norm": 1872.5902010347115,
+      "learning_rate": 1.39073893121466e-05,
+      "loss": 308.783,
+      "step": 19680
+    },
+    {
+      "epoch": 0.3786884378861531,
+      "grad_norm": 1937.8186631922715,
+      "learning_rate": 1.3901770632605546e-05,
+      "loss": 306.2152,
+      "step": 19690
+    },
+    {
+      "epoch": 0.3788807631466336,
+      "grad_norm": 1725.4803130162502,
+      "learning_rate": 1.3896150499719372e-05,
+      "loss": 301.3231,
+      "step": 19700
+    },
+    {
+      "epoch": 0.3790730884071141,
+      "grad_norm": 1912.7927087868466,
+      "learning_rate": 1.3890528915581482e-05,
+      "loss": 308.968,
+      "step": 19710
+    },
+    {
+      "epoch": 0.37926541366759464,
+      "grad_norm": 1833.9923013133835,
+      "learning_rate": 1.3884905882285829e-05,
+      "loss": 308.7003,
+      "step": 19720
+    },
+    {
+      "epoch": 0.37945773892807516,
+      "grad_norm": 1628.4517013746747,
+      "learning_rate": 1.3879281401926894e-05,
+      "loss": 302.7566,
+      "step": 19730
+    },
+    {
+      "epoch": 0.3796500641885557,
+      "grad_norm": 1851.9579537275933,
+      "learning_rate": 1.3873655476599707e-05,
+      "loss": 302.5998,
+      "step": 19740
+    },
+    {
+      "epoch": 0.3798423894490362,
+      "grad_norm": 1871.898105679082,
+      "learning_rate": 1.3868028108399829e-05,
+      "loss": 304.5054,
+      "step": 19750
+    },
+    {
+      "epoch": 0.3800347147095167,
+      "grad_norm": 1777.4171810886094,
+      "learning_rate": 1.3862399299423364e-05,
+      "loss": 306.4808,
+      "step": 19760
+    },
+    {
+      "epoch": 0.38022703996999724,
+      "grad_norm": 1784.2512467844372,
+      "learning_rate": 1.3856769051766947e-05,
+      "loss": 307.1631,
+      "step": 19770
+    },
+    {
+      "epoch": 0.38041936523047776,
+      "grad_norm": 1792.7844449582904,
+      "learning_rate": 1.3851137367527757e-05,
+      "loss": 295.8958,
+      "step": 19780
+    },
+    {
+      "epoch": 0.38061169049095833,
+      "grad_norm": 1943.2985094579828,
+      "learning_rate": 1.3845504248803501e-05,
+      "loss": 309.6948,
+      "step": 19790
+    },
+    {
+      "epoch": 0.38080401575143885,
+      "grad_norm": 1795.7383849342336,
+      "learning_rate": 1.383986969769242e-05,
+      "loss": 318.3569,
+      "step": 19800
+    },
+    {
+      "epoch": 0.38099634101191937,
+      "grad_norm": 1790.1327686269296,
+      "learning_rate": 1.38342337162933e-05,
+      "loss": 311.1551,
+      "step": 19810
+    },
+    {
+      "epoch": 0.3811886662723999,
+      "grad_norm": 1814.746142774624,
+      "learning_rate": 1.3828596306705442e-05,
+      "loss": 303.1561,
+      "step": 19820
+    },
+    {
+      "epoch": 0.3813809915328804,
+      "grad_norm": 2045.3185718760863,
+      "learning_rate": 1.3822957471028693e-05,
+      "loss": 296.6726,
+      "step": 19830
+    },
+    {
+      "epoch": 0.38157331679336093,
+      "grad_norm": 1870.208715880873,
+      "learning_rate": 1.3817317211363422e-05,
+      "loss": 297.9434,
+      "step": 19840
+    },
+    {
+      "epoch": 0.38176564205384145,
+      "grad_norm": 2449.6777574980124,
+      "learning_rate": 1.3811675529810535e-05,
+      "loss": 303.1532,
+      "step": 19850
+    },
+    {
+      "epoch": 0.38195796731432197,
+      "grad_norm": 2092.790781881497,
+      "learning_rate": 1.3806032428471463e-05,
+      "loss": 306.2771,
+      "step": 19860
+    },
+    {
+      "epoch": 0.3821502925748025,
+      "grad_norm": 2165.8330746346837,
+      "learning_rate": 1.3800387909448171e-05,
+      "loss": 308.9358,
+      "step": 19870
+    },
+    {
+      "epoch": 0.382342617835283,
+      "grad_norm": 2036.3089414824317,
+      "learning_rate": 1.3794741974843154e-05,
+      "loss": 312.2573,
+      "step": 19880
+    },
+    {
+      "epoch": 0.3825349430957636,
+      "grad_norm": 1758.9101366434552,
+      "learning_rate": 1.3789094626759419e-05,
+      "loss": 301.9613,
+      "step": 19890
+    },
+    {
+      "epoch": 0.3827272683562441,
+      "grad_norm": 1721.577430651585,
+      "learning_rate": 1.3783445867300515e-05,
+      "loss": 311.2537,
+      "step": 19900
+    },
+    {
+      "epoch": 0.3829195936167246,
+      "grad_norm": 2152.709150467604,
+      "learning_rate": 1.3777795698570511e-05,
+      "loss": 312.099,
+      "step": 19910
+    },
+    {
+      "epoch": 0.38311191887720514,
+      "grad_norm": 1718.4757220843696,
+      "learning_rate": 1.3772144122674e-05,
+      "loss": 301.4711,
+      "step": 19920
+    },
+    {
+      "epoch": 0.38330424413768566,
+      "grad_norm": 1753.7455137870545,
+      "learning_rate": 1.3766491141716103e-05,
+      "loss": 303.6883,
+      "step": 19930
+    },
+    {
+      "epoch": 0.3834965693981662,
+      "grad_norm": 1921.620070446975,
+      "learning_rate": 1.3760836757802462e-05,
+      "loss": 308.2743,
+      "step": 19940
+    },
+    {
+      "epoch": 0.3836888946586467,
+      "grad_norm": 1820.7438708008497,
+      "learning_rate": 1.3755180973039241e-05,
+      "loss": 308.7924,
+      "step": 19950
+    },
+    {
+      "epoch": 0.3838812199191272,
+      "grad_norm": 2025.6634304031434,
+      "learning_rate": 1.3749523789533128e-05,
+      "loss": 304.7339,
+      "step": 19960
+    },
+    {
+      "epoch": 0.38407354517960773,
+      "grad_norm": 2104.3083708910913,
+      "learning_rate": 1.374386520939133e-05,
+      "loss": 299.9552,
+      "step": 19970
+    },
+    {
+      "epoch": 0.38426587044008825,
+      "grad_norm": 2070.0041175750175,
+      "learning_rate": 1.3738205234721568e-05,
+      "loss": 319.6469,
+      "step": 19980
+    },
+    {
+      "epoch": 0.3844581957005688,
+      "grad_norm": 1869.4763599286875,
+      "learning_rate": 1.3732543867632098e-05,
+      "loss": 306.2007,
+      "step": 19990
+    },
+    {
+      "epoch": 0.38465052096104935,
+      "grad_norm": 1892.170127635212,
+      "learning_rate": 1.3726881110231682e-05,
+      "loss": 302.9237,
+      "step": 20000
+    },
+    {
+      "epoch": 0.38484284622152987,
+      "grad_norm": 1968.1574306751745,
+      "learning_rate": 1.3721216964629605e-05,
+      "loss": 308.6295,
+      "step": 20010
+    },
+    {
+      "epoch": 0.3850351714820104,
+      "grad_norm": 1816.0456579383317,
+      "learning_rate": 1.3715551432935664e-05,
+      "loss": 317.6802,
+      "step": 20020
+    },
+    {
+      "epoch": 0.3852274967424909,
+      "grad_norm": 1864.5578866243363,
+      "learning_rate": 1.3709884517260178e-05,
+      "loss": 318.0706,
+      "step": 20030
+    },
+    {
+      "epoch": 0.3854198220029714,
+      "grad_norm": 1839.697498226589,
+      "learning_rate": 1.3704216219713984e-05,
+      "loss": 300.5747,
+      "step": 20040
+    },
+    {
+      "epoch": 0.38561214726345194,
+      "grad_norm": 2010.0250361594676,
+      "learning_rate": 1.3698546542408424e-05,
+      "loss": 307.6041,
+      "step": 20050
+    },
+    {
+      "epoch": 0.38580447252393246,
+      "grad_norm": 1887.5334117631078,
+      "learning_rate": 1.369287548745536e-05,
+      "loss": 304.0994,
+      "step": 20060
+    },
+    {
+      "epoch": 0.385996797784413,
+      "grad_norm": 3058.863678189641,
+      "learning_rate": 1.3687203056967165e-05,
+      "loss": 309.2307,
+      "step": 20070
+    },
+    {
+      "epoch": 0.3861891230448935,
+      "grad_norm": 1924.364075007915,
+      "learning_rate": 1.368152925305673e-05,
+      "loss": 312.305,
+      "step": 20080
+    },
+    {
+      "epoch": 0.3863814483053741,
+      "grad_norm": 1694.1889732630975,
+      "learning_rate": 1.3675854077837449e-05,
+      "loss": 307.0336,
+      "step": 20090
+    },
+    {
+      "epoch": 0.3865737735658546,
+      "grad_norm": 1735.3038576083945,
+      "learning_rate": 1.3670177533423234e-05,
+      "loss": 296.0693,
+      "step": 20100
+    },
+    {
+      "epoch": 0.3867660988263351,
+      "grad_norm": 1800.3112961267825,
+      "learning_rate": 1.3664499621928502e-05,
+      "loss": 302.2557,
+      "step": 20110
+    },
+    {
+      "epoch": 0.38695842408681563,
+      "grad_norm": 1886.0789788519592,
+      "learning_rate": 1.3658820345468183e-05,
+      "loss": 309.6297,
+      "step": 20120
+    },
+    {
+      "epoch": 0.38715074934729615,
+      "grad_norm": 1860.5563989674151,
+      "learning_rate": 1.365313970615771e-05,
+      "loss": 299.3849,
+      "step": 20130
+    },
+    {
+      "epoch": 0.38734307460777667,
+      "grad_norm": 1933.7444355192472,
+      "learning_rate": 1.3647457706113031e-05,
+      "loss": 310.0622,
+      "step": 20140
+    },
+    {
+      "epoch": 0.3875353998682572,
+      "grad_norm": 1923.294080177032,
+      "learning_rate": 1.36417743474506e-05,
+      "loss": 311.0535,
+      "step": 20150
+    },
+    {
+      "epoch": 0.3877277251287377,
+      "grad_norm": 1742.9458139277679,
+      "learning_rate": 1.3636089632287369e-05,
+      "loss": 315.1836,
+      "step": 20160
+    },
+    {
+      "epoch": 0.38792005038921823,
+      "grad_norm": 1699.483128600211,
+      "learning_rate": 1.36304035627408e-05,
+      "loss": 310.0409,
+      "step": 20170
+    },
+    {
+      "epoch": 0.38811237564969875,
+      "grad_norm": 1838.9495473511888,
+      "learning_rate": 1.3624716140928861e-05,
+      "loss": 316.7457,
+      "step": 20180
+    },
+    {
+      "epoch": 0.38830470091017927,
+      "grad_norm": 1872.3533495535078,
+      "learning_rate": 1.3619027368970025e-05,
+      "loss": 309.5434,
+      "step": 20190
+    },
+    {
+      "epoch": 0.38849702617065984,
+      "grad_norm": 1910.4574769206988,
+      "learning_rate": 1.3613337248983265e-05,
+      "loss": 302.2189,
+      "step": 20200
+    },
+    {
+      "epoch": 0.38868935143114036,
+      "grad_norm": 1999.723835217515,
+      "learning_rate": 1.3607645783088055e-05,
+      "loss": 308.666,
+      "step": 20210
+    },
+    {
+      "epoch": 0.3888816766916209,
+      "grad_norm": 1820.57260452842,
+      "learning_rate": 1.3601952973404376e-05,
+      "loss": 314.4062,
+      "step": 20220
+    },
+    {
+      "epoch": 0.3890740019521014,
+      "grad_norm": 1830.3909934952903,
+      "learning_rate": 1.35962588220527e-05,
+      "loss": 306.443,
+      "step": 20230
+    },
+    {
+      "epoch": 0.3892663272125819,
+      "grad_norm": 2191.5605644904786,
+      "learning_rate": 1.3590563331154008e-05,
+      "loss": 316.5143,
+      "step": 20240
+    },
+    {
+      "epoch": 0.38945865247306244,
+      "grad_norm": 1963.1167099716085,
+      "learning_rate": 1.3584866502829774e-05,
+      "loss": 309.769,
+      "step": 20250
+    },
+    {
+      "epoch": 0.38965097773354296,
+      "grad_norm": 1885.9349248379274,
+      "learning_rate": 1.3579168339201975e-05,
+      "loss": 307.6866,
+      "step": 20260
+    },
+    {
+      "epoch": 0.3898433029940235,
+      "grad_norm": 1982.2356328691396,
+      "learning_rate": 1.3573468842393077e-05,
+      "loss": 302.0568,
+      "step": 20270
+    },
+    {
+      "epoch": 0.390035628254504,
+      "grad_norm": 1787.9098675560642,
+      "learning_rate": 1.356776801452606e-05,
+      "loss": 305.8773,
+      "step": 20280
+    },
+    {
+      "epoch": 0.3902279535149845,
+      "grad_norm": 1900.9447895241024,
+      "learning_rate": 1.3562065857724378e-05,
+      "loss": 306.7134,
+      "step": 20290
+    },
+    {
+      "epoch": 0.3904202787754651,
+      "grad_norm": 2310.223757347156,
+      "learning_rate": 1.3556362374111993e-05,
+      "loss": 304.0658,
+      "step": 20300
+    },
+    {
+      "epoch": 0.3906126040359456,
+      "grad_norm": 2283.91585825219,
+      "learning_rate": 1.3550657565813362e-05,
+      "loss": 311.3708,
+      "step": 20310
+    },
+    {
+      "epoch": 0.39080492929642613,
+      "grad_norm": 1843.021944476782,
+      "learning_rate": 1.3544951434953423e-05,
+      "loss": 299.4825,
+      "step": 20320
+    },
+    {
+      "epoch": 0.39099725455690665,
+      "grad_norm": 1968.392839143164,
+      "learning_rate": 1.3539243983657627e-05,
+      "loss": 304.3815,
+      "step": 20330
+    },
+    {
+      "epoch": 0.39118957981738717,
+      "grad_norm": 1935.425234679553,
+      "learning_rate": 1.3533535214051896e-05,
+      "loss": 304.0946,
+      "step": 20340
+    },
+    {
+      "epoch": 0.3913819050778677,
+      "grad_norm": 2016.4306690262665,
+      "learning_rate": 1.3527825128262656e-05,
+      "loss": 305.2499,
+      "step": 20350
+    },
+    {
+      "epoch": 0.3915742303383482,
+      "grad_norm": 1738.8058022016662,
+      "learning_rate": 1.3522113728416821e-05,
+      "loss": 311.6364,
+      "step": 20360
+    },
+    {
+      "epoch": 0.3917665555988287,
+      "grad_norm": 1792.621179012536,
+      "learning_rate": 1.3516401016641793e-05,
+      "loss": 299.9791,
+      "step": 20370
+    },
+    {
+      "epoch": 0.39195888085930924,
+      "grad_norm": 1919.9712146204274,
+      "learning_rate": 1.351068699506546e-05,
+      "loss": 305.6713,
+      "step": 20380
+    },
+    {
+      "epoch": 0.39215120611978976,
+      "grad_norm": 1753.0533645800076,
+      "learning_rate": 1.3504971665816202e-05,
+      "loss": 301.0382,
+      "step": 20390
+    },
+    {
+      "epoch": 0.39234353138027034,
+      "grad_norm": 1790.4699338073806,
+      "learning_rate": 1.3499255031022887e-05,
+      "loss": 294.2801,
+      "step": 20400
+    },
+    {
+      "epoch": 0.39253585664075086,
+      "grad_norm": 1716.6242004302876,
+      "learning_rate": 1.3493537092814863e-05,
+      "loss": 307.2205,
+      "step": 20410
+    },
+    {
+      "epoch": 0.3927281819012314,
+      "grad_norm": 2143.055767097122,
+      "learning_rate": 1.348781785332197e-05,
+      "loss": 303.4742,
+      "step": 20420
+    },
+    {
+      "epoch": 0.3929205071617119,
+      "grad_norm": 1811.5101931208928,
+      "learning_rate": 1.3482097314674526e-05,
+      "loss": 300.5256,
+      "step": 20430
+    },
+    {
+      "epoch": 0.3931128324221924,
+      "grad_norm": 1908.5382972739787,
+      "learning_rate": 1.3476375479003347e-05,
+      "loss": 300.7552,
+      "step": 20440
+    },
+    {
+      "epoch": 0.39330515768267293,
+      "grad_norm": 1934.5156499404786,
+      "learning_rate": 1.3470652348439715e-05,
+      "loss": 302.4374,
+      "step": 20450
+    },
+    {
+      "epoch": 0.39349748294315345,
+      "grad_norm": 1944.0607000815519,
+      "learning_rate": 1.3464927925115405e-05,
+      "loss": 318.1395,
+      "step": 20460
+    },
+    {
+      "epoch": 0.393689808203634,
+      "grad_norm": 1816.2609733133675,
+      "learning_rate": 1.3459202211162663e-05,
+      "loss": 300.7997,
+      "step": 20470
+    },
+    {
+      "epoch": 0.3938821334641145,
+      "grad_norm": 2159.638351922511,
+      "learning_rate": 1.345347520871423e-05,
+      "loss": 300.4586,
+      "step": 20480
+    },
+    {
+      "epoch": 0.394074458724595,
+      "grad_norm": 1877.6505123313825,
+      "learning_rate": 1.3447746919903318e-05,
+      "loss": 297.6736,
+      "step": 20490
+    },
+    {
+      "epoch": 0.3942667839850756,
+      "grad_norm": 1693.4089025440492,
+      "learning_rate": 1.3442017346863618e-05,
+      "loss": 306.1064,
+      "step": 20500
+    },
+    {
+      "epoch": 0.3944591092455561,
+      "grad_norm": 1928.2149955214938,
+      "learning_rate": 1.3436286491729306e-05,
+      "loss": 312.0971,
+      "step": 20510
+    },
+    {
+      "epoch": 0.3946514345060366,
+      "grad_norm": 1944.4569307096185,
+      "learning_rate": 1.3430554356635029e-05,
+      "loss": 307.6115,
+      "step": 20520
+    },
+    {
+      "epoch": 0.39484375976651714,
+      "grad_norm": 1820.6749193770745,
+      "learning_rate": 1.342482094371591e-05,
+      "loss": 300.1809,
+      "step": 20530
+    },
+    {
+      "epoch": 0.39503608502699766,
+      "grad_norm": 1897.183396809748,
+      "learning_rate": 1.341908625510755e-05,
+      "loss": 311.0384,
+      "step": 20540
+    },
+    {
+      "epoch": 0.3952284102874782,
+      "grad_norm": 1738.8602129288627,
+      "learning_rate": 1.341335029294603e-05,
+      "loss": 298.289,
+      "step": 20550
+    },
+    {
+      "epoch": 0.3954207355479587,
+      "grad_norm": 1850.6868360886565,
+      "learning_rate": 1.3407613059367898e-05,
+      "loss": 301.0768,
+      "step": 20560
+    },
+    {
+      "epoch": 0.3956130608084392,
+      "grad_norm": 1897.2260623714128,
+      "learning_rate": 1.3401874556510182e-05,
+      "loss": 301.7463,
+      "step": 20570
+    },
+    {
+      "epoch": 0.39580538606891974,
+      "grad_norm": 1802.2790799115212,
+      "learning_rate": 1.3396134786510375e-05,
+      "loss": 308.0902,
+      "step": 20580
+    },
+    {
+      "epoch": 0.39599771132940026,
+      "grad_norm": 2107.27764567609,
+      "learning_rate": 1.3390393751506452e-05,
+      "loss": 304.4175,
+      "step": 20590
+    },
+    {
+      "epoch": 0.39619003658988083,
+      "grad_norm": 1924.5927188210346,
+      "learning_rate": 1.338465145363685e-05,
+      "loss": 294.9072,
+      "step": 20600
+    },
+    {
+      "epoch": 0.39638236185036135,
+      "grad_norm": 2123.723157473039,
+      "learning_rate": 1.3378907895040478e-05,
+      "loss": 306.6204,
+      "step": 20610
+    },
+    {
+      "epoch": 0.3965746871108419,
+      "grad_norm": 1978.9194895967685,
+      "learning_rate": 1.3373163077856724e-05,
+      "loss": 301.3621,
+      "step": 20620
+    },
+    {
+      "epoch": 0.3967670123713224,
+      "grad_norm": 2008.498190228556,
+      "learning_rate": 1.3367417004225429e-05,
+      "loss": 307.2951,
+      "step": 20630
+    },
+    {
+      "epoch": 0.3969593376318029,
+      "grad_norm": 1983.9487977245653,
+      "learning_rate": 1.3361669676286919e-05,
+      "loss": 306.8815,
+      "step": 20640
+    },
+    {
+      "epoch": 0.39715166289228343,
+      "grad_norm": 1885.2817705791942,
+      "learning_rate": 1.335592109618197e-05,
+      "loss": 291.2721,
+      "step": 20650
+    },
+    {
+      "epoch": 0.39734398815276395,
+      "grad_norm": 1978.2404699626795,
+      "learning_rate": 1.335017126605184e-05,
+      "loss": 307.5826,
+      "step": 20660
+    },
+    {
+      "epoch": 0.39753631341324447,
+      "grad_norm": 1711.536637295286,
+      "learning_rate": 1.3344420188038243e-05,
+      "loss": 295.397,
+      "step": 20670
+    },
+    {
+      "epoch": 0.397728638673725,
+      "grad_norm": 2072.1110820578065,
+      "learning_rate": 1.333866786428336e-05,
+      "loss": 305.0892,
+      "step": 20680
+    },
+    {
+      "epoch": 0.3979209639342055,
+      "grad_norm": 1986.9482166882167,
+      "learning_rate": 1.3332914296929838e-05,
+      "loss": 308.2779,
+      "step": 20690
+    },
+    {
+      "epoch": 0.398113289194686,
+      "grad_norm": 1979.9944757036844,
+      "learning_rate": 1.3327159488120784e-05,
+      "loss": 299.6597,
+      "step": 20700
+    },
+    {
+      "epoch": 0.3983056144551666,
+      "grad_norm": 1970.288215913745,
+      "learning_rate": 1.3321403439999775e-05,
+      "loss": 302.8106,
+      "step": 20710
+    },
+    {
+      "epoch": 0.3984979397156471,
+      "grad_norm": 2021.9350583544017,
+      "learning_rate": 1.3315646154710835e-05,
+      "loss": 309.0129,
+      "step": 20720
+    },
+    {
+      "epoch": 0.39869026497612764,
+      "grad_norm": 1902.0656693466192,
+      "learning_rate": 1.3309887634398466e-05,
+      "loss": 310.4726,
+      "step": 20730
+    },
+    {
+      "epoch": 0.39888259023660816,
+      "grad_norm": 1849.287582506347,
+      "learning_rate": 1.3304127881207614e-05,
+      "loss": 297.1906,
+      "step": 20740
+    },
+    {
+      "epoch": 0.3990749154970887,
+      "grad_norm": 2095.0836566519965,
+      "learning_rate": 1.3298366897283697e-05,
+      "loss": 308.3663,
+      "step": 20750
+    },
+    {
+      "epoch": 0.3992672407575692,
+      "grad_norm": 1842.6313532937406,
+      "learning_rate": 1.3292604684772585e-05,
+      "loss": 301.0415,
+      "step": 20760
+    },
+    {
+      "epoch": 0.3994595660180497,
+      "grad_norm": 1924.561840897817,
+      "learning_rate": 1.3286841245820605e-05,
+      "loss": 299.9735,
+      "step": 20770
+    },
+    {
+      "epoch": 0.39965189127853024,
+      "grad_norm": 1953.7262204634574,
+      "learning_rate": 1.3281076582574548e-05,
+      "loss": 299.2476,
+      "step": 20780
+    },
+    {
+      "epoch": 0.39984421653901075,
+      "grad_norm": 1771.9099335540877,
+      "learning_rate": 1.3275310697181652e-05,
+      "loss": 297.1938,
+      "step": 20790
+    },
+    {
+      "epoch": 0.4000365417994913,
+      "grad_norm": 1885.1043841854043,
+      "learning_rate": 1.3269543591789616e-05,
+      "loss": 294.3169,
+      "step": 20800
+    },
+    {
+      "epoch": 0.40022886705997185,
+      "grad_norm": 1833.7788057586558,
+      "learning_rate": 1.3263775268546588e-05,
+      "loss": 304.3086,
+      "step": 20810
+    },
+    {
+      "epoch": 0.40042119232045237,
+      "grad_norm": 1917.3282746074783,
+      "learning_rate": 1.3258005729601178e-05,
+      "loss": 306.5842,
+      "step": 20820
+    },
+    {
+      "epoch": 0.4006135175809329,
+      "grad_norm": 2121.8848780096496,
+      "learning_rate": 1.325223497710244e-05,
+      "loss": 307.2001,
+      "step": 20830
+    },
+    {
+      "epoch": 0.4008058428414134,
+      "grad_norm": 1960.7797063608668,
+      "learning_rate": 1.3246463013199882e-05,
+      "loss": 308.4385,
+      "step": 20840
+    },
+    {
+      "epoch": 0.4009981681018939,
+      "grad_norm": 1834.4042510846705,
+      "learning_rate": 1.3240689840043475e-05,
+      "loss": 302.6972,
+      "step": 20850
+    },
+    {
+      "epoch": 0.40119049336237445,
+      "grad_norm": 2089.5596117458117,
+      "learning_rate": 1.323491545978362e-05,
+      "loss": 297.695,
+      "step": 20860
+    },
+    {
+      "epoch": 0.40138281862285496,
+      "grad_norm": 1804.7775436660631,
+      "learning_rate": 1.3229139874571186e-05,
+      "loss": 299.2692,
+      "step": 20870
+    },
+    {
+      "epoch": 0.4015751438833355,
+      "grad_norm": 2222.277733768969,
+      "learning_rate": 1.3223363086557477e-05,
+      "loss": 296.3286,
+      "step": 20880
+    },
+    {
+      "epoch": 0.401767469143816,
+      "grad_norm": 1854.8714144108574,
+      "learning_rate": 1.3217585097894255e-05,
+      "loss": 290.2188,
+      "step": 20890
+    },
+    {
+      "epoch": 0.4019597944042965,
+      "grad_norm": 1811.285512528795,
+      "learning_rate": 1.3211805910733724e-05,
+      "loss": 308.2051,
+      "step": 20900
+    },
+    {
+      "epoch": 0.4021521196647771,
+      "grad_norm": 1660.2408208230754,
+      "learning_rate": 1.3206025527228535e-05,
+      "loss": 298.5014,
+      "step": 20910
+    },
+    {
+      "epoch": 0.4023444449252576,
+      "grad_norm": 1792.3885659954376,
+      "learning_rate": 1.3200243949531788e-05,
+      "loss": 306.2331,
+      "step": 20920
+    },
+    {
+      "epoch": 0.40253677018573814,
+      "grad_norm": 1845.8499639942563,
+      "learning_rate": 1.3194461179797023e-05,
+      "loss": 297.2011,
+      "step": 20930
+    },
+    {
+      "epoch": 0.40272909544621865,
+      "grad_norm": 1914.73668626949,
+      "learning_rate": 1.3188677220178225e-05,
+      "loss": 287.7009,
+      "step": 20940
+    },
+    {
+      "epoch": 0.4029214207066992,
+      "grad_norm": 2056.1501768728363,
+      "learning_rate": 1.3182892072829828e-05,
+      "loss": 287.964,
+      "step": 20950
+    },
+    {
+      "epoch": 0.4031137459671797,
+      "grad_norm": 2399.747600124868,
+      "learning_rate": 1.3177105739906702e-05,
+      "loss": 290.4619,
+      "step": 20960
+    },
+    {
+      "epoch": 0.4033060712276602,
+      "grad_norm": 1773.4294751260477,
+      "learning_rate": 1.3171318223564156e-05,
+      "loss": 297.6337,
+      "step": 20970
+    },
+    {
+      "epoch": 0.40349839648814073,
+      "grad_norm": 1786.5106947788709,
+      "learning_rate": 1.3165529525957947e-05,
+      "loss": 302.6508,
+      "step": 20980
+    },
+    {
+      "epoch": 0.40369072174862125,
+      "grad_norm": 1928.64473627544,
+      "learning_rate": 1.3159739649244271e-05,
+      "loss": 300.5086,
+      "step": 20990
+    },
+    {
+      "epoch": 0.40388304700910177,
+      "grad_norm": 1869.8660979602523,
+      "learning_rate": 1.3153948595579764e-05,
+      "loss": 303.0783,
+      "step": 21000
+    },
+    {
+      "epoch": 0.40407537226958234,
+      "grad_norm": 1920.4635089054786,
+      "learning_rate": 1.3148156367121491e-05,
+      "loss": 309.3283,
+      "step": 21010
+    },
+    {
+      "epoch": 0.40426769753006286,
+      "grad_norm": 2272.931885824946,
+      "learning_rate": 1.3142362966026967e-05,
+      "loss": 300.1064,
+      "step": 21020
+    },
+    {
+      "epoch": 0.4044600227905434,
+      "grad_norm": 1858.2356739937206,
+      "learning_rate": 1.3136568394454136e-05,
+      "loss": 300.9337,
+      "step": 21030
+    },
+    {
+      "epoch": 0.4046523480510239,
+      "grad_norm": 1615.2452610517569,
+      "learning_rate": 1.313077265456138e-05,
+      "loss": 294.8649,
+      "step": 21040
+    },
+    {
+      "epoch": 0.4048446733115044,
+      "grad_norm": 1877.643361455623,
+      "learning_rate": 1.3124975748507514e-05,
+      "loss": 303.0341,
+      "step": 21050
+    },
+    {
+      "epoch": 0.40503699857198494,
+      "grad_norm": 1802.6308984113073,
+      "learning_rate": 1.3119177678451793e-05,
+      "loss": 308.1125,
+      "step": 21060
+    },
+    {
+      "epoch": 0.40522932383246546,
+      "grad_norm": 1901.1710674314595,
+      "learning_rate": 1.3113378446553903e-05,
+      "loss": 307.1291,
+      "step": 21070
+    },
+    {
+      "epoch": 0.405421649092946,
+      "grad_norm": 1838.5408841054675,
+      "learning_rate": 1.3107578054973962e-05,
+      "loss": 304.0225,
+      "step": 21080
+    },
+    {
+      "epoch": 0.4056139743534265,
+      "grad_norm": 1714.6882619276512,
+      "learning_rate": 1.3101776505872516e-05,
+      "loss": 301.9726,
+      "step": 21090
+    },
+    {
+      "epoch": 0.405806299613907,
+      "grad_norm": 2140.1223557205826,
+      "learning_rate": 1.309597380141055e-05,
+      "loss": 298.8791,
+      "step": 21100
+    },
+    {
+      "epoch": 0.40599862487438754,
+      "grad_norm": 1714.7941183356802,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 297.6602,
+      "step": 21110
+    },
+    {
+      "epoch": 0.4061909501348681,
+      "grad_norm": 1948.562309954345,
+      "learning_rate": 1.3084364935051132e-05,
+      "loss": 298.1119,
+      "step": 21120
+    },
+    {
+      "epoch": 0.40638327539534863,
+      "grad_norm": 1972.0588281432786,
+      "learning_rate": 1.307855877747779e-05,
+      "loss": 308.4491,
+      "step": 21130
+    },
+    {
+      "epoch": 0.40657560065582915,
+      "grad_norm": 1708.7027356324083,
+      "learning_rate": 1.307275147319215e-05,
+      "loss": 289.4312,
+      "step": 21140
+    },
+    {
+      "epoch": 0.40676792591630967,
+      "grad_norm": 1920.7208460277584,
+      "learning_rate": 1.3066943024357333e-05,
+      "loss": 290.8358,
+      "step": 21150
+    },
+    {
+      "epoch": 0.4069602511767902,
+      "grad_norm": 2134.1834226023575,
+      "learning_rate": 1.306113343313689e-05,
+      "loss": 301.6934,
+      "step": 21160
+    },
+    {
+      "epoch": 0.4071525764372707,
+      "grad_norm": 1747.1162765035936,
+      "learning_rate": 1.3055322701694801e-05,
+      "loss": 307.3142,
+      "step": 21170
+    },
+    {
+      "epoch": 0.4073449016977512,
+      "grad_norm": 2325.287656446818,
+      "learning_rate": 1.3049510832195466e-05,
+      "loss": 304.0766,
+      "step": 21180
+    },
+    {
+      "epoch": 0.40753722695823175,
+      "grad_norm": 1728.5381080702648,
+      "learning_rate": 1.3043697826803707e-05,
+      "loss": 294.9708,
+      "step": 21190
+    },
+    {
+      "epoch": 0.40772955221871227,
+      "grad_norm": 1811.3401316095064,
+      "learning_rate": 1.303788368768478e-05,
+      "loss": 287.6326,
+      "step": 21200
+    },
+    {
+      "epoch": 0.4079218774791928,
+      "grad_norm": 1945.005544787725,
+      "learning_rate": 1.3032068417004351e-05,
+      "loss": 302.9698,
+      "step": 21210
+    },
+    {
+      "epoch": 0.40811420273967336,
+      "grad_norm": 1974.8318234836372,
+      "learning_rate": 1.302625201692851e-05,
+      "loss": 299.1098,
+      "step": 21220
+    },
+    {
+      "epoch": 0.4083065280001539,
+      "grad_norm": 1721.9382760643002,
+      "learning_rate": 1.302043448962378e-05,
+      "loss": 292.9459,
+      "step": 21230
+    },
+    {
+      "epoch": 0.4084988532606344,
+      "grad_norm": 2057.284639654734,
+      "learning_rate": 1.301461583725708e-05,
+      "loss": 304.0902,
+      "step": 21240
+    },
+    {
+      "epoch": 0.4086911785211149,
+      "grad_norm": 2448.143948204674,
+      "learning_rate": 1.3008796061995772e-05,
+      "loss": 303.6405,
+      "step": 21250
+    },
+    {
+      "epoch": 0.40888350378159544,
+      "grad_norm": 2091.8278752606043,
+      "learning_rate": 1.3002975166007618e-05,
+      "loss": 295.297,
+      "step": 21260
+    },
+    {
+      "epoch": 0.40907582904207596,
+      "grad_norm": 1952.5980407949598,
+      "learning_rate": 1.2997153151460814e-05,
+      "loss": 299.297,
+      "step": 21270
+    },
+    {
+      "epoch": 0.4092681543025565,
+      "grad_norm": 1742.8795948181055,
+      "learning_rate": 1.299133002052396e-05,
+      "loss": 295.3625,
+      "step": 21280
+    },
+    {
+      "epoch": 0.409460479563037,
+      "grad_norm": 1778.9719901360481,
+      "learning_rate": 1.2985505775366079e-05,
+      "loss": 301.1708,
+      "step": 21290
+    },
+    {
+      "epoch": 0.4096528048235175,
+      "grad_norm": 2001.3645957659803,
+      "learning_rate": 1.2979680418156604e-05,
+      "loss": 307.0015,
+      "step": 21300
+    },
+    {
+      "epoch": 0.40984513008399803,
+      "grad_norm": 1931.998494168223,
+      "learning_rate": 1.297385395106538e-05,
+      "loss": 304.0401,
+      "step": 21310
+    },
+    {
+      "epoch": 0.4100374553444786,
+      "grad_norm": 1928.5697027821686,
+      "learning_rate": 1.2968026376262679e-05,
+      "loss": 286.0681,
+      "step": 21320
+    },
+    {
+      "epoch": 0.4102297806049591,
+      "grad_norm": 1879.6835077467238,
+      "learning_rate": 1.2962197695919167e-05,
+      "loss": 304.8239,
+      "step": 21330
+    },
+    {
+      "epoch": 0.41042210586543965,
+      "grad_norm": 1806.144973724193,
+      "learning_rate": 1.2956367912205939e-05,
+      "loss": 301.0433,
+      "step": 21340
+    },
+    {
+      "epoch": 0.41061443112592017,
+      "grad_norm": 1724.298103341707,
+      "learning_rate": 1.2950537027294487e-05,
+      "loss": 294.6038,
+      "step": 21350
+    },
+    {
+      "epoch": 0.4108067563864007,
+      "grad_norm": 1857.6353073166906,
+      "learning_rate": 1.2944705043356722e-05,
+      "loss": 297.4807,
+      "step": 21360
+    },
+    {
+      "epoch": 0.4109990816468812,
+      "grad_norm": 1900.9297964014493,
+      "learning_rate": 1.2938871962564965e-05,
+      "loss": 292.8086,
+      "step": 21370
+    },
+    {
+      "epoch": 0.4111914069073617,
+      "grad_norm": 1930.3994623822425,
+      "learning_rate": 1.2933037787091935e-05,
+      "loss": 305.0081,
+      "step": 21380
+    },
+    {
+      "epoch": 0.41138373216784224,
+      "grad_norm": 2107.1891030578204,
+      "learning_rate": 1.2927202519110775e-05,
+      "loss": 308.3201,
+      "step": 21390
+    },
+    {
+      "epoch": 0.41157605742832276,
+      "grad_norm": 1970.3479751807247,
+      "learning_rate": 1.2921366160795017e-05,
+      "loss": 296.5559,
+      "step": 21400
+    },
+    {
+      "epoch": 0.4117683826888033,
+      "grad_norm": 1719.1955993990437,
+      "learning_rate": 1.2915528714318612e-05,
+      "loss": 306.5933,
+      "step": 21410
+    },
+    {
+      "epoch": 0.41196070794928386,
+      "grad_norm": 1813.5370900230837,
+      "learning_rate": 1.2909690181855914e-05,
+      "loss": 295.4979,
+      "step": 21420
+    },
+    {
+      "epoch": 0.4121530332097644,
+      "grad_norm": 1926.807755797552,
+      "learning_rate": 1.290385056558168e-05,
+      "loss": 298.9159,
+      "step": 21430
+    },
+    {
+      "epoch": 0.4123453584702449,
+      "grad_norm": 1727.0178089454132,
+      "learning_rate": 1.2898009867671066e-05,
+      "loss": 295.8993,
+      "step": 21440
+    },
+    {
+      "epoch": 0.4125376837307254,
+      "grad_norm": 1960.7607981142648,
+      "learning_rate": 1.2892168090299639e-05,
+      "loss": 290.4833,
+      "step": 21450
+    },
+    {
+      "epoch": 0.41273000899120593,
+      "grad_norm": 2084.632148230582,
+      "learning_rate": 1.2886325235643367e-05,
+      "loss": 293.5019,
+      "step": 21460
+    },
+    {
+      "epoch": 0.41292233425168645,
+      "grad_norm": 1892.695566989941,
+      "learning_rate": 1.288048130587861e-05,
+      "loss": 291.6956,
+      "step": 21470
+    },
+    {
+      "epoch": 0.41311465951216697,
+      "grad_norm": 1956.193304032728,
+      "learning_rate": 1.287463630318214e-05,
+      "loss": 296.915,
+      "step": 21480
+    },
+    {
+      "epoch": 0.4133069847726475,
+      "grad_norm": 1957.0029367183324,
+      "learning_rate": 1.2868790229731123e-05,
+      "loss": 298.4166,
+      "step": 21490
+    },
+    {
+      "epoch": 0.413499310033128,
+      "grad_norm": 1936.4240602095022,
+      "learning_rate": 1.2862943087703127e-05,
+      "loss": 294.1665,
+      "step": 21500
+    },
+    {
+      "epoch": 0.41369163529360853,
+      "grad_norm": 1900.9234262394655,
+      "learning_rate": 1.2857094879276115e-05,
+      "loss": 301.6358,
+      "step": 21510
+    },
+    {
+      "epoch": 0.4138839605540891,
+      "grad_norm": 2008.8984612329446,
+      "learning_rate": 1.2851245606628447e-05,
+      "loss": 297.9974,
+      "step": 21520
+    },
+    {
+      "epoch": 0.4140762858145696,
+      "grad_norm": 2062.091397879479,
+      "learning_rate": 1.2845395271938876e-05,
+      "loss": 294.9293,
+      "step": 21530
+    },
+    {
+      "epoch": 0.41426861107505014,
+      "grad_norm": 1970.1115120327734,
+      "learning_rate": 1.2839543877386562e-05,
+      "loss": 290.6354,
+      "step": 21540
+    },
+    {
+      "epoch": 0.41446093633553066,
+      "grad_norm": 1866.7710249497457,
+      "learning_rate": 1.283369142515105e-05,
+      "loss": 306.8433,
+      "step": 21550
+    },
+    {
+      "epoch": 0.4146532615960112,
+      "grad_norm": 2121.5235414930767,
+      "learning_rate": 1.282783791741228e-05,
+      "loss": 308.4364,
+      "step": 21560
+    },
+    {
+      "epoch": 0.4148455868564917,
+      "grad_norm": 5781.957453285093,
+      "learning_rate": 1.2821983356350593e-05,
+      "loss": 316.6105,
+      "step": 21570
+    },
+    {
+      "epoch": 0.4150379121169722,
+      "grad_norm": 1955.5440952715064,
+      "learning_rate": 1.2816127744146711e-05,
+      "loss": 297.5697,
+      "step": 21580
+    },
+    {
+      "epoch": 0.41523023737745274,
+      "grad_norm": 1869.1392561426703,
+      "learning_rate": 1.2810271082981755e-05,
+      "loss": 297.6666,
+      "step": 21590
+    },
+    {
+      "epoch": 0.41542256263793326,
+      "grad_norm": 1755.9972261241396,
+      "learning_rate": 1.2804413375037232e-05,
+      "loss": 289.4357,
+      "step": 21600
+    },
+    {
+      "epoch": 0.4156148878984138,
+      "grad_norm": 1923.5048604579317,
+      "learning_rate": 1.2798554622495042e-05,
+      "loss": 294.3641,
+      "step": 21610
+    },
+    {
+      "epoch": 0.4158072131588943,
+      "grad_norm": 1709.1281205509256,
+      "learning_rate": 1.2792694827537477e-05,
+      "loss": 300.6993,
+      "step": 21620
+    },
+    {
+      "epoch": 0.41599953841937487,
+      "grad_norm": 2116.160721850635,
+      "learning_rate": 1.2786833992347212e-05,
+      "loss": 293.2287,
+      "step": 21630
+    },
+    {
+      "epoch": 0.4161918636798554,
+      "grad_norm": 1815.0267312970952,
+      "learning_rate": 1.2780972119107312e-05,
+      "loss": 299.084,
+      "step": 21640
+    },
+    {
+      "epoch": 0.4163841889403359,
+      "grad_norm": 1764.3040461231517,
+      "learning_rate": 1.2775109210001225e-05,
+      "loss": 288.4564,
+      "step": 21650
+    },
+    {
+      "epoch": 0.41657651420081643,
+      "grad_norm": 1834.684604356838,
+      "learning_rate": 1.2769245267212789e-05,
+      "loss": 296.033,
+      "step": 21660
+    },
+    {
+      "epoch": 0.41676883946129695,
+      "grad_norm": 1926.984096764738,
+      "learning_rate": 1.2763380292926227e-05,
+      "loss": 294.0977,
+      "step": 21670
+    },
+    {
+      "epoch": 0.41696116472177747,
+      "grad_norm": 1782.2764671826128,
+      "learning_rate": 1.2757514289326146e-05,
+      "loss": 293.2214,
+      "step": 21680
+    },
+    {
+      "epoch": 0.417153489982258,
+      "grad_norm": 1790.3535213323487,
+      "learning_rate": 1.2751647258597531e-05,
+      "loss": 295.3477,
+      "step": 21690
+    },
+    {
+      "epoch": 0.4173458152427385,
+      "grad_norm": 1735.5075479791724,
+      "learning_rate": 1.274577920292576e-05,
+      "loss": 285.6985,
+      "step": 21700
+    },
+    {
+      "epoch": 0.417538140503219,
+      "grad_norm": 1883.5984831296273,
+      "learning_rate": 1.2739910124496585e-05,
+      "loss": 299.5036,
+      "step": 21710
+    },
+    {
+      "epoch": 0.41773046576369954,
+      "grad_norm": 1738.302029158757,
+      "learning_rate": 1.2734040025496137e-05,
+      "loss": 293.0357,
+      "step": 21720
+    },
+    {
+      "epoch": 0.4179227910241801,
+      "grad_norm": 1887.6161143895417,
+      "learning_rate": 1.2728168908110937e-05,
+      "loss": 294.5104,
+      "step": 21730
+    },
+    {
+      "epoch": 0.41811511628466064,
+      "grad_norm": 1738.6950897243764,
+      "learning_rate": 1.2722296774527871e-05,
+      "loss": 285.5324,
+      "step": 21740
+    },
+    {
+      "epoch": 0.41830744154514116,
+      "grad_norm": 1559.3244221477023,
+      "learning_rate": 1.271642362693422e-05,
+      "loss": 306.2529,
+      "step": 21750
+    },
+    {
+      "epoch": 0.4184997668056217,
+      "grad_norm": 1916.9145767723796,
+      "learning_rate": 1.2710549467517628e-05,
+      "loss": 292.2417,
+      "step": 21760
+    },
+    {
+      "epoch": 0.4186920920661022,
+      "grad_norm": 1635.4967774731526,
+      "learning_rate": 1.270467429846613e-05,
+      "loss": 280.6471,
+      "step": 21770
+    },
+    {
+      "epoch": 0.4188844173265827,
+      "grad_norm": 1710.6036949396544,
+      "learning_rate": 1.2698798121968122e-05,
+      "loss": 292.6485,
+      "step": 21780
+    },
+    {
+      "epoch": 0.41907674258706323,
+      "grad_norm": 1725.0476783710178,
+      "learning_rate": 1.2692920940212387e-05,
+      "loss": 304.322,
+      "step": 21790
+    },
+    {
+      "epoch": 0.41926906784754375,
+      "grad_norm": 1682.6806608105035,
+      "learning_rate": 1.2687042755388077e-05,
+      "loss": 287.9139,
+      "step": 21800
+    },
+    {
+      "epoch": 0.41946139310802427,
+      "grad_norm": 1717.7820827129565,
+      "learning_rate": 1.2681163569684718e-05,
+      "loss": 297.2933,
+      "step": 21810
+    },
+    {
+      "epoch": 0.4196537183685048,
+      "grad_norm": 1681.5071264004243,
+      "learning_rate": 1.2675283385292212e-05,
+      "loss": 297.7878,
+      "step": 21820
+    },
+    {
+      "epoch": 0.41984604362898537,
+      "grad_norm": 1983.29666861836,
+      "learning_rate": 1.2669402204400825e-05,
+      "loss": 289.575,
+      "step": 21830
+    },
+    {
+      "epoch": 0.4200383688894659,
+      "grad_norm": 1901.4875038480411,
+      "learning_rate": 1.2663520029201206e-05,
+      "loss": 286.8185,
+      "step": 21840
+    },
+    {
+      "epoch": 0.4202306941499464,
+      "grad_norm": 2029.6549501739983,
+      "learning_rate": 1.2657636861884363e-05,
+      "loss": 289.6501,
+      "step": 21850
+    },
+    {
+      "epoch": 0.4204230194104269,
+      "grad_norm": 2115.6233724820286,
+      "learning_rate": 1.2651752704641686e-05,
+      "loss": 294.1346,
+      "step": 21860
+    },
+    {
+      "epoch": 0.42061534467090744,
+      "grad_norm": 1708.5996574100225,
+      "learning_rate": 1.2645867559664918e-05,
+      "loss": 296.3937,
+      "step": 21870
+    },
+    {
+      "epoch": 0.42080766993138796,
+      "grad_norm": 1717.8765160615162,
+      "learning_rate": 1.2639981429146184e-05,
+      "loss": 284.5227,
+      "step": 21880
+    },
+    {
+      "epoch": 0.4209999951918685,
+      "grad_norm": 1909.8869030102762,
+      "learning_rate": 1.2634094315277967e-05,
+      "loss": 290.336,
+      "step": 21890
+    },
+    {
+      "epoch": 0.421192320452349,
+      "grad_norm": 2021.3751083371153,
+      "learning_rate": 1.262820622025312e-05,
+      "loss": 300.9393,
+      "step": 21900
+    },
+    {
+      "epoch": 0.4213846457128295,
+      "grad_norm": 1873.8783271538828,
+      "learning_rate": 1.2622317146264864e-05,
+      "loss": 296.5559,
+      "step": 21910
+    },
+    {
+      "epoch": 0.42157697097331004,
+      "grad_norm": 1882.235048962817,
+      "learning_rate": 1.2616427095506779e-05,
+      "loss": 295.0742,
+      "step": 21920
+    },
+    {
+      "epoch": 0.4217692962337906,
+      "grad_norm": 1750.0323455953414,
+      "learning_rate": 1.2610536070172815e-05,
+      "loss": 296.5745,
+      "step": 21930
+    },
+    {
+      "epoch": 0.42196162149427113,
+      "grad_norm": 1803.931236936814,
+      "learning_rate": 1.2604644072457277e-05,
+      "loss": 292.0239,
+      "step": 21940
+    },
+    {
+      "epoch": 0.42215394675475165,
+      "grad_norm": 1818.828242287104,
+      "learning_rate": 1.259875110455484e-05,
+      "loss": 294.9897,
+      "step": 21950
+    },
+    {
+      "epoch": 0.42234627201523217,
+      "grad_norm": 1632.7115684899397,
+      "learning_rate": 1.2592857168660535e-05,
+      "loss": 294.3602,
+      "step": 21960
+    },
+    {
+      "epoch": 0.4225385972757127,
+      "grad_norm": 2032.3467186529804,
+      "learning_rate": 1.2586962266969758e-05,
+      "loss": 295.6209,
+      "step": 21970
+    },
+    {
+      "epoch": 0.4227309225361932,
+      "grad_norm": 2031.1830937625148,
+      "learning_rate": 1.2581066401678261e-05,
+      "loss": 307.6854,
+      "step": 21980
+    },
+    {
+      "epoch": 0.42292324779667373,
+      "grad_norm": 1714.389078398596,
+      "learning_rate": 1.2575169574982158e-05,
+      "loss": 290.2839,
+      "step": 21990
+    },
+    {
+      "epoch": 0.42311557305715425,
+      "grad_norm": 1773.1484100021128,
+      "learning_rate": 1.2569271789077919e-05,
+      "loss": 298.0011,
+      "step": 22000
+    },
+    {
+      "epoch": 0.42330789831763477,
+      "grad_norm": 1852.364034041167,
+      "learning_rate": 1.256337304616237e-05,
+      "loss": 287.7284,
+      "step": 22010
+    },
+    {
+      "epoch": 0.4235002235781153,
+      "grad_norm": 1801.260639873809,
+      "learning_rate": 1.25574733484327e-05,
+      "loss": 289.85,
+      "step": 22020
+    },
+    {
+      "epoch": 0.4236925488385958,
+      "grad_norm": 1993.5948357991335,
+      "learning_rate": 1.2551572698086446e-05,
+      "loss": 296.8924,
+      "step": 22030
+    },
+    {
+      "epoch": 0.4238848740990764,
+      "grad_norm": 1768.890574758428,
+      "learning_rate": 1.2545671097321497e-05,
+      "loss": 292.4219,
+      "step": 22040
+    },
+    {
+      "epoch": 0.4240771993595569,
+      "grad_norm": 1729.1688215889747,
+      "learning_rate": 1.2539768548336112e-05,
+      "loss": 294.4896,
+      "step": 22050
+    },
+    {
+      "epoch": 0.4242695246200374,
+      "grad_norm": 1634.3742657371652,
+      "learning_rate": 1.2533865053328886e-05,
+      "loss": 292.933,
+      "step": 22060
+    },
+    {
+      "epoch": 0.42446184988051794,
+      "grad_norm": 1865.495961916923,
+      "learning_rate": 1.2527960614498778e-05,
+      "loss": 299.1005,
+      "step": 22070
+    },
+    {
+      "epoch": 0.42465417514099846,
+      "grad_norm": 1926.8130712866237,
+      "learning_rate": 1.252205523404509e-05,
+      "loss": 299.8442,
+      "step": 22080
+    },
+    {
+      "epoch": 0.424846500401479,
+      "grad_norm": 1801.0788168391084,
+      "learning_rate": 1.2516148914167481e-05,
+      "loss": 299.7834,
+      "step": 22090
+    },
+    {
+      "epoch": 0.4250388256619595,
+      "grad_norm": 1906.336234849879,
+      "learning_rate": 1.2510241657065958e-05,
+      "loss": 284.22,
+      "step": 22100
+    },
+    {
+      "epoch": 0.42523115092244,
+      "grad_norm": 1854.4176695067338,
+      "learning_rate": 1.2504333464940874e-05,
+      "loss": 284.0715,
+      "step": 22110
+    },
+    {
+      "epoch": 0.42542347618292053,
+      "grad_norm": 2191.795101238567,
+      "learning_rate": 1.2498424339992934e-05,
+      "loss": 289.333,
+      "step": 22120
+    },
+    {
+      "epoch": 0.42561580144340105,
+      "grad_norm": 1915.6945693128519,
+      "learning_rate": 1.2492514284423194e-05,
+      "loss": 294.488,
+      "step": 22130
+    },
+    {
+      "epoch": 0.42580812670388163,
+      "grad_norm": 1861.301185668194,
+      "learning_rate": 1.2486603300433045e-05,
+      "loss": 290.9472,
+      "step": 22140
+    },
+    {
+      "epoch": 0.42600045196436215,
+      "grad_norm": 1772.4191910270065,
+      "learning_rate": 1.2480691390224238e-05,
+      "loss": 297.0469,
+      "step": 22150
+    },
+    {
+      "epoch": 0.42619277722484267,
+      "grad_norm": 1754.227022755561,
+      "learning_rate": 1.2474778555998855e-05,
+      "loss": 278.9458,
+      "step": 22160
+    },
+    {
+      "epoch": 0.4263851024853232,
+      "grad_norm": 1846.1901970919557,
+      "learning_rate": 1.246886479995933e-05,
+      "loss": 298.2821,
+      "step": 22170
+    },
+    {
+      "epoch": 0.4265774277458037,
+      "grad_norm": 1906.5567654606382,
+      "learning_rate": 1.2462950124308444e-05,
+      "loss": 296.0458,
+      "step": 22180
+    },
+    {
+      "epoch": 0.4267697530062842,
+      "grad_norm": 1927.1360362454009,
+      "learning_rate": 1.2457034531249313e-05,
+      "loss": 290.0084,
+      "step": 22190
+    },
+    {
+      "epoch": 0.42696207826676474,
+      "grad_norm": 1932.2271998168626,
+      "learning_rate": 1.2451118022985402e-05,
+      "loss": 291.821,
+      "step": 22200
+    },
+    {
+      "epoch": 0.42715440352724526,
+      "grad_norm": 1802.4321113262986,
+      "learning_rate": 1.2445200601720504e-05,
+      "loss": 284.2274,
+      "step": 22210
+    },
+    {
+      "epoch": 0.4273467287877258,
+      "grad_norm": 1843.828194362936,
+      "learning_rate": 1.243928226965877e-05,
+      "loss": 289.67,
+      "step": 22220
+    },
+    {
+      "epoch": 0.4275390540482063,
+      "grad_norm": 1745.8425061119474,
+      "learning_rate": 1.2433363029004675e-05,
+      "loss": 289.6597,
+      "step": 22230
+    },
+    {
+      "epoch": 0.4277313793086869,
+      "grad_norm": 1720.8414756983705,
+      "learning_rate": 1.2427442881963042e-05,
+      "loss": 290.4238,
+      "step": 22240
+    },
+    {
+      "epoch": 0.4279237045691674,
+      "grad_norm": 1861.0984929042972,
+      "learning_rate": 1.242152183073902e-05,
+      "loss": 294.6113,
+      "step": 22250
+    },
+    {
+      "epoch": 0.4281160298296479,
+      "grad_norm": 1764.0034169805558,
+      "learning_rate": 1.2415599877538111e-05,
+      "loss": 283.536,
+      "step": 22260
+    },
+    {
+      "epoch": 0.42830835509012843,
+      "grad_norm": 1803.4229172536875,
+      "learning_rate": 1.2409677024566145e-05,
+      "loss": 301.8564,
+      "step": 22270
+    },
+    {
+      "epoch": 0.42850068035060895,
+      "grad_norm": 2321.4059910452024,
+      "learning_rate": 1.2403753274029281e-05,
+      "loss": 295.7579,
+      "step": 22280
+    },
+    {
+      "epoch": 0.4286930056110895,
+      "grad_norm": 1855.341243688981,
+      "learning_rate": 1.2397828628134028e-05,
+      "loss": 283.3616,
+      "step": 22290
+    },
+    {
+      "epoch": 0.42888533087157,
+      "grad_norm": 1755.6097579861346,
+      "learning_rate": 1.2391903089087208e-05,
+      "loss": 291.0702,
+      "step": 22300
+    },
+    {
+      "epoch": 0.4290776561320505,
+      "grad_norm": 1735.3184787380108,
+      "learning_rate": 1.2385976659095993e-05,
+      "loss": 289.4383,
+      "step": 22310
+    },
+    {
+      "epoch": 0.42926998139253103,
+      "grad_norm": 1654.0841749641856,
+      "learning_rate": 1.2380049340367876e-05,
+      "loss": 285.3722,
+      "step": 22320
+    },
+    {
+      "epoch": 0.42946230665301155,
+      "grad_norm": 1937.8697323262231,
+      "learning_rate": 1.2374121135110688e-05,
+      "loss": 299.4644,
+      "step": 22330
+    },
+    {
+      "epoch": 0.4296546319134921,
+      "grad_norm": 1885.82108573323,
+      "learning_rate": 1.2368192045532586e-05,
+      "loss": 292.687,
+      "step": 22340
+    },
+    {
+      "epoch": 0.42984695717397264,
+      "grad_norm": 1746.53780413847,
+      "learning_rate": 1.236226207384206e-05,
+      "loss": 288.4655,
+      "step": 22350
+    },
+    {
+      "epoch": 0.43003928243445316,
+      "grad_norm": 1912.713094679185,
+      "learning_rate": 1.2356331222247929e-05,
+      "loss": 295.9019,
+      "step": 22360
+    },
+    {
+      "epoch": 0.4302316076949337,
+      "grad_norm": 1860.961584211503,
+      "learning_rate": 1.2350399492959328e-05,
+      "loss": 297.3277,
+      "step": 22370
+    },
+    {
+      "epoch": 0.4304239329554142,
+      "grad_norm": 1996.6017135778345,
+      "learning_rate": 1.234446688818574e-05,
+      "loss": 293.0891,
+      "step": 22380
+    },
+    {
+      "epoch": 0.4306162582158947,
+      "grad_norm": 1920.130615923429,
+      "learning_rate": 1.233853341013695e-05,
+      "loss": 291.4277,
+      "step": 22390
+    },
+    {
+      "epoch": 0.43080858347637524,
+      "grad_norm": 1734.5926270743116,
+      "learning_rate": 1.233259906102309e-05,
+      "loss": 288.4785,
+      "step": 22400
+    },
+    {
+      "epoch": 0.43100090873685576,
+      "grad_norm": 1671.5169904532902,
+      "learning_rate": 1.2326663843054603e-05,
+      "loss": 287.3598,
+      "step": 22410
+    },
+    {
+      "epoch": 0.4311932339973363,
+      "grad_norm": 1878.0200892778894,
+      "learning_rate": 1.2320727758442264e-05,
+      "loss": 295.8173,
+      "step": 22420
+    },
+    {
+      "epoch": 0.4313855592578168,
+      "grad_norm": 1826.1557725608554,
+      "learning_rate": 1.231479080939716e-05,
+      "loss": 301.0681,
+      "step": 22430
+    },
+    {
+      "epoch": 0.43157788451829737,
+      "grad_norm": 1624.1604849581604,
+      "learning_rate": 1.230885299813071e-05,
+      "loss": 300.6742,
+      "step": 22440
+    },
+    {
+      "epoch": 0.4317702097787779,
+      "grad_norm": 1989.9818544823925,
+      "learning_rate": 1.2302914326854651e-05,
+      "loss": 291.5841,
+      "step": 22450
+    },
+    {
+      "epoch": 0.4319625350392584,
+      "grad_norm": 1713.6037141490233,
+      "learning_rate": 1.2296974797781036e-05,
+      "loss": 299.4552,
+      "step": 22460
+    },
+    {
+      "epoch": 0.43215486029973893,
+      "grad_norm": 1932.6392639156013,
+      "learning_rate": 1.2291034413122247e-05,
+      "loss": 303.7552,
+      "step": 22470
+    },
+    {
+      "epoch": 0.43234718556021945,
+      "grad_norm": 1819.0821810113941,
+      "learning_rate": 1.2285093175090976e-05,
+      "loss": 289.8333,
+      "step": 22480
+    },
+    {
+      "epoch": 0.43253951082069997,
+      "grad_norm": 1824.8548153464003,
+      "learning_rate": 1.227915108590024e-05,
+      "loss": 285.3101,
+      "step": 22490
+    },
+    {
+      "epoch": 0.4327318360811805,
+      "grad_norm": 1747.1056902099006,
+      "learning_rate": 1.2273208147763363e-05,
+      "loss": 279.0743,
+      "step": 22500
+    },
+    {
+      "epoch": 0.432924161341661,
+      "grad_norm": 1719.7361446852408,
+      "learning_rate": 1.2267264362893995e-05,
+      "loss": 292.7379,
+      "step": 22510
+    },
+    {
+      "epoch": 0.4331164866021415,
+      "grad_norm": 1833.2847314860103,
+      "learning_rate": 1.2261319733506096e-05,
+      "loss": 290.1692,
+      "step": 22520
+    },
+    {
+      "epoch": 0.43330881186262205,
+      "grad_norm": 1779.5976041555969,
+      "learning_rate": 1.2255374261813944e-05,
+      "loss": 297.8661,
+      "step": 22530
+    },
+    {
+      "epoch": 0.43350113712310256,
+      "grad_norm": 1805.6607169983013,
+      "learning_rate": 1.2249427950032127e-05,
+      "loss": 294.5839,
+      "step": 22540
+    },
+    {
+      "epoch": 0.43369346238358314,
+      "grad_norm": 2916.730721179416,
+      "learning_rate": 1.224348080037555e-05,
+      "loss": 302.201,
+      "step": 22550
+    },
+    {
+      "epoch": 0.43388578764406366,
+      "grad_norm": 1699.518198099036,
+      "learning_rate": 1.2237532815059427e-05,
+      "loss": 289.6796,
+      "step": 22560
+    },
+    {
+      "epoch": 0.4340781129045442,
+      "grad_norm": 1795.4791786912174,
+      "learning_rate": 1.2231583996299285e-05,
+      "loss": 288.5549,
+      "step": 22570
+    },
+    {
+      "epoch": 0.4342704381650247,
+      "grad_norm": 1738.0039835942123,
+      "learning_rate": 1.2225634346310962e-05,
+      "loss": 285.4092,
+      "step": 22580
+    },
+    {
+      "epoch": 0.4344627634255052,
+      "grad_norm": 1664.6936558757486,
+      "learning_rate": 1.22196838673106e-05,
+      "loss": 294.4807,
+      "step": 22590
+    },
+    {
+      "epoch": 0.43465508868598574,
+      "grad_norm": 1738.6415660923024,
+      "learning_rate": 1.2213732561514657e-05,
+      "loss": 296.1545,
+      "step": 22600
+    },
+    {
+      "epoch": 0.43484741394646625,
+      "grad_norm": 1868.8050573850862,
+      "learning_rate": 1.2207780431139894e-05,
+      "loss": 295.5609,
+      "step": 22610
+    },
+    {
+      "epoch": 0.4350397392069468,
+      "grad_norm": 2003.515807142469,
+      "learning_rate": 1.2201827478403385e-05,
+      "loss": 289.1937,
+      "step": 22620
+    },
+    {
+      "epoch": 0.4352320644674273,
+      "grad_norm": 1656.9276532432098,
+      "learning_rate": 1.2195873705522508e-05,
+      "loss": 276.789,
+      "step": 22630
+    },
+    {
+      "epoch": 0.4354243897279078,
+      "grad_norm": 10790.686677964712,
+      "learning_rate": 1.2189919114714936e-05,
+      "loss": 281.9262,
+      "step": 22640
+    },
+    {
+      "epoch": 0.4356167149883884,
+      "grad_norm": 1764.029363039468,
+      "learning_rate": 1.2183963708198668e-05,
+      "loss": 301.1744,
+      "step": 22650
+    },
+    {
+      "epoch": 0.4358090402488689,
+      "grad_norm": 1914.4914198195202,
+      "learning_rate": 1.2178007488191983e-05,
+      "loss": 289.2565,
+      "step": 22660
+    },
+    {
+      "epoch": 0.4360013655093494,
+      "grad_norm": 1621.3208806654015,
+      "learning_rate": 1.2172050456913482e-05,
+      "loss": 293.0277,
+      "step": 22670
+    },
+    {
+      "epoch": 0.43619369076982994,
+      "grad_norm": 1748.3025486960762,
+      "learning_rate": 1.2166092616582055e-05,
+      "loss": 279.6763,
+      "step": 22680
+    },
+    {
+      "epoch": 0.43638601603031046,
+      "grad_norm": 1908.817111333349,
+      "learning_rate": 1.2160133969416903e-05,
+      "loss": 298.3905,
+      "step": 22690
+    },
+    {
+      "epoch": 0.436578341290791,
+      "grad_norm": 1757.4049454032338,
+      "learning_rate": 1.2154174517637526e-05,
+      "loss": 289.2108,
+      "step": 22700
+    },
+    {
+      "epoch": 0.4367706665512715,
+      "grad_norm": 1775.150754177834,
+      "learning_rate": 1.2148214263463718e-05,
+      "loss": 280.9745,
+      "step": 22710
+    },
+    {
+      "epoch": 0.436962991811752,
+      "grad_norm": 1960.9874272261552,
+      "learning_rate": 1.2142253209115577e-05,
+      "loss": 288.9633,
+      "step": 22720
+    },
+    {
+      "epoch": 0.43715531707223254,
+      "grad_norm": 1910.6484392043762,
+      "learning_rate": 1.2136291356813494e-05,
+      "loss": 300.4317,
+      "step": 22730
+    },
+    {
+      "epoch": 0.43734764233271306,
+      "grad_norm": 1973.128983668307,
+      "learning_rate": 1.2130328708778162e-05,
+      "loss": 287.7798,
+      "step": 22740
+    },
+    {
+      "epoch": 0.43753996759319363,
+      "grad_norm": 1780.8425845537547,
+      "learning_rate": 1.2124365267230571e-05,
+      "loss": 293.087,
+      "step": 22750
+    },
+    {
+      "epoch": 0.43773229285367415,
+      "grad_norm": 1995.534795662323,
+      "learning_rate": 1.2118401034392003e-05,
+      "loss": 293.0213,
+      "step": 22760
+    },
+    {
+      "epoch": 0.4379246181141547,
+      "grad_norm": 1691.903001368237,
+      "learning_rate": 1.2112436012484035e-05,
+      "loss": 283.7337,
+      "step": 22770
+    },
+    {
+      "epoch": 0.4381169433746352,
+      "grad_norm": 1746.81748547236,
+      "learning_rate": 1.210647020372854e-05,
+      "loss": 293.1092,
+      "step": 22780
+    },
+    {
+      "epoch": 0.4383092686351157,
+      "grad_norm": 1764.215225084088,
+      "learning_rate": 1.2100503610347686e-05,
+      "loss": 286.2631,
+      "step": 22790
+    },
+    {
+      "epoch": 0.43850159389559623,
+      "grad_norm": 1879.1926409448206,
+      "learning_rate": 1.2094536234563927e-05,
+      "loss": 293.0421,
+      "step": 22800
+    },
+    {
+      "epoch": 0.43869391915607675,
+      "grad_norm": 1790.099066455191,
+      "learning_rate": 1.2088568078600013e-05,
+      "loss": 288.7428,
+      "step": 22810
+    },
+    {
+      "epoch": 0.43888624441655727,
+      "grad_norm": 1849.6678853059964,
+      "learning_rate": 1.2082599144678983e-05,
+      "loss": 296.7231,
+      "step": 22820
+    },
+    {
+      "epoch": 0.4390785696770378,
+      "grad_norm": 1801.5874065013898,
+      "learning_rate": 1.2076629435024168e-05,
+      "loss": 291.6802,
+      "step": 22830
+    },
+    {
+      "epoch": 0.4392708949375183,
+      "grad_norm": 1898.256536186525,
+      "learning_rate": 1.2070658951859183e-05,
+      "loss": 302.706,
+      "step": 22840
+    },
+    {
+      "epoch": 0.4394632201979989,
+      "grad_norm": 1615.3688707153901,
+      "learning_rate": 1.2064687697407939e-05,
+      "loss": 300.7423,
+      "step": 22850
+    },
+    {
+      "epoch": 0.4396555454584794,
+      "grad_norm": 1799.5121917128301,
+      "learning_rate": 1.2058715673894625e-05,
+      "loss": 283.2971,
+      "step": 22860
+    },
+    {
+      "epoch": 0.4398478707189599,
+      "grad_norm": 2187.5173816919432,
+      "learning_rate": 1.2052742883543724e-05,
+      "loss": 303.7761,
+      "step": 22870
+    },
+    {
+      "epoch": 0.44004019597944044,
+      "grad_norm": 1543.7112171266506,
+      "learning_rate": 1.2046769328580004e-05,
+      "loss": 283.914,
+      "step": 22880
+    },
+    {
+      "epoch": 0.44023252123992096,
+      "grad_norm": 1938.9783279348553,
+      "learning_rate": 1.204079501122851e-05,
+      "loss": 289.9621,
+      "step": 22890
+    },
+    {
+      "epoch": 0.4404248465004015,
+      "grad_norm": 1854.0355565012467,
+      "learning_rate": 1.2034819933714576e-05,
+      "loss": 286.943,
+      "step": 22900
+    },
+    {
+      "epoch": 0.440617171760882,
+      "grad_norm": 1720.5437543112669,
+      "learning_rate": 1.2028844098263827e-05,
+      "loss": 291.134,
+      "step": 22910
+    },
+    {
+      "epoch": 0.4408094970213625,
+      "grad_norm": 1988.576575048972,
+      "learning_rate": 1.2022867507102159e-05,
+      "loss": 295.0517,
+      "step": 22920
+    },
+    {
+      "epoch": 0.44100182228184304,
+      "grad_norm": 1673.8616203733277,
+      "learning_rate": 1.2016890162455752e-05,
+      "loss": 289.4667,
+      "step": 22930
+    },
+    {
+      "epoch": 0.44119414754232356,
+      "grad_norm": 1873.379645644224,
+      "learning_rate": 1.2010912066551072e-05,
+      "loss": 283.8355,
+      "step": 22940
+    },
+    {
+      "epoch": 0.4413864728028041,
+      "grad_norm": 1770.043212712193,
+      "learning_rate": 1.2004933221614854e-05,
+      "loss": 281.9152,
+      "step": 22950
+    },
+    {
+      "epoch": 0.44157879806328465,
+      "grad_norm": 2606.506205559988,
+      "learning_rate": 1.1998953629874126e-05,
+      "loss": 282.2,
+      "step": 22960
+    },
+    {
+      "epoch": 0.44177112332376517,
+      "grad_norm": 1732.050881790884,
+      "learning_rate": 1.1992973293556184e-05,
+      "loss": 297.1208,
+      "step": 22970
+    },
+    {
+      "epoch": 0.4419634485842457,
+      "grad_norm": 1973.102295231436,
+      "learning_rate": 1.1986992214888607e-05,
+      "loss": 296.9179,
+      "step": 22980
+    },
+    {
+      "epoch": 0.4421557738447262,
+      "grad_norm": 1811.2823120979465,
+      "learning_rate": 1.1981010396099244e-05,
+      "loss": 288.4724,
+      "step": 22990
+    },
+    {
+      "epoch": 0.4423480991052067,
+      "grad_norm": 1867.996128919048,
+      "learning_rate": 1.1975027839416227e-05,
+      "loss": 287.2646,
+      "step": 23000
+    },
+    {
+      "epoch": 0.44254042436568725,
+      "grad_norm": 1913.4885113246035,
+      "learning_rate": 1.1969044547067961e-05,
+      "loss": 304.0793,
+      "step": 23010
+    },
+    {
+      "epoch": 0.44273274962616777,
+      "grad_norm": 1780.974856084601,
+      "learning_rate": 1.1963060521283117e-05,
+      "loss": 281.992,
+      "step": 23020
+    },
+    {
+      "epoch": 0.4429250748866483,
+      "grad_norm": 1683.1468078575483,
+      "learning_rate": 1.1957075764290652e-05,
+      "loss": 286.047,
+      "step": 23030
+    },
+    {
+      "epoch": 0.4431174001471288,
+      "grad_norm": 1734.3989975089044,
+      "learning_rate": 1.1951090278319784e-05,
+      "loss": 297.9622,
+      "step": 23040
+    },
+    {
+      "epoch": 0.4433097254076093,
+      "grad_norm": 1701.7504968822498,
+      "learning_rate": 1.1945104065600013e-05,
+      "loss": 281.6445,
+      "step": 23050
+    },
+    {
+      "epoch": 0.4435020506680899,
+      "grad_norm": 1704.3040180379824,
+      "learning_rate": 1.1939117128361101e-05,
+      "loss": 287.8655,
+      "step": 23060
+    },
+    {
+      "epoch": 0.4436943759285704,
+      "grad_norm": 2258.7872575451747,
+      "learning_rate": 1.1933129468833087e-05,
+      "loss": 281.5917,
+      "step": 23070
+    },
+    {
+      "epoch": 0.44388670118905094,
+      "grad_norm": 1928.2623730642117,
+      "learning_rate": 1.192714108924627e-05,
+      "loss": 283.3175,
+      "step": 23080
+    },
+    {
+      "epoch": 0.44407902644953146,
+      "grad_norm": 2073.9861314727987,
+      "learning_rate": 1.1921151991831225e-05,
+      "loss": 291.7861,
+      "step": 23090
+    },
+    {
+      "epoch": 0.444271351710012,
+      "grad_norm": 1715.9587300330145,
+      "learning_rate": 1.1915162178818793e-05,
+      "loss": 295.6284,
+      "step": 23100
+    },
+    {
+      "epoch": 0.4444636769704925,
+      "grad_norm": 1837.7060794350339,
+      "learning_rate": 1.1909171652440079e-05,
+      "loss": 290.2767,
+      "step": 23110
+    },
+    {
+      "epoch": 0.444656002230973,
+      "grad_norm": 1797.4906408081708,
+      "learning_rate": 1.1903180414926457e-05,
+      "loss": 286.0442,
+      "step": 23120
+    },
+    {
+      "epoch": 0.44484832749145353,
+      "grad_norm": 1759.1466779667753,
+      "learning_rate": 1.189718846850956e-05,
+      "loss": 282.2319,
+      "step": 23130
+    },
+    {
+      "epoch": 0.44504065275193405,
+      "grad_norm": 1704.9890405638937,
+      "learning_rate": 1.1891195815421293e-05,
+      "loss": 273.0284,
+      "step": 23140
+    },
+    {
+      "epoch": 0.44523297801241457,
+      "grad_norm": 1812.7961728657122,
+      "learning_rate": 1.1885202457893819e-05,
+      "loss": 286.3382,
+      "step": 23150
+    },
+    {
+      "epoch": 0.44542530327289515,
+      "grad_norm": 1710.1556883789276,
+      "learning_rate": 1.1879208398159563e-05,
+      "loss": 288.4673,
+      "step": 23160
+    },
+    {
+      "epoch": 0.44561762853337566,
+      "grad_norm": 1678.5718853197513,
+      "learning_rate": 1.1873213638451215e-05,
+      "loss": 279.5454,
+      "step": 23170
+    },
+    {
+      "epoch": 0.4458099537938562,
+      "grad_norm": 1874.0836182268522,
+      "learning_rate": 1.1867218181001725e-05,
+      "loss": 288.6174,
+      "step": 23180
+    },
+    {
+      "epoch": 0.4460022790543367,
+      "grad_norm": 2670.9328264712767,
+      "learning_rate": 1.1861222028044301e-05,
+      "loss": 296.0979,
+      "step": 23190
+    },
+    {
+      "epoch": 0.4461946043148172,
+      "grad_norm": 1665.7025297889513,
+      "learning_rate": 1.1855225181812408e-05,
+      "loss": 280.7955,
+      "step": 23200
+    },
+    {
+      "epoch": 0.44638692957529774,
+      "grad_norm": 1692.5609582789361,
+      "learning_rate": 1.1849227644539781e-05,
+      "loss": 280.3237,
+      "step": 23210
+    },
+    {
+      "epoch": 0.44657925483577826,
+      "grad_norm": 1516.822947210408,
+      "learning_rate": 1.1843229418460391e-05,
+      "loss": 279.9126,
+      "step": 23220
+    },
+    {
+      "epoch": 0.4467715800962588,
+      "grad_norm": 1635.8471836627796,
+      "learning_rate": 1.1837230505808485e-05,
+      "loss": 278.7231,
+      "step": 23230
+    },
+    {
+      "epoch": 0.4469639053567393,
+      "grad_norm": 1604.722505836218,
+      "learning_rate": 1.1831230908818563e-05,
+      "loss": 281.3688,
+      "step": 23240
+    },
+    {
+      "epoch": 0.4471562306172198,
+      "grad_norm": 1812.974517169146,
+      "learning_rate": 1.1825230629725366e-05,
+      "loss": 293.6629,
+      "step": 23250
+    },
+    {
+      "epoch": 0.4473485558777004,
+      "grad_norm": 1774.454309007271,
+      "learning_rate": 1.1819229670763908e-05,
+      "loss": 289.6647,
+      "step": 23260
+    },
+    {
+      "epoch": 0.4475408811381809,
+      "grad_norm": 2079.130727773057,
+      "learning_rate": 1.1813228034169442e-05,
+      "loss": 296.6703,
+      "step": 23270
+    },
+    {
+      "epoch": 0.44773320639866143,
+      "grad_norm": 1775.0031349756969,
+      "learning_rate": 1.180722572217748e-05,
+      "loss": 291.4848,
+      "step": 23280
+    },
+    {
+      "epoch": 0.44792553165914195,
+      "grad_norm": 1835.0883911093642,
+      "learning_rate": 1.1801222737023783e-05,
+      "loss": 291.9565,
+      "step": 23290
+    },
+    {
+      "epoch": 0.44811785691962247,
+      "grad_norm": 1732.2632603627926,
+      "learning_rate": 1.1795219080944366e-05,
+      "loss": 287.1501,
+      "step": 23300
+    },
+    {
+      "epoch": 0.448310182180103,
+      "grad_norm": 1760.7354771369664,
+      "learning_rate": 1.178921475617549e-05,
+      "loss": 285.9664,
+      "step": 23310
+    },
+    {
+      "epoch": 0.4485025074405835,
+      "grad_norm": 1826.707406558517,
+      "learning_rate": 1.1783209764953666e-05,
+      "loss": 292.7454,
+      "step": 23320
+    },
+    {
+      "epoch": 0.44869483270106403,
+      "grad_norm": 1970.5524085679651,
+      "learning_rate": 1.1777204109515653e-05,
+      "loss": 295.9027,
+      "step": 23330
+    },
+    {
+      "epoch": 0.44888715796154455,
+      "grad_norm": 1909.2435260195157,
+      "learning_rate": 1.1771197792098465e-05,
+      "loss": 289.9696,
+      "step": 23340
+    },
+    {
+      "epoch": 0.44907948322202507,
+      "grad_norm": 1683.089366796813,
+      "learning_rate": 1.1765190814939351e-05,
+      "loss": 287.4251,
+      "step": 23350
+    },
+    {
+      "epoch": 0.4492718084825056,
+      "grad_norm": 1795.7155474563117,
+      "learning_rate": 1.175918318027581e-05,
+      "loss": 280.0622,
+      "step": 23360
+    },
+    {
+      "epoch": 0.44946413374298616,
+      "grad_norm": 1756.949841339035,
+      "learning_rate": 1.1753174890345591e-05,
+      "loss": 287.0618,
+      "step": 23370
+    },
+    {
+      "epoch": 0.4496564590034667,
+      "grad_norm": 1910.6755429209422,
+      "learning_rate": 1.174716594738668e-05,
+      "loss": 289.142,
+      "step": 23380
+    },
+    {
+      "epoch": 0.4498487842639472,
+      "grad_norm": 1911.7015281491028,
+      "learning_rate": 1.1741156353637304e-05,
+      "loss": 280.1322,
+      "step": 23390
+    },
+    {
+      "epoch": 0.4500411095244277,
+      "grad_norm": 1773.9867806509214,
+      "learning_rate": 1.1735146111335945e-05,
+      "loss": 282.747,
+      "step": 23400
+    },
+    {
+      "epoch": 0.45023343478490824,
+      "grad_norm": 1676.6937716961493,
+      "learning_rate": 1.1729135222721315e-05,
+      "loss": 282.6137,
+      "step": 23410
+    },
+    {
+      "epoch": 0.45042576004538876,
+      "grad_norm": 1808.9680360244736,
+      "learning_rate": 1.1723123690032376e-05,
+      "loss": 272.8415,
+      "step": 23420
+    },
+    {
+      "epoch": 0.4506180853058693,
+      "grad_norm": 1653.1305496027244,
+      "learning_rate": 1.171711151550832e-05,
+      "loss": 299.3093,
+      "step": 23430
+    },
+    {
+      "epoch": 0.4508104105663498,
+      "grad_norm": 1812.7404402262237,
+      "learning_rate": 1.1711098701388581e-05,
+      "loss": 285.1163,
+      "step": 23440
+    },
+    {
+      "epoch": 0.4510027358268303,
+      "grad_norm": 1702.8119340459302,
+      "learning_rate": 1.1705085249912837e-05,
+      "loss": 282.1939,
+      "step": 23450
+    },
+    {
+      "epoch": 0.45119506108731083,
+      "grad_norm": 1583.8576373399394,
+      "learning_rate": 1.1699071163320997e-05,
+      "loss": 289.1969,
+      "step": 23460
+    },
+    {
+      "epoch": 0.4513873863477914,
+      "grad_norm": 1734.128409584299,
+      "learning_rate": 1.169305644385321e-05,
+      "loss": 281.8136,
+      "step": 23470
+    },
+    {
+      "epoch": 0.4515797116082719,
+      "grad_norm": 1722.558155028593,
+      "learning_rate": 1.168704109374986e-05,
+      "loss": 287.1599,
+      "step": 23480
+    },
+    {
+      "epoch": 0.45177203686875245,
+      "grad_norm": 1710.6931929672164,
+      "learning_rate": 1.1681025115251566e-05,
+      "loss": 288.8959,
+      "step": 23490
+    },
+    {
+      "epoch": 0.45196436212923297,
+      "grad_norm": 2047.1026638475935,
+      "learning_rate": 1.1675008510599176e-05,
+      "loss": 278.7546,
+      "step": 23500
+    },
+    {
+      "epoch": 0.4521566873897135,
+      "grad_norm": 1851.0792011260548,
+      "learning_rate": 1.1668991282033784e-05,
+      "loss": 284.0779,
+      "step": 23510
+    },
+    {
+      "epoch": 0.452349012650194,
+      "grad_norm": 1846.816831283124,
+      "learning_rate": 1.16629734317967e-05,
+      "loss": 290.6807,
+      "step": 23520
+    },
+    {
+      "epoch": 0.4525413379106745,
+      "grad_norm": 1878.5260776922232,
+      "learning_rate": 1.165695496212948e-05,
+      "loss": 295.8176,
+      "step": 23530
+    },
+    {
+      "epoch": 0.45273366317115504,
+      "grad_norm": 1816.7253291343181,
+      "learning_rate": 1.1650935875273901e-05,
+      "loss": 288.7826,
+      "step": 23540
+    },
+    {
+      "epoch": 0.45292598843163556,
+      "grad_norm": 1586.9737205937136,
+      "learning_rate": 1.1644916173471976e-05,
+      "loss": 283.6865,
+      "step": 23550
+    },
+    {
+      "epoch": 0.4531183136921161,
+      "grad_norm": 1703.918771905539,
+      "learning_rate": 1.1638895858965942e-05,
+      "loss": 283.1265,
+      "step": 23560
+    },
+    {
+      "epoch": 0.45331063895259666,
+      "grad_norm": 1742.0075842146127,
+      "learning_rate": 1.1632874933998268e-05,
+      "loss": 282.4831,
+      "step": 23570
+    },
+    {
+      "epoch": 0.4535029642130772,
+      "grad_norm": 1810.1424471119528,
+      "learning_rate": 1.1626853400811649e-05,
+      "loss": 286.5923,
+      "step": 23580
+    },
+    {
+      "epoch": 0.4536952894735577,
+      "grad_norm": 1494.5794457017953,
+      "learning_rate": 1.1620831261649003e-05,
+      "loss": 279.275,
+      "step": 23590
+    },
+    {
+      "epoch": 0.4538876147340382,
+      "grad_norm": 1587.080233581412,
+      "learning_rate": 1.1614808518753485e-05,
+      "loss": 285.0605,
+      "step": 23600
+    },
+    {
+      "epoch": 0.45407993999451873,
+      "grad_norm": 1655.8667909162855,
+      "learning_rate": 1.1608785174368461e-05,
+      "loss": 286.6222,
+      "step": 23610
+    },
+    {
+      "epoch": 0.45427226525499925,
+      "grad_norm": 1764.1897700555548,
+      "learning_rate": 1.1602761230737531e-05,
+      "loss": 297.7672,
+      "step": 23620
+    },
+    {
+      "epoch": 0.45446459051547977,
+      "grad_norm": 1780.384728045414,
+      "learning_rate": 1.1596736690104514e-05,
+      "loss": 300.0785,
+      "step": 23630
+    },
+    {
+      "epoch": 0.4546569157759603,
+      "grad_norm": 1755.779092767589,
+      "learning_rate": 1.1590711554713452e-05,
+      "loss": 280.6965,
+      "step": 23640
+    },
+    {
+      "epoch": 0.4548492410364408,
+      "grad_norm": 1807.9899391052184,
+      "learning_rate": 1.1584685826808604e-05,
+      "loss": 295.9619,
+      "step": 23650
+    },
+    {
+      "epoch": 0.45504156629692133,
+      "grad_norm": 2064.8159441450043,
+      "learning_rate": 1.157865950863446e-05,
+      "loss": 287.0157,
+      "step": 23660
+    },
+    {
+      "epoch": 0.4552338915574019,
+      "grad_norm": 2137.3376159337927,
+      "learning_rate": 1.1572632602435717e-05,
+      "loss": 293.902,
+      "step": 23670
+    },
+    {
+      "epoch": 0.4554262168178824,
+      "grad_norm": 1569.9931877270574,
+      "learning_rate": 1.1566605110457305e-05,
+      "loss": 280.1166,
+      "step": 23680
+    },
+    {
+      "epoch": 0.45561854207836294,
+      "grad_norm": 2007.340270423367,
+      "learning_rate": 1.1560577034944364e-05,
+      "loss": 289.0956,
+      "step": 23690
+    },
+    {
+      "epoch": 0.45581086733884346,
+      "grad_norm": 1931.3024264232283,
+      "learning_rate": 1.1554548378142249e-05,
+      "loss": 288.4179,
+      "step": 23700
+    },
+    {
+      "epoch": 0.456003192599324,
+      "grad_norm": 1944.1780294249734,
+      "learning_rate": 1.1548519142296541e-05,
+      "loss": 286.3278,
+      "step": 23710
+    },
+    {
+      "epoch": 0.4561955178598045,
+      "grad_norm": 1689.7150106930471,
+      "learning_rate": 1.1542489329653024e-05,
+      "loss": 291.3526,
+      "step": 23720
+    },
+    {
+      "epoch": 0.456387843120285,
+      "grad_norm": 1807.9609317369523,
+      "learning_rate": 1.153645894245771e-05,
+      "loss": 275.6158,
+      "step": 23730
+    },
+    {
+      "epoch": 0.45658016838076554,
+      "grad_norm": 1822.6861262073455,
+      "learning_rate": 1.1530427982956813e-05,
+      "loss": 292.7068,
+      "step": 23740
+    },
+    {
+      "epoch": 0.45677249364124606,
+      "grad_norm": 1798.8724708841796,
+      "learning_rate": 1.1524396453396767e-05,
+      "loss": 284.5452,
+      "step": 23750
+    },
+    {
+      "epoch": 0.4569648189017266,
+      "grad_norm": 2078.4341054267998,
+      "learning_rate": 1.1518364356024219e-05,
+      "loss": 289.6863,
+      "step": 23760
+    },
+    {
+      "epoch": 0.45715714416220715,
+      "grad_norm": 2248.5845611815375,
+      "learning_rate": 1.1512331693086025e-05,
+      "loss": 277.8321,
+      "step": 23770
+    },
+    {
+      "epoch": 0.45734946942268767,
+      "grad_norm": 1927.9092338464047,
+      "learning_rate": 1.1506298466829256e-05,
+      "loss": 292.2804,
+      "step": 23780
+    },
+    {
+      "epoch": 0.4575417946831682,
+      "grad_norm": 1850.887892926861,
+      "learning_rate": 1.1500264679501181e-05,
+      "loss": 283.8637,
+      "step": 23790
+    },
+    {
+      "epoch": 0.4577341199436487,
+      "grad_norm": 1693.6868742509823,
+      "learning_rate": 1.1494230333349292e-05,
+      "loss": 288.8228,
+      "step": 23800
+    },
+    {
+      "epoch": 0.45792644520412923,
+      "grad_norm": 1936.015090643325,
+      "learning_rate": 1.1488195430621284e-05,
+      "loss": 284.9828,
+      "step": 23810
+    },
+    {
+      "epoch": 0.45811877046460975,
+      "grad_norm": 1682.8866890385339,
+      "learning_rate": 1.1482159973565051e-05,
+      "loss": 293.2018,
+      "step": 23820
+    },
+    {
+      "epoch": 0.45831109572509027,
+      "grad_norm": 1938.7381966171101,
+      "learning_rate": 1.147612396442871e-05,
+      "loss": 285.0724,
+      "step": 23830
+    },
+    {
+      "epoch": 0.4585034209855708,
+      "grad_norm": 1641.6893164987168,
+      "learning_rate": 1.1470087405460572e-05,
+      "loss": 277.8538,
+      "step": 23840
+    },
+    {
+      "epoch": 0.4586957462460513,
+      "grad_norm": 1850.2389463573038,
+      "learning_rate": 1.1464050298909153e-05,
+      "loss": 277.7248,
+      "step": 23850
+    },
+    {
+      "epoch": 0.4588880715065318,
+      "grad_norm": 1544.748625565167,
+      "learning_rate": 1.1458012647023178e-05,
+      "loss": 267.8215,
+      "step": 23860
+    },
+    {
+      "epoch": 0.45908039676701234,
+      "grad_norm": 1835.520167322254,
+      "learning_rate": 1.1451974452051572e-05,
+      "loss": 287.3512,
+      "step": 23870
+    },
+    {
+      "epoch": 0.4592727220274929,
+      "grad_norm": 2124.162675866568,
+      "learning_rate": 1.1445935716243463e-05,
+      "loss": 282.1791,
+      "step": 23880
+    },
+    {
+      "epoch": 0.45946504728797344,
+      "grad_norm": 1864.3869241400778,
+      "learning_rate": 1.143989644184818e-05,
+      "loss": 286.0018,
+      "step": 23890
+    },
+    {
+      "epoch": 0.45965737254845396,
+      "grad_norm": 1823.1616715390342,
+      "learning_rate": 1.1433856631115252e-05,
+      "loss": 292.7189,
+      "step": 23900
+    },
+    {
+      "epoch": 0.4598496978089345,
+      "grad_norm": 1659.4276317327933,
+      "learning_rate": 1.142781628629441e-05,
+      "loss": 279.31,
+      "step": 23910
+    },
+    {
+      "epoch": 0.460042023069415,
+      "grad_norm": 1648.992431356757,
+      "learning_rate": 1.1421775409635585e-05,
+      "loss": 283.0493,
+      "step": 23920
+    },
+    {
+      "epoch": 0.4602343483298955,
+      "grad_norm": 2809.6358145128875,
+      "learning_rate": 1.1415734003388899e-05,
+      "loss": 289.6218,
+      "step": 23930
+    },
+    {
+      "epoch": 0.46042667359037603,
+      "grad_norm": 1761.138388944903,
+      "learning_rate": 1.1409692069804678e-05,
+      "loss": 290.4586,
+      "step": 23940
+    },
+    {
+      "epoch": 0.46061899885085655,
+      "grad_norm": 1699.1472868601832,
+      "learning_rate": 1.1403649611133444e-05,
+      "loss": 275.4669,
+      "step": 23950
+    },
+    {
+      "epoch": 0.4608113241113371,
+      "grad_norm": 1790.2134656203066,
+      "learning_rate": 1.1397606629625913e-05,
+      "loss": 289.6756,
+      "step": 23960
+    },
+    {
+      "epoch": 0.4610036493718176,
+      "grad_norm": 1724.0363318964176,
+      "learning_rate": 1.1391563127532992e-05,
+      "loss": 282.2445,
+      "step": 23970
+    },
+    {
+      "epoch": 0.46119597463229817,
+      "grad_norm": 1709.0754815797438,
+      "learning_rate": 1.1385519107105791e-05,
+      "loss": 283.2302,
+      "step": 23980
+    },
+    {
+      "epoch": 0.4613882998927787,
+      "grad_norm": 1967.664449518948,
+      "learning_rate": 1.1379474570595604e-05,
+      "loss": 287.0517,
+      "step": 23990
+    },
+    {
+      "epoch": 0.4615806251532592,
+      "grad_norm": 1903.6430453002488,
+      "learning_rate": 1.1373429520253922e-05,
+      "loss": 275.3917,
+      "step": 24000
+    },
+    {
+      "epoch": 0.4617729504137397,
+      "grad_norm": 1953.0533945189911,
+      "learning_rate": 1.1367383958332427e-05,
+      "loss": 278.4549,
+      "step": 24010
+    },
+    {
+      "epoch": 0.46196527567422024,
+      "grad_norm": 1901.03403093669,
+      "learning_rate": 1.1361337887082991e-05,
+      "loss": 282.2804,
+      "step": 24020
+    },
+    {
+      "epoch": 0.46215760093470076,
+      "grad_norm": 1905.0695224264493,
+      "learning_rate": 1.1355291308757672e-05,
+      "loss": 287.2164,
+      "step": 24030
+    },
+    {
+      "epoch": 0.4623499261951813,
+      "grad_norm": 1725.2841825427975,
+      "learning_rate": 1.1349244225608727e-05,
+      "loss": 283.5318,
+      "step": 24040
+    },
+    {
+      "epoch": 0.4625422514556618,
+      "grad_norm": 1868.2311153844091,
+      "learning_rate": 1.1343196639888591e-05,
+      "loss": 286.734,
+      "step": 24050
+    },
+    {
+      "epoch": 0.4627345767161423,
+      "grad_norm": 1888.79687534906,
+      "learning_rate": 1.133714855384989e-05,
+      "loss": 289.3315,
+      "step": 24060
+    },
+    {
+      "epoch": 0.46292690197662284,
+      "grad_norm": 1879.3003434705133,
+      "learning_rate": 1.1331099969745439e-05,
+      "loss": 282.837,
+      "step": 24070
+    },
+    {
+      "epoch": 0.4631192272371034,
+      "grad_norm": 1886.172658678069,
+      "learning_rate": 1.132505088982823e-05,
+      "loss": 296.6889,
+      "step": 24080
+    },
+    {
+      "epoch": 0.46331155249758393,
+      "grad_norm": 1743.3502478122045,
+      "learning_rate": 1.131900131635145e-05,
+      "loss": 277.7363,
+      "step": 24090
+    },
+    {
+      "epoch": 0.46350387775806445,
+      "grad_norm": 1830.5505160450712,
+      "learning_rate": 1.1312951251568461e-05,
+      "loss": 285.4822,
+      "step": 24100
+    },
+    {
+      "epoch": 0.463696203018545,
+      "grad_norm": 1655.2835366167606,
+      "learning_rate": 1.1306900697732816e-05,
+      "loss": 277.2251,
+      "step": 24110
+    },
+    {
+      "epoch": 0.4638885282790255,
+      "grad_norm": 1804.5381177198728,
+      "learning_rate": 1.1300849657098248e-05,
+      "loss": 283.2133,
+      "step": 24120
+    },
+    {
+      "epoch": 0.464080853539506,
+      "grad_norm": 1984.019623330842,
+      "learning_rate": 1.1294798131918665e-05,
+      "loss": 277.3018,
+      "step": 24130
+    },
+    {
+      "epoch": 0.46427317879998653,
+      "grad_norm": 1767.2289431539864,
+      "learning_rate": 1.1288746124448164e-05,
+      "loss": 280.9935,
+      "step": 24140
+    },
+    {
+      "epoch": 0.46446550406046705,
+      "grad_norm": 1732.5699189471168,
+      "learning_rate": 1.1282693636941013e-05,
+      "loss": 273.8635,
+      "step": 24150
+    },
+    {
+      "epoch": 0.46465782932094757,
+      "grad_norm": 1655.3718854805231,
+      "learning_rate": 1.1276640671651671e-05,
+      "loss": 292.1583,
+      "step": 24160
+    },
+    {
+      "epoch": 0.4648501545814281,
+      "grad_norm": 1757.5040467921374,
+      "learning_rate": 1.1270587230834757e-05,
+      "loss": 286.4842,
+      "step": 24170
+    },
+    {
+      "epoch": 0.46504247984190866,
+      "grad_norm": 1718.2535882858665,
+      "learning_rate": 1.1264533316745088e-05,
+      "loss": 280.9207,
+      "step": 24180
+    },
+    {
+      "epoch": 0.4652348051023892,
+      "grad_norm": 1646.2519027718342,
+      "learning_rate": 1.1258478931637641e-05,
+      "loss": 280.6315,
+      "step": 24190
+    },
+    {
+      "epoch": 0.4654271303628697,
+      "grad_norm": 1735.7570762363196,
+      "learning_rate": 1.1252424077767577e-05,
+      "loss": 267.6875,
+      "step": 24200
+    },
+    {
+      "epoch": 0.4656194556233502,
+      "grad_norm": 1794.7969598830223,
+      "learning_rate": 1.1246368757390231e-05,
+      "loss": 277.728,
+      "step": 24210
+    },
+    {
+      "epoch": 0.46581178088383074,
+      "grad_norm": 1839.0279282002361,
+      "learning_rate": 1.1240312972761105e-05,
+      "loss": 276.9731,
+      "step": 24220
+    },
+    {
+      "epoch": 0.46600410614431126,
+      "grad_norm": 1829.699830664948,
+      "learning_rate": 1.1234256726135882e-05,
+      "loss": 281.7794,
+      "step": 24230
+    },
+    {
+      "epoch": 0.4661964314047918,
+      "grad_norm": 1604.2336744731938,
+      "learning_rate": 1.1228200019770412e-05,
+      "loss": 267.9988,
+      "step": 24240
+    },
+    {
+      "epoch": 0.4663887566652723,
+      "grad_norm": 2740.5041147937695,
+      "learning_rate": 1.1222142855920719e-05,
+      "loss": 286.5131,
+      "step": 24250
+    },
+    {
+      "epoch": 0.4665810819257528,
+      "grad_norm": 1768.08052342765,
+      "learning_rate": 1.1216085236842997e-05,
+      "loss": 279.2538,
+      "step": 24260
+    },
+    {
+      "epoch": 0.46677340718623334,
+      "grad_norm": 1682.2930516996632,
+      "learning_rate": 1.1210027164793609e-05,
+      "loss": 287.3463,
+      "step": 24270
+    },
+    {
+      "epoch": 0.46696573244671385,
+      "grad_norm": 1745.260429319336,
+      "learning_rate": 1.1203968642029086e-05,
+      "loss": 287.067,
+      "step": 24280
+    },
+    {
+      "epoch": 0.46715805770719443,
+      "grad_norm": 2030.0964252395556,
+      "learning_rate": 1.1197909670806126e-05,
+      "loss": 281.6308,
+      "step": 24290
+    },
+    {
+      "epoch": 0.46735038296767495,
+      "grad_norm": 2263.056414037873,
+      "learning_rate": 1.1191850253381602e-05,
+      "loss": 292.0448,
+      "step": 24300
+    },
+    {
+      "epoch": 0.46754270822815547,
+      "grad_norm": 1608.135946897927,
+      "learning_rate": 1.1185790392012538e-05,
+      "loss": 280.7332,
+      "step": 24310
+    },
+    {
+      "epoch": 0.467735033488636,
+      "grad_norm": 1841.1976983480774,
+      "learning_rate": 1.117973008895614e-05,
+      "loss": 282.3082,
+      "step": 24320
+    },
+    {
+      "epoch": 0.4679273587491165,
+      "grad_norm": 1687.6509817266824,
+      "learning_rate": 1.1173669346469767e-05,
+      "loss": 288.2399,
+      "step": 24330
+    },
+    {
+      "epoch": 0.468119684009597,
+      "grad_norm": 1615.3864068301991,
+      "learning_rate": 1.1167608166810948e-05,
+      "loss": 279.0982,
+      "step": 24340
+    },
+    {
+      "epoch": 0.46831200927007754,
+      "grad_norm": 1957.4190617024055,
+      "learning_rate": 1.1161546552237368e-05,
+      "loss": 294.8661,
+      "step": 24350
+    },
+    {
+      "epoch": 0.46850433453055806,
+      "grad_norm": 1943.3842650418649,
+      "learning_rate": 1.1155484505006884e-05,
+      "loss": 283.2988,
+      "step": 24360
+    },
+    {
+      "epoch": 0.4686966597910386,
+      "grad_norm": 1892.399811926419,
+      "learning_rate": 1.1149422027377501e-05,
+      "loss": 284.6024,
+      "step": 24370
+    },
+    {
+      "epoch": 0.4688889850515191,
+      "grad_norm": 1720.6943931501082,
+      "learning_rate": 1.1143359121607397e-05,
+      "loss": 278.5779,
+      "step": 24380
+    },
+    {
+      "epoch": 0.4690813103119997,
+      "grad_norm": 1693.030082106743,
+      "learning_rate": 1.1137295789954904e-05,
+      "loss": 275.7291,
+      "step": 24390
+    },
+    {
+      "epoch": 0.4692736355724802,
+      "grad_norm": 1663.5824178421692,
+      "learning_rate": 1.1131232034678513e-05,
+      "loss": 280.3416,
+      "step": 24400
+    },
+    {
+      "epoch": 0.4694659608329607,
+      "grad_norm": 1633.5569561764394,
+      "learning_rate": 1.1125167858036874e-05,
+      "loss": 281.3934,
+      "step": 24410
+    },
+    {
+      "epoch": 0.46965828609344124,
+      "grad_norm": 3232.357659216158,
+      "learning_rate": 1.1119103262288788e-05,
+      "loss": 280.0552,
+      "step": 24420
+    },
+    {
+      "epoch": 0.46985061135392175,
+      "grad_norm": 1792.070896933448,
+      "learning_rate": 1.1113038249693221e-05,
+      "loss": 278.2444,
+      "step": 24430
+    },
+    {
+      "epoch": 0.4700429366144023,
+      "grad_norm": 2616.900151591412,
+      "learning_rate": 1.1106972822509287e-05,
+      "loss": 279.481,
+      "step": 24440
+    },
+    {
+      "epoch": 0.4702352618748828,
+      "grad_norm": 2391.309370893597,
+      "learning_rate": 1.1100906982996257e-05,
+      "loss": 291.4892,
+      "step": 24450
+    },
+    {
+      "epoch": 0.4704275871353633,
+      "grad_norm": 1729.293782646985,
+      "learning_rate": 1.109484073341356e-05,
+      "loss": 281.5636,
+      "step": 24460
+    },
+    {
+      "epoch": 0.47061991239584383,
+      "grad_norm": 1787.1199819749916,
+      "learning_rate": 1.1088774076020772e-05,
+      "loss": 271.1173,
+      "step": 24470
+    },
+    {
+      "epoch": 0.47081223765632435,
+      "grad_norm": 1748.516395411248,
+      "learning_rate": 1.1082707013077625e-05,
+      "loss": 276.029,
+      "step": 24480
+    },
+    {
+      "epoch": 0.4710045629168049,
+      "grad_norm": 1806.315034393149,
+      "learning_rate": 1.1076639546843993e-05,
+      "loss": 282.0608,
+      "step": 24490
+    },
+    {
+      "epoch": 0.47119688817728544,
+      "grad_norm": 1714.881780505176,
+      "learning_rate": 1.1070571679579913e-05,
+      "loss": 283.65,
+      "step": 24500
+    },
+    {
+      "epoch": 0.47138921343776596,
+      "grad_norm": 1725.475171527444,
+      "learning_rate": 1.1064503413545565e-05,
+      "loss": 285.5003,
+      "step": 24510
+    },
+    {
+      "epoch": 0.4715815386982465,
+      "grad_norm": 1921.2247765509198,
+      "learning_rate": 1.1058434751001272e-05,
+      "loss": 281.6898,
+      "step": 24520
+    },
+    {
+      "epoch": 0.471773863958727,
+      "grad_norm": 1757.3312740082472,
+      "learning_rate": 1.1052365694207516e-05,
+      "loss": 277.4069,
+      "step": 24530
+    },
+    {
+      "epoch": 0.4719661892192075,
+      "grad_norm": 1573.0580372973177,
+      "learning_rate": 1.1046296245424922e-05,
+      "loss": 273.5803,
+      "step": 24540
+    },
+    {
+      "epoch": 0.47215851447968804,
+      "grad_norm": 1684.4195762894176,
+      "learning_rate": 1.1040226406914254e-05,
+      "loss": 275.6242,
+      "step": 24550
+    },
+    {
+      "epoch": 0.47235083974016856,
+      "grad_norm": 1709.783815174474,
+      "learning_rate": 1.103415618093643e-05,
+      "loss": 289.4204,
+      "step": 24560
+    },
+    {
+      "epoch": 0.4725431650006491,
+      "grad_norm": 1866.0051373995607,
+      "learning_rate": 1.1028085569752512e-05,
+      "loss": 281.4208,
+      "step": 24570
+    },
+    {
+      "epoch": 0.4727354902611296,
+      "grad_norm": 1581.4615809336772,
+      "learning_rate": 1.1022014575623695e-05,
+      "loss": 283.6304,
+      "step": 24580
+    },
+    {
+      "epoch": 0.4729278155216102,
+      "grad_norm": 1879.0791094284268,
+      "learning_rate": 1.1015943200811325e-05,
+      "loss": 277.4734,
+      "step": 24590
+    },
+    {
+      "epoch": 0.4731201407820907,
+      "grad_norm": 1661.6447599006524,
+      "learning_rate": 1.1009871447576894e-05,
+      "loss": 272.8037,
+      "step": 24600
+    },
+    {
+      "epoch": 0.4733124660425712,
+      "grad_norm": 1725.0222049205809,
+      "learning_rate": 1.100379931818203e-05,
+      "loss": 284.0248,
+      "step": 24610
+    },
+    {
+      "epoch": 0.47350479130305173,
+      "grad_norm": 1942.6387207529522,
+      "learning_rate": 1.0997726814888497e-05,
+      "loss": 293.1799,
+      "step": 24620
+    },
+    {
+      "epoch": 0.47369711656353225,
+      "grad_norm": 1743.039914004875,
+      "learning_rate": 1.0991653939958203e-05,
+      "loss": 271.784,
+      "step": 24630
+    },
+    {
+      "epoch": 0.47388944182401277,
+      "grad_norm": 1686.7357424112322,
+      "learning_rate": 1.0985580695653193e-05,
+      "loss": 284.0414,
+      "step": 24640
+    },
+    {
+      "epoch": 0.4740817670844933,
+      "grad_norm": 1842.0619794252013,
+      "learning_rate": 1.0979507084235653e-05,
+      "loss": 283.0204,
+      "step": 24650
+    },
+    {
+      "epoch": 0.4742740923449738,
+      "grad_norm": 1585.062797187579,
+      "learning_rate": 1.0973433107967901e-05,
+      "loss": 276.5904,
+      "step": 24660
+    },
+    {
+      "epoch": 0.4744664176054543,
+      "grad_norm": 2080.3863754800336,
+      "learning_rate": 1.0967358769112389e-05,
+      "loss": 282.0562,
+      "step": 24670
+    },
+    {
+      "epoch": 0.47465874286593485,
+      "grad_norm": 1716.4446207902547,
+      "learning_rate": 1.0961284069931717e-05,
+      "loss": 274.7074,
+      "step": 24680
+    },
+    {
+      "epoch": 0.4748510681264154,
+      "grad_norm": 1728.7992809529417,
+      "learning_rate": 1.0955209012688602e-05,
+      "loss": 279.8301,
+      "step": 24690
+    },
+    {
+      "epoch": 0.47504339338689594,
+      "grad_norm": 1667.5814763707388,
+      "learning_rate": 1.094913359964591e-05,
+      "loss": 274.3013,
+      "step": 24700
+    },
+    {
+      "epoch": 0.47523571864737646,
+      "grad_norm": 2065.739433043234,
+      "learning_rate": 1.0943057833066622e-05,
+      "loss": 285.344,
+      "step": 24710
+    },
+    {
+      "epoch": 0.475428043907857,
+      "grad_norm": 1673.6681994801904,
+      "learning_rate": 1.093698171521387e-05,
+      "loss": 279.6878,
+      "step": 24720
+    },
+    {
+      "epoch": 0.4756203691683375,
+      "grad_norm": 1943.167747370926,
+      "learning_rate": 1.0930905248350903e-05,
+      "loss": 269.1851,
+      "step": 24730
+    },
+    {
+      "epoch": 0.475812694428818,
+      "grad_norm": 1763.5127728618127,
+      "learning_rate": 1.0924828434741101e-05,
+      "loss": 280.9409,
+      "step": 24740
+    },
+    {
+      "epoch": 0.47600501968929854,
+      "grad_norm": 1737.2621383584503,
+      "learning_rate": 1.0918751276647988e-05,
+      "loss": 282.1235,
+      "step": 24750
+    },
+    {
+      "epoch": 0.47619734494977906,
+      "grad_norm": 1689.3505373660075,
+      "learning_rate": 1.0912673776335194e-05,
+      "loss": 275.5593,
+      "step": 24760
+    },
+    {
+      "epoch": 0.4763896702102596,
+      "grad_norm": 1750.2103559510556,
+      "learning_rate": 1.0906595936066496e-05,
+      "loss": 279.0149,
+      "step": 24770
+    },
+    {
+      "epoch": 0.4765819954707401,
+      "grad_norm": 1574.4191111061614,
+      "learning_rate": 1.090051775810578e-05,
+      "loss": 276.1497,
+      "step": 24780
+    },
+    {
+      "epoch": 0.4767743207312206,
+      "grad_norm": 1777.465945592367,
+      "learning_rate": 1.0894439244717075e-05,
+      "loss": 285.6875,
+      "step": 24790
+    },
+    {
+      "epoch": 0.4769666459917012,
+      "grad_norm": 1870.3403189173935,
+      "learning_rate": 1.0888360398164521e-05,
+      "loss": 291.24,
+      "step": 24800
+    },
+    {
+      "epoch": 0.4771589712521817,
+      "grad_norm": 1657.42980807789,
+      "learning_rate": 1.088228122071239e-05,
+      "loss": 282.8729,
+      "step": 24810
+    },
+    {
+      "epoch": 0.4773512965126622,
+      "grad_norm": 1662.491142966301,
+      "learning_rate": 1.0876201714625076e-05,
+      "loss": 281.3687,
+      "step": 24820
+    },
+    {
+      "epoch": 0.47754362177314275,
+      "grad_norm": 1753.9557880837808,
+      "learning_rate": 1.0870121882167095e-05,
+      "loss": 279.9903,
+      "step": 24830
+    },
+    {
+      "epoch": 0.47773594703362326,
+      "grad_norm": 1725.727367314517,
+      "learning_rate": 1.0864041725603085e-05,
+      "loss": 276.5499,
+      "step": 24840
+    },
+    {
+      "epoch": 0.4779282722941038,
+      "grad_norm": 1801.003114626843,
+      "learning_rate": 1.08579612471978e-05,
+      "loss": 273.0663,
+      "step": 24850
+    },
+    {
+      "epoch": 0.4781205975545843,
+      "grad_norm": 2355.5563038069777,
+      "learning_rate": 1.0851880449216123e-05,
+      "loss": 272.5237,
+      "step": 24860
+    },
+    {
+      "epoch": 0.4783129228150648,
+      "grad_norm": 1653.960207940239,
+      "learning_rate": 1.0845799333923045e-05,
+      "loss": 283.3812,
+      "step": 24870
+    },
+    {
+      "epoch": 0.47850524807554534,
+      "grad_norm": 1701.787098663378,
+      "learning_rate": 1.0839717903583684e-05,
+      "loss": 276.0204,
+      "step": 24880
+    },
+    {
+      "epoch": 0.47869757333602586,
+      "grad_norm": 1913.5958147198544,
+      "learning_rate": 1.0833636160463273e-05,
+      "loss": 277.9247,
+      "step": 24890
+    },
+    {
+      "epoch": 0.47888989859650644,
+      "grad_norm": 1777.9577341763338,
+      "learning_rate": 1.0827554106827162e-05,
+      "loss": 280.5851,
+      "step": 24900
+    },
+    {
+      "epoch": 0.47908222385698696,
+      "grad_norm": 1703.7082344355422,
+      "learning_rate": 1.0821471744940812e-05,
+      "loss": 274.7295,
+      "step": 24910
+    },
+    {
+      "epoch": 0.4792745491174675,
+      "grad_norm": 1587.6434780646814,
+      "learning_rate": 1.0815389077069805e-05,
+      "loss": 273.788,
+      "step": 24920
+    },
+    {
+      "epoch": 0.479466874377948,
+      "grad_norm": 2014.059943348886,
+      "learning_rate": 1.0809306105479834e-05,
+      "loss": 281.809,
+      "step": 24930
+    },
+    {
+      "epoch": 0.4796591996384285,
+      "grad_norm": 1655.0458377770333,
+      "learning_rate": 1.0803222832436701e-05,
+      "loss": 272.5006,
+      "step": 24940
+    },
+    {
+      "epoch": 0.47985152489890903,
+      "grad_norm": 2044.3995723079227,
+      "learning_rate": 1.0797139260206331e-05,
+      "loss": 278.8814,
+      "step": 24950
+    },
+    {
+      "epoch": 0.48004385015938955,
+      "grad_norm": 1703.1689266297653,
+      "learning_rate": 1.079105539105475e-05,
+      "loss": 283.2675,
+      "step": 24960
+    },
+    {
+      "epoch": 0.48023617541987007,
+      "grad_norm": 2852.4551116791026,
+      "learning_rate": 1.0784971227248104e-05,
+      "loss": 285.7957,
+      "step": 24970
+    },
+    {
+      "epoch": 0.4804285006803506,
+      "grad_norm": 1584.9577317053706,
+      "learning_rate": 1.077888677105264e-05,
+      "loss": 284.5297,
+      "step": 24980
+    },
+    {
+      "epoch": 0.4806208259408311,
+      "grad_norm": 1608.1657154046318,
+      "learning_rate": 1.0772802024734716e-05,
+      "loss": 274.6846,
+      "step": 24990
+    },
+    {
+      "epoch": 0.4808131512013117,
+      "grad_norm": 1684.024173161361,
+      "learning_rate": 1.07667169905608e-05,
+      "loss": 284.1644,
+      "step": 25000
+    },
+    {
+      "epoch": 0.4810054764617922,
+      "grad_norm": 1790.628843096983,
+      "learning_rate": 1.0760631670797468e-05,
+      "loss": 274.1171,
+      "step": 25010
+    },
+    {
+      "epoch": 0.4811978017222727,
+      "grad_norm": 1842.0984959429345,
+      "learning_rate": 1.07545460677114e-05,
+      "loss": 270.6828,
+      "step": 25020
+    },
+    {
+      "epoch": 0.48139012698275324,
+      "grad_norm": 1928.2675026309344,
+      "learning_rate": 1.0748460183569385e-05,
+      "loss": 280.9262,
+      "step": 25030
+    },
+    {
+      "epoch": 0.48158245224323376,
+      "grad_norm": 1770.4835210400588,
+      "learning_rate": 1.0742374020638315e-05,
+      "loss": 277.8285,
+      "step": 25040
+    },
+    {
+      "epoch": 0.4817747775037143,
+      "grad_norm": 1783.6631353171954,
+      "learning_rate": 1.073628758118518e-05,
+      "loss": 286.9687,
+      "step": 25050
+    },
+    {
+      "epoch": 0.4819671027641948,
+      "grad_norm": 1761.2541630268242,
+      "learning_rate": 1.0730200867477083e-05,
+      "loss": 275.5293,
+      "step": 25060
+    },
+    {
+      "epoch": 0.4821594280246753,
+      "grad_norm": 1959.9590857632304,
+      "learning_rate": 1.0724113881781222e-05,
+      "loss": 277.5024,
+      "step": 25070
+    },
+    {
+      "epoch": 0.48235175328515584,
+      "grad_norm": 2043.7565981906453,
+      "learning_rate": 1.0718026626364902e-05,
+      "loss": 288.158,
+      "step": 25080
+    },
+    {
+      "epoch": 0.48254407854563636,
+      "grad_norm": 1748.9083129201,
+      "learning_rate": 1.0711939103495515e-05,
+      "loss": 276.0483,
+      "step": 25090
+    },
+    {
+      "epoch": 0.48273640380611693,
+      "grad_norm": 1934.4530760864109,
+      "learning_rate": 1.0705851315440575e-05,
+      "loss": 278.8729,
+      "step": 25100
+    },
+    {
+      "epoch": 0.48292872906659745,
+      "grad_norm": 1777.035986055601,
+      "learning_rate": 1.0699763264467675e-05,
+      "loss": 277.8787,
+      "step": 25110
+    },
+    {
+      "epoch": 0.48312105432707797,
+      "grad_norm": 1881.9017495627606,
+      "learning_rate": 1.069367495284452e-05,
+      "loss": 276.2537,
+      "step": 25120
+    },
+    {
+      "epoch": 0.4833133795875585,
+      "grad_norm": 1664.6538773127083,
+      "learning_rate": 1.06875863828389e-05,
+      "loss": 267.9381,
+      "step": 25130
+    },
+    {
+      "epoch": 0.483505704848039,
+      "grad_norm": 1815.7195932497489,
+      "learning_rate": 1.0681497556718706e-05,
+      "loss": 264.0365,
+      "step": 25140
+    },
+    {
+      "epoch": 0.4836980301085195,
+      "grad_norm": 1719.857512381996,
+      "learning_rate": 1.067540847675193e-05,
+      "loss": 301.7791,
+      "step": 25150
+    },
+    {
+      "epoch": 0.48389035536900005,
+      "grad_norm": 1671.2205614469908,
+      "learning_rate": 1.0669319145206646e-05,
+      "loss": 279.6609,
+      "step": 25160
+    },
+    {
+      "epoch": 0.48408268062948057,
+      "grad_norm": 1991.3580821715914,
+      "learning_rate": 1.066322956435104e-05,
+      "loss": 293.9594,
+      "step": 25170
+    },
+    {
+      "epoch": 0.4842750058899611,
+      "grad_norm": 1797.9206966496258,
+      "learning_rate": 1.0657139736453375e-05,
+      "loss": 287.1266,
+      "step": 25180
+    },
+    {
+      "epoch": 0.4844673311504416,
+      "grad_norm": 1802.0197193014135,
+      "learning_rate": 1.0651049663782007e-05,
+      "loss": 286.4166,
+      "step": 25190
+    },
+    {
+      "epoch": 0.4846596564109221,
+      "grad_norm": 1774.6020667486755,
+      "learning_rate": 1.0644959348605397e-05,
+      "loss": 278.7697,
+      "step": 25200
+    },
+    {
+      "epoch": 0.4848519816714027,
+      "grad_norm": 1821.700872460919,
+      "learning_rate": 1.0638868793192079e-05,
+      "loss": 283.0953,
+      "step": 25210
+    },
+    {
+      "epoch": 0.4850443069318832,
+      "grad_norm": 1807.7977795813345,
+      "learning_rate": 1.0632777999810685e-05,
+      "loss": 270.5381,
+      "step": 25220
+    },
+    {
+      "epoch": 0.48523663219236374,
+      "grad_norm": 1659.3755943189267,
+      "learning_rate": 1.062668697072994e-05,
+      "loss": 276.1015,
+      "step": 25230
+    },
+    {
+      "epoch": 0.48542895745284426,
+      "grad_norm": 1793.1507979005469,
+      "learning_rate": 1.0620595708218646e-05,
+      "loss": 274.4119,
+      "step": 25240
+    },
+    {
+      "epoch": 0.4856212827133248,
+      "grad_norm": 1649.2407056113195,
+      "learning_rate": 1.06145042145457e-05,
+      "loss": 282.1158,
+      "step": 25250
+    },
+    {
+      "epoch": 0.4858136079738053,
+      "grad_norm": 1836.4639436265707,
+      "learning_rate": 1.0608412491980084e-05,
+      "loss": 269.347,
+      "step": 25260
+    },
+    {
+      "epoch": 0.4860059332342858,
+      "grad_norm": 1771.3113262949394,
+      "learning_rate": 1.0602320542790866e-05,
+      "loss": 276.8815,
+      "step": 25270
+    },
+    {
+      "epoch": 0.48619825849476633,
+      "grad_norm": 1880.434198165726,
+      "learning_rate": 1.0596228369247188e-05,
+      "loss": 284.8107,
+      "step": 25280
+    },
+    {
+      "epoch": 0.48639058375524685,
+      "grad_norm": 1716.0271201448497,
+      "learning_rate": 1.0590135973618294e-05,
+      "loss": 273.0092,
+      "step": 25290
+    },
+    {
+      "epoch": 0.48658290901572737,
+      "grad_norm": 1816.324271525453,
+      "learning_rate": 1.0584043358173492e-05,
+      "loss": 275.8546,
+      "step": 25300
+    },
+    {
+      "epoch": 0.48677523427620795,
+      "grad_norm": 1700.7637828142767,
+      "learning_rate": 1.0577950525182189e-05,
+      "loss": 277.8783,
+      "step": 25310
+    },
+    {
+      "epoch": 0.48696755953668847,
+      "grad_norm": 1689.9329002120899,
+      "learning_rate": 1.0571857476913856e-05,
+      "loss": 267.9604,
+      "step": 25320
+    },
+    {
+      "epoch": 0.487159884797169,
+      "grad_norm": 1833.7864871145116,
+      "learning_rate": 1.056576421563806e-05,
+      "loss": 280.6204,
+      "step": 25330
+    },
+    {
+      "epoch": 0.4873522100576495,
+      "grad_norm": 1629.7841715102484,
+      "learning_rate": 1.0559670743624439e-05,
+      "loss": 266.1679,
+      "step": 25340
+    },
+    {
+      "epoch": 0.48754453531813,
+      "grad_norm": 1885.512054481977,
+      "learning_rate": 1.0553577063142705e-05,
+      "loss": 285.2523,
+      "step": 25350
+    },
+    {
+      "epoch": 0.48773686057861054,
+      "grad_norm": 1814.4307838412424,
+      "learning_rate": 1.0547483176462662e-05,
+      "loss": 282.387,
+      "step": 25360
+    },
+    {
+      "epoch": 0.48792918583909106,
+      "grad_norm": 8283.33005281376,
+      "learning_rate": 1.0541389085854177e-05,
+      "loss": 280.0253,
+      "step": 25370
+    },
+    {
+      "epoch": 0.4881215110995716,
+      "grad_norm": 1914.0782489842227,
+      "learning_rate": 1.0535294793587197e-05,
+      "loss": 279.8125,
+      "step": 25380
+    },
+    {
+      "epoch": 0.4883138363600521,
+      "grad_norm": 1697.7411988739045,
+      "learning_rate": 1.0529200301931747e-05,
+      "loss": 267.1816,
+      "step": 25390
+    },
+    {
+      "epoch": 0.4885061616205326,
+      "grad_norm": 1607.8453944014698,
+      "learning_rate": 1.052310561315793e-05,
+      "loss": 275.4394,
+      "step": 25400
+    },
+    {
+      "epoch": 0.4886984868810132,
+      "grad_norm": 1717.7467186318952,
+      "learning_rate": 1.0517010729535903e-05,
+      "loss": 281.75,
+      "step": 25410
+    },
+    {
+      "epoch": 0.4888908121414937,
+      "grad_norm": 1914.089401758374,
+      "learning_rate": 1.0510915653335925e-05,
+      "loss": 273.0778,
+      "step": 25420
+    },
+    {
+      "epoch": 0.48908313740197423,
+      "grad_norm": 1741.6983232242,
+      "learning_rate": 1.05048203868283e-05,
+      "loss": 285.0327,
+      "step": 25430
+    },
+    {
+      "epoch": 0.48927546266245475,
+      "grad_norm": 1975.1225339249038,
+      "learning_rate": 1.0498724932283419e-05,
+      "loss": 275.4808,
+      "step": 25440
+    },
+    {
+      "epoch": 0.48946778792293527,
+      "grad_norm": 1832.3843377850574,
+      "learning_rate": 1.0492629291971738e-05,
+      "loss": 279.7704,
+      "step": 25450
+    },
+    {
+      "epoch": 0.4896601131834158,
+      "grad_norm": 1545.4531884640614,
+      "learning_rate": 1.0486533468163782e-05,
+      "loss": 273.2373,
+      "step": 25460
+    },
+    {
+      "epoch": 0.4898524384438963,
+      "grad_norm": 1555.5081987489202,
+      "learning_rate": 1.0480437463130145e-05,
+      "loss": 270.6621,
+      "step": 25470
+    },
+    {
+      "epoch": 0.49004476370437683,
+      "grad_norm": 1828.9902942376457,
+      "learning_rate": 1.0474341279141486e-05,
+      "loss": 282.0716,
+      "step": 25480
+    },
+    {
+      "epoch": 0.49023708896485735,
+      "grad_norm": 1687.9795908257415,
+      "learning_rate": 1.0468244918468538e-05,
+      "loss": 279.4738,
+      "step": 25490
+    },
+    {
+      "epoch": 0.49042941422533787,
+      "grad_norm": 1769.5628320782544,
+      "learning_rate": 1.0462148383382086e-05,
+      "loss": 273.6405,
+      "step": 25500
+    },
+    {
+      "epoch": 0.49062173948581844,
+      "grad_norm": 1799.4230786330918,
+      "learning_rate": 1.0456051676152996e-05,
+      "loss": 272.1144,
+      "step": 25510
+    },
+    {
+      "epoch": 0.49081406474629896,
+      "grad_norm": 2529.581452200891,
+      "learning_rate": 1.0449954799052189e-05,
+      "loss": 281.015,
+      "step": 25520
+    },
+    {
+      "epoch": 0.4910063900067795,
+      "grad_norm": 2142.8966451075707,
+      "learning_rate": 1.044385775435065e-05,
+      "loss": 283.0552,
+      "step": 25530
+    },
+    {
+      "epoch": 0.49119871526726,
+      "grad_norm": 1722.2710304151983,
+      "learning_rate": 1.043776054431943e-05,
+      "loss": 274.963,
+      "step": 25540
+    },
+    {
+      "epoch": 0.4913910405277405,
+      "grad_norm": 1664.9203377356225,
+      "learning_rate": 1.0431663171229636e-05,
+      "loss": 275.5726,
+      "step": 25550
+    },
+    {
+      "epoch": 0.49158336578822104,
+      "grad_norm": 1857.1865724237516,
+      "learning_rate": 1.0425565637352441e-05,
+      "loss": 278.0228,
+      "step": 25560
+    },
+    {
+      "epoch": 0.49177569104870156,
+      "grad_norm": 1971.1161842338283,
+      "learning_rate": 1.0419467944959072e-05,
+      "loss": 279.5196,
+      "step": 25570
+    },
+    {
+      "epoch": 0.4919680163091821,
+      "grad_norm": 1880.027071619858,
+      "learning_rate": 1.0413370096320823e-05,
+      "loss": 266.9959,
+      "step": 25580
+    },
+    {
+      "epoch": 0.4921603415696626,
+      "grad_norm": 1655.724603171346,
+      "learning_rate": 1.0407272093709038e-05,
+      "loss": 273.6131,
+      "step": 25590
+    },
+    {
+      "epoch": 0.4923526668301431,
+      "grad_norm": 1776.7069566842088,
+      "learning_rate": 1.0401173939395128e-05,
+      "loss": 272.8131,
+      "step": 25600
+    },
+    {
+      "epoch": 0.4925449920906237,
+      "grad_norm": 1844.6656748396456,
+      "learning_rate": 1.0395075635650549e-05,
+      "loss": 270.34,
+      "step": 25610
+    },
+    {
+      "epoch": 0.4927373173511042,
+      "grad_norm": 1572.2616924510403,
+      "learning_rate": 1.038897718474682e-05,
+      "loss": 275.1192,
+      "step": 25620
+    },
+    {
+      "epoch": 0.49292964261158473,
+      "grad_norm": 1788.643043149432,
+      "learning_rate": 1.0382878588955517e-05,
+      "loss": 280.5417,
+      "step": 25630
+    },
+    {
+      "epoch": 0.49312196787206525,
+      "grad_norm": 1636.245284263101,
+      "learning_rate": 1.0376779850548257e-05,
+      "loss": 275.8214,
+      "step": 25640
+    },
+    {
+      "epoch": 0.49331429313254577,
+      "grad_norm": 1752.808939684616,
+      "learning_rate": 1.0370680971796732e-05,
+      "loss": 269.9422,
+      "step": 25650
+    },
+    {
+      "epoch": 0.4935066183930263,
+      "grad_norm": 1747.1700432609318,
+      "learning_rate": 1.0364581954972662e-05,
+      "loss": 269.1088,
+      "step": 25660
+    },
+    {
+      "epoch": 0.4936989436535068,
+      "grad_norm": 1830.7041813039798,
+      "learning_rate": 1.0358482802347838e-05,
+      "loss": 277.7571,
+      "step": 25670
+    },
+    {
+      "epoch": 0.4938912689139873,
+      "grad_norm": 2036.4426659652127,
+      "learning_rate": 1.0352383516194088e-05,
+      "loss": 277.2694,
+      "step": 25680
+    },
+    {
+      "epoch": 0.49408359417446784,
+      "grad_norm": 1772.4798752197973,
+      "learning_rate": 1.0346284098783304e-05,
+      "loss": 276.9773,
+      "step": 25690
+    },
+    {
+      "epoch": 0.49427591943494836,
+      "grad_norm": 1604.2219490851046,
+      "learning_rate": 1.0340184552387406e-05,
+      "loss": 274.673,
+      "step": 25700
+    },
+    {
+      "epoch": 0.4944682446954289,
+      "grad_norm": 1990.425648617706,
+      "learning_rate": 1.0334084879278381e-05,
+      "loss": 275.3094,
+      "step": 25710
+    },
+    {
+      "epoch": 0.49466056995590946,
+      "grad_norm": 1580.7324982115563,
+      "learning_rate": 1.032798508172826e-05,
+      "loss": 275.8235,
+      "step": 25720
+    },
+    {
+      "epoch": 0.49485289521639,
+      "grad_norm": 1580.1661572774167,
+      "learning_rate": 1.0321885162009111e-05,
+      "loss": 266.4813,
+      "step": 25730
+    },
+    {
+      "epoch": 0.4950452204768705,
+      "grad_norm": 1664.8125716677737,
+      "learning_rate": 1.0315785122393053e-05,
+      "loss": 270.8755,
+      "step": 25740
+    },
+    {
+      "epoch": 0.495237545737351,
+      "grad_norm": 1913.6285760673086,
+      "learning_rate": 1.0309684965152254e-05,
+      "loss": 278.3399,
+      "step": 25750
+    },
+    {
+      "epoch": 0.49542987099783153,
+      "grad_norm": 1672.6336172762456,
+      "learning_rate": 1.030358469255892e-05,
+      "loss": 279.4013,
+      "step": 25760
+    },
+    {
+      "epoch": 0.49562219625831205,
+      "grad_norm": 1894.4729594011826,
+      "learning_rate": 1.0297484306885304e-05,
+      "loss": 273.5778,
+      "step": 25770
+    },
+    {
+      "epoch": 0.4958145215187926,
+      "grad_norm": 1486.428909526691,
+      "learning_rate": 1.0291383810403697e-05,
+      "loss": 274.573,
+      "step": 25780
+    },
+    {
+      "epoch": 0.4960068467792731,
+      "grad_norm": 1636.4533208711755,
+      "learning_rate": 1.028528320538643e-05,
+      "loss": 276.0129,
+      "step": 25790
+    },
+    {
+      "epoch": 0.4961991720397536,
+      "grad_norm": 1735.0772739518704,
+      "learning_rate": 1.0279182494105879e-05,
+      "loss": 269.5101,
+      "step": 25800
+    },
+    {
+      "epoch": 0.49639149730023413,
+      "grad_norm": 1657.1296433778853,
+      "learning_rate": 1.0273081678834462e-05,
+      "loss": 269.3597,
+      "step": 25810
+    },
+    {
+      "epoch": 0.4965838225607147,
+      "grad_norm": 1677.74422572379,
+      "learning_rate": 1.026698076184463e-05,
+      "loss": 267.2675,
+      "step": 25820
+    },
+    {
+      "epoch": 0.4967761478211952,
+      "grad_norm": 1723.5692957252045,
+      "learning_rate": 1.0260879745408876e-05,
+      "loss": 276.1706,
+      "step": 25830
+    },
+    {
+      "epoch": 0.49696847308167574,
+      "grad_norm": 1980.9613986069344,
+      "learning_rate": 1.0254778631799722e-05,
+      "loss": 265.838,
+      "step": 25840
+    },
+    {
+      "epoch": 0.49716079834215626,
+      "grad_norm": 1712.3810455210344,
+      "learning_rate": 1.024867742328974e-05,
+      "loss": 280.8373,
+      "step": 25850
+    },
+    {
+      "epoch": 0.4973531236026368,
+      "grad_norm": 1844.5937338169515,
+      "learning_rate": 1.024257612215152e-05,
+      "loss": 276.4784,
+      "step": 25860
+    },
+    {
+      "epoch": 0.4975454488631173,
+      "grad_norm": 1770.8577750523227,
+      "learning_rate": 1.02364747306577e-05,
+      "loss": 276.7522,
+      "step": 25870
+    },
+    {
+      "epoch": 0.4977377741235978,
+      "grad_norm": 1938.988621651512,
+      "learning_rate": 1.023037325108095e-05,
+      "loss": 272.975,
+      "step": 25880
+    },
+    {
+      "epoch": 0.49793009938407834,
+      "grad_norm": 1811.7810293825148,
+      "learning_rate": 1.022427168569397e-05,
+      "loss": 268.9581,
+      "step": 25890
+    },
+    {
+      "epoch": 0.49812242464455886,
+      "grad_norm": 1591.4568115613229,
+      "learning_rate": 1.021817003676949e-05,
+      "loss": 275.2877,
+      "step": 25900
+    },
+    {
+      "epoch": 0.4983147499050394,
+      "grad_norm": 1962.51063734328,
+      "learning_rate": 1.0212068306580274e-05,
+      "loss": 279.2284,
+      "step": 25910
+    },
+    {
+      "epoch": 0.49850707516551995,
+      "grad_norm": 1689.4127112193344,
+      "learning_rate": 1.0205966497399118e-05,
+      "loss": 263.6699,
+      "step": 25920
+    },
+    {
+      "epoch": 0.49869940042600047,
+      "grad_norm": 1880.813468352572,
+      "learning_rate": 1.0199864611498841e-05,
+      "loss": 279.2016,
+      "step": 25930
+    },
+    {
+      "epoch": 0.498891725686481,
+      "grad_norm": 1607.8035022145914,
+      "learning_rate": 1.0193762651152299e-05,
+      "loss": 279.518,
+      "step": 25940
+    },
+    {
+      "epoch": 0.4990840509469615,
+      "grad_norm": 1864.2411447773225,
+      "learning_rate": 1.018766061863237e-05,
+      "loss": 271.3069,
+      "step": 25950
+    },
+    {
+      "epoch": 0.49927637620744203,
+      "grad_norm": 2091.9565295944117,
+      "learning_rate": 1.018155851621196e-05,
+      "loss": 273.8984,
+      "step": 25960
+    },
+    {
+      "epoch": 0.49946870146792255,
+      "grad_norm": 1964.398008919894,
+      "learning_rate": 1.0175456346164e-05,
+      "loss": 270.9842,
+      "step": 25970
+    },
+    {
+      "epoch": 0.49966102672840307,
+      "grad_norm": 1672.5696589957342,
+      "learning_rate": 1.0169354110761447e-05,
+      "loss": 271.7555,
+      "step": 25980
+    },
+    {
+      "epoch": 0.4998533519888836,
+      "grad_norm": 1513.4552497496738,
+      "learning_rate": 1.0163251812277289e-05,
+      "loss": 269.53,
+      "step": 25990
+    },
+    {
+      "epoch": 0.5000456772493641,
+      "grad_norm": 1524.6941834301404,
+      "learning_rate": 1.0157149452984523e-05,
+      "loss": 272.5135,
+      "step": 26000
+    },
+    {
+      "epoch": 0.5002380025098446,
+      "grad_norm": 1683.4352601114072,
+      "learning_rate": 1.0151047035156182e-05,
+      "loss": 273.8935,
+      "step": 26010
+    },
+    {
+      "epoch": 0.5004303277703251,
+      "grad_norm": 2080.4936321093123,
+      "learning_rate": 1.0144944561065316e-05,
+      "loss": 270.1687,
+      "step": 26020
+    },
+    {
+      "epoch": 0.5006226530308057,
+      "grad_norm": 1632.5934581019233,
+      "learning_rate": 1.0138842032984996e-05,
+      "loss": 286.5729,
+      "step": 26030
+    },
+    {
+      "epoch": 0.5008149782912862,
+      "grad_norm": 1680.7945305414196,
+      "learning_rate": 1.013273945318831e-05,
+      "loss": 279.0429,
+      "step": 26040
+    },
+    {
+      "epoch": 0.5010073035517667,
+      "grad_norm": 1664.2913124351896,
+      "learning_rate": 1.0126636823948373e-05,
+      "loss": 268.6135,
+      "step": 26050
+    },
+    {
+      "epoch": 0.5011996288122472,
+      "grad_norm": 1848.315319928652,
+      "learning_rate": 1.0120534147538305e-05,
+      "loss": 276.1851,
+      "step": 26060
+    },
+    {
+      "epoch": 0.5013919540727279,
+      "grad_norm": 1558.6579765155225,
+      "learning_rate": 1.011443142623126e-05,
+      "loss": 280.8799,
+      "step": 26070
+    },
+    {
+      "epoch": 0.5015842793332084,
+      "grad_norm": 1793.5497181113903,
+      "learning_rate": 1.0108328662300399e-05,
+      "loss": 271.3325,
+      "step": 26080
+    },
+    {
+      "epoch": 0.5017766045936889,
+      "grad_norm": 2003.0828263051126,
+      "learning_rate": 1.0102225858018902e-05,
+      "loss": 266.1945,
+      "step": 26090
+    },
+    {
+      "epoch": 0.5019689298541694,
+      "grad_norm": 1789.1073389557582,
+      "learning_rate": 1.009612301565996e-05,
+      "loss": 274.8918,
+      "step": 26100
+    },
+    {
+      "epoch": 0.5021612551146499,
+      "grad_norm": 1603.861578608274,
+      "learning_rate": 1.0090020137496783e-05,
+      "loss": 269.6036,
+      "step": 26110
+    },
+    {
+      "epoch": 0.5023535803751304,
+      "grad_norm": 2100.0472697596374,
+      "learning_rate": 1.008391722580259e-05,
+      "loss": 283.053,
+      "step": 26120
+    },
+    {
+      "epoch": 0.502545905635611,
+      "grad_norm": 1780.7006841160278,
+      "learning_rate": 1.0077814282850617e-05,
+      "loss": 290.6751,
+      "step": 26130
+    },
+    {
+      "epoch": 0.5027382308960915,
+      "grad_norm": 1702.6634408727787,
+      "learning_rate": 1.0071711310914111e-05,
+      "loss": 273.0348,
+      "step": 26140
+    },
+    {
+      "epoch": 0.502930556156572,
+      "grad_norm": 1692.7111133172946,
+      "learning_rate": 1.0065608312266324e-05,
+      "loss": 275.5359,
+      "step": 26150
+    },
+    {
+      "epoch": 0.5031228814170525,
+      "grad_norm": 1798.8909913614705,
+      "learning_rate": 1.005950528918052e-05,
+      "loss": 274.5655,
+      "step": 26160
+    },
+    {
+      "epoch": 0.503315206677533,
+      "grad_norm": 1733.10211000895,
+      "learning_rate": 1.0053402243929986e-05,
+      "loss": 280.753,
+      "step": 26170
+    },
+    {
+      "epoch": 0.5035075319380136,
+      "grad_norm": 1902.6321891726207,
+      "learning_rate": 1.0047299178787993e-05,
+      "loss": 274.4896,
+      "step": 26180
+    },
+    {
+      "epoch": 0.5036998571984941,
+      "grad_norm": 1700.5182253775538,
+      "learning_rate": 1.0041196096027841e-05,
+      "loss": 269.7237,
+      "step": 26190
+    },
+    {
+      "epoch": 0.5038921824589746,
+      "grad_norm": 1635.4956676626955,
+      "learning_rate": 1.003509299792282e-05,
+      "loss": 271.1262,
+      "step": 26200
+    },
+    {
+      "epoch": 0.5040845077194551,
+      "grad_norm": 1752.5666472258445,
+      "learning_rate": 1.0028989886746241e-05,
+      "loss": 262.8223,
+      "step": 26210
+    },
+    {
+      "epoch": 0.5042768329799356,
+      "grad_norm": 1784.563574089057,
+      "learning_rate": 1.0022886764771405e-05,
+      "loss": 270.4692,
+      "step": 26220
+    },
+    {
+      "epoch": 0.5044691582404162,
+      "grad_norm": 1870.167366417679,
+      "learning_rate": 1.0016783634271626e-05,
+      "loss": 269.1117,
+      "step": 26230
+    },
+    {
+      "epoch": 0.5046614835008967,
+      "grad_norm": 1729.0544402515077,
+      "learning_rate": 1.001068049752022e-05,
+      "loss": 272.5295,
+      "step": 26240
+    },
+    {
+      "epoch": 0.5048538087613772,
+      "grad_norm": 1643.144669762963,
+      "learning_rate": 1.0004577356790506e-05,
+      "loss": 268.1054,
+      "step": 26250
+    },
+    {
+      "epoch": 0.5050461340218577,
+      "grad_norm": 1740.6069494075255,
+      "learning_rate": 9.998474214355805e-06,
+      "loss": 271.8513,
+      "step": 26260
+    },
+    {
+      "epoch": 0.5052384592823382,
+      "grad_norm": 1992.7094120220727,
+      "learning_rate": 9.992371072489434e-06,
+      "loss": 278.6317,
+      "step": 26270
+    },
+    {
+      "epoch": 0.5054307845428189,
+      "grad_norm": 1708.1682303439318,
+      "learning_rate": 9.986267933464707e-06,
+      "loss": 281.2483,
+      "step": 26280
+    },
+    {
+      "epoch": 0.5056231098032994,
+      "grad_norm": 1647.0960935851654,
+      "learning_rate": 9.980164799554953e-06,
+      "loss": 277.364,
+      "step": 26290
+    },
+    {
+      "epoch": 0.5058154350637799,
+      "grad_norm": 1831.5940511449849,
+      "learning_rate": 9.97406167303348e-06,
+      "loss": 272.5185,
+      "step": 26300
+    },
+    {
+      "epoch": 0.5060077603242604,
+      "grad_norm": 1850.4748919082288,
+      "learning_rate": 9.967958556173612e-06,
+      "loss": 275.872,
+      "step": 26310
+    },
+    {
+      "epoch": 0.5062000855847409,
+      "grad_norm": 1981.073526222446,
+      "learning_rate": 9.961855451248645e-06,
+      "loss": 267.4932,
+      "step": 26320
+    },
+    {
+      "epoch": 0.5063924108452215,
+      "grad_norm": 1713.9122190963226,
+      "learning_rate": 9.955752360531896e-06,
+      "loss": 279.3994,
+      "step": 26330
+    },
+    {
+      "epoch": 0.506584736105702,
+      "grad_norm": 1988.6807078523354,
+      "learning_rate": 9.949649286296663e-06,
+      "loss": 272.955,
+      "step": 26340
+    },
+    {
+      "epoch": 0.5067770613661825,
+      "grad_norm": 1795.7312095104523,
+      "learning_rate": 9.943546230816236e-06,
+      "loss": 272.6052,
+      "step": 26350
+    },
+    {
+      "epoch": 0.506969386626663,
+      "grad_norm": 1891.0151326283483,
+      "learning_rate": 9.937443196363908e-06,
+      "loss": 279.6853,
+      "step": 26360
+    },
+    {
+      "epoch": 0.5071617118871435,
+      "grad_norm": 1702.973385216345,
+      "learning_rate": 9.931340185212955e-06,
+      "loss": 265.7823,
+      "step": 26370
+    },
+    {
+      "epoch": 0.5073540371476241,
+      "grad_norm": 1795.5546094161111,
+      "learning_rate": 9.925237199636652e-06,
+      "loss": 264.735,
+      "step": 26380
+    },
+    {
+      "epoch": 0.5075463624081046,
+      "grad_norm": 1740.8570507590339,
+      "learning_rate": 9.919134241908252e-06,
+      "loss": 269.8005,
+      "step": 26390
+    },
+    {
+      "epoch": 0.5077386876685851,
+      "grad_norm": 1863.9503903518575,
+      "learning_rate": 9.913031314301016e-06,
+      "loss": 278.6392,
+      "step": 26400
+    },
+    {
+      "epoch": 0.5079310129290656,
+      "grad_norm": 1638.866319568942,
+      "learning_rate": 9.906928419088178e-06,
+      "loss": 266.4949,
+      "step": 26410
+    },
+    {
+      "epoch": 0.5081233381895461,
+      "grad_norm": 1886.6857498121033,
+      "learning_rate": 9.900825558542965e-06,
+      "loss": 276.7004,
+      "step": 26420
+    },
+    {
+      "epoch": 0.5083156634500267,
+      "grad_norm": 1782.98923090286,
+      "learning_rate": 9.894722734938595e-06,
+      "loss": 282.0657,
+      "step": 26430
+    },
+    {
+      "epoch": 0.5085079887105072,
+      "grad_norm": 2400.2128392775367,
+      "learning_rate": 9.888619950548267e-06,
+      "loss": 269.9959,
+      "step": 26440
+    },
+    {
+      "epoch": 0.5087003139709877,
+      "grad_norm": 1693.9313596502593,
+      "learning_rate": 9.88251720764517e-06,
+      "loss": 265.1197,
+      "step": 26450
+    },
+    {
+      "epoch": 0.5088926392314682,
+      "grad_norm": 1729.9994544333172,
+      "learning_rate": 9.876414508502468e-06,
+      "loss": 270.7896,
+      "step": 26460
+    },
+    {
+      "epoch": 0.5090849644919487,
+      "grad_norm": 1742.3993622554844,
+      "learning_rate": 9.870311855393324e-06,
+      "loss": 275.8884,
+      "step": 26470
+    },
+    {
+      "epoch": 0.5092772897524294,
+      "grad_norm": 1715.8933194161416,
+      "learning_rate": 9.864209250590875e-06,
+      "loss": 263.0368,
+      "step": 26480
+    },
+    {
+      "epoch": 0.5094696150129099,
+      "grad_norm": 1640.5133977051426,
+      "learning_rate": 9.858106696368235e-06,
+      "loss": 274.8562,
+      "step": 26490
+    },
+    {
+      "epoch": 0.5096619402733904,
+      "grad_norm": 1588.4028118143278,
+      "learning_rate": 9.852004194998503e-06,
+      "loss": 268.0271,
+      "step": 26500
+    },
+    {
+      "epoch": 0.5098542655338709,
+      "grad_norm": 1600.8963906933461,
+      "learning_rate": 9.845901748754767e-06,
+      "loss": 268.6611,
+      "step": 26510
+    },
+    {
+      "epoch": 0.5100465907943514,
+      "grad_norm": 1681.4632120338292,
+      "learning_rate": 9.839799359910087e-06,
+      "loss": 273.9835,
+      "step": 26520
+    },
+    {
+      "epoch": 0.510238916054832,
+      "grad_norm": 1637.923273314008,
+      "learning_rate": 9.833697030737495e-06,
+      "loss": 283.8083,
+      "step": 26530
+    },
+    {
+      "epoch": 0.5104312413153125,
+      "grad_norm": 1714.3455707896221,
+      "learning_rate": 9.827594763510016e-06,
+      "loss": 275.4779,
+      "step": 26540
+    },
+    {
+      "epoch": 0.510623566575793,
+      "grad_norm": 1747.434790453476,
+      "learning_rate": 9.821492560500641e-06,
+      "loss": 268.2246,
+      "step": 26550
+    },
+    {
+      "epoch": 0.5108158918362735,
+      "grad_norm": 1592.2270746319073,
+      "learning_rate": 9.815390423982339e-06,
+      "loss": 270.3903,
+      "step": 26560
+    },
+    {
+      "epoch": 0.511008217096754,
+      "grad_norm": 1742.4783152382627,
+      "learning_rate": 9.809288356228051e-06,
+      "loss": 268.1029,
+      "step": 26570
+    },
+    {
+      "epoch": 0.5112005423572346,
+      "grad_norm": 1671.141526900796,
+      "learning_rate": 9.803186359510701e-06,
+      "loss": 268.3448,
+      "step": 26580
+    },
+    {
+      "epoch": 0.5113928676177151,
+      "grad_norm": 1570.6436232196118,
+      "learning_rate": 9.797084436103186e-06,
+      "loss": 271.0514,
+      "step": 26590
+    },
+    {
+      "epoch": 0.5115851928781956,
+      "grad_norm": 1693.8866383781922,
+      "learning_rate": 9.790982588278362e-06,
+      "loss": 273.5804,
+      "step": 26600
+    },
+    {
+      "epoch": 0.5117775181386761,
+      "grad_norm": 2046.825733085198,
+      "learning_rate": 9.784880818309076e-06,
+      "loss": 268.9866,
+      "step": 26610
+    },
+    {
+      "epoch": 0.5119698433991566,
+      "grad_norm": 1754.154892170759,
+      "learning_rate": 9.778779128468133e-06,
+      "loss": 288.0217,
+      "step": 26620
+    },
+    {
+      "epoch": 0.5121621686596372,
+      "grad_norm": 1762.6089269051984,
+      "learning_rate": 9.77267752102831e-06,
+      "loss": 271.5234,
+      "step": 26630
+    },
+    {
+      "epoch": 0.5123544939201177,
+      "grad_norm": 1568.8738217503458,
+      "learning_rate": 9.766575998262353e-06,
+      "loss": 266.7213,
+      "step": 26640
+    },
+    {
+      "epoch": 0.5125468191805982,
+      "grad_norm": 1547.7107254803711,
+      "learning_rate": 9.760474562442984e-06,
+      "loss": 285.0681,
+      "step": 26650
+    },
+    {
+      "epoch": 0.5127391444410787,
+      "grad_norm": 1516.5694113966097,
+      "learning_rate": 9.754373215842884e-06,
+      "loss": 272.9505,
+      "step": 26660
+    },
+    {
+      "epoch": 0.5129314697015592,
+      "grad_norm": 1885.127714472673,
+      "learning_rate": 9.748271960734708e-06,
+      "loss": 279.3075,
+      "step": 26670
+    },
+    {
+      "epoch": 0.5131237949620397,
+      "grad_norm": 1704.6934684909331,
+      "learning_rate": 9.742170799391063e-06,
+      "loss": 262.489,
+      "step": 26680
+    },
+    {
+      "epoch": 0.5133161202225204,
+      "grad_norm": 1565.6576422339674,
+      "learning_rate": 9.736069734084541e-06,
+      "loss": 261.1578,
+      "step": 26690
+    },
+    {
+      "epoch": 0.5135084454830009,
+      "grad_norm": 1626.064689973387,
+      "learning_rate": 9.729968767087685e-06,
+      "loss": 267.2632,
+      "step": 26700
+    },
+    {
+      "epoch": 0.5137007707434814,
+      "grad_norm": 1708.6445641992425,
+      "learning_rate": 9.723867900673e-06,
+      "loss": 263.0513,
+      "step": 26710
+    },
+    {
+      "epoch": 0.5138930960039619,
+      "grad_norm": 1662.7940086333215,
+      "learning_rate": 9.717767137112964e-06,
+      "loss": 270.2977,
+      "step": 26720
+    },
+    {
+      "epoch": 0.5140854212644425,
+      "grad_norm": 1692.3392392486355,
+      "learning_rate": 9.711666478680007e-06,
+      "loss": 275.7062,
+      "step": 26730
+    },
+    {
+      "epoch": 0.514277746524923,
+      "grad_norm": 1664.5381189162295,
+      "learning_rate": 9.705565927646526e-06,
+      "loss": 268.1979,
+      "step": 26740
+    },
+    {
+      "epoch": 0.5144700717854035,
+      "grad_norm": 1777.8573271768778,
+      "learning_rate": 9.699465486284871e-06,
+      "loss": 268.7613,
+      "step": 26750
+    },
+    {
+      "epoch": 0.514662397045884,
+      "grad_norm": 1745.6112233650072,
+      "learning_rate": 9.693365156867363e-06,
+      "loss": 263.1743,
+      "step": 26760
+    },
+    {
+      "epoch": 0.5148547223063645,
+      "grad_norm": 1612.7770271871843,
+      "learning_rate": 9.68726494166627e-06,
+      "loss": 265.2632,
+      "step": 26770
+    },
+    {
+      "epoch": 0.515047047566845,
+      "grad_norm": 1763.8031608827034,
+      "learning_rate": 9.681164842953816e-06,
+      "loss": 268.9233,
+      "step": 26780
+    },
+    {
+      "epoch": 0.5152393728273256,
+      "grad_norm": 1690.798057848904,
+      "learning_rate": 9.675064863002196e-06,
+      "loss": 270.2367,
+      "step": 26790
+    },
+    {
+      "epoch": 0.5154316980878061,
+      "grad_norm": 1737.4487933297435,
+      "learning_rate": 9.668965004083549e-06,
+      "loss": 275.0438,
+      "step": 26800
+    },
+    {
+      "epoch": 0.5156240233482866,
+      "grad_norm": 1887.430022879798,
+      "learning_rate": 9.66286526846997e-06,
+      "loss": 272.8342,
+      "step": 26810
+    },
+    {
+      "epoch": 0.5158163486087671,
+      "grad_norm": 1691.1122652435636,
+      "learning_rate": 9.656765658433507e-06,
+      "loss": 265.7619,
+      "step": 26820
+    },
+    {
+      "epoch": 0.5160086738692476,
+      "grad_norm": 1991.6050596638497,
+      "learning_rate": 9.650666176246171e-06,
+      "loss": 268.2168,
+      "step": 26830
+    },
+    {
+      "epoch": 0.5162009991297282,
+      "grad_norm": 1803.3879335031136,
+      "learning_rate": 9.644566824179916e-06,
+      "loss": 262.2074,
+      "step": 26840
+    },
+    {
+      "epoch": 0.5163933243902087,
+      "grad_norm": 1669.6728527011023,
+      "learning_rate": 9.638467604506648e-06,
+      "loss": 271.3996,
+      "step": 26850
+    },
+    {
+      "epoch": 0.5165856496506892,
+      "grad_norm": 1532.422202290953,
+      "learning_rate": 9.632368519498224e-06,
+      "loss": 266.5838,
+      "step": 26860
+    },
+    {
+      "epoch": 0.5167779749111697,
+      "grad_norm": 1746.4595464416727,
+      "learning_rate": 9.626269571426456e-06,
+      "loss": 276.9618,
+      "step": 26870
+    },
+    {
+      "epoch": 0.5169703001716502,
+      "grad_norm": 1552.2117536956355,
+      "learning_rate": 9.620170762563103e-06,
+      "loss": 262.0165,
+      "step": 26880
+    },
+    {
+      "epoch": 0.5171626254321309,
+      "grad_norm": 1611.71030835006,
+      "learning_rate": 9.614072095179862e-06,
+      "loss": 267.8775,
+      "step": 26890
+    },
+    {
+      "epoch": 0.5173549506926114,
+      "grad_norm": 1793.0356958685372,
+      "learning_rate": 9.607973571548396e-06,
+      "loss": 277.4816,
+      "step": 26900
+    },
+    {
+      "epoch": 0.5175472759530919,
+      "grad_norm": 1694.433011116257,
+      "learning_rate": 9.601875193940301e-06,
+      "loss": 261.8774,
+      "step": 26910
+    },
+    {
+      "epoch": 0.5177396012135724,
+      "grad_norm": 1742.9504962111776,
+      "learning_rate": 9.595776964627119e-06,
+      "loss": 264.9292,
+      "step": 26920
+    },
+    {
+      "epoch": 0.517931926474053,
+      "grad_norm": 1567.7765524754527,
+      "learning_rate": 9.589678885880342e-06,
+      "loss": 271.9219,
+      "step": 26930
+    },
+    {
+      "epoch": 0.5181242517345335,
+      "grad_norm": 1795.1615088469457,
+      "learning_rate": 9.583580959971402e-06,
+      "loss": 268.8843,
+      "step": 26940
+    },
+    {
+      "epoch": 0.518316576995014,
+      "grad_norm": 1707.4568846423876,
+      "learning_rate": 9.577483189171681e-06,
+      "loss": 275.2421,
+      "step": 26950
+    },
+    {
+      "epoch": 0.5185089022554945,
+      "grad_norm": 1689.4044028785813,
+      "learning_rate": 9.571385575752487e-06,
+      "loss": 263.9782,
+      "step": 26960
+    },
+    {
+      "epoch": 0.518701227515975,
+      "grad_norm": 1892.2200129508271,
+      "learning_rate": 9.565288121985093e-06,
+      "loss": 276.4928,
+      "step": 26970
+    },
+    {
+      "epoch": 0.5188935527764555,
+      "grad_norm": 1697.2964052702898,
+      "learning_rate": 9.559190830140695e-06,
+      "loss": 272.9482,
+      "step": 26980
+    },
+    {
+      "epoch": 0.5190858780369361,
+      "grad_norm": 1695.1652072817276,
+      "learning_rate": 9.553093702490433e-06,
+      "loss": 264.6189,
+      "step": 26990
+    },
+    {
+      "epoch": 0.5192782032974166,
+      "grad_norm": 1700.957498664333,
+      "learning_rate": 9.54699674130538e-06,
+      "loss": 266.9093,
+      "step": 27000
+    },
+    {
+      "epoch": 0.5194705285578971,
+      "grad_norm": 1644.3914203519191,
+      "learning_rate": 9.540899948856561e-06,
+      "loss": 259.7535,
+      "step": 27010
+    },
+    {
+      "epoch": 0.5196628538183776,
+      "grad_norm": 1505.4662412764394,
+      "learning_rate": 9.534803327414931e-06,
+      "loss": 272.6865,
+      "step": 27020
+    },
+    {
+      "epoch": 0.5198551790788581,
+      "grad_norm": 1612.389307080163,
+      "learning_rate": 9.52870687925138e-06,
+      "loss": 269.3465,
+      "step": 27030
+    },
+    {
+      "epoch": 0.5200475043393387,
+      "grad_norm": 1757.804421178999,
+      "learning_rate": 9.522610606636728e-06,
+      "loss": 270.4947,
+      "step": 27040
+    },
+    {
+      "epoch": 0.5202398295998192,
+      "grad_norm": 1679.9345432220853,
+      "learning_rate": 9.516514511841745e-06,
+      "loss": 270.8916,
+      "step": 27050
+    },
+    {
+      "epoch": 0.5204321548602997,
+      "grad_norm": 1642.4428909824912,
+      "learning_rate": 9.51041859713712e-06,
+      "loss": 258.8593,
+      "step": 27060
+    },
+    {
+      "epoch": 0.5206244801207802,
+      "grad_norm": 1735.534557700401,
+      "learning_rate": 9.50432286479348e-06,
+      "loss": 266.2346,
+      "step": 27070
+    },
+    {
+      "epoch": 0.5208168053812607,
+      "grad_norm": 1812.4163287774832,
+      "learning_rate": 9.498227317081387e-06,
+      "loss": 268.837,
+      "step": 27080
+    },
+    {
+      "epoch": 0.5210091306417413,
+      "grad_norm": 1964.8421362879276,
+      "learning_rate": 9.492131956271334e-06,
+      "loss": 267.5575,
+      "step": 27090
+    },
+    {
+      "epoch": 0.5212014559022219,
+      "grad_norm": 1825.0592922536978,
+      "learning_rate": 9.486036784633738e-06,
+      "loss": 269.7385,
+      "step": 27100
+    },
+    {
+      "epoch": 0.5213937811627024,
+      "grad_norm": 1655.977637532272,
+      "learning_rate": 9.47994180443895e-06,
+      "loss": 262.6288,
+      "step": 27110
+    },
+    {
+      "epoch": 0.5215861064231829,
+      "grad_norm": 1753.050934234177,
+      "learning_rate": 9.473847017957254e-06,
+      "loss": 267.9633,
+      "step": 27120
+    },
+    {
+      "epoch": 0.5217784316836634,
+      "grad_norm": 1665.7673290698224,
+      "learning_rate": 9.467752427458851e-06,
+      "loss": 269.7881,
+      "step": 27130
+    },
+    {
+      "epoch": 0.521970756944144,
+      "grad_norm": 1700.701063654924,
+      "learning_rate": 9.461658035213878e-06,
+      "loss": 266.743,
+      "step": 27140
+    },
+    {
+      "epoch": 0.5221630822046245,
+      "grad_norm": 1630.676990055864,
+      "learning_rate": 9.455563843492397e-06,
+      "loss": 269.311,
+      "step": 27150
+    },
+    {
+      "epoch": 0.522355407465105,
+      "grad_norm": 1798.3596045117974,
+      "learning_rate": 9.449469854564393e-06,
+      "loss": 262.1842,
+      "step": 27160
+    },
+    {
+      "epoch": 0.5225477327255855,
+      "grad_norm": 1506.0457521575593,
+      "learning_rate": 9.443376070699778e-06,
+      "loss": 270.4075,
+      "step": 27170
+    },
+    {
+      "epoch": 0.522740057986066,
+      "grad_norm": 1784.7343947685292,
+      "learning_rate": 9.437282494168379e-06,
+      "loss": 264.4419,
+      "step": 27180
+    },
+    {
+      "epoch": 0.5229323832465466,
+      "grad_norm": 1702.8371620709388,
+      "learning_rate": 9.431189127239962e-06,
+      "loss": 276.2573,
+      "step": 27190
+    },
+    {
+      "epoch": 0.5231247085070271,
+      "grad_norm": 1666.6848796432125,
+      "learning_rate": 9.4250959721842e-06,
+      "loss": 264.0803,
+      "step": 27200
+    },
+    {
+      "epoch": 0.5233170337675076,
+      "grad_norm": 1813.2448375008926,
+      "learning_rate": 9.419003031270692e-06,
+      "loss": 267.787,
+      "step": 27210
+    },
+    {
+      "epoch": 0.5235093590279881,
+      "grad_norm": 1609.1142744358856,
+      "learning_rate": 9.412910306768959e-06,
+      "loss": 261.8443,
+      "step": 27220
+    },
+    {
+      "epoch": 0.5237016842884686,
+      "grad_norm": 1924.4247481088046,
+      "learning_rate": 9.40681780094844e-06,
+      "loss": 260.3859,
+      "step": 27230
+    },
+    {
+      "epoch": 0.5238940095489492,
+      "grad_norm": 1674.2312576088304,
+      "learning_rate": 9.400725516078496e-06,
+      "loss": 266.449,
+      "step": 27240
+    },
+    {
+      "epoch": 0.5240863348094297,
+      "grad_norm": 1722.3043112679425,
+      "learning_rate": 9.394633454428396e-06,
+      "loss": 273.4871,
+      "step": 27250
+    },
+    {
+      "epoch": 0.5242786600699102,
+      "grad_norm": 1522.6320537332201,
+      "learning_rate": 9.388541618267341e-06,
+      "loss": 270.7666,
+      "step": 27260
+    },
+    {
+      "epoch": 0.5244709853303907,
+      "grad_norm": 1659.1718924580953,
+      "learning_rate": 9.382450009864434e-06,
+      "loss": 271.6879,
+      "step": 27270
+    },
+    {
+      "epoch": 0.5246633105908712,
+      "grad_norm": 1570.790918449188,
+      "learning_rate": 9.376358631488697e-06,
+      "loss": 257.0079,
+      "step": 27280
+    },
+    {
+      "epoch": 0.5248556358513518,
+      "grad_norm": 1573.7836391516046,
+      "learning_rate": 9.37026748540907e-06,
+      "loss": 270.6402,
+      "step": 27290
+    },
+    {
+      "epoch": 0.5250479611118324,
+      "grad_norm": 1642.2766486115481,
+      "learning_rate": 9.364176573894404e-06,
+      "loss": 260.5158,
+      "step": 27300
+    },
+    {
+      "epoch": 0.5252402863723129,
+      "grad_norm": 1835.4146849417075,
+      "learning_rate": 9.358085899213467e-06,
+      "loss": 267.7711,
+      "step": 27310
+    },
+    {
+      "epoch": 0.5254326116327934,
+      "grad_norm": 1719.6920763262058,
+      "learning_rate": 9.351995463634925e-06,
+      "loss": 266.0158,
+      "step": 27320
+    },
+    {
+      "epoch": 0.5256249368932739,
+      "grad_norm": 1679.1312513435219,
+      "learning_rate": 9.345905269427374e-06,
+      "loss": 264.1468,
+      "step": 27330
+    },
+    {
+      "epoch": 0.5258172621537545,
+      "grad_norm": 1626.3173692713149,
+      "learning_rate": 9.33981531885931e-06,
+      "loss": 270.7995,
+      "step": 27340
+    },
+    {
+      "epoch": 0.526009587414235,
+      "grad_norm": 1811.6392817061899,
+      "learning_rate": 9.333725614199132e-06,
+      "loss": 269.1323,
+      "step": 27350
+    },
+    {
+      "epoch": 0.5262019126747155,
+      "grad_norm": 2241.589188191697,
+      "learning_rate": 9.32763615771516e-06,
+      "loss": 271.5982,
+      "step": 27360
+    },
+    {
+      "epoch": 0.526394237935196,
+      "grad_norm": 2055.8316599772043,
+      "learning_rate": 9.321546951675616e-06,
+      "loss": 269.8291,
+      "step": 27370
+    },
+    {
+      "epoch": 0.5265865631956765,
+      "grad_norm": 1653.4648633453337,
+      "learning_rate": 9.315457998348627e-06,
+      "loss": 263.0201,
+      "step": 27380
+    },
+    {
+      "epoch": 0.5267788884561571,
+      "grad_norm": 1539.8485747905363,
+      "learning_rate": 9.309369300002224e-06,
+      "loss": 276.6529,
+      "step": 27390
+    },
+    {
+      "epoch": 0.5269712137166376,
+      "grad_norm": 1764.5718062380618,
+      "learning_rate": 9.303280858904356e-06,
+      "loss": 261.8161,
+      "step": 27400
+    },
+    {
+      "epoch": 0.5271635389771181,
+      "grad_norm": 2128.8481479222705,
+      "learning_rate": 9.297192677322862e-06,
+      "loss": 273.249,
+      "step": 27410
+    },
+    {
+      "epoch": 0.5273558642375986,
+      "grad_norm": 1617.9349758229725,
+      "learning_rate": 9.291104757525486e-06,
+      "loss": 272.4803,
+      "step": 27420
+    },
+    {
+      "epoch": 0.5275481894980791,
+      "grad_norm": 1757.5701618188102,
+      "learning_rate": 9.28501710177988e-06,
+      "loss": 258.0056,
+      "step": 27430
+    },
+    {
+      "epoch": 0.5277405147585597,
+      "grad_norm": 1739.9126733629178,
+      "learning_rate": 9.278929712353595e-06,
+      "loss": 272.746,
+      "step": 27440
+    },
+    {
+      "epoch": 0.5279328400190402,
+      "grad_norm": 1649.6529111185948,
+      "learning_rate": 9.272842591514083e-06,
+      "loss": 265.0695,
+      "step": 27450
+    },
+    {
+      "epoch": 0.5281251652795207,
+      "grad_norm": 1636.081173647623,
+      "learning_rate": 9.2667557415287e-06,
+      "loss": 269.9739,
+      "step": 27460
+    },
+    {
+      "epoch": 0.5283174905400012,
+      "grad_norm": 1833.3392132795411,
+      "learning_rate": 9.260669164664687e-06,
+      "loss": 272.6387,
+      "step": 27470
+    },
+    {
+      "epoch": 0.5285098158004817,
+      "grad_norm": 1689.9206501238227,
+      "learning_rate": 9.254582863189205e-06,
+      "loss": 274.3722,
+      "step": 27480
+    },
+    {
+      "epoch": 0.5287021410609622,
+      "grad_norm": 2022.3859797909827,
+      "learning_rate": 9.248496839369293e-06,
+      "loss": 260.403,
+      "step": 27490
+    },
+    {
+      "epoch": 0.5288944663214428,
+      "grad_norm": 1779.9215541039039,
+      "learning_rate": 9.242411095471897e-06,
+      "loss": 285.2193,
+      "step": 27500
+    },
+    {
+      "epoch": 0.5290867915819234,
+      "grad_norm": 1766.390633558896,
+      "learning_rate": 9.236325633763856e-06,
+      "loss": 265.0325,
+      "step": 27510
+    },
+    {
+      "epoch": 0.5292791168424039,
+      "grad_norm": 1569.3923986919492,
+      "learning_rate": 9.230240456511905e-06,
+      "loss": 278.332,
+      "step": 27520
+    },
+    {
+      "epoch": 0.5294714421028844,
+      "grad_norm": 1873.8427468584869,
+      "learning_rate": 9.224155565982673e-06,
+      "loss": 262.1785,
+      "step": 27530
+    },
+    {
+      "epoch": 0.529663767363365,
+      "grad_norm": 1705.317041833594,
+      "learning_rate": 9.218070964442673e-06,
+      "loss": 267.993,
+      "step": 27540
+    },
+    {
+      "epoch": 0.5298560926238455,
+      "grad_norm": 1818.2525733384166,
+      "learning_rate": 9.21198665415833e-06,
+      "loss": 266.7808,
+      "step": 27550
+    },
+    {
+      "epoch": 0.530048417884326,
+      "grad_norm": 1640.422556601716,
+      "learning_rate": 9.205902637395943e-06,
+      "loss": 268.128,
+      "step": 27560
+    },
+    {
+      "epoch": 0.5302407431448065,
+      "grad_norm": 1760.3109550929307,
+      "learning_rate": 9.199818916421706e-06,
+      "loss": 261.4038,
+      "step": 27570
+    },
+    {
+      "epoch": 0.530433068405287,
+      "grad_norm": 1620.3411373485635,
+      "learning_rate": 9.193735493501707e-06,
+      "loss": 266.1088,
+      "step": 27580
+    },
+    {
+      "epoch": 0.5306253936657676,
+      "grad_norm": 1605.0512944882896,
+      "learning_rate": 9.187652370901925e-06,
+      "loss": 272.0439,
+      "step": 27590
+    },
+    {
+      "epoch": 0.5308177189262481,
+      "grad_norm": 1665.9501984803335,
+      "learning_rate": 9.181569550888217e-06,
+      "loss": 271.1273,
+      "step": 27600
+    },
+    {
+      "epoch": 0.5310100441867286,
+      "grad_norm": 1696.5075718828227,
+      "learning_rate": 9.175487035726332e-06,
+      "loss": 263.8659,
+      "step": 27610
+    },
+    {
+      "epoch": 0.5312023694472091,
+      "grad_norm": 1672.7171767690058,
+      "learning_rate": 9.169404827681912e-06,
+      "loss": 261.8372,
+      "step": 27620
+    },
+    {
+      "epoch": 0.5313946947076896,
+      "grad_norm": 1611.2612620256748,
+      "learning_rate": 9.163322929020476e-06,
+      "loss": 261.2308,
+      "step": 27630
+    },
+    {
+      "epoch": 0.5315870199681701,
+      "grad_norm": 1581.610396039253,
+      "learning_rate": 9.157241342007428e-06,
+      "loss": 265.8153,
+      "step": 27640
+    },
+    {
+      "epoch": 0.5317793452286507,
+      "grad_norm": 1721.8160664638826,
+      "learning_rate": 9.15116006890806e-06,
+      "loss": 268.9988,
+      "step": 27650
+    },
+    {
+      "epoch": 0.5319716704891312,
+      "grad_norm": 1605.6097117632758,
+      "learning_rate": 9.145079111987552e-06,
+      "loss": 270.3846,
+      "step": 27660
+    },
+    {
+      "epoch": 0.5321639957496117,
+      "grad_norm": 1590.0273875726666,
+      "learning_rate": 9.138998473510953e-06,
+      "loss": 264.0372,
+      "step": 27670
+    },
+    {
+      "epoch": 0.5323563210100922,
+      "grad_norm": 1595.7658820030006,
+      "learning_rate": 9.1329181557432e-06,
+      "loss": 267.6364,
+      "step": 27680
+    },
+    {
+      "epoch": 0.5325486462705727,
+      "grad_norm": 1679.5579486409333,
+      "learning_rate": 9.126838160949119e-06,
+      "loss": 266.3184,
+      "step": 27690
+    },
+    {
+      "epoch": 0.5327409715310533,
+      "grad_norm": 1539.515659871773,
+      "learning_rate": 9.120758491393402e-06,
+      "loss": 272.8407,
+      "step": 27700
+    },
+    {
+      "epoch": 0.5329332967915339,
+      "grad_norm": 1479.431659387387,
+      "learning_rate": 9.114679149340623e-06,
+      "loss": 273.1751,
+      "step": 27710
+    },
+    {
+      "epoch": 0.5331256220520144,
+      "grad_norm": 1575.3753378800268,
+      "learning_rate": 9.10860013705524e-06,
+      "loss": 264.8448,
+      "step": 27720
+    },
+    {
+      "epoch": 0.5333179473124949,
+      "grad_norm": 1676.6388846882585,
+      "learning_rate": 9.102521456801582e-06,
+      "loss": 259.9461,
+      "step": 27730
+    },
+    {
+      "epoch": 0.5335102725729755,
+      "grad_norm": 1705.92009299794,
+      "learning_rate": 9.096443110843864e-06,
+      "loss": 264.2077,
+      "step": 27740
+    },
+    {
+      "epoch": 0.533702597833456,
+      "grad_norm": 1612.866193190612,
+      "learning_rate": 9.090365101446157e-06,
+      "loss": 265.0348,
+      "step": 27750
+    },
+    {
+      "epoch": 0.5338949230939365,
+      "grad_norm": 1681.3447580738216,
+      "learning_rate": 9.084287430872435e-06,
+      "loss": 265.8832,
+      "step": 27760
+    },
+    {
+      "epoch": 0.534087248354417,
+      "grad_norm": 1651.3258990916645,
+      "learning_rate": 9.078210101386518e-06,
+      "loss": 261.6132,
+      "step": 27770
+    },
+    {
+      "epoch": 0.5342795736148975,
+      "grad_norm": 1773.57898334408,
+      "learning_rate": 9.072133115252113e-06,
+      "loss": 264.9888,
+      "step": 27780
+    },
+    {
+      "epoch": 0.534471898875378,
+      "grad_norm": 1744.3508470384886,
+      "learning_rate": 9.066056474732798e-06,
+      "loss": 264.6408,
+      "step": 27790
+    },
+    {
+      "epoch": 0.5346642241358586,
+      "grad_norm": 1595.6669729080709,
+      "learning_rate": 9.059980182092022e-06,
+      "loss": 259.3286,
+      "step": 27800
+    },
+    {
+      "epoch": 0.5348565493963391,
+      "grad_norm": 1709.518386029998,
+      "learning_rate": 9.053904239593106e-06,
+      "loss": 266.3236,
+      "step": 27810
+    },
+    {
+      "epoch": 0.5350488746568196,
+      "grad_norm": 2429.771177624152,
+      "learning_rate": 9.047828649499236e-06,
+      "loss": 273.9908,
+      "step": 27820
+    },
+    {
+      "epoch": 0.5352411999173001,
+      "grad_norm": 1686.4281521790188,
+      "learning_rate": 9.041753414073463e-06,
+      "loss": 262.8393,
+      "step": 27830
+    },
+    {
+      "epoch": 0.5354335251777806,
+      "grad_norm": 1658.942456732611,
+      "learning_rate": 9.035678535578723e-06,
+      "loss": 264.2115,
+      "step": 27840
+    },
+    {
+      "epoch": 0.5356258504382612,
+      "grad_norm": 1581.2670375800524,
+      "learning_rate": 9.029604016277798e-06,
+      "loss": 264.842,
+      "step": 27850
+    },
+    {
+      "epoch": 0.5358181756987417,
+      "grad_norm": 1767.3090528996986,
+      "learning_rate": 9.02352985843335e-06,
+      "loss": 271.3129,
+      "step": 27860
+    },
+    {
+      "epoch": 0.5360105009592222,
+      "grad_norm": 1493.242922428013,
+      "learning_rate": 9.017456064307904e-06,
+      "loss": 266.3733,
+      "step": 27870
+    },
+    {
+      "epoch": 0.5362028262197027,
+      "grad_norm": 1804.544461083966,
+      "learning_rate": 9.01138263616385e-06,
+      "loss": 266.0222,
+      "step": 27880
+    },
+    {
+      "epoch": 0.5363951514801832,
+      "grad_norm": 1657.9095763685818,
+      "learning_rate": 9.005309576263436e-06,
+      "loss": 263.0513,
+      "step": 27890
+    },
+    {
+      "epoch": 0.5365874767406638,
+      "grad_norm": 1744.940196801218,
+      "learning_rate": 8.999236886868772e-06,
+      "loss": 259.8919,
+      "step": 27900
+    },
+    {
+      "epoch": 0.5367798020011444,
+      "grad_norm": 1733.4525089973727,
+      "learning_rate": 8.993164570241844e-06,
+      "loss": 261.6144,
+      "step": 27910
+    },
+    {
+      "epoch": 0.5369721272616249,
+      "grad_norm": 1752.5580786166774,
+      "learning_rate": 8.987092628644483e-06,
+      "loss": 276.0592,
+      "step": 27920
+    },
+    {
+      "epoch": 0.5371644525221054,
+      "grad_norm": 1704.7271751714843,
+      "learning_rate": 8.981021064338388e-06,
+      "loss": 263.5728,
+      "step": 27930
+    },
+    {
+      "epoch": 0.537356777782586,
+      "grad_norm": 1745.2344509657614,
+      "learning_rate": 8.974949879585118e-06,
+      "loss": 266.2813,
+      "step": 27940
+    },
+    {
+      "epoch": 0.5375491030430665,
+      "grad_norm": 1947.4409417900872,
+      "learning_rate": 8.968879076646093e-06,
+      "loss": 263.7487,
+      "step": 27950
+    },
+    {
+      "epoch": 0.537741428303547,
+      "grad_norm": 1546.9149549898798,
+      "learning_rate": 8.96280865778258e-06,
+      "loss": 262.628,
+      "step": 27960
+    },
+    {
+      "epoch": 0.5379337535640275,
+      "grad_norm": 1658.7955073371197,
+      "learning_rate": 8.956738625255709e-06,
+      "loss": 268.2876,
+      "step": 27970
+    },
+    {
+      "epoch": 0.538126078824508,
+      "grad_norm": 1608.7908764996746,
+      "learning_rate": 8.950668981326473e-06,
+      "loss": 260.7712,
+      "step": 27980
+    },
+    {
+      "epoch": 0.5383184040849885,
+      "grad_norm": 1814.1484165008458,
+      "learning_rate": 8.94459972825571e-06,
+      "loss": 263.5938,
+      "step": 27990
+    },
+    {
+      "epoch": 0.5385107293454691,
+      "grad_norm": 1639.1999684789391,
+      "learning_rate": 8.938530868304121e-06,
+      "loss": 261.185,
+      "step": 28000
+    },
+    {
+      "epoch": 0.5387030546059496,
+      "grad_norm": 1715.0675156113984,
+      "learning_rate": 8.932462403732248e-06,
+      "loss": 263.6565,
+      "step": 28010
+    },
+    {
+      "epoch": 0.5388953798664301,
+      "grad_norm": 1641.1183656892497,
+      "learning_rate": 8.926394336800502e-06,
+      "loss": 271.369,
+      "step": 28020
+    },
+    {
+      "epoch": 0.5390877051269106,
+      "grad_norm": 1667.5529741573303,
+      "learning_rate": 8.920326669769134e-06,
+      "loss": 269.715,
+      "step": 28030
+    },
+    {
+      "epoch": 0.5392800303873911,
+      "grad_norm": 1905.3607460298292,
+      "learning_rate": 8.914259404898247e-06,
+      "loss": 264.3525,
+      "step": 28040
+    },
+    {
+      "epoch": 0.5394723556478717,
+      "grad_norm": 1607.8311224232446,
+      "learning_rate": 8.908192544447803e-06,
+      "loss": 264.5651,
+      "step": 28050
+    },
+    {
+      "epoch": 0.5396646809083522,
+      "grad_norm": 1607.0847150935763,
+      "learning_rate": 8.902126090677605e-06,
+      "loss": 253.4698,
+      "step": 28060
+    },
+    {
+      "epoch": 0.5398570061688327,
+      "grad_norm": 1551.4778511500265,
+      "learning_rate": 8.896060045847305e-06,
+      "loss": 264.4631,
+      "step": 28070
+    },
+    {
+      "epoch": 0.5400493314293132,
+      "grad_norm": 1601.4403829853502,
+      "learning_rate": 8.889994412216403e-06,
+      "loss": 261.2042,
+      "step": 28080
+    },
+    {
+      "epoch": 0.5402416566897937,
+      "grad_norm": 1571.7127505384396,
+      "learning_rate": 8.883929192044254e-06,
+      "loss": 258.9222,
+      "step": 28090
+    },
+    {
+      "epoch": 0.5404339819502743,
+      "grad_norm": 1580.4622930781718,
+      "learning_rate": 8.877864387590049e-06,
+      "loss": 267.5097,
+      "step": 28100
+    },
+    {
+      "epoch": 0.5406263072107548,
+      "grad_norm": 1770.563420245373,
+      "learning_rate": 8.871800001112822e-06,
+      "loss": 266.7638,
+      "step": 28110
+    },
+    {
+      "epoch": 0.5408186324712354,
+      "grad_norm": 1744.5356029850248,
+      "learning_rate": 8.865736034871468e-06,
+      "loss": 271.1773,
+      "step": 28120
+    },
+    {
+      "epoch": 0.5410109577317159,
+      "grad_norm": 1803.642666608406,
+      "learning_rate": 8.859672491124706e-06,
+      "loss": 265.9221,
+      "step": 28130
+    },
+    {
+      "epoch": 0.5412032829921964,
+      "grad_norm": 1669.1577635476885,
+      "learning_rate": 8.853609372131105e-06,
+      "loss": 267.0815,
+      "step": 28140
+    },
+    {
+      "epoch": 0.541395608252677,
+      "grad_norm": 1723.4174579656947,
+      "learning_rate": 8.84754668014908e-06,
+      "loss": 264.052,
+      "step": 28150
+    },
+    {
+      "epoch": 0.5415879335131575,
+      "grad_norm": 1730.0654044504843,
+      "learning_rate": 8.841484417436886e-06,
+      "loss": 260.9908,
+      "step": 28160
+    },
+    {
+      "epoch": 0.541780258773638,
+      "grad_norm": 1673.2819251610888,
+      "learning_rate": 8.835422586252613e-06,
+      "loss": 271.824,
+      "step": 28170
+    },
+    {
+      "epoch": 0.5419725840341185,
+      "grad_norm": 1659.9596457926832,
+      "learning_rate": 8.829361188854194e-06,
+      "loss": 253.8619,
+      "step": 28180
+    },
+    {
+      "epoch": 0.542164909294599,
+      "grad_norm": 1716.7488308107304,
+      "learning_rate": 8.823300227499393e-06,
+      "loss": 260.1895,
+      "step": 28190
+    },
+    {
+      "epoch": 0.5423572345550796,
+      "grad_norm": 1846.4098048384983,
+      "learning_rate": 8.817239704445827e-06,
+      "loss": 266.1777,
+      "step": 28200
+    },
+    {
+      "epoch": 0.5425495598155601,
+      "grad_norm": 1755.0675953207456,
+      "learning_rate": 8.811179621950937e-06,
+      "loss": 257.8344,
+      "step": 28210
+    },
+    {
+      "epoch": 0.5427418850760406,
+      "grad_norm": 1897.4077009929483,
+      "learning_rate": 8.805119982272001e-06,
+      "loss": 264.3596,
+      "step": 28220
+    },
+    {
+      "epoch": 0.5429342103365211,
+      "grad_norm": 1718.2129387106945,
+      "learning_rate": 8.799060787666142e-06,
+      "loss": 267.1837,
+      "step": 28230
+    },
+    {
+      "epoch": 0.5431265355970016,
+      "grad_norm": 1721.6662863786794,
+      "learning_rate": 8.793002040390304e-06,
+      "loss": 267.4376,
+      "step": 28240
+    },
+    {
+      "epoch": 0.5433188608574822,
+      "grad_norm": 1516.9921656670845,
+      "learning_rate": 8.786943742701273e-06,
+      "loss": 261.2276,
+      "step": 28250
+    },
+    {
+      "epoch": 0.5435111861179627,
+      "grad_norm": 1839.8244921803764,
+      "learning_rate": 8.780885896855659e-06,
+      "loss": 262.1788,
+      "step": 28260
+    },
+    {
+      "epoch": 0.5437035113784432,
+      "grad_norm": 2024.4472195977369,
+      "learning_rate": 8.774828505109918e-06,
+      "loss": 271.277,
+      "step": 28270
+    },
+    {
+      "epoch": 0.5438958366389237,
+      "grad_norm": 1809.448040103689,
+      "learning_rate": 8.768771569720324e-06,
+      "loss": 264.9433,
+      "step": 28280
+    },
+    {
+      "epoch": 0.5440881618994042,
+      "grad_norm": 1755.5076131439093,
+      "learning_rate": 8.762715092942983e-06,
+      "loss": 273.2599,
+      "step": 28290
+    },
+    {
+      "epoch": 0.5442804871598848,
+      "grad_norm": 1507.6318245205412,
+      "learning_rate": 8.756659077033838e-06,
+      "loss": 264.8037,
+      "step": 28300
+    },
+    {
+      "epoch": 0.5444728124203653,
+      "grad_norm": 1655.4461040879078,
+      "learning_rate": 8.750603524248653e-06,
+      "loss": 271.1872,
+      "step": 28310
+    },
+    {
+      "epoch": 0.5446651376808459,
+      "grad_norm": 1549.3562031492345,
+      "learning_rate": 8.744548436843021e-06,
+      "loss": 266.1502,
+      "step": 28320
+    },
+    {
+      "epoch": 0.5448574629413264,
+      "grad_norm": 2123.7346842577267,
+      "learning_rate": 8.738493817072359e-06,
+      "loss": 264.7796,
+      "step": 28330
+    },
+    {
+      "epoch": 0.5450497882018069,
+      "grad_norm": 1771.267338219201,
+      "learning_rate": 8.73243966719192e-06,
+      "loss": 266.8416,
+      "step": 28340
+    },
+    {
+      "epoch": 0.5452421134622875,
+      "grad_norm": 1542.6316820384213,
+      "learning_rate": 8.726385989456764e-06,
+      "loss": 261.7626,
+      "step": 28350
+    },
+    {
+      "epoch": 0.545434438722768,
+      "grad_norm": 1732.3366773590503,
+      "learning_rate": 8.7203327861218e-06,
+      "loss": 265.4007,
+      "step": 28360
+    },
+    {
+      "epoch": 0.5456267639832485,
+      "grad_norm": 1666.2656426740543,
+      "learning_rate": 8.71428005944173e-06,
+      "loss": 263.2684,
+      "step": 28370
+    },
+    {
+      "epoch": 0.545819089243729,
+      "grad_norm": 1696.6021335501719,
+      "learning_rate": 8.708227811671112e-06,
+      "loss": 260.3924,
+      "step": 28380
+    },
+    {
+      "epoch": 0.5460114145042095,
+      "grad_norm": 1723.0235671410373,
+      "learning_rate": 8.702176045064296e-06,
+      "loss": 263.2037,
+      "step": 28390
+    },
+    {
+      "epoch": 0.54620373976469,
+      "grad_norm": 1539.2516405996203,
+      "learning_rate": 8.696124761875467e-06,
+      "loss": 265.2386,
+      "step": 28400
+    },
+    {
+      "epoch": 0.5463960650251706,
+      "grad_norm": 1562.5699357508377,
+      "learning_rate": 8.690073964358635e-06,
+      "loss": 271.1584,
+      "step": 28410
+    },
+    {
+      "epoch": 0.5465883902856511,
+      "grad_norm": 1771.977996789139,
+      "learning_rate": 8.684023654767613e-06,
+      "loss": 263.7209,
+      "step": 28420
+    },
+    {
+      "epoch": 0.5467807155461316,
+      "grad_norm": 1630.0735219365288,
+      "learning_rate": 8.677973835356048e-06,
+      "loss": 258.8074,
+      "step": 28430
+    },
+    {
+      "epoch": 0.5469730408066121,
+      "grad_norm": 1512.5582064074229,
+      "learning_rate": 8.671924508377392e-06,
+      "loss": 259.4953,
+      "step": 28440
+    },
+    {
+      "epoch": 0.5471653660670927,
+      "grad_norm": 1719.477551194441,
+      "learning_rate": 8.665875676084927e-06,
+      "loss": 263.0208,
+      "step": 28450
+    },
+    {
+      "epoch": 0.5473576913275732,
+      "grad_norm": 1723.6964463058296,
+      "learning_rate": 8.659827340731738e-06,
+      "loss": 267.5635,
+      "step": 28460
+    },
+    {
+      "epoch": 0.5475500165880537,
+      "grad_norm": 1746.722074664869,
+      "learning_rate": 8.653779504570728e-06,
+      "loss": 263.6304,
+      "step": 28470
+    },
+    {
+      "epoch": 0.5477423418485342,
+      "grad_norm": 1565.9587145262878,
+      "learning_rate": 8.647732169854622e-06,
+      "loss": 265.3575,
+      "step": 28480
+    },
+    {
+      "epoch": 0.5479346671090147,
+      "grad_norm": 1518.788829151476,
+      "learning_rate": 8.641685338835947e-06,
+      "loss": 262.618,
+      "step": 28490
+    },
+    {
+      "epoch": 0.5481269923694952,
+      "grad_norm": 1491.4182713254686,
+      "learning_rate": 8.635639013767053e-06,
+      "loss": 263.9698,
+      "step": 28500
+    },
+    {
+      "epoch": 0.5483193176299758,
+      "grad_norm": 1700.1457657151984,
+      "learning_rate": 8.629593196900088e-06,
+      "loss": 261.8302,
+      "step": 28510
+    },
+    {
+      "epoch": 0.5485116428904563,
+      "grad_norm": 1753.6427483796228,
+      "learning_rate": 8.62354789048703e-06,
+      "loss": 262.7962,
+      "step": 28520
+    },
+    {
+      "epoch": 0.5487039681509369,
+      "grad_norm": 1655.5452149103307,
+      "learning_rate": 8.617503096779648e-06,
+      "loss": 270.4195,
+      "step": 28530
+    },
+    {
+      "epoch": 0.5488962934114174,
+      "grad_norm": 1661.5759968315624,
+      "learning_rate": 8.61145881802953e-06,
+      "loss": 265.9621,
+      "step": 28540
+    },
+    {
+      "epoch": 0.549088618671898,
+      "grad_norm": 1603.5227939985791,
+      "learning_rate": 8.605415056488067e-06,
+      "loss": 263.2188,
+      "step": 28550
+    },
+    {
+      "epoch": 0.5492809439323785,
+      "grad_norm": 1702.672125098128,
+      "learning_rate": 8.599371814406465e-06,
+      "loss": 256.8539,
+      "step": 28560
+    },
+    {
+      "epoch": 0.549473269192859,
+      "grad_norm": 1708.7709976273409,
+      "learning_rate": 8.59332909403573e-06,
+      "loss": 287.2967,
+      "step": 28570
+    },
+    {
+      "epoch": 0.5496655944533395,
+      "grad_norm": 1910.401314671144,
+      "learning_rate": 8.587286897626672e-06,
+      "loss": 262.7298,
+      "step": 28580
+    },
+    {
+      "epoch": 0.54985791971382,
+      "grad_norm": 1755.2194390445322,
+      "learning_rate": 8.581245227429918e-06,
+      "loss": 267.085,
+      "step": 28590
+    },
+    {
+      "epoch": 0.5500502449743006,
+      "grad_norm": 1711.487715647836,
+      "learning_rate": 8.575204085695887e-06,
+      "loss": 256.2586,
+      "step": 28600
+    },
+    {
+      "epoch": 0.5502425702347811,
+      "grad_norm": 1774.5002180338,
+      "learning_rate": 8.5691634746748e-06,
+      "loss": 267.2271,
+      "step": 28610
+    },
+    {
+      "epoch": 0.5504348954952616,
+      "grad_norm": 1628.19399877106,
+      "learning_rate": 8.563123396616683e-06,
+      "loss": 255.9816,
+      "step": 28620
+    },
+    {
+      "epoch": 0.5506272207557421,
+      "grad_norm": 1626.4219694223898,
+      "learning_rate": 8.557083853771377e-06,
+      "loss": 260.0803,
+      "step": 28630
+    },
+    {
+      "epoch": 0.5508195460162226,
+      "grad_norm": 2346.0292739618067,
+      "learning_rate": 8.551044848388502e-06,
+      "loss": 265.1259,
+      "step": 28640
+    },
+    {
+      "epoch": 0.5510118712767031,
+      "grad_norm": 1645.7422381334331,
+      "learning_rate": 8.545006382717487e-06,
+      "loss": 263.926,
+      "step": 28650
+    },
+    {
+      "epoch": 0.5512041965371837,
+      "grad_norm": 1641.5596737815654,
+      "learning_rate": 8.538968459007569e-06,
+      "loss": 271.0264,
+      "step": 28660
+    },
+    {
+      "epoch": 0.5513965217976642,
+      "grad_norm": 1504.2272061625495,
+      "learning_rate": 8.532931079507772e-06,
+      "loss": 261.5937,
+      "step": 28670
+    },
+    {
+      "epoch": 0.5515888470581447,
+      "grad_norm": 1717.4515410807696,
+      "learning_rate": 8.526894246466916e-06,
+      "loss": 259.6705,
+      "step": 28680
+    },
+    {
+      "epoch": 0.5517811723186252,
+      "grad_norm": 1679.497250678857,
+      "learning_rate": 8.520857962133623e-06,
+      "loss": 258.9284,
+      "step": 28690
+    },
+    {
+      "epoch": 0.5519734975791057,
+      "grad_norm": 1859.0425088816849,
+      "learning_rate": 8.514822228756311e-06,
+      "loss": 266.5152,
+      "step": 28700
+    },
+    {
+      "epoch": 0.5521658228395863,
+      "grad_norm": 1601.5290974606924,
+      "learning_rate": 8.508787048583191e-06,
+      "loss": 262.185,
+      "step": 28710
+    },
+    {
+      "epoch": 0.5523581481000668,
+      "grad_norm": 1604.3446769992377,
+      "learning_rate": 8.502752423862264e-06,
+      "loss": 265.8423,
+      "step": 28720
+    },
+    {
+      "epoch": 0.5525504733605474,
+      "grad_norm": 1621.264889435611,
+      "learning_rate": 8.496718356841335e-06,
+      "loss": 256.4238,
+      "step": 28730
+    },
+    {
+      "epoch": 0.5527427986210279,
+      "grad_norm": 1549.595125678023,
+      "learning_rate": 8.49068484976799e-06,
+      "loss": 262.9498,
+      "step": 28740
+    },
+    {
+      "epoch": 0.5529351238815085,
+      "grad_norm": 1656.3031253152483,
+      "learning_rate": 8.484651904889614e-06,
+      "loss": 257.4689,
+      "step": 28750
+    },
+    {
+      "epoch": 0.553127449141989,
+      "grad_norm": 1641.9218524564199,
+      "learning_rate": 8.478619524453369e-06,
+      "loss": 258.6582,
+      "step": 28760
+    },
+    {
+      "epoch": 0.5533197744024695,
+      "grad_norm": 1759.0087233263441,
+      "learning_rate": 8.472587710706232e-06,
+      "loss": 255.9802,
+      "step": 28770
+    },
+    {
+      "epoch": 0.55351209966295,
+      "grad_norm": 1636.1809978566096,
+      "learning_rate": 8.466556465894942e-06,
+      "loss": 263.2994,
+      "step": 28780
+    },
+    {
+      "epoch": 0.5537044249234305,
+      "grad_norm": 1808.53240251678,
+      "learning_rate": 8.460525792266046e-06,
+      "loss": 267.5326,
+      "step": 28790
+    },
+    {
+      "epoch": 0.553896750183911,
+      "grad_norm": 1747.4212950937745,
+      "learning_rate": 8.454495692065862e-06,
+      "loss": 270.017,
+      "step": 28800
+    },
+    {
+      "epoch": 0.5540890754443916,
+      "grad_norm": 1669.61785488322,
+      "learning_rate": 8.448466167540514e-06,
+      "loss": 257.3167,
+      "step": 28810
+    },
+    {
+      "epoch": 0.5542814007048721,
+      "grad_norm": 1695.3462001112866,
+      "learning_rate": 8.442437220935893e-06,
+      "loss": 269.7112,
+      "step": 28820
+    },
+    {
+      "epoch": 0.5544737259653526,
+      "grad_norm": 1603.8919651161252,
+      "learning_rate": 8.436408854497679e-06,
+      "loss": 264.117,
+      "step": 28830
+    },
+    {
+      "epoch": 0.5546660512258331,
+      "grad_norm": 1679.5842600495218,
+      "learning_rate": 8.430381070471348e-06,
+      "loss": 264.923,
+      "step": 28840
+    },
+    {
+      "epoch": 0.5548583764863136,
+      "grad_norm": 1751.3710382623956,
+      "learning_rate": 8.424353871102144e-06,
+      "loss": 258.5021,
+      "step": 28850
+    },
+    {
+      "epoch": 0.5550507017467942,
+      "grad_norm": 1860.8253738714443,
+      "learning_rate": 8.4183272586351e-06,
+      "loss": 263.138,
+      "step": 28860
+    },
+    {
+      "epoch": 0.5552430270072747,
+      "grad_norm": 1719.623460434945,
+      "learning_rate": 8.412301235315026e-06,
+      "loss": 272.6122,
+      "step": 28870
+    },
+    {
+      "epoch": 0.5554353522677552,
+      "grad_norm": 1771.4754301382038,
+      "learning_rate": 8.406275803386525e-06,
+      "loss": 263.0697,
+      "step": 28880
+    },
+    {
+      "epoch": 0.5556276775282357,
+      "grad_norm": 1623.082697350942,
+      "learning_rate": 8.400250965093968e-06,
+      "loss": 263.7715,
+      "step": 28890
+    },
+    {
+      "epoch": 0.5558200027887162,
+      "grad_norm": 1712.7251600085062,
+      "learning_rate": 8.394226722681498e-06,
+      "loss": 268.2861,
+      "step": 28900
+    },
+    {
+      "epoch": 0.5560123280491968,
+      "grad_norm": 1690.7204189287131,
+      "learning_rate": 8.38820307839306e-06,
+      "loss": 261.4057,
+      "step": 28910
+    },
+    {
+      "epoch": 0.5562046533096773,
+      "grad_norm": 1540.101827431106,
+      "learning_rate": 8.382180034472353e-06,
+      "loss": 252.8094,
+      "step": 28920
+    },
+    {
+      "epoch": 0.5563969785701578,
+      "grad_norm": 1534.4748097839247,
+      "learning_rate": 8.376157593162867e-06,
+      "loss": 257.1021,
+      "step": 28930
+    },
+    {
+      "epoch": 0.5565893038306384,
+      "grad_norm": 1866.5748165768205,
+      "learning_rate": 8.370135756707853e-06,
+      "loss": 262.1258,
+      "step": 28940
+    },
+    {
+      "epoch": 0.556781629091119,
+      "grad_norm": 1626.0533696306002,
+      "learning_rate": 8.364114527350357e-06,
+      "loss": 260.1841,
+      "step": 28950
+    },
+    {
+      "epoch": 0.5569739543515995,
+      "grad_norm": 1706.7859715741658,
+      "learning_rate": 8.358093907333182e-06,
+      "loss": 256.0894,
+      "step": 28960
+    },
+    {
+      "epoch": 0.55716627961208,
+      "grad_norm": 1773.8256385724405,
+      "learning_rate": 8.35207389889891e-06,
+      "loss": 264.1106,
+      "step": 28970
+    },
+    {
+      "epoch": 0.5573586048725605,
+      "grad_norm": 1713.528596028282,
+      "learning_rate": 8.346054504289888e-06,
+      "loss": 267.1071,
+      "step": 28980
+    },
+    {
+      "epoch": 0.557550930133041,
+      "grad_norm": 1617.3470108224242,
+      "learning_rate": 8.34003572574825e-06,
+      "loss": 257.4415,
+      "step": 28990
+    },
+    {
+      "epoch": 0.5577432553935215,
+      "grad_norm": 1908.921435238911,
+      "learning_rate": 8.334017565515892e-06,
+      "loss": 263.7025,
+      "step": 29000
+    },
+    {
+      "epoch": 0.5579355806540021,
+      "grad_norm": 1812.21547786451,
+      "learning_rate": 8.328000025834472e-06,
+      "loss": 251.759,
+      "step": 29010
+    },
+    {
+      "epoch": 0.5581279059144826,
+      "grad_norm": 1791.3729993213485,
+      "learning_rate": 8.321983108945431e-06,
+      "loss": 261.9572,
+      "step": 29020
+    },
+    {
+      "epoch": 0.5583202311749631,
+      "grad_norm": 1855.5173768517516,
+      "learning_rate": 8.315966817089972e-06,
+      "loss": 267.1535,
+      "step": 29030
+    },
+    {
+      "epoch": 0.5585125564354436,
+      "grad_norm": 1645.0745532454027,
+      "learning_rate": 8.309951152509057e-06,
+      "loss": 262.0804,
+      "step": 29040
+    },
+    {
+      "epoch": 0.5587048816959241,
+      "grad_norm": 1535.6144313916911,
+      "learning_rate": 8.303936117443422e-06,
+      "loss": 268.5148,
+      "step": 29050
+    },
+    {
+      "epoch": 0.5588972069564047,
+      "grad_norm": 1516.5967306550074,
+      "learning_rate": 8.297921714133576e-06,
+      "loss": 263.3952,
+      "step": 29060
+    },
+    {
+      "epoch": 0.5590895322168852,
+      "grad_norm": 1750.4754603088422,
+      "learning_rate": 8.291907944819782e-06,
+      "loss": 256.8598,
+      "step": 29070
+    },
+    {
+      "epoch": 0.5592818574773657,
+      "grad_norm": 2737.4582033448974,
+      "learning_rate": 8.285894811742065e-06,
+      "loss": 250.8736,
+      "step": 29080
+    },
+    {
+      "epoch": 0.5594741827378462,
+      "grad_norm": 1549.4098703538348,
+      "learning_rate": 8.279882317140224e-06,
+      "loss": 254.5251,
+      "step": 29090
+    },
+    {
+      "epoch": 0.5596665079983267,
+      "grad_norm": 1780.4013207779396,
+      "learning_rate": 8.273870463253813e-06,
+      "loss": 260.6593,
+      "step": 29100
+    },
+    {
+      "epoch": 0.5598588332588073,
+      "grad_norm": 1710.0118800741498,
+      "learning_rate": 8.267859252322144e-06,
+      "loss": 264.3909,
+      "step": 29110
+    },
+    {
+      "epoch": 0.5600511585192878,
+      "grad_norm": 1680.5628552528506,
+      "learning_rate": 8.261848686584293e-06,
+      "loss": 255.6524,
+      "step": 29120
+    },
+    {
+      "epoch": 0.5602434837797683,
+      "grad_norm": 1791.1790269775383,
+      "learning_rate": 8.255838768279106e-06,
+      "loss": 259.3277,
+      "step": 29130
+    },
+    {
+      "epoch": 0.5604358090402489,
+      "grad_norm": 1665.6806488812938,
+      "learning_rate": 8.249829499645167e-06,
+      "loss": 260.3279,
+      "step": 29140
+    },
+    {
+      "epoch": 0.5606281343007294,
+      "grad_norm": 1600.8695300019212,
+      "learning_rate": 8.243820882920837e-06,
+      "loss": 254.6754,
+      "step": 29150
+    },
+    {
+      "epoch": 0.56082045956121,
+      "grad_norm": 1706.1281344378983,
+      "learning_rate": 8.23781292034422e-06,
+      "loss": 258.3431,
+      "step": 29160
+    },
+    {
+      "epoch": 0.5610127848216905,
+      "grad_norm": 1500.2418716396585,
+      "learning_rate": 8.231805614153192e-06,
+      "loss": 256.5448,
+      "step": 29170
+    },
+    {
+      "epoch": 0.561205110082171,
+      "grad_norm": 1740.0418185125993,
+      "learning_rate": 8.22579896658537e-06,
+      "loss": 256.3152,
+      "step": 29180
+    },
+    {
+      "epoch": 0.5613974353426515,
+      "grad_norm": 1669.5863696687672,
+      "learning_rate": 8.219792979878126e-06,
+      "loss": 254.8009,
+      "step": 29190
+    },
+    {
+      "epoch": 0.561589760603132,
+      "grad_norm": 1549.772760257509,
+      "learning_rate": 8.213787656268599e-06,
+      "loss": 263.0438,
+      "step": 29200
+    },
+    {
+      "epoch": 0.5617820858636126,
+      "grad_norm": 1615.7651737854992,
+      "learning_rate": 8.20778299799367e-06,
+      "loss": 259.0657,
+      "step": 29210
+    },
+    {
+      "epoch": 0.5619744111240931,
+      "grad_norm": 1624.1853145065897,
+      "learning_rate": 8.201779007289975e-06,
+      "loss": 257.0698,
+      "step": 29220
+    },
+    {
+      "epoch": 0.5621667363845736,
+      "grad_norm": 1826.796077227274,
+      "learning_rate": 8.195775686393898e-06,
+      "loss": 267.6748,
+      "step": 29230
+    },
+    {
+      "epoch": 0.5623590616450541,
+      "grad_norm": 1727.4431291280343,
+      "learning_rate": 8.189773037541585e-06,
+      "loss": 262.4996,
+      "step": 29240
+    },
+    {
+      "epoch": 0.5625513869055346,
+      "grad_norm": 1587.6399997450299,
+      "learning_rate": 8.183771062968917e-06,
+      "loss": 265.5079,
+      "step": 29250
+    },
+    {
+      "epoch": 0.5627437121660152,
+      "grad_norm": 1883.7327725145358,
+      "learning_rate": 8.177769764911528e-06,
+      "loss": 263.0093,
+      "step": 29260
+    },
+    {
+      "epoch": 0.5629360374264957,
+      "grad_norm": 1694.7681624801678,
+      "learning_rate": 8.171769145604812e-06,
+      "loss": 262.3144,
+      "step": 29270
+    },
+    {
+      "epoch": 0.5631283626869762,
+      "grad_norm": 1619.7653774560392,
+      "learning_rate": 8.165769207283891e-06,
+      "loss": 256.7964,
+      "step": 29280
+    },
+    {
+      "epoch": 0.5633206879474567,
+      "grad_norm": 1749.3642936956164,
+      "learning_rate": 8.15976995218365e-06,
+      "loss": 261.9936,
+      "step": 29290
+    },
+    {
+      "epoch": 0.5635130132079372,
+      "grad_norm": 1572.1873160877105,
+      "learning_rate": 8.153771382538706e-06,
+      "loss": 252.2075,
+      "step": 29300
+    },
+    {
+      "epoch": 0.5637053384684177,
+      "grad_norm": 1574.1147980318397,
+      "learning_rate": 8.147773500583434e-06,
+      "loss": 260.9694,
+      "step": 29310
+    },
+    {
+      "epoch": 0.5638976637288983,
+      "grad_norm": 1639.7355153236501,
+      "learning_rate": 8.141776308551942e-06,
+      "loss": 259.9039,
+      "step": 29320
+    },
+    {
+      "epoch": 0.5640899889893788,
+      "grad_norm": 1733.5032381774695,
+      "learning_rate": 8.135779808678084e-06,
+      "loss": 262.8336,
+      "step": 29330
+    },
+    {
+      "epoch": 0.5642823142498593,
+      "grad_norm": 1698.116775113115,
+      "learning_rate": 8.129784003195458e-06,
+      "loss": 252.238,
+      "step": 29340
+    },
+    {
+      "epoch": 0.5644746395103399,
+      "grad_norm": 1832.8736596042636,
+      "learning_rate": 8.123788894337405e-06,
+      "loss": 256.7703,
+      "step": 29350
+    },
+    {
+      "epoch": 0.5646669647708205,
+      "grad_norm": 1722.3295017627631,
+      "learning_rate": 8.117794484337003e-06,
+      "loss": 259.3589,
+      "step": 29360
+    },
+    {
+      "epoch": 0.564859290031301,
+      "grad_norm": 1744.9682866985647,
+      "learning_rate": 8.111800775427066e-06,
+      "loss": 258.7911,
+      "step": 29370
+    },
+    {
+      "epoch": 0.5650516152917815,
+      "grad_norm": 1763.443636224553,
+      "learning_rate": 8.10580776984016e-06,
+      "loss": 258.3529,
+      "step": 29380
+    },
+    {
+      "epoch": 0.565243940552262,
+      "grad_norm": 1529.8340065435077,
+      "learning_rate": 8.099815469808573e-06,
+      "loss": 257.7073,
+      "step": 29390
+    },
+    {
+      "epoch": 0.5654362658127425,
+      "grad_norm": 1562.8055147644613,
+      "learning_rate": 8.093823877564343e-06,
+      "loss": 259.259,
+      "step": 29400
+    },
+    {
+      "epoch": 0.565628591073223,
+      "grad_norm": 1819.3072836856472,
+      "learning_rate": 8.087832995339236e-06,
+      "loss": 257.0694,
+      "step": 29410
+    },
+    {
+      "epoch": 0.5658209163337036,
+      "grad_norm": 1543.5473407925751,
+      "learning_rate": 8.081842825364756e-06,
+      "loss": 251.8829,
+      "step": 29420
+    },
+    {
+      "epoch": 0.5660132415941841,
+      "grad_norm": 1661.6098921606247,
+      "learning_rate": 8.075853369872149e-06,
+      "loss": 258.9299,
+      "step": 29430
+    },
+    {
+      "epoch": 0.5662055668546646,
+      "grad_norm": 1529.2277228028574,
+      "learning_rate": 8.069864631092377e-06,
+      "loss": 259.2561,
+      "step": 29440
+    },
+    {
+      "epoch": 0.5663978921151451,
+      "grad_norm": 1596.6967262722217,
+      "learning_rate": 8.063876611256158e-06,
+      "loss": 260.8965,
+      "step": 29450
+    },
+    {
+      "epoch": 0.5665902173756256,
+      "grad_norm": 1470.6336917309131,
+      "learning_rate": 8.057889312593924e-06,
+      "loss": 264.1809,
+      "step": 29460
+    },
+    {
+      "epoch": 0.5667825426361062,
+      "grad_norm": 9776.630849906853,
+      "learning_rate": 8.051902737335847e-06,
+      "loss": 273.7714,
+      "step": 29470
+    },
+    {
+      "epoch": 0.5669748678965867,
+      "grad_norm": 1777.4158813809197,
+      "learning_rate": 8.045916887711822e-06,
+      "loss": 263.6798,
+      "step": 29480
+    },
+    {
+      "epoch": 0.5671671931570672,
+      "grad_norm": 1687.0903027100449,
+      "learning_rate": 8.039931765951485e-06,
+      "loss": 268.1132,
+      "step": 29490
+    },
+    {
+      "epoch": 0.5673595184175477,
+      "grad_norm": 1633.984428211593,
+      "learning_rate": 8.033947374284196e-06,
+      "loss": 253.8765,
+      "step": 29500
+    },
+    {
+      "epoch": 0.5675518436780282,
+      "grad_norm": 1662.2432551323188,
+      "learning_rate": 8.02796371493904e-06,
+      "loss": 263.5848,
+      "step": 29510
+    },
+    {
+      "epoch": 0.5677441689385088,
+      "grad_norm": 1758.8880776455665,
+      "learning_rate": 8.021980790144828e-06,
+      "loss": 250.4654,
+      "step": 29520
+    },
+    {
+      "epoch": 0.5679364941989893,
+      "grad_norm": 1542.2274407858008,
+      "learning_rate": 8.015998602130107e-06,
+      "loss": 256.701,
+      "step": 29530
+    },
+    {
+      "epoch": 0.5681288194594698,
+      "grad_norm": 1685.170516342027,
+      "learning_rate": 8.01001715312314e-06,
+      "loss": 252.7837,
+      "step": 29540
+    },
+    {
+      "epoch": 0.5683211447199504,
+      "grad_norm": 1591.2610085534864,
+      "learning_rate": 8.00403644535191e-06,
+      "loss": 257.4487,
+      "step": 29550
+    },
+    {
+      "epoch": 0.568513469980431,
+      "grad_norm": 1758.5769584172062,
+      "learning_rate": 7.998056481044146e-06,
+      "loss": 260.9813,
+      "step": 29560
+    },
+    {
+      "epoch": 0.5687057952409115,
+      "grad_norm": 1804.1089571958166,
+      "learning_rate": 7.992077262427282e-06,
+      "loss": 269.32,
+      "step": 29570
+    },
+    {
+      "epoch": 0.568898120501392,
+      "grad_norm": 1606.5731674684068,
+      "learning_rate": 7.986098791728475e-06,
+      "loss": 256.1864,
+      "step": 29580
+    },
+    {
+      "epoch": 0.5690904457618725,
+      "grad_norm": 1793.0990921106627,
+      "learning_rate": 7.9801210711746e-06,
+      "loss": 263.5898,
+      "step": 29590
+    },
+    {
+      "epoch": 0.569282771022353,
+      "grad_norm": 1556.6486471242072,
+      "learning_rate": 7.974144102992273e-06,
+      "loss": 257.2144,
+      "step": 29600
+    },
+    {
+      "epoch": 0.5694750962828335,
+      "grad_norm": 1876.4372825534501,
+      "learning_rate": 7.968167889407813e-06,
+      "loss": 254.0837,
+      "step": 29610
+    },
+    {
+      "epoch": 0.5696674215433141,
+      "grad_norm": 1659.321834989625,
+      "learning_rate": 7.96219243264725e-06,
+      "loss": 260.9888,
+      "step": 29620
+    },
+    {
+      "epoch": 0.5698597468037946,
+      "grad_norm": 1702.4693842484958,
+      "learning_rate": 7.956217734936353e-06,
+      "loss": 256.2627,
+      "step": 29630
+    },
+    {
+      "epoch": 0.5700520720642751,
+      "grad_norm": 1582.4475567609404,
+      "learning_rate": 7.950243798500593e-06,
+      "loss": 253.6847,
+      "step": 29640
+    },
+    {
+      "epoch": 0.5702443973247556,
+      "grad_norm": 1655.4415922933936,
+      "learning_rate": 7.94427062556517e-06,
+      "loss": 261.9364,
+      "step": 29650
+    },
+    {
+      "epoch": 0.5704367225852361,
+      "grad_norm": 1568.66228727692,
+      "learning_rate": 7.938298218354985e-06,
+      "loss": 255.769,
+      "step": 29660
+    },
+    {
+      "epoch": 0.5706290478457167,
+      "grad_norm": 1742.6320506014838,
+      "learning_rate": 7.932326579094665e-06,
+      "loss": 261.6543,
+      "step": 29670
+    },
+    {
+      "epoch": 0.5708213731061972,
+      "grad_norm": 1907.4995829468237,
+      "learning_rate": 7.926355710008545e-06,
+      "loss": 258.1616,
+      "step": 29680
+    },
+    {
+      "epoch": 0.5710136983666777,
+      "grad_norm": 1755.1787331280811,
+      "learning_rate": 7.920385613320675e-06,
+      "loss": 254.7824,
+      "step": 29690
+    },
+    {
+      "epoch": 0.5712060236271582,
+      "grad_norm": 1666.4852055395288,
+      "learning_rate": 7.914416291254817e-06,
+      "loss": 248.0865,
+      "step": 29700
+    },
+    {
+      "epoch": 0.5713983488876387,
+      "grad_norm": 1610.2507901329434,
+      "learning_rate": 7.908447746034447e-06,
+      "loss": 251.9105,
+      "step": 29710
+    },
+    {
+      "epoch": 0.5715906741481193,
+      "grad_norm": 1511.8163434725564,
+      "learning_rate": 7.902479979882749e-06,
+      "loss": 263.9274,
+      "step": 29720
+    },
+    {
+      "epoch": 0.5717829994085998,
+      "grad_norm": 1785.3980715498606,
+      "learning_rate": 7.896512995022614e-06,
+      "loss": 260.5769,
+      "step": 29730
+    },
+    {
+      "epoch": 0.5719753246690803,
+      "grad_norm": 2264.234484715675,
+      "learning_rate": 7.890546793676652e-06,
+      "loss": 255.2928,
+      "step": 29740
+    },
+    {
+      "epoch": 0.5721676499295609,
+      "grad_norm": 1640.052917357427,
+      "learning_rate": 7.884581378067171e-06,
+      "loss": 260.5069,
+      "step": 29750
+    },
+    {
+      "epoch": 0.5723599751900414,
+      "grad_norm": 1596.1775922252384,
+      "learning_rate": 7.878616750416186e-06,
+      "loss": 255.1843,
+      "step": 29760
+    },
+    {
+      "epoch": 0.572552300450522,
+      "grad_norm": 1701.8181508222356,
+      "learning_rate": 7.872652912945426e-06,
+      "loss": 255.3512,
+      "step": 29770
+    },
+    {
+      "epoch": 0.5727446257110025,
+      "grad_norm": 1625.8288873639642,
+      "learning_rate": 7.866689867876323e-06,
+      "loss": 260.3132,
+      "step": 29780
+    },
+    {
+      "epoch": 0.572936950971483,
+      "grad_norm": 1685.9981537972806,
+      "learning_rate": 7.860727617430013e-06,
+      "loss": 257.2799,
+      "step": 29790
+    },
+    {
+      "epoch": 0.5731292762319635,
+      "grad_norm": 1530.6522926872287,
+      "learning_rate": 7.85476616382733e-06,
+      "loss": 248.3945,
+      "step": 29800
+    },
+    {
+      "epoch": 0.573321601492444,
+      "grad_norm": 1822.0703356421443,
+      "learning_rate": 7.848805509288824e-06,
+      "loss": 264.8996,
+      "step": 29810
+    },
+    {
+      "epoch": 0.5735139267529246,
+      "grad_norm": 1543.0483468032942,
+      "learning_rate": 7.84284565603474e-06,
+      "loss": 255.3366,
+      "step": 29820
+    },
+    {
+      "epoch": 0.5737062520134051,
+      "grad_norm": 1452.1735410792028,
+      "learning_rate": 7.83688660628502e-06,
+      "loss": 257.8694,
+      "step": 29830
+    },
+    {
+      "epoch": 0.5738985772738856,
+      "grad_norm": 1854.4592069948071,
+      "learning_rate": 7.830928362259313e-06,
+      "loss": 257.7086,
+      "step": 29840
+    },
+    {
+      "epoch": 0.5740909025343661,
+      "grad_norm": 1583.422677836044,
+      "learning_rate": 7.824970926176967e-06,
+      "loss": 260.0587,
+      "step": 29850
+    },
+    {
+      "epoch": 0.5742832277948466,
+      "grad_norm": 1709.211584185582,
+      "learning_rate": 7.819014300257033e-06,
+      "loss": 255.1465,
+      "step": 29860
+    },
+    {
+      "epoch": 0.5744755530553272,
+      "grad_norm": 1682.8532648989647,
+      "learning_rate": 7.813058486718252e-06,
+      "loss": 260.2551,
+      "step": 29870
+    },
+    {
+      "epoch": 0.5746678783158077,
+      "grad_norm": 1587.5191109815617,
+      "learning_rate": 7.80710348777906e-06,
+      "loss": 258.077,
+      "step": 29880
+    },
+    {
+      "epoch": 0.5748602035762882,
+      "grad_norm": 1716.47193098543,
+      "learning_rate": 7.801149305657609e-06,
+      "loss": 257.1473,
+      "step": 29890
+    },
+    {
+      "epoch": 0.5750525288367687,
+      "grad_norm": 1622.459882534187,
+      "learning_rate": 7.795195942571722e-06,
+      "loss": 267.1598,
+      "step": 29900
+    },
+    {
+      "epoch": 0.5752448540972492,
+      "grad_norm": 1633.6140126424693,
+      "learning_rate": 7.789243400738934e-06,
+      "loss": 255.4259,
+      "step": 29910
+    },
+    {
+      "epoch": 0.5754371793577298,
+      "grad_norm": 1693.6210912418912,
+      "learning_rate": 7.783291682376465e-06,
+      "loss": 265.9042,
+      "step": 29920
+    },
+    {
+      "epoch": 0.5756295046182103,
+      "grad_norm": 1658.3411836017206,
+      "learning_rate": 7.77734078970124e-06,
+      "loss": 265.6731,
+      "step": 29930
+    },
+    {
+      "epoch": 0.5758218298786908,
+      "grad_norm": 1684.4254026975589,
+      "learning_rate": 7.77139072492986e-06,
+      "loss": 255.4352,
+      "step": 29940
+    },
+    {
+      "epoch": 0.5760141551391713,
+      "grad_norm": 1544.7323611779777,
+      "learning_rate": 7.765441490278625e-06,
+      "loss": 256.1602,
+      "step": 29950
+    },
+    {
+      "epoch": 0.5762064803996519,
+      "grad_norm": 1623.63606167973,
+      "learning_rate": 7.759493087963535e-06,
+      "loss": 260.5616,
+      "step": 29960
+    },
+    {
+      "epoch": 0.5763988056601325,
+      "grad_norm": 1545.0933427402958,
+      "learning_rate": 7.753545520200264e-06,
+      "loss": 261.6,
+      "step": 29970
+    },
+    {
+      "epoch": 0.576591130920613,
+      "grad_norm": 1488.7030586598682,
+      "learning_rate": 7.747598789204183e-06,
+      "loss": 254.9581,
+      "step": 29980
+    },
+    {
+      "epoch": 0.5767834561810935,
+      "grad_norm": 1583.2270031910834,
+      "learning_rate": 7.741652897190355e-06,
+      "loss": 260.2494,
+      "step": 29990
+    },
+    {
+      "epoch": 0.576975781441574,
+      "grad_norm": 1642.5285050128755,
+      "learning_rate": 7.735707846373527e-06,
+      "loss": 253.3322,
+      "step": 30000
+    },
+    {
+      "epoch": 0.5771681067020545,
+      "grad_norm": 1634.2302429630959,
+      "learning_rate": 7.72976363896813e-06,
+      "loss": 258.2329,
+      "step": 30010
+    },
+    {
+      "epoch": 0.5773604319625351,
+      "grad_norm": 1724.195099879509,
+      "learning_rate": 7.723820277188278e-06,
+      "loss": 260.7178,
+      "step": 30020
+    },
+    {
+      "epoch": 0.5775527572230156,
+      "grad_norm": 1593.5780539050602,
+      "learning_rate": 7.717877763247787e-06,
+      "loss": 255.6292,
+      "step": 30030
+    },
+    {
+      "epoch": 0.5777450824834961,
+      "grad_norm": 1924.6729037815921,
+      "learning_rate": 7.711936099360138e-06,
+      "loss": 262.5923,
+      "step": 30040
+    },
+    {
+      "epoch": 0.5779374077439766,
+      "grad_norm": 1500.2251309822552,
+      "learning_rate": 7.705995287738498e-06,
+      "loss": 256.1763,
+      "step": 30050
+    },
+    {
+      "epoch": 0.5781297330044571,
+      "grad_norm": 1567.5068155497322,
+      "learning_rate": 7.70005533059573e-06,
+      "loss": 257.8809,
+      "step": 30060
+    },
+    {
+      "epoch": 0.5783220582649377,
+      "grad_norm": 1586.9441191292162,
+      "learning_rate": 7.694116230144368e-06,
+      "loss": 256.0111,
+      "step": 30070
+    },
+    {
+      "epoch": 0.5785143835254182,
+      "grad_norm": 1653.2544937911218,
+      "learning_rate": 7.688177988596628e-06,
+      "loss": 258.6795,
+      "step": 30080
+    },
+    {
+      "epoch": 0.5787067087858987,
+      "grad_norm": 1866.6799021531012,
+      "learning_rate": 7.682240608164401e-06,
+      "loss": 250.3602,
+      "step": 30090
+    },
+    {
+      "epoch": 0.5788990340463792,
+      "grad_norm": 1455.2311803550322,
+      "learning_rate": 7.676304091059273e-06,
+      "loss": 256.5646,
+      "step": 30100
+    },
+    {
+      "epoch": 0.5790913593068597,
+      "grad_norm": 1708.5237439901796,
+      "learning_rate": 7.670368439492495e-06,
+      "loss": 257.9868,
+      "step": 30110
+    },
+    {
+      "epoch": 0.5792836845673403,
+      "grad_norm": 1543.8437704096577,
+      "learning_rate": 7.664433655674995e-06,
+      "loss": 242.7472,
+      "step": 30120
+    },
+    {
+      "epoch": 0.5794760098278208,
+      "grad_norm": 1635.6584103819475,
+      "learning_rate": 7.658499741817383e-06,
+      "loss": 263.0051,
+      "step": 30130
+    },
+    {
+      "epoch": 0.5796683350883013,
+      "grad_norm": 1682.0024499297333,
+      "learning_rate": 7.65256670012995e-06,
+      "loss": 252.0004,
+      "step": 30140
+    },
+    {
+      "epoch": 0.5798606603487818,
+      "grad_norm": 1740.574253156307,
+      "learning_rate": 7.646634532822652e-06,
+      "loss": 258.3138,
+      "step": 30150
+    },
+    {
+      "epoch": 0.5800529856092624,
+      "grad_norm": 1706.1034769315715,
+      "learning_rate": 7.64070324210512e-06,
+      "loss": 261.4026,
+      "step": 30160
+    },
+    {
+      "epoch": 0.580245310869743,
+      "grad_norm": 1722.00151312128,
+      "learning_rate": 7.634772830186668e-06,
+      "loss": 249.8261,
+      "step": 30170
+    },
+    {
+      "epoch": 0.5804376361302235,
+      "grad_norm": 1870.2121998216276,
+      "learning_rate": 7.628843299276276e-06,
+      "loss": 262.8084,
+      "step": 30180
+    },
+    {
+      "epoch": 0.580629961390704,
+      "grad_norm": 1549.6376563168453,
+      "learning_rate": 7.62291465158259e-06,
+      "loss": 252.0434,
+      "step": 30190
+    },
+    {
+      "epoch": 0.5808222866511845,
+      "grad_norm": 1520.1811528740461,
+      "learning_rate": 7.616986889313939e-06,
+      "loss": 249.7399,
+      "step": 30200
+    },
+    {
+      "epoch": 0.581014611911665,
+      "grad_norm": 1750.1855122388404,
+      "learning_rate": 7.611060014678313e-06,
+      "loss": 252.6227,
+      "step": 30210
+    },
+    {
+      "epoch": 0.5812069371721456,
+      "grad_norm": 1569.694157718902,
+      "learning_rate": 7.605134029883381e-06,
+      "loss": 254.9713,
+      "step": 30220
+    },
+    {
+      "epoch": 0.5813992624326261,
+      "grad_norm": 1638.9948737839395,
+      "learning_rate": 7.599208937136465e-06,
+      "loss": 254.3394,
+      "step": 30230
+    },
+    {
+      "epoch": 0.5815915876931066,
+      "grad_norm": 1989.2750455640175,
+      "learning_rate": 7.593284738644574e-06,
+      "loss": 261.4517,
+      "step": 30240
+    },
+    {
+      "epoch": 0.5817839129535871,
+      "grad_norm": 1495.2031069416976,
+      "learning_rate": 7.58736143661437e-06,
+      "loss": 255.4625,
+      "step": 30250
+    },
+    {
+      "epoch": 0.5819762382140676,
+      "grad_norm": 1513.314032026537,
+      "learning_rate": 7.5814390332521824e-06,
+      "loss": 255.932,
+      "step": 30260
+    },
+    {
+      "epoch": 0.5821685634745482,
+      "grad_norm": 1670.8325677670348,
+      "learning_rate": 7.575517530764011e-06,
+      "loss": 260.4452,
+      "step": 30270
+    },
+    {
+      "epoch": 0.5823608887350287,
+      "grad_norm": 1472.943918337438,
+      "learning_rate": 7.569596931355517e-06,
+      "loss": 255.0141,
+      "step": 30280
+    },
+    {
+      "epoch": 0.5825532139955092,
+      "grad_norm": 1519.1418263985556,
+      "learning_rate": 7.56367723723203e-06,
+      "loss": 253.5092,
+      "step": 30290
+    },
+    {
+      "epoch": 0.5827455392559897,
+      "grad_norm": 1769.2208171709449,
+      "learning_rate": 7.557758450598534e-06,
+      "loss": 259.7952,
+      "step": 30300
+    },
+    {
+      "epoch": 0.5829378645164702,
+      "grad_norm": 1647.2983558351118,
+      "learning_rate": 7.551840573659677e-06,
+      "loss": 257.521,
+      "step": 30310
+    },
+    {
+      "epoch": 0.5831301897769507,
+      "grad_norm": 1746.1721663956646,
+      "learning_rate": 7.5459236086197775e-06,
+      "loss": 260.4395,
+      "step": 30320
+    },
+    {
+      "epoch": 0.5833225150374313,
+      "grad_norm": 1505.8451451374303,
+      "learning_rate": 7.5400075576828e-06,
+      "loss": 254.8534,
+      "step": 30330
+    },
+    {
+      "epoch": 0.5835148402979118,
+      "grad_norm": 1884.5504649503973,
+      "learning_rate": 7.534092423052382e-06,
+      "loss": 253.4212,
+      "step": 30340
+    },
+    {
+      "epoch": 0.5837071655583923,
+      "grad_norm": 1733.6211561157456,
+      "learning_rate": 7.5281782069318075e-06,
+      "loss": 260.8099,
+      "step": 30350
+    },
+    {
+      "epoch": 0.5838994908188728,
+      "grad_norm": 1507.4167716955735,
+      "learning_rate": 7.522264911524031e-06,
+      "loss": 249.6041,
+      "step": 30360
+    },
+    {
+      "epoch": 0.5840918160793535,
+      "grad_norm": 1655.6587040136014,
+      "learning_rate": 7.516352539031654e-06,
+      "loss": 255.6054,
+      "step": 30370
+    },
+    {
+      "epoch": 0.584284141339834,
+      "grad_norm": 3409.4158939378403,
+      "learning_rate": 7.510441091656933e-06,
+      "loss": 254.867,
+      "step": 30380
+    },
+    {
+      "epoch": 0.5844764666003145,
+      "grad_norm": 1626.0703420585514,
+      "learning_rate": 7.504530571601792e-06,
+      "loss": 257.5422,
+      "step": 30390
+    },
+    {
+      "epoch": 0.584668791860795,
+      "grad_norm": 1840.8861436924435,
+      "learning_rate": 7.498620981067799e-06,
+      "loss": 256.4879,
+      "step": 30400
+    },
+    {
+      "epoch": 0.5848611171212755,
+      "grad_norm": 1674.8549404658115,
+      "learning_rate": 7.492712322256177e-06,
+      "loss": 259.2995,
+      "step": 30410
+    },
+    {
+      "epoch": 0.585053442381756,
+      "grad_norm": 1637.4692718380159,
+      "learning_rate": 7.486804597367807e-06,
+      "loss": 250.6786,
+      "step": 30420
+    },
+    {
+      "epoch": 0.5852457676422366,
+      "grad_norm": 1693.3821592007978,
+      "learning_rate": 7.480897808603219e-06,
+      "loss": 250.0638,
+      "step": 30430
+    },
+    {
+      "epoch": 0.5854380929027171,
+      "grad_norm": 1577.6699252860035,
+      "learning_rate": 7.474991958162594e-06,
+      "loss": 252.9034,
+      "step": 30440
+    },
+    {
+      "epoch": 0.5856304181631976,
+      "grad_norm": 1551.5076809213983,
+      "learning_rate": 7.469087048245758e-06,
+      "loss": 251.5118,
+      "step": 30450
+    },
+    {
+      "epoch": 0.5858227434236781,
+      "grad_norm": 2040.7377270627353,
+      "learning_rate": 7.463183081052201e-06,
+      "loss": 267.5035,
+      "step": 30460
+    },
+    {
+      "epoch": 0.5860150686841586,
+      "grad_norm": 1936.2896028491634,
+      "learning_rate": 7.457280058781049e-06,
+      "loss": 259.5283,
+      "step": 30470
+    },
+    {
+      "epoch": 0.5862073939446392,
+      "grad_norm": 1561.9808574916208,
+      "learning_rate": 7.451377983631078e-06,
+      "loss": 263.088,
+      "step": 30480
+    },
+    {
+      "epoch": 0.5863997192051197,
+      "grad_norm": 1666.215732910661,
+      "learning_rate": 7.445476857800717e-06,
+      "loss": 254.9014,
+      "step": 30490
+    },
+    {
+      "epoch": 0.5865920444656002,
+      "grad_norm": 1616.762714608203,
+      "learning_rate": 7.439576683488039e-06,
+      "loss": 260.8711,
+      "step": 30500
+    },
+    {
+      "epoch": 0.5867843697260807,
+      "grad_norm": 1587.6657423708082,
+      "learning_rate": 7.4336774628907604e-06,
+      "loss": 257.7554,
+      "step": 30510
+    },
+    {
+      "epoch": 0.5869766949865612,
+      "grad_norm": 1597.979394881119,
+      "learning_rate": 7.427779198206238e-06,
+      "loss": 250.839,
+      "step": 30520
+    },
+    {
+      "epoch": 0.5871690202470418,
+      "grad_norm": 1783.0718997255078,
+      "learning_rate": 7.421881891631487e-06,
+      "loss": 254.9956,
+      "step": 30530
+    },
+    {
+      "epoch": 0.5873613455075223,
+      "grad_norm": 1710.507265803589,
+      "learning_rate": 7.415985545363152e-06,
+      "loss": 250.5494,
+      "step": 30540
+    },
+    {
+      "epoch": 0.5875536707680028,
+      "grad_norm": 1623.6243047475557,
+      "learning_rate": 7.410090161597523e-06,
+      "loss": 252.9712,
+      "step": 30550
+    },
+    {
+      "epoch": 0.5877459960284833,
+      "grad_norm": 1721.4330559834823,
+      "learning_rate": 7.404195742530533e-06,
+      "loss": 252.7263,
+      "step": 30560
+    },
+    {
+      "epoch": 0.587938321288964,
+      "grad_norm": 1609.3097671558262,
+      "learning_rate": 7.398302290357763e-06,
+      "loss": 253.4286,
+      "step": 30570
+    },
+    {
+      "epoch": 0.5881306465494445,
+      "grad_norm": 1588.2475167830564,
+      "learning_rate": 7.392409807274421e-06,
+      "loss": 249.8665,
+      "step": 30580
+    },
+    {
+      "epoch": 0.588322971809925,
+      "grad_norm": 1667.5022025425314,
+      "learning_rate": 7.386518295475355e-06,
+      "loss": 255.5411,
+      "step": 30590
+    },
+    {
+      "epoch": 0.5885152970704055,
+      "grad_norm": 1637.3017647446482,
+      "learning_rate": 7.380627757155065e-06,
+      "loss": 257.2046,
+      "step": 30600
+    },
+    {
+      "epoch": 0.588707622330886,
+      "grad_norm": 1575.354829647761,
+      "learning_rate": 7.374738194507675e-06,
+      "loss": 251.7738,
+      "step": 30610
+    },
+    {
+      "epoch": 0.5888999475913665,
+      "grad_norm": 1664.3297491301137,
+      "learning_rate": 7.3688496097269494e-06,
+      "loss": 253.2013,
+      "step": 30620
+    },
+    {
+      "epoch": 0.5890922728518471,
+      "grad_norm": 1751.6774219865351,
+      "learning_rate": 7.362962005006286e-06,
+      "loss": 265.0654,
+      "step": 30630
+    },
+    {
+      "epoch": 0.5892845981123276,
+      "grad_norm": 1584.3565108544628,
+      "learning_rate": 7.3570753825387275e-06,
+      "loss": 246.735,
+      "step": 30640
+    },
+    {
+      "epoch": 0.5894769233728081,
+      "grad_norm": 1619.6574234851712,
+      "learning_rate": 7.35118974451694e-06,
+      "loss": 246.4986,
+      "step": 30650
+    },
+    {
+      "epoch": 0.5896692486332886,
+      "grad_norm": 1857.8790454709745,
+      "learning_rate": 7.345305093133226e-06,
+      "loss": 250.8391,
+      "step": 30660
+    },
+    {
+      "epoch": 0.5898615738937691,
+      "grad_norm": 1530.3763290482568,
+      "learning_rate": 7.3394214305795175e-06,
+      "loss": 255.2492,
+      "step": 30670
+    },
+    {
+      "epoch": 0.5900538991542497,
+      "grad_norm": 1637.8511492093023,
+      "learning_rate": 7.33353875904739e-06,
+      "loss": 252.3276,
+      "step": 30680
+    },
+    {
+      "epoch": 0.5902462244147302,
+      "grad_norm": 1602.444571441276,
+      "learning_rate": 7.327657080728032e-06,
+      "loss": 249.006,
+      "step": 30690
+    },
+    {
+      "epoch": 0.5904385496752107,
+      "grad_norm": 1713.9192670303632,
+      "learning_rate": 7.321776397812279e-06,
+      "loss": 252.6097,
+      "step": 30700
+    },
+    {
+      "epoch": 0.5906308749356912,
+      "grad_norm": 1656.1174715814968,
+      "learning_rate": 7.315896712490584e-06,
+      "loss": 257.8542,
+      "step": 30710
+    },
+    {
+      "epoch": 0.5908232001961717,
+      "grad_norm": 1590.4843269849346,
+      "learning_rate": 7.310018026953036e-06,
+      "loss": 254.9456,
+      "step": 30720
+    },
+    {
+      "epoch": 0.5910155254566523,
+      "grad_norm": 1566.7090224771084,
+      "learning_rate": 7.304140343389348e-06,
+      "loss": 253.2903,
+      "step": 30730
+    },
+    {
+      "epoch": 0.5912078507171328,
+      "grad_norm": 1566.0103004199314,
+      "learning_rate": 7.298263663988853e-06,
+      "loss": 260.1873,
+      "step": 30740
+    },
+    {
+      "epoch": 0.5914001759776133,
+      "grad_norm": 2092.2012335686536,
+      "learning_rate": 7.292387990940526e-06,
+      "loss": 244.4104,
+      "step": 30750
+    },
+    {
+      "epoch": 0.5915925012380938,
+      "grad_norm": 1650.419203500856,
+      "learning_rate": 7.286513326432953e-06,
+      "loss": 252.8488,
+      "step": 30760
+    },
+    {
+      "epoch": 0.5917848264985743,
+      "grad_norm": 1578.28047741487,
+      "learning_rate": 7.2806396726543526e-06,
+      "loss": 262.9985,
+      "step": 30770
+    },
+    {
+      "epoch": 0.591977151759055,
+      "grad_norm": 1845.8480121646596,
+      "learning_rate": 7.2747670317925625e-06,
+      "loss": 257.6338,
+      "step": 30780
+    },
+    {
+      "epoch": 0.5921694770195355,
+      "grad_norm": 1974.4470116573864,
+      "learning_rate": 7.268895406035046e-06,
+      "loss": 251.438,
+      "step": 30790
+    },
+    {
+      "epoch": 0.592361802280016,
+      "grad_norm": 1841.9674024691687,
+      "learning_rate": 7.263024797568884e-06,
+      "loss": 261.3318,
+      "step": 30800
+    },
+    {
+      "epoch": 0.5925541275404965,
+      "grad_norm": 1485.6737540030588,
+      "learning_rate": 7.257155208580778e-06,
+      "loss": 251.8768,
+      "step": 30810
+    },
+    {
+      "epoch": 0.592746452800977,
+      "grad_norm": 1706.8842117862596,
+      "learning_rate": 7.251286641257062e-06,
+      "loss": 257.6578,
+      "step": 30820
+    },
+    {
+      "epoch": 0.5929387780614576,
+      "grad_norm": 1663.8095588862873,
+      "learning_rate": 7.245419097783674e-06,
+      "loss": 262.0871,
+      "step": 30830
+    },
+    {
+      "epoch": 0.5931311033219381,
+      "grad_norm": 1573.6637705795938,
+      "learning_rate": 7.239552580346181e-06,
+      "loss": 245.1458,
+      "step": 30840
+    },
+    {
+      "epoch": 0.5933234285824186,
+      "grad_norm": 1699.2428161549244,
+      "learning_rate": 7.233687091129757e-06,
+      "loss": 265.6187,
+      "step": 30850
+    },
+    {
+      "epoch": 0.5935157538428991,
+      "grad_norm": 1973.7300314829045,
+      "learning_rate": 7.227822632319208e-06,
+      "loss": 264.8671,
+      "step": 30860
+    },
+    {
+      "epoch": 0.5937080791033796,
+      "grad_norm": 2116.200925766933,
+      "learning_rate": 7.221959206098945e-06,
+      "loss": 269.4615,
+      "step": 30870
+    },
+    {
+      "epoch": 0.5939004043638602,
+      "grad_norm": 1820.7836993861838,
+      "learning_rate": 7.216096814652992e-06,
+      "loss": 267.4089,
+      "step": 30880
+    },
+    {
+      "epoch": 0.5940927296243407,
+      "grad_norm": 1615.8204308074812,
+      "learning_rate": 7.210235460165002e-06,
+      "loss": 254.4707,
+      "step": 30890
+    },
+    {
+      "epoch": 0.5942850548848212,
+      "grad_norm": 1658.3097901525714,
+      "learning_rate": 7.2043751448182275e-06,
+      "loss": 252.7119,
+      "step": 30900
+    },
+    {
+      "epoch": 0.5944773801453017,
+      "grad_norm": 1673.4034565115521,
+      "learning_rate": 7.198515870795542e-06,
+      "loss": 250.9427,
+      "step": 30910
+    },
+    {
+      "epoch": 0.5946697054057822,
+      "grad_norm": 1547.5950616964494,
+      "learning_rate": 7.192657640279421e-06,
+      "loss": 259.1345,
+      "step": 30920
+    },
+    {
+      "epoch": 0.5948620306662628,
+      "grad_norm": 1675.760765028026,
+      "learning_rate": 7.18680045545197e-06,
+      "loss": 258.7112,
+      "step": 30930
+    },
+    {
+      "epoch": 0.5950543559267433,
+      "grad_norm": 2205.419779580163,
+      "learning_rate": 7.180944318494888e-06,
+      "loss": 257.2583,
+      "step": 30940
+    },
+    {
+      "epoch": 0.5952466811872238,
+      "grad_norm": 1528.8485652231484,
+      "learning_rate": 7.175089231589485e-06,
+      "loss": 258.0885,
+      "step": 30950
+    },
+    {
+      "epoch": 0.5954390064477043,
+      "grad_norm": 1669.1332566979506,
+      "learning_rate": 7.1692351969166905e-06,
+      "loss": 266.1266,
+      "step": 30960
+    },
+    {
+      "epoch": 0.5956313317081848,
+      "grad_norm": 1705.1472426130142,
+      "learning_rate": 7.163382216657033e-06,
+      "loss": 250.7686,
+      "step": 30970
+    },
+    {
+      "epoch": 0.5958236569686655,
+      "grad_norm": 1550.030714470702,
+      "learning_rate": 7.157530292990654e-06,
+      "loss": 252.4722,
+      "step": 30980
+    },
+    {
+      "epoch": 0.596015982229146,
+      "grad_norm": 1587.797767818733,
+      "learning_rate": 7.151679428097291e-06,
+      "loss": 244.6052,
+      "step": 30990
+    },
+    {
+      "epoch": 0.5962083074896265,
+      "grad_norm": 1718.5129403705446,
+      "learning_rate": 7.145829624156304e-06,
+      "loss": 255.8966,
+      "step": 31000
+    },
+    {
+      "epoch": 0.596400632750107,
+      "grad_norm": 1588.9529401264854,
+      "learning_rate": 7.1399808833466445e-06,
+      "loss": 252.2865,
+      "step": 31010
+    },
+    {
+      "epoch": 0.5965929580105875,
+      "grad_norm": 1476.7402336192167,
+      "learning_rate": 7.134133207846869e-06,
+      "loss": 254.0667,
+      "step": 31020
+    },
+    {
+      "epoch": 0.5967852832710681,
+      "grad_norm": 1657.7247551425467,
+      "learning_rate": 7.128286599835139e-06,
+      "loss": 258.9685,
+      "step": 31030
+    },
+    {
+      "epoch": 0.5969776085315486,
+      "grad_norm": 1493.3089829784653,
+      "learning_rate": 7.122441061489228e-06,
+      "loss": 244.1329,
+      "step": 31040
+    },
+    {
+      "epoch": 0.5971699337920291,
+      "grad_norm": 1697.6064276542609,
+      "learning_rate": 7.1165965949864934e-06,
+      "loss": 254.1651,
+      "step": 31050
+    },
+    {
+      "epoch": 0.5973622590525096,
+      "grad_norm": 1598.173350250402,
+      "learning_rate": 7.110753202503906e-06,
+      "loss": 252.9963,
+      "step": 31060
+    },
+    {
+      "epoch": 0.5975545843129901,
+      "grad_norm": 1715.0095474392097,
+      "learning_rate": 7.104910886218036e-06,
+      "loss": 255.701,
+      "step": 31070
+    },
+    {
+      "epoch": 0.5977469095734707,
+      "grad_norm": 1797.1796430432103,
+      "learning_rate": 7.0990696483050466e-06,
+      "loss": 254.4337,
+      "step": 31080
+    },
+    {
+      "epoch": 0.5979392348339512,
+      "grad_norm": 1543.2766174927401,
+      "learning_rate": 7.093229490940704e-06,
+      "loss": 255.3483,
+      "step": 31090
+    },
+    {
+      "epoch": 0.5981315600944317,
+      "grad_norm": 1530.8538209123392,
+      "learning_rate": 7.087390416300364e-06,
+      "loss": 246.355,
+      "step": 31100
+    },
+    {
+      "epoch": 0.5983238853549122,
+      "grad_norm": 1534.7340270855257,
+      "learning_rate": 7.081552426558995e-06,
+      "loss": 261.3595,
+      "step": 31110
+    },
+    {
+      "epoch": 0.5985162106153927,
+      "grad_norm": 1597.4918208427741,
+      "learning_rate": 7.075715523891146e-06,
+      "loss": 247.4333,
+      "step": 31120
+    },
+    {
+      "epoch": 0.5987085358758732,
+      "grad_norm": 1619.4496582796928,
+      "learning_rate": 7.069879710470965e-06,
+      "loss": 249.2795,
+      "step": 31130
+    },
+    {
+      "epoch": 0.5989008611363538,
+      "grad_norm": 1766.288958669557,
+      "learning_rate": 7.064044988472204e-06,
+      "loss": 245.3983,
+      "step": 31140
+    },
+    {
+      "epoch": 0.5990931863968343,
+      "grad_norm": 1715.6246582145068,
+      "learning_rate": 7.058211360068196e-06,
+      "loss": 248.6759,
+      "step": 31150
+    },
+    {
+      "epoch": 0.5992855116573148,
+      "grad_norm": 1600.9913551559735,
+      "learning_rate": 7.052378827431871e-06,
+      "loss": 254.0407,
+      "step": 31160
+    },
+    {
+      "epoch": 0.5994778369177953,
+      "grad_norm": 1631.445811614129,
+      "learning_rate": 7.046547392735747e-06,
+      "loss": 247.8378,
+      "step": 31170
+    },
+    {
+      "epoch": 0.5996701621782758,
+      "grad_norm": 1841.4276324290997,
+      "learning_rate": 7.040717058151945e-06,
+      "loss": 249.4191,
+      "step": 31180
+    },
+    {
+      "epoch": 0.5998624874387565,
+      "grad_norm": 1668.9245369334262,
+      "learning_rate": 7.034887825852164e-06,
+      "loss": 253.9356,
+      "step": 31190
+    },
+    {
+      "epoch": 0.600054812699237,
+      "grad_norm": 1617.808100854206,
+      "learning_rate": 7.029059698007699e-06,
+      "loss": 245.587,
+      "step": 31200
+    },
+    {
+      "epoch": 0.6002471379597175,
+      "grad_norm": 1861.6967300436138,
+      "learning_rate": 7.023232676789424e-06,
+      "loss": 262.5603,
+      "step": 31210
+    },
+    {
+      "epoch": 0.600439463220198,
+      "grad_norm": 1984.2303963111485,
+      "learning_rate": 7.01740676436782e-06,
+      "loss": 252.9359,
+      "step": 31220
+    },
+    {
+      "epoch": 0.6006317884806786,
+      "grad_norm": 1644.3810825427179,
+      "learning_rate": 7.011581962912936e-06,
+      "loss": 256.1246,
+      "step": 31230
+    },
+    {
+      "epoch": 0.6008241137411591,
+      "grad_norm": 1539.2303624853166,
+      "learning_rate": 7.005758274594412e-06,
+      "loss": 252.4768,
+      "step": 31240
+    },
+    {
+      "epoch": 0.6010164390016396,
+      "grad_norm": 1697.7370864596169,
+      "learning_rate": 6.999935701581482e-06,
+      "loss": 251.9505,
+      "step": 31250
+    },
+    {
+      "epoch": 0.6012087642621201,
+      "grad_norm": 1692.0694242967468,
+      "learning_rate": 6.9941142460429555e-06,
+      "loss": 251.524,
+      "step": 31260
+    },
+    {
+      "epoch": 0.6014010895226006,
+      "grad_norm": 1514.3251569804493,
+      "learning_rate": 6.988293910147229e-06,
+      "loss": 252.5521,
+      "step": 31270
+    },
+    {
+      "epoch": 0.6015934147830811,
+      "grad_norm": 1629.8771022634385,
+      "learning_rate": 6.982474696062278e-06,
+      "loss": 253.2377,
+      "step": 31280
+    },
+    {
+      "epoch": 0.6017857400435617,
+      "grad_norm": 1507.48910827372,
+      "learning_rate": 6.97665660595567e-06,
+      "loss": 252.0084,
+      "step": 31290
+    },
+    {
+      "epoch": 0.6019780653040422,
+      "grad_norm": 1642.313027731901,
+      "learning_rate": 6.970839641994545e-06,
+      "loss": 249.8814,
+      "step": 31300
+    },
+    {
+      "epoch": 0.6021703905645227,
+      "grad_norm": 1656.1947154348657,
+      "learning_rate": 6.965023806345619e-06,
+      "loss": 249.3657,
+      "step": 31310
+    },
+    {
+      "epoch": 0.6023627158250032,
+      "grad_norm": 1389.1206807563376,
+      "learning_rate": 6.959209101175206e-06,
+      "loss": 248.1517,
+      "step": 31320
+    },
+    {
+      "epoch": 0.6025550410854837,
+      "grad_norm": 1668.3207880499845,
+      "learning_rate": 6.9533955286491805e-06,
+      "loss": 257.285,
+      "step": 31330
+    },
+    {
+      "epoch": 0.6027473663459643,
+      "grad_norm": 1427.8934830695273,
+      "learning_rate": 6.947583090933008e-06,
+      "loss": 245.3844,
+      "step": 31340
+    },
+    {
+      "epoch": 0.6029396916064448,
+      "grad_norm": 1489.9132665636912,
+      "learning_rate": 6.941771790191716e-06,
+      "loss": 247.1532,
+      "step": 31350
+    },
+    {
+      "epoch": 0.6031320168669253,
+      "grad_norm": 1563.59563281525,
+      "learning_rate": 6.9359616285899266e-06,
+      "loss": 253.3851,
+      "step": 31360
+    },
+    {
+      "epoch": 0.6033243421274058,
+      "grad_norm": 1878.7687245126763,
+      "learning_rate": 6.930152608291829e-06,
+      "loss": 259.8763,
+      "step": 31370
+    },
+    {
+      "epoch": 0.6035166673878863,
+      "grad_norm": 1539.902923288817,
+      "learning_rate": 6.924344731461179e-06,
+      "loss": 245.4633,
+      "step": 31380
+    },
+    {
+      "epoch": 0.603708992648367,
+      "grad_norm": 1550.7855968338188,
+      "learning_rate": 6.918538000261325e-06,
+      "loss": 249.0463,
+      "step": 31390
+    },
+    {
+      "epoch": 0.6039013179088475,
+      "grad_norm": 1988.9150506110027,
+      "learning_rate": 6.912732416855171e-06,
+      "loss": 256.2311,
+      "step": 31400
+    },
+    {
+      "epoch": 0.604093643169328,
+      "grad_norm": 1878.9682886092892,
+      "learning_rate": 6.906927983405207e-06,
+      "loss": 257.5113,
+      "step": 31410
+    },
+    {
+      "epoch": 0.6042859684298085,
+      "grad_norm": 1642.4895769428017,
+      "learning_rate": 6.901124702073481e-06,
+      "loss": 249.8725,
+      "step": 31420
+    },
+    {
+      "epoch": 0.604478293690289,
+      "grad_norm": 1876.8221362242273,
+      "learning_rate": 6.895322575021628e-06,
+      "loss": 255.899,
+      "step": 31430
+    },
+    {
+      "epoch": 0.6046706189507696,
+      "grad_norm": 1773.895536924106,
+      "learning_rate": 6.88952160441084e-06,
+      "loss": 256.9432,
+      "step": 31440
+    },
+    {
+      "epoch": 0.6048629442112501,
+      "grad_norm": 1608.965136540702,
+      "learning_rate": 6.8837217924018825e-06,
+      "loss": 250.8332,
+      "step": 31450
+    },
+    {
+      "epoch": 0.6050552694717306,
+      "grad_norm": 1537.358247161311,
+      "learning_rate": 6.877923141155087e-06,
+      "loss": 257.739,
+      "step": 31460
+    },
+    {
+      "epoch": 0.6052475947322111,
+      "grad_norm": 1663.2277440354994,
+      "learning_rate": 6.87212565283036e-06,
+      "loss": 246.3145,
+      "step": 31470
+    },
+    {
+      "epoch": 0.6054399199926916,
+      "grad_norm": 1802.6395685323575,
+      "learning_rate": 6.8663293295871715e-06,
+      "loss": 246.6193,
+      "step": 31480
+    },
+    {
+      "epoch": 0.6056322452531722,
+      "grad_norm": 1522.459516268851,
+      "learning_rate": 6.86053417358455e-06,
+      "loss": 255.4797,
+      "step": 31490
+    },
+    {
+      "epoch": 0.6058245705136527,
+      "grad_norm": 1624.380718400352,
+      "learning_rate": 6.854740186981102e-06,
+      "loss": 251.4926,
+      "step": 31500
+    },
+    {
+      "epoch": 0.6060168957741332,
+      "grad_norm": 1574.675381804736,
+      "learning_rate": 6.848947371934989e-06,
+      "loss": 255.4592,
+      "step": 31510
+    },
+    {
+      "epoch": 0.6062092210346137,
+      "grad_norm": 1681.2388359024635,
+      "learning_rate": 6.843155730603939e-06,
+      "loss": 257.7308,
+      "step": 31520
+    },
+    {
+      "epoch": 0.6064015462950942,
+      "grad_norm": 1530.8075190777176,
+      "learning_rate": 6.837365265145237e-06,
+      "loss": 247.6673,
+      "step": 31530
+    },
+    {
+      "epoch": 0.6065938715555748,
+      "grad_norm": 1628.2402185766443,
+      "learning_rate": 6.831575977715745e-06,
+      "loss": 254.0782,
+      "step": 31540
+    },
+    {
+      "epoch": 0.6067861968160553,
+      "grad_norm": 1544.8008847477113,
+      "learning_rate": 6.825787870471873e-06,
+      "loss": 248.9658,
+      "step": 31550
+    },
+    {
+      "epoch": 0.6069785220765358,
+      "grad_norm": 1867.8608113699281,
+      "learning_rate": 6.820000945569592e-06,
+      "loss": 250.3517,
+      "step": 31560
+    },
+    {
+      "epoch": 0.6071708473370163,
+      "grad_norm": 2162.486715266018,
+      "learning_rate": 6.814215205164444e-06,
+      "loss": 261.1754,
+      "step": 31570
+    },
+    {
+      "epoch": 0.6073631725974968,
+      "grad_norm": 1577.9628754088033,
+      "learning_rate": 6.808430651411518e-06,
+      "loss": 248.5343,
+      "step": 31580
+    },
+    {
+      "epoch": 0.6075554978579775,
+      "grad_norm": 1709.8006448517604,
+      "learning_rate": 6.802647286465461e-06,
+      "loss": 251.4245,
+      "step": 31590
+    },
+    {
+      "epoch": 0.607747823118458,
+      "grad_norm": 1705.761315145192,
+      "learning_rate": 6.796865112480482e-06,
+      "loss": 258.8202,
+      "step": 31600
+    },
+    {
+      "epoch": 0.6079401483789385,
+      "grad_norm": 1484.2652345066156,
+      "learning_rate": 6.79108413161035e-06,
+      "loss": 253.5714,
+      "step": 31610
+    },
+    {
+      "epoch": 0.608132473639419,
+      "grad_norm": 1799.1816316610466,
+      "learning_rate": 6.785304346008381e-06,
+      "loss": 252.6374,
+      "step": 31620
+    },
+    {
+      "epoch": 0.6083247988998995,
+      "grad_norm": 1538.6895950374667,
+      "learning_rate": 6.779525757827452e-06,
+      "loss": 246.7217,
+      "step": 31630
+    },
+    {
+      "epoch": 0.6085171241603801,
+      "grad_norm": 1596.60695218881,
+      "learning_rate": 6.773748369219986e-06,
+      "loss": 255.4618,
+      "step": 31640
+    },
+    {
+      "epoch": 0.6087094494208606,
+      "grad_norm": 1839.616090262443,
+      "learning_rate": 6.767972182337974e-06,
+      "loss": 260.0553,
+      "step": 31650
+    },
+    {
+      "epoch": 0.6089017746813411,
+      "grad_norm": 1613.0026459049998,
+      "learning_rate": 6.762197199332945e-06,
+      "loss": 250.3196,
+      "step": 31660
+    },
+    {
+      "epoch": 0.6090940999418216,
+      "grad_norm": 1491.9956726457312,
+      "learning_rate": 6.756423422355981e-06,
+      "loss": 245.0071,
+      "step": 31670
+    },
+    {
+      "epoch": 0.6092864252023021,
+      "grad_norm": 1555.0605083988942,
+      "learning_rate": 6.750650853557728e-06,
+      "loss": 245.5285,
+      "step": 31680
+    },
+    {
+      "epoch": 0.6094787504627827,
+      "grad_norm": 1657.9337461123548,
+      "learning_rate": 6.744879495088364e-06,
+      "loss": 254.6239,
+      "step": 31690
+    },
+    {
+      "epoch": 0.6096710757232632,
+      "grad_norm": 1903.515552856546,
+      "learning_rate": 6.7391093490976285e-06,
+      "loss": 242.8989,
+      "step": 31700
+    },
+    {
+      "epoch": 0.6098634009837437,
+      "grad_norm": 1601.9733261037613,
+      "learning_rate": 6.7333404177348036e-06,
+      "loss": 243.1104,
+      "step": 31710
+    },
+    {
+      "epoch": 0.6100557262442242,
+      "grad_norm": 1578.7571563074657,
+      "learning_rate": 6.727572703148726e-06,
+      "loss": 245.3339,
+      "step": 31720
+    },
+    {
+      "epoch": 0.6102480515047047,
+      "grad_norm": 1512.1713138921884,
+      "learning_rate": 6.721806207487769e-06,
+      "loss": 252.2113,
+      "step": 31730
+    },
+    {
+      "epoch": 0.6104403767651853,
+      "grad_norm": 1734.8967999615904,
+      "learning_rate": 6.716040932899857e-06,
+      "loss": 258.8036,
+      "step": 31740
+    },
+    {
+      "epoch": 0.6106327020256658,
+      "grad_norm": 1628.0268229886353,
+      "learning_rate": 6.710276881532463e-06,
+      "loss": 251.443,
+      "step": 31750
+    },
+    {
+      "epoch": 0.6108250272861463,
+      "grad_norm": 1549.991520920231,
+      "learning_rate": 6.704514055532597e-06,
+      "loss": 252.4328,
+      "step": 31760
+    },
+    {
+      "epoch": 0.6110173525466268,
+      "grad_norm": 1774.0970598385702,
+      "learning_rate": 6.698752457046822e-06,
+      "loss": 245.6301,
+      "step": 31770
+    },
+    {
+      "epoch": 0.6112096778071073,
+      "grad_norm": 1632.1267733035593,
+      "learning_rate": 6.692992088221231e-06,
+      "loss": 256.3446,
+      "step": 31780
+    },
+    {
+      "epoch": 0.6114020030675879,
+      "grad_norm": 1502.057169906419,
+      "learning_rate": 6.687232951201473e-06,
+      "loss": 246.7599,
+      "step": 31790
+    },
+    {
+      "epoch": 0.6115943283280685,
+      "grad_norm": 1620.8729984258923,
+      "learning_rate": 6.681475048132729e-06,
+      "loss": 259.1275,
+      "step": 31800
+    },
+    {
+      "epoch": 0.611786653588549,
+      "grad_norm": 1744.82182574893,
+      "learning_rate": 6.675718381159719e-06,
+      "loss": 248.6843,
+      "step": 31810
+    },
+    {
+      "epoch": 0.6119789788490295,
+      "grad_norm": 1580.456265980805,
+      "learning_rate": 6.6699629524267114e-06,
+      "loss": 246.2566,
+      "step": 31820
+    },
+    {
+      "epoch": 0.61217130410951,
+      "grad_norm": 1525.056289089862,
+      "learning_rate": 6.664208764077507e-06,
+      "loss": 253.225,
+      "step": 31830
+    },
+    {
+      "epoch": 0.6123636293699906,
+      "grad_norm": 1649.851850165884,
+      "learning_rate": 6.658455818255445e-06,
+      "loss": 243.1506,
+      "step": 31840
+    },
+    {
+      "epoch": 0.6125559546304711,
+      "grad_norm": 1597.9424416727297,
+      "learning_rate": 6.652704117103401e-06,
+      "loss": 244.7898,
+      "step": 31850
+    },
+    {
+      "epoch": 0.6127482798909516,
+      "grad_norm": 1602.6126692859373,
+      "learning_rate": 6.646953662763796e-06,
+      "loss": 261.2959,
+      "step": 31860
+    },
+    {
+      "epoch": 0.6129406051514321,
+      "grad_norm": 1549.4641554031066,
+      "learning_rate": 6.6412044573785725e-06,
+      "loss": 250.3392,
+      "step": 31870
+    },
+    {
+      "epoch": 0.6131329304119126,
+      "grad_norm": 1582.955431582562,
+      "learning_rate": 6.635456503089217e-06,
+      "loss": 249.266,
+      "step": 31880
+    },
+    {
+      "epoch": 0.6133252556723932,
+      "grad_norm": 1792.818896541898,
+      "learning_rate": 6.6297098020367435e-06,
+      "loss": 265.8789,
+      "step": 31890
+    },
+    {
+      "epoch": 0.6135175809328737,
+      "grad_norm": 1719.069564256674,
+      "learning_rate": 6.623964356361707e-06,
+      "loss": 264.3555,
+      "step": 31900
+    },
+    {
+      "epoch": 0.6137099061933542,
+      "grad_norm": 1726.9317400608413,
+      "learning_rate": 6.618220168204193e-06,
+      "loss": 247.0801,
+      "step": 31910
+    },
+    {
+      "epoch": 0.6139022314538347,
+      "grad_norm": 1510.7470892740862,
+      "learning_rate": 6.6124772397038115e-06,
+      "loss": 250.4055,
+      "step": 31920
+    },
+    {
+      "epoch": 0.6140945567143152,
+      "grad_norm": 1613.3241614756466,
+      "learning_rate": 6.606735572999714e-06,
+      "loss": 256.8448,
+      "step": 31930
+    },
+    {
+      "epoch": 0.6142868819747958,
+      "grad_norm": 1454.3553455606593,
+      "learning_rate": 6.600995170230575e-06,
+      "loss": 239.996,
+      "step": 31940
+    },
+    {
+      "epoch": 0.6144792072352763,
+      "grad_norm": 1589.8753590643007,
+      "learning_rate": 6.595256033534598e-06,
+      "loss": 245.4918,
+      "step": 31950
+    },
+    {
+      "epoch": 0.6146715324957568,
+      "grad_norm": 1788.2335614252272,
+      "learning_rate": 6.589518165049514e-06,
+      "loss": 246.8885,
+      "step": 31960
+    },
+    {
+      "epoch": 0.6148638577562373,
+      "grad_norm": 1608.8035911286963,
+      "learning_rate": 6.5837815669125906e-06,
+      "loss": 252.2777,
+      "step": 31970
+    },
+    {
+      "epoch": 0.6150561830167178,
+      "grad_norm": 1471.079507708029,
+      "learning_rate": 6.5780462412606124e-06,
+      "loss": 238.6989,
+      "step": 31980
+    },
+    {
+      "epoch": 0.6152485082771983,
+      "grad_norm": 1511.0155152253,
+      "learning_rate": 6.572312190229895e-06,
+      "loss": 247.5496,
+      "step": 31990
+    },
+    {
+      "epoch": 0.615440833537679,
+      "grad_norm": 1614.8713740986534,
+      "learning_rate": 6.5665794159562734e-06,
+      "loss": 246.021,
+      "step": 32000
+    },
+    {
+      "epoch": 0.6156331587981595,
+      "grad_norm": 1623.9992160658344,
+      "learning_rate": 6.560847920575118e-06,
+      "loss": 242.5339,
+      "step": 32010
+    },
+    {
+      "epoch": 0.61582548405864,
+      "grad_norm": 1594.1458627122008,
+      "learning_rate": 6.5551177062213126e-06,
+      "loss": 255.3999,
+      "step": 32020
+    },
+    {
+      "epoch": 0.6160178093191205,
+      "grad_norm": 1628.471693623422,
+      "learning_rate": 6.5493887750292616e-06,
+      "loss": 246.7218,
+      "step": 32030
+    },
+    {
+      "epoch": 0.616210134579601,
+      "grad_norm": 1476.5325397406718,
+      "learning_rate": 6.54366112913291e-06,
+      "loss": 257.9446,
+      "step": 32040
+    },
+    {
+      "epoch": 0.6164024598400816,
+      "grad_norm": 1857.1994670994266,
+      "learning_rate": 6.537934770665701e-06,
+      "loss": 256.731,
+      "step": 32050
+    },
+    {
+      "epoch": 0.6165947851005621,
+      "grad_norm": 1629.5665638543069,
+      "learning_rate": 6.532209701760615e-06,
+      "loss": 260.7297,
+      "step": 32060
+    },
+    {
+      "epoch": 0.6167871103610426,
+      "grad_norm": 1568.8003140103235,
+      "learning_rate": 6.526485924550138e-06,
+      "loss": 254.8337,
+      "step": 32070
+    },
+    {
+      "epoch": 0.6169794356215231,
+      "grad_norm": 2038.3809150575157,
+      "learning_rate": 6.520763441166291e-06,
+      "loss": 253.361,
+      "step": 32080
+    },
+    {
+      "epoch": 0.6171717608820037,
+      "grad_norm": 1476.8358441094674,
+      "learning_rate": 6.515042253740601e-06,
+      "loss": 237.9503,
+      "step": 32090
+    },
+    {
+      "epoch": 0.6173640861424842,
+      "grad_norm": 1765.9492939359543,
+      "learning_rate": 6.509322364404112e-06,
+      "loss": 252.383,
+      "step": 32100
+    },
+    {
+      "epoch": 0.6175564114029647,
+      "grad_norm": 1488.644221621674,
+      "learning_rate": 6.503603775287395e-06,
+      "loss": 249.6691,
+      "step": 32110
+    },
+    {
+      "epoch": 0.6177487366634452,
+      "grad_norm": 1583.3130400077239,
+      "learning_rate": 6.497886488520524e-06,
+      "loss": 245.5295,
+      "step": 32120
+    },
+    {
+      "epoch": 0.6179410619239257,
+      "grad_norm": 1683.3685894768932,
+      "learning_rate": 6.4921705062331e-06,
+      "loss": 257.5813,
+      "step": 32130
+    },
+    {
+      "epoch": 0.6181333871844062,
+      "grad_norm": 1682.7045364901157,
+      "learning_rate": 6.486455830554224e-06,
+      "loss": 244.7124,
+      "step": 32140
+    },
+    {
+      "epoch": 0.6183257124448868,
+      "grad_norm": 1877.0490159193469,
+      "learning_rate": 6.4807424636125285e-06,
+      "loss": 249.3643,
+      "step": 32150
+    },
+    {
+      "epoch": 0.6185180377053673,
+      "grad_norm": 1700.0408400252504,
+      "learning_rate": 6.475030407536141e-06,
+      "loss": 251.4053,
+      "step": 32160
+    },
+    {
+      "epoch": 0.6187103629658478,
+      "grad_norm": 1493.8667077487155,
+      "learning_rate": 6.469319664452709e-06,
+      "loss": 249.9095,
+      "step": 32170
+    },
+    {
+      "epoch": 0.6189026882263283,
+      "grad_norm": 1520.7891904512587,
+      "learning_rate": 6.463610236489391e-06,
+      "loss": 242.9274,
+      "step": 32180
+    },
+    {
+      "epoch": 0.6190950134868088,
+      "grad_norm": 1751.6989459226484,
+      "learning_rate": 6.457902125772854e-06,
+      "loss": 249.4639,
+      "step": 32190
+    },
+    {
+      "epoch": 0.6192873387472894,
+      "grad_norm": 1928.0840012032722,
+      "learning_rate": 6.452195334429277e-06,
+      "loss": 252.8523,
+      "step": 32200
+    },
+    {
+      "epoch": 0.61947966400777,
+      "grad_norm": 1546.1189550341815,
+      "learning_rate": 6.446489864584341e-06,
+      "loss": 246.3375,
+      "step": 32210
+    },
+    {
+      "epoch": 0.6196719892682505,
+      "grad_norm": 1570.5477599560918,
+      "learning_rate": 6.440785718363245e-06,
+      "loss": 245.6131,
+      "step": 32220
+    },
+    {
+      "epoch": 0.619864314528731,
+      "grad_norm": 1739.4972611254748,
+      "learning_rate": 6.435082897890688e-06,
+      "loss": 246.1498,
+      "step": 32230
+    },
+    {
+      "epoch": 0.6200566397892115,
+      "grad_norm": 1678.4469029802467,
+      "learning_rate": 6.429381405290873e-06,
+      "loss": 246.2843,
+      "step": 32240
+    },
+    {
+      "epoch": 0.6202489650496921,
+      "grad_norm": 1634.775324897167,
+      "learning_rate": 6.4236812426875124e-06,
+      "loss": 250.7838,
+      "step": 32250
+    },
+    {
+      "epoch": 0.6204412903101726,
+      "grad_norm": 1589.5214058834304,
+      "learning_rate": 6.4179824122038244e-06,
+      "loss": 249.8279,
+      "step": 32260
+    },
+    {
+      "epoch": 0.6206336155706531,
+      "grad_norm": 1758.1687057063032,
+      "learning_rate": 6.412284915962532e-06,
+      "loss": 250.6428,
+      "step": 32270
+    },
+    {
+      "epoch": 0.6208259408311336,
+      "grad_norm": 1590.9530138235439,
+      "learning_rate": 6.406588756085849e-06,
+      "loss": 244.9568,
+      "step": 32280
+    },
+    {
+      "epoch": 0.6210182660916141,
+      "grad_norm": 1608.8235312338588,
+      "learning_rate": 6.400893934695514e-06,
+      "loss": 247.0817,
+      "step": 32290
+    },
+    {
+      "epoch": 0.6212105913520947,
+      "grad_norm": 1598.3996115572095,
+      "learning_rate": 6.395200453912747e-06,
+      "loss": 253.0403,
+      "step": 32300
+    },
+    {
+      "epoch": 0.6214029166125752,
+      "grad_norm": 1485.0043715057932,
+      "learning_rate": 6.389508315858272e-06,
+      "loss": 250.6169,
+      "step": 32310
+    },
+    {
+      "epoch": 0.6215952418730557,
+      "grad_norm": 1706.0132389886498,
+      "learning_rate": 6.38381752265232e-06,
+      "loss": 247.6966,
+      "step": 32320
+    },
+    {
+      "epoch": 0.6217875671335362,
+      "grad_norm": 1940.961766426073,
+      "learning_rate": 6.378128076414619e-06,
+      "loss": 252.8314,
+      "step": 32330
+    },
+    {
+      "epoch": 0.6219798923940167,
+      "grad_norm": 1751.9731911783938,
+      "learning_rate": 6.372439979264393e-06,
+      "loss": 253.346,
+      "step": 32340
+    },
+    {
+      "epoch": 0.6221722176544973,
+      "grad_norm": 1580.5134141952267,
+      "learning_rate": 6.3667532333203655e-06,
+      "loss": 243.3838,
+      "step": 32350
+    },
+    {
+      "epoch": 0.6223645429149778,
+      "grad_norm": 1534.8565407098354,
+      "learning_rate": 6.361067840700747e-06,
+      "loss": 245.5675,
+      "step": 32360
+    },
+    {
+      "epoch": 0.6225568681754583,
+      "grad_norm": 1535.1015959866318,
+      "learning_rate": 6.355383803523265e-06,
+      "loss": 246.4744,
+      "step": 32370
+    },
+    {
+      "epoch": 0.6227491934359388,
+      "grad_norm": 1521.4844756247937,
+      "learning_rate": 6.349701123905123e-06,
+      "loss": 248.2752,
+      "step": 32380
+    },
+    {
+      "epoch": 0.6229415186964193,
+      "grad_norm": 1585.2650509553055,
+      "learning_rate": 6.344019803963021e-06,
+      "loss": 250.1241,
+      "step": 32390
+    },
+    {
+      "epoch": 0.6231338439568999,
+      "grad_norm": 1618.6602149788982,
+      "learning_rate": 6.338339845813164e-06,
+      "loss": 245.747,
+      "step": 32400
+    },
+    {
+      "epoch": 0.6233261692173805,
+      "grad_norm": 1601.0337287187801,
+      "learning_rate": 6.332661251571241e-06,
+      "loss": 245.6743,
+      "step": 32410
+    },
+    {
+      "epoch": 0.623518494477861,
+      "grad_norm": 1538.1357091455461,
+      "learning_rate": 6.326984023352435e-06,
+      "loss": 252.085,
+      "step": 32420
+    },
+    {
+      "epoch": 0.6237108197383415,
+      "grad_norm": 1689.5525904868528,
+      "learning_rate": 6.321308163271413e-06,
+      "loss": 247.09,
+      "step": 32430
+    },
+    {
+      "epoch": 0.623903144998822,
+      "grad_norm": 1607.63037700626,
+      "learning_rate": 6.315633673442349e-06,
+      "loss": 248.4497,
+      "step": 32440
+    },
+    {
+      "epoch": 0.6240954702593026,
+      "grad_norm": 1688.4707478131659,
+      "learning_rate": 6.309960555978894e-06,
+      "loss": 243.1952,
+      "step": 32450
+    },
+    {
+      "epoch": 0.6242877955197831,
+      "grad_norm": 1589.6763783257345,
+      "learning_rate": 6.304288812994183e-06,
+      "loss": 245.1268,
+      "step": 32460
+    },
+    {
+      "epoch": 0.6244801207802636,
+      "grad_norm": 1567.0129232821203,
+      "learning_rate": 6.298618446600856e-06,
+      "loss": 256.4447,
+      "step": 32470
+    },
+    {
+      "epoch": 0.6246724460407441,
+      "grad_norm": 1567.832361528433,
+      "learning_rate": 6.292949458911029e-06,
+      "loss": 248.8332,
+      "step": 32480
+    },
+    {
+      "epoch": 0.6248647713012246,
+      "grad_norm": 1622.7786841280042,
+      "learning_rate": 6.287281852036304e-06,
+      "loss": 242.8957,
+      "step": 32490
+    },
+    {
+      "epoch": 0.6250570965617052,
+      "grad_norm": 1574.65614154125,
+      "learning_rate": 6.2816156280877675e-06,
+      "loss": 245.2473,
+      "step": 32500
+    },
+    {
+      "epoch": 0.6252494218221857,
+      "grad_norm": 1450.0720051573194,
+      "learning_rate": 6.275950789176002e-06,
+      "loss": 248.8614,
+      "step": 32510
+    },
+    {
+      "epoch": 0.6254417470826662,
+      "grad_norm": 2114.7601680932303,
+      "learning_rate": 6.270287337411064e-06,
+      "loss": 244.607,
+      "step": 32520
+    },
+    {
+      "epoch": 0.6256340723431467,
+      "grad_norm": 1648.1336613282983,
+      "learning_rate": 6.264625274902492e-06,
+      "loss": 251.1316,
+      "step": 32530
+    },
+    {
+      "epoch": 0.6258263976036272,
+      "grad_norm": 1705.3451392061618,
+      "learning_rate": 6.258964603759311e-06,
+      "loss": 253.2709,
+      "step": 32540
+    },
+    {
+      "epoch": 0.6260187228641078,
+      "grad_norm": 1757.0387141401993,
+      "learning_rate": 6.2533053260900345e-06,
+      "loss": 245.0642,
+      "step": 32550
+    },
+    {
+      "epoch": 0.6262110481245883,
+      "grad_norm": 1727.526741065685,
+      "learning_rate": 6.247647444002644e-06,
+      "loss": 248.3847,
+      "step": 32560
+    },
+    {
+      "epoch": 0.6264033733850688,
+      "grad_norm": 1566.666573390075,
+      "learning_rate": 6.241990959604607e-06,
+      "loss": 254.7402,
+      "step": 32570
+    },
+    {
+      "epoch": 0.6265956986455493,
+      "grad_norm": 1579.8128116168036,
+      "learning_rate": 6.2363358750028745e-06,
+      "loss": 246.6831,
+      "step": 32580
+    },
+    {
+      "epoch": 0.6267880239060298,
+      "grad_norm": 1567.6092283979997,
+      "learning_rate": 6.23068219230387e-06,
+      "loss": 242.8022,
+      "step": 32590
+    },
+    {
+      "epoch": 0.6269803491665104,
+      "grad_norm": 1494.8954769744298,
+      "learning_rate": 6.2250299136134925e-06,
+      "loss": 242.8416,
+      "step": 32600
+    },
+    {
+      "epoch": 0.6271726744269909,
+      "grad_norm": 1507.4813117435835,
+      "learning_rate": 6.219379041037128e-06,
+      "loss": 248.8065,
+      "step": 32610
+    },
+    {
+      "epoch": 0.6273649996874715,
+      "grad_norm": 1506.6896920838308,
+      "learning_rate": 6.213729576679632e-06,
+      "loss": 246.772,
+      "step": 32620
+    },
+    {
+      "epoch": 0.627557324947952,
+      "grad_norm": 1626.2599369080062,
+      "learning_rate": 6.208081522645339e-06,
+      "loss": 241.8009,
+      "step": 32630
+    },
+    {
+      "epoch": 0.6277496502084325,
+      "grad_norm": 1647.035360325793,
+      "learning_rate": 6.202434881038048e-06,
+      "loss": 245.0505,
+      "step": 32640
+    },
+    {
+      "epoch": 0.6279419754689131,
+      "grad_norm": 1507.5611481247427,
+      "learning_rate": 6.196789653961048e-06,
+      "loss": 243.504,
+      "step": 32650
+    },
+    {
+      "epoch": 0.6281343007293936,
+      "grad_norm": 1442.3763540100265,
+      "learning_rate": 6.191145843517093e-06,
+      "loss": 244.2661,
+      "step": 32660
+    },
+    {
+      "epoch": 0.6283266259898741,
+      "grad_norm": 1768.6358707894049,
+      "learning_rate": 6.185503451808401e-06,
+      "loss": 248.468,
+      "step": 32670
+    },
+    {
+      "epoch": 0.6285189512503546,
+      "grad_norm": 1537.5944360458418,
+      "learning_rate": 6.1798624809366755e-06,
+      "loss": 254.1811,
+      "step": 32680
+    },
+    {
+      "epoch": 0.6287112765108351,
+      "grad_norm": 1708.2752417297472,
+      "learning_rate": 6.174222933003084e-06,
+      "loss": 248.4348,
+      "step": 32690
+    },
+    {
+      "epoch": 0.6289036017713157,
+      "grad_norm": 1619.463492778449,
+      "learning_rate": 6.168584810108269e-06,
+      "loss": 249.2467,
+      "step": 32700
+    },
+    {
+      "epoch": 0.6290959270317962,
+      "grad_norm": 1816.4470533788578,
+      "learning_rate": 6.162948114352328e-06,
+      "loss": 254.6911,
+      "step": 32710
+    },
+    {
+      "epoch": 0.6292882522922767,
+      "grad_norm": 1624.2993518406572,
+      "learning_rate": 6.157312847834848e-06,
+      "loss": 248.102,
+      "step": 32720
+    },
+    {
+      "epoch": 0.6294805775527572,
+      "grad_norm": 1599.3784062367943,
+      "learning_rate": 6.1516790126548695e-06,
+      "loss": 245.2853,
+      "step": 32730
+    },
+    {
+      "epoch": 0.6296729028132377,
+      "grad_norm": 1589.5099275240586,
+      "learning_rate": 6.146046610910899e-06,
+      "loss": 248.4512,
+      "step": 32740
+    },
+    {
+      "epoch": 0.6298652280737183,
+      "grad_norm": 1536.989171143397,
+      "learning_rate": 6.140415644700915e-06,
+      "loss": 257.6639,
+      "step": 32750
+    },
+    {
+      "epoch": 0.6300575533341988,
+      "grad_norm": 1837.6683537179802,
+      "learning_rate": 6.13478611612236e-06,
+      "loss": 249.9978,
+      "step": 32760
+    },
+    {
+      "epoch": 0.6302498785946793,
+      "grad_norm": 1539.785434588748,
+      "learning_rate": 6.129158027272144e-06,
+      "loss": 246.3008,
+      "step": 32770
+    },
+    {
+      "epoch": 0.6304422038551598,
+      "grad_norm": 1828.6763877484193,
+      "learning_rate": 6.123531380246632e-06,
+      "loss": 243.6701,
+      "step": 32780
+    },
+    {
+      "epoch": 0.6306345291156403,
+      "grad_norm": 1518.9369019170892,
+      "learning_rate": 6.117906177141657e-06,
+      "loss": 250.4036,
+      "step": 32790
+    },
+    {
+      "epoch": 0.6308268543761208,
+      "grad_norm": 1770.785546728867,
+      "learning_rate": 6.112282420052518e-06,
+      "loss": 243.2777,
+      "step": 32800
+    },
+    {
+      "epoch": 0.6310191796366014,
+      "grad_norm": 1607.8479079552997,
+      "learning_rate": 6.106660111073969e-06,
+      "loss": 247.6805,
+      "step": 32810
+    },
+    {
+      "epoch": 0.631211504897082,
+      "grad_norm": 1532.8465484716394,
+      "learning_rate": 6.101039252300227e-06,
+      "loss": 250.477,
+      "step": 32820
+    },
+    {
+      "epoch": 0.6314038301575625,
+      "grad_norm": 1583.6556477459974,
+      "learning_rate": 6.095419845824971e-06,
+      "loss": 245.806,
+      "step": 32830
+    },
+    {
+      "epoch": 0.631596155418043,
+      "grad_norm": 1565.7516005574798,
+      "learning_rate": 6.089801893741338e-06,
+      "loss": 247.4883,
+      "step": 32840
+    },
+    {
+      "epoch": 0.6317884806785236,
+      "grad_norm": 1578.300650843023,
+      "learning_rate": 6.084185398141921e-06,
+      "loss": 246.5214,
+      "step": 32850
+    },
+    {
+      "epoch": 0.6319808059390041,
+      "grad_norm": 1534.5244133975293,
+      "learning_rate": 6.078570361118768e-06,
+      "loss": 249.1956,
+      "step": 32860
+    },
+    {
+      "epoch": 0.6321731311994846,
+      "grad_norm": 1595.3661320643646,
+      "learning_rate": 6.072956784763393e-06,
+      "loss": 246.858,
+      "step": 32870
+    },
+    {
+      "epoch": 0.6323654564599651,
+      "grad_norm": 1477.1977708577524,
+      "learning_rate": 6.067344671166757e-06,
+      "loss": 249.2065,
+      "step": 32880
+    },
+    {
+      "epoch": 0.6325577817204456,
+      "grad_norm": 1663.6308689361103,
+      "learning_rate": 6.061734022419281e-06,
+      "loss": 246.4378,
+      "step": 32890
+    },
+    {
+      "epoch": 0.6327501069809262,
+      "grad_norm": 1576.8311054849837,
+      "learning_rate": 6.056124840610839e-06,
+      "loss": 236.941,
+      "step": 32900
+    },
+    {
+      "epoch": 0.6329424322414067,
+      "grad_norm": 1718.6030339996457,
+      "learning_rate": 6.050517127830761e-06,
+      "loss": 248.3223,
+      "step": 32910
+    },
+    {
+      "epoch": 0.6331347575018872,
+      "grad_norm": 1720.5924619205668,
+      "learning_rate": 6.044910886167825e-06,
+      "loss": 255.572,
+      "step": 32920
+    },
+    {
+      "epoch": 0.6333270827623677,
+      "grad_norm": 1711.6160043458594,
+      "learning_rate": 6.03930611771026e-06,
+      "loss": 243.622,
+      "step": 32930
+    },
+    {
+      "epoch": 0.6335194080228482,
+      "grad_norm": 1727.5870155952293,
+      "learning_rate": 6.033702824545755e-06,
+      "loss": 240.5427,
+      "step": 32940
+    },
+    {
+      "epoch": 0.6337117332833287,
+      "grad_norm": 1638.667753915611,
+      "learning_rate": 6.028101008761445e-06,
+      "loss": 236.3573,
+      "step": 32950
+    },
+    {
+      "epoch": 0.6339040585438093,
+      "grad_norm": 1629.6333799236143,
+      "learning_rate": 6.022500672443907e-06,
+      "loss": 244.3131,
+      "step": 32960
+    },
+    {
+      "epoch": 0.6340963838042898,
+      "grad_norm": 1663.290947145034,
+      "learning_rate": 6.016901817679177e-06,
+      "loss": 252.1628,
+      "step": 32970
+    },
+    {
+      "epoch": 0.6342887090647703,
+      "grad_norm": 1412.7271375235703,
+      "learning_rate": 6.011304446552741e-06,
+      "loss": 241.2882,
+      "step": 32980
+    },
+    {
+      "epoch": 0.6344810343252508,
+      "grad_norm": 1594.4454825745604,
+      "learning_rate": 6.005708561149523e-06,
+      "loss": 246.0481,
+      "step": 32990
+    },
+    {
+      "epoch": 0.6346733595857313,
+      "grad_norm": 1707.5138202281103,
+      "learning_rate": 6.000114163553894e-06,
+      "loss": 249.547,
+      "step": 33000
+    },
+    {
+      "epoch": 0.6348656848462119,
+      "grad_norm": 1528.2134605789608,
+      "learning_rate": 5.994521255849684e-06,
+      "loss": 251.3665,
+      "step": 33010
+    },
+    {
+      "epoch": 0.6350580101066924,
+      "grad_norm": 1478.1877711859377,
+      "learning_rate": 5.988929840120151e-06,
+      "loss": 248.4078,
+      "step": 33020
+    },
+    {
+      "epoch": 0.635250335367173,
+      "grad_norm": 1769.7989304924388,
+      "learning_rate": 5.983339918448008e-06,
+      "loss": 246.3821,
+      "step": 33030
+    },
+    {
+      "epoch": 0.6354426606276535,
+      "grad_norm": 1623.5398650186996,
+      "learning_rate": 5.977751492915404e-06,
+      "loss": 241.6955,
+      "step": 33040
+    },
+    {
+      "epoch": 0.635634985888134,
+      "grad_norm": 1568.3239354920047,
+      "learning_rate": 5.972164565603944e-06,
+      "loss": 246.2752,
+      "step": 33050
+    },
+    {
+      "epoch": 0.6358273111486146,
+      "grad_norm": 1620.8237772235327,
+      "learning_rate": 5.966579138594661e-06,
+      "loss": 253.7861,
+      "step": 33060
+    },
+    {
+      "epoch": 0.6360196364090951,
+      "grad_norm": 1539.9819747996405,
+      "learning_rate": 5.960995213968033e-06,
+      "loss": 248.4565,
+      "step": 33070
+    },
+    {
+      "epoch": 0.6362119616695756,
+      "grad_norm": 1596.4702876001725,
+      "learning_rate": 5.955412793803983e-06,
+      "loss": 242.4913,
+      "step": 33080
+    },
+    {
+      "epoch": 0.6364042869300561,
+      "grad_norm": 1980.138686396594,
+      "learning_rate": 5.949831880181869e-06,
+      "loss": 259.1184,
+      "step": 33090
+    },
+    {
+      "epoch": 0.6365966121905366,
+      "grad_norm": 1562.2473903526125,
+      "learning_rate": 5.944252475180487e-06,
+      "loss": 256.8562,
+      "step": 33100
+    },
+    {
+      "epoch": 0.6367889374510172,
+      "grad_norm": 1515.1407736467438,
+      "learning_rate": 5.938674580878077e-06,
+      "loss": 246.3608,
+      "step": 33110
+    },
+    {
+      "epoch": 0.6369812627114977,
+      "grad_norm": 1441.905515966365,
+      "learning_rate": 5.933098199352307e-06,
+      "loss": 250.6053,
+      "step": 33120
+    },
+    {
+      "epoch": 0.6371735879719782,
+      "grad_norm": 1629.1783300601244,
+      "learning_rate": 5.927523332680296e-06,
+      "loss": 250.9615,
+      "step": 33130
+    },
+    {
+      "epoch": 0.6373659132324587,
+      "grad_norm": 1641.1701206727084,
+      "learning_rate": 5.921949982938583e-06,
+      "loss": 245.559,
+      "step": 33140
+    },
+    {
+      "epoch": 0.6375582384929392,
+      "grad_norm": 1596.1086023443413,
+      "learning_rate": 5.916378152203146e-06,
+      "loss": 249.9641,
+      "step": 33150
+    },
+    {
+      "epoch": 0.6377505637534198,
+      "grad_norm": 1427.0600192349104,
+      "learning_rate": 5.910807842549409e-06,
+      "loss": 244.6634,
+      "step": 33160
+    },
+    {
+      "epoch": 0.6379428890139003,
+      "grad_norm": 1662.3631962510083,
+      "learning_rate": 5.905239056052212e-06,
+      "loss": 242.7421,
+      "step": 33170
+    },
+    {
+      "epoch": 0.6381352142743808,
+      "grad_norm": 1665.114998134074,
+      "learning_rate": 5.8996717947858395e-06,
+      "loss": 245.7042,
+      "step": 33180
+    },
+    {
+      "epoch": 0.6383275395348613,
+      "grad_norm": 1519.6548917706268,
+      "learning_rate": 5.894106060824005e-06,
+      "loss": 245.6331,
+      "step": 33190
+    },
+    {
+      "epoch": 0.6385198647953418,
+      "grad_norm": 1549.21597966117,
+      "learning_rate": 5.888541856239854e-06,
+      "loss": 241.5423,
+      "step": 33200
+    },
+    {
+      "epoch": 0.6387121900558224,
+      "grad_norm": 1652.5725857712496,
+      "learning_rate": 5.882979183105959e-06,
+      "loss": 245.7441,
+      "step": 33210
+    },
+    {
+      "epoch": 0.6389045153163029,
+      "grad_norm": 1384.0270437995769,
+      "learning_rate": 5.8774180434943184e-06,
+      "loss": 242.0473,
+      "step": 33220
+    },
+    {
+      "epoch": 0.6390968405767835,
+      "grad_norm": 1653.192009309887,
+      "learning_rate": 5.871858439476374e-06,
+      "loss": 245.5735,
+      "step": 33230
+    },
+    {
+      "epoch": 0.639289165837264,
+      "grad_norm": 1547.3408590604622,
+      "learning_rate": 5.86630037312298e-06,
+      "loss": 244.2901,
+      "step": 33240
+    },
+    {
+      "epoch": 0.6394814910977445,
+      "grad_norm": 1613.1810826595165,
+      "learning_rate": 5.860743846504429e-06,
+      "loss": 249.6994,
+      "step": 33250
+    },
+    {
+      "epoch": 0.6396738163582251,
+      "grad_norm": 1473.0402502223385,
+      "learning_rate": 5.8551888616904305e-06,
+      "loss": 246.9429,
+      "step": 33260
+    },
+    {
+      "epoch": 0.6398661416187056,
+      "grad_norm": 1517.465889778779,
+      "learning_rate": 5.849635420750131e-06,
+      "loss": 255.199,
+      "step": 33270
+    },
+    {
+      "epoch": 0.6400584668791861,
+      "grad_norm": 1592.9883626456392,
+      "learning_rate": 5.84408352575209e-06,
+      "loss": 245.4556,
+      "step": 33280
+    },
+    {
+      "epoch": 0.6402507921396666,
+      "grad_norm": 1729.6262112804789,
+      "learning_rate": 5.8385331787642956e-06,
+      "loss": 238.5987,
+      "step": 33290
+    },
+    {
+      "epoch": 0.6404431174001471,
+      "grad_norm": 1580.4064431986308,
+      "learning_rate": 5.8329843818541665e-06,
+      "loss": 247.3982,
+      "step": 33300
+    },
+    {
+      "epoch": 0.6406354426606277,
+      "grad_norm": 1785.1346877032258,
+      "learning_rate": 5.827437137088535e-06,
+      "loss": 236.1925,
+      "step": 33310
+    },
+    {
+      "epoch": 0.6408277679211082,
+      "grad_norm": 1765.173873381017,
+      "learning_rate": 5.8218914465336585e-06,
+      "loss": 244.7444,
+      "step": 33320
+    },
+    {
+      "epoch": 0.6410200931815887,
+      "grad_norm": 1687.537795474892,
+      "learning_rate": 5.816347312255209e-06,
+      "loss": 242.8977,
+      "step": 33330
+    },
+    {
+      "epoch": 0.6412124184420692,
+      "grad_norm": 1580.8500233371835,
+      "learning_rate": 5.810804736318295e-06,
+      "loss": 239.8719,
+      "step": 33340
+    },
+    {
+      "epoch": 0.6414047437025497,
+      "grad_norm": 1704.6797554911557,
+      "learning_rate": 5.805263720787426e-06,
+      "loss": 235.9653,
+      "step": 33350
+    },
+    {
+      "epoch": 0.6415970689630303,
+      "grad_norm": 1534.0302203848453,
+      "learning_rate": 5.799724267726547e-06,
+      "loss": 252.2206,
+      "step": 33360
+    },
+    {
+      "epoch": 0.6417893942235108,
+      "grad_norm": 1427.7885876304517,
+      "learning_rate": 5.794186379199004e-06,
+      "loss": 245.9651,
+      "step": 33370
+    },
+    {
+      "epoch": 0.6419817194839913,
+      "grad_norm": 1441.674886182865,
+      "learning_rate": 5.7886500572675774e-06,
+      "loss": 251.7738,
+      "step": 33380
+    },
+    {
+      "epoch": 0.6421740447444718,
+      "grad_norm": 1562.9216476791923,
+      "learning_rate": 5.783115303994451e-06,
+      "loss": 250.4023,
+      "step": 33390
+    },
+    {
+      "epoch": 0.6423663700049523,
+      "grad_norm": 1920.4788053768436,
+      "learning_rate": 5.777582121441227e-06,
+      "loss": 251.1658,
+      "step": 33400
+    },
+    {
+      "epoch": 0.6425586952654329,
+      "grad_norm": 1941.7802402947027,
+      "learning_rate": 5.772050511668931e-06,
+      "loss": 246.5123,
+      "step": 33410
+    },
+    {
+      "epoch": 0.6427510205259134,
+      "grad_norm": 1568.5024031214668,
+      "learning_rate": 5.766520476737993e-06,
+      "loss": 249.1046,
+      "step": 33420
+    },
+    {
+      "epoch": 0.642943345786394,
+      "grad_norm": 1567.923840090729,
+      "learning_rate": 5.760992018708253e-06,
+      "loss": 243.7918,
+      "step": 33430
+    },
+    {
+      "epoch": 0.6431356710468745,
+      "grad_norm": 1486.9123075394398,
+      "learning_rate": 5.755465139638983e-06,
+      "loss": 232.4583,
+      "step": 33440
+    },
+    {
+      "epoch": 0.643327996307355,
+      "grad_norm": 1570.8365556836536,
+      "learning_rate": 5.749939841588846e-06,
+      "loss": 255.1207,
+      "step": 33450
+    },
+    {
+      "epoch": 0.6435203215678356,
+      "grad_norm": 1407.0351624994057,
+      "learning_rate": 5.744416126615926e-06,
+      "loss": 243.4232,
+      "step": 33460
+    },
+    {
+      "epoch": 0.6437126468283161,
+      "grad_norm": 1580.1842310231264,
+      "learning_rate": 5.738893996777713e-06,
+      "loss": 249.9519,
+      "step": 33470
+    },
+    {
+      "epoch": 0.6439049720887966,
+      "grad_norm": 1593.3669439472326,
+      "learning_rate": 5.7333734541311144e-06,
+      "loss": 240.081,
+      "step": 33480
+    },
+    {
+      "epoch": 0.6440972973492771,
+      "grad_norm": 2612.285670751348,
+      "learning_rate": 5.727854500732435e-06,
+      "loss": 250.6562,
+      "step": 33490
+    },
+    {
+      "epoch": 0.6442896226097576,
+      "grad_norm": 1539.7385637260627,
+      "learning_rate": 5.722337138637402e-06,
+      "loss": 240.3438,
+      "step": 33500
+    },
+    {
+      "epoch": 0.6444819478702382,
+      "grad_norm": 1403.7126688946553,
+      "learning_rate": 5.716821369901131e-06,
+      "loss": 243.7391,
+      "step": 33510
+    },
+    {
+      "epoch": 0.6446742731307187,
+      "grad_norm": 1528.1058627803038,
+      "learning_rate": 5.711307196578166e-06,
+      "loss": 238.9891,
+      "step": 33520
+    },
+    {
+      "epoch": 0.6448665983911992,
+      "grad_norm": 1563.8369730986801,
+      "learning_rate": 5.70579462072244e-06,
+      "loss": 241.4954,
+      "step": 33530
+    },
+    {
+      "epoch": 0.6450589236516797,
+      "grad_norm": 1537.7667829895724,
+      "learning_rate": 5.700283644387292e-06,
+      "loss": 245.0168,
+      "step": 33540
+    },
+    {
+      "epoch": 0.6452512489121602,
+      "grad_norm": 1595.3647465891213,
+      "learning_rate": 5.6947742696254794e-06,
+      "loss": 249.0262,
+      "step": 33550
+    },
+    {
+      "epoch": 0.6454435741726408,
+      "grad_norm": 1668.3365718955092,
+      "learning_rate": 5.689266498489148e-06,
+      "loss": 245.9052,
+      "step": 33560
+    },
+    {
+      "epoch": 0.6456358994331213,
+      "grad_norm": 1498.947090389856,
+      "learning_rate": 5.683760333029851e-06,
+      "loss": 248.3906,
+      "step": 33570
+    },
+    {
+      "epoch": 0.6458282246936018,
+      "grad_norm": 1701.087014436955,
+      "learning_rate": 5.678255775298542e-06,
+      "loss": 240.099,
+      "step": 33580
+    },
+    {
+      "epoch": 0.6460205499540823,
+      "grad_norm": 1896.5405180718585,
+      "learning_rate": 5.672752827345584e-06,
+      "loss": 241.0282,
+      "step": 33590
+    },
+    {
+      "epoch": 0.6462128752145628,
+      "grad_norm": 1743.6078933233418,
+      "learning_rate": 5.667251491220731e-06,
+      "loss": 237.3629,
+      "step": 33600
+    },
+    {
+      "epoch": 0.6464052004750434,
+      "grad_norm": 1681.7415906054146,
+      "learning_rate": 5.661751768973136e-06,
+      "loss": 241.7321,
+      "step": 33610
+    },
+    {
+      "epoch": 0.6465975257355239,
+      "grad_norm": 1721.9831247416032,
+      "learning_rate": 5.656253662651362e-06,
+      "loss": 248.6104,
+      "step": 33620
+    },
+    {
+      "epoch": 0.6467898509960044,
+      "grad_norm": 1580.6107418236597,
+      "learning_rate": 5.650757174303356e-06,
+      "loss": 237.5195,
+      "step": 33630
+    },
+    {
+      "epoch": 0.646982176256485,
+      "grad_norm": 1564.6415974446613,
+      "learning_rate": 5.645262305976476e-06,
+      "loss": 245.6259,
+      "step": 33640
+    },
+    {
+      "epoch": 0.6471745015169655,
+      "grad_norm": 1659.3163505740938,
+      "learning_rate": 5.639769059717462e-06,
+      "loss": 240.3887,
+      "step": 33650
+    },
+    {
+      "epoch": 0.6473668267774461,
+      "grad_norm": 1782.4986364043527,
+      "learning_rate": 5.634277437572466e-06,
+      "loss": 243.8076,
+      "step": 33660
+    },
+    {
+      "epoch": 0.6475591520379266,
+      "grad_norm": 1543.5231035426991,
+      "learning_rate": 5.6287874415870225e-06,
+      "loss": 243.0699,
+      "step": 33670
+    },
+    {
+      "epoch": 0.6477514772984071,
+      "grad_norm": 1716.2955970962194,
+      "learning_rate": 5.623299073806063e-06,
+      "loss": 240.1385,
+      "step": 33680
+    },
+    {
+      "epoch": 0.6479438025588876,
+      "grad_norm": 1688.011132883271,
+      "learning_rate": 5.617812336273912e-06,
+      "loss": 244.3975,
+      "step": 33690
+    },
+    {
+      "epoch": 0.6481361278193681,
+      "grad_norm": 1563.549686292041,
+      "learning_rate": 5.612327231034296e-06,
+      "loss": 239.5934,
+      "step": 33700
+    },
+    {
+      "epoch": 0.6483284530798487,
+      "grad_norm": 1600.1830411580322,
+      "learning_rate": 5.606843760130321e-06,
+      "loss": 240.8768,
+      "step": 33710
+    },
+    {
+      "epoch": 0.6485207783403292,
+      "grad_norm": 1822.9876700064274,
+      "learning_rate": 5.601361925604485e-06,
+      "loss": 243.2996,
+      "step": 33720
+    },
+    {
+      "epoch": 0.6487131036008097,
+      "grad_norm": 1646.714053356861,
+      "learning_rate": 5.595881729498691e-06,
+      "loss": 238.8543,
+      "step": 33730
+    },
+    {
+      "epoch": 0.6489054288612902,
+      "grad_norm": 1663.792240115043,
+      "learning_rate": 5.590403173854215e-06,
+      "loss": 246.3007,
+      "step": 33740
+    },
+    {
+      "epoch": 0.6490977541217707,
+      "grad_norm": 1544.8035930567348,
+      "learning_rate": 5.584926260711732e-06,
+      "loss": 245.8514,
+      "step": 33750
+    },
+    {
+      "epoch": 0.6492900793822513,
+      "grad_norm": 1449.8989005270469,
+      "learning_rate": 5.579450992111294e-06,
+      "loss": 246.5289,
+      "step": 33760
+    },
+    {
+      "epoch": 0.6494824046427318,
+      "grad_norm": 1516.2265236731903,
+      "learning_rate": 5.573977370092358e-06,
+      "loss": 237.2555,
+      "step": 33770
+    },
+    {
+      "epoch": 0.6496747299032123,
+      "grad_norm": 1598.6996037292274,
+      "learning_rate": 5.568505396693749e-06,
+      "loss": 246.269,
+      "step": 33780
+    },
+    {
+      "epoch": 0.6498670551636928,
+      "grad_norm": 1643.7361251387388,
+      "learning_rate": 5.563035073953691e-06,
+      "loss": 249.9002,
+      "step": 33790
+    },
+    {
+      "epoch": 0.6500593804241733,
+      "grad_norm": 1534.8969275845736,
+      "learning_rate": 5.557566403909794e-06,
+      "loss": 236.6114,
+      "step": 33800
+    },
+    {
+      "epoch": 0.6502517056846538,
+      "grad_norm": 1583.1704136875312,
+      "learning_rate": 5.552099388599042e-06,
+      "loss": 245.5881,
+      "step": 33810
+    },
+    {
+      "epoch": 0.6504440309451344,
+      "grad_norm": 1628.260049827327,
+      "learning_rate": 5.5466340300578095e-06,
+      "loss": 237.9173,
+      "step": 33820
+    },
+    {
+      "epoch": 0.6506363562056149,
+      "grad_norm": 1594.8747911753942,
+      "learning_rate": 5.541170330321845e-06,
+      "loss": 241.4152,
+      "step": 33830
+    },
+    {
+      "epoch": 0.6508286814660955,
+      "grad_norm": 1535.6191317398784,
+      "learning_rate": 5.535708291426297e-06,
+      "loss": 244.6761,
+      "step": 33840
+    },
+    {
+      "epoch": 0.651021006726576,
+      "grad_norm": 1552.608541969007,
+      "learning_rate": 5.53024791540568e-06,
+      "loss": 239.3801,
+      "step": 33850
+    },
+    {
+      "epoch": 0.6512133319870566,
+      "grad_norm": 1543.8663426498124,
+      "learning_rate": 5.5247892042938944e-06,
+      "loss": 236.4334,
+      "step": 33860
+    },
+    {
+      "epoch": 0.6514056572475371,
+      "grad_norm": 1540.959945513679,
+      "learning_rate": 5.519332160124215e-06,
+      "loss": 242.037,
+      "step": 33870
+    },
+    {
+      "epoch": 0.6515979825080176,
+      "grad_norm": 1607.608344707393,
+      "learning_rate": 5.513876784929311e-06,
+      "loss": 237.8388,
+      "step": 33880
+    },
+    {
+      "epoch": 0.6517903077684981,
+      "grad_norm": 1655.9616162185432,
+      "learning_rate": 5.5084230807412135e-06,
+      "loss": 236.6411,
+      "step": 33890
+    },
+    {
+      "epoch": 0.6519826330289786,
+      "grad_norm": 1994.690495480855,
+      "learning_rate": 5.502971049591332e-06,
+      "loss": 249.3606,
+      "step": 33900
+    },
+    {
+      "epoch": 0.6521749582894592,
+      "grad_norm": 1486.854638150478,
+      "learning_rate": 5.497520693510469e-06,
+      "loss": 242.6046,
+      "step": 33910
+    },
+    {
+      "epoch": 0.6523672835499397,
+      "grad_norm": 1488.9397508013353,
+      "learning_rate": 5.492072014528783e-06,
+      "loss": 245.5518,
+      "step": 33920
+    },
+    {
+      "epoch": 0.6525596088104202,
+      "grad_norm": 1577.3181891064764,
+      "learning_rate": 5.4866250146758235e-06,
+      "loss": 243.3775,
+      "step": 33930
+    },
+    {
+      "epoch": 0.6527519340709007,
+      "grad_norm": 1440.5795182481272,
+      "learning_rate": 5.481179695980503e-06,
+      "loss": 243.203,
+      "step": 33940
+    },
+    {
+      "epoch": 0.6529442593313812,
+      "grad_norm": 1368.608328644047,
+      "learning_rate": 5.475736060471117e-06,
+      "loss": 247.5795,
+      "step": 33950
+    },
+    {
+      "epoch": 0.6531365845918617,
+      "grad_norm": 1500.5440019994876,
+      "learning_rate": 5.470294110175329e-06,
+      "loss": 242.9794,
+      "step": 33960
+    },
+    {
+      "epoch": 0.6533289098523423,
+      "grad_norm": 1505.3898057259446,
+      "learning_rate": 5.464853847120169e-06,
+      "loss": 237.1644,
+      "step": 33970
+    },
+    {
+      "epoch": 0.6535212351128228,
+      "grad_norm": 1565.3298648135217,
+      "learning_rate": 5.459415273332056e-06,
+      "loss": 244.4879,
+      "step": 33980
+    },
+    {
+      "epoch": 0.6537135603733033,
+      "grad_norm": 1558.1025582137224,
+      "learning_rate": 5.453978390836763e-06,
+      "loss": 236.0704,
+      "step": 33990
+    },
+    {
+      "epoch": 0.6539058856337838,
+      "grad_norm": 1480.863230972188,
+      "learning_rate": 5.44854320165944e-06,
+      "loss": 248.3781,
+      "step": 34000
+    },
+    {
+      "epoch": 0.6540982108942643,
+      "grad_norm": 1624.917563891583,
+      "learning_rate": 5.443109707824599e-06,
+      "loss": 249.1022,
+      "step": 34010
+    },
+    {
+      "epoch": 0.6542905361547449,
+      "grad_norm": 1540.8689369688175,
+      "learning_rate": 5.437677911356137e-06,
+      "loss": 241.2371,
+      "step": 34020
+    },
+    {
+      "epoch": 0.6544828614152254,
+      "grad_norm": 1661.2790109400378,
+      "learning_rate": 5.432247814277305e-06,
+      "loss": 236.8804,
+      "step": 34030
+    },
+    {
+      "epoch": 0.6546751866757059,
+      "grad_norm": 1910.3639870614024,
+      "learning_rate": 5.426819418610718e-06,
+      "loss": 248.7827,
+      "step": 34040
+    },
+    {
+      "epoch": 0.6548675119361865,
+      "grad_norm": 1558.603472536026,
+      "learning_rate": 5.4213927263783725e-06,
+      "loss": 236.9976,
+      "step": 34050
+    },
+    {
+      "epoch": 0.655059837196667,
+      "grad_norm": 1421.7470280716952,
+      "learning_rate": 5.415967739601616e-06,
+      "loss": 234.0622,
+      "step": 34060
+    },
+    {
+      "epoch": 0.6552521624571476,
+      "grad_norm": 1553.646499313457,
+      "learning_rate": 5.41054446030117e-06,
+      "loss": 241.8123,
+      "step": 34070
+    },
+    {
+      "epoch": 0.6554444877176281,
+      "grad_norm": 1851.6982961388287,
+      "learning_rate": 5.405122890497114e-06,
+      "loss": 250.3723,
+      "step": 34080
+    },
+    {
+      "epoch": 0.6556368129781086,
+      "grad_norm": 1561.6896288143896,
+      "learning_rate": 5.399703032208896e-06,
+      "loss": 243.1483,
+      "step": 34090
+    },
+    {
+      "epoch": 0.6558291382385891,
+      "grad_norm": 1577.121166693554,
+      "learning_rate": 5.3942848874553235e-06,
+      "loss": 240.305,
+      "step": 34100
+    },
+    {
+      "epoch": 0.6560214634990696,
+      "grad_norm": 1475.4204529224933,
+      "learning_rate": 5.388868458254565e-06,
+      "loss": 233.7801,
+      "step": 34110
+    },
+    {
+      "epoch": 0.6562137887595502,
+      "grad_norm": 1507.669425403277,
+      "learning_rate": 5.3834537466241455e-06,
+      "loss": 238.5951,
+      "step": 34120
+    },
+    {
+      "epoch": 0.6564061140200307,
+      "grad_norm": 1383.2948893470773,
+      "learning_rate": 5.378040754580964e-06,
+      "loss": 234.3419,
+      "step": 34130
+    },
+    {
+      "epoch": 0.6565984392805112,
+      "grad_norm": 1587.0554629770497,
+      "learning_rate": 5.37262948414127e-06,
+      "loss": 244.7607,
+      "step": 34140
+    },
+    {
+      "epoch": 0.6567907645409917,
+      "grad_norm": 1728.0160670652772,
+      "learning_rate": 5.367219937320663e-06,
+      "loss": 236.5773,
+      "step": 34150
+    },
+    {
+      "epoch": 0.6569830898014722,
+      "grad_norm": 1497.7001376726575,
+      "learning_rate": 5.361812116134122e-06,
+      "loss": 240.3799,
+      "step": 34160
+    },
+    {
+      "epoch": 0.6571754150619528,
+      "grad_norm": 1649.5838296742736,
+      "learning_rate": 5.356406022595963e-06,
+      "loss": 245.3542,
+      "step": 34170
+    },
+    {
+      "epoch": 0.6573677403224333,
+      "grad_norm": 1558.1627200630019,
+      "learning_rate": 5.351001658719872e-06,
+      "loss": 244.3351,
+      "step": 34180
+    },
+    {
+      "epoch": 0.6575600655829138,
+      "grad_norm": 1353.5702789873073,
+      "learning_rate": 5.345599026518877e-06,
+      "loss": 242.9741,
+      "step": 34190
+    },
+    {
+      "epoch": 0.6577523908433943,
+      "grad_norm": 1537.4491165065745,
+      "learning_rate": 5.3401981280053745e-06,
+      "loss": 250.0492,
+      "step": 34200
+    },
+    {
+      "epoch": 0.6579447161038748,
+      "grad_norm": 1556.1483851977846,
+      "learning_rate": 5.334798965191115e-06,
+      "loss": 241.9707,
+      "step": 34210
+    },
+    {
+      "epoch": 0.6581370413643554,
+      "grad_norm": 1591.265777770099,
+      "learning_rate": 5.329401540087188e-06,
+      "loss": 237.3961,
+      "step": 34220
+    },
+    {
+      "epoch": 0.6583293666248359,
+      "grad_norm": 1527.0948515868372,
+      "learning_rate": 5.3240058547040525e-06,
+      "loss": 234.2487,
+      "step": 34230
+    },
+    {
+      "epoch": 0.6585216918853164,
+      "grad_norm": 1463.990784982395,
+      "learning_rate": 5.318611911051512e-06,
+      "loss": 239.1191,
+      "step": 34240
+    },
+    {
+      "epoch": 0.658714017145797,
+      "grad_norm": 1430.5563353294062,
+      "learning_rate": 5.313219711138717e-06,
+      "loss": 240.6146,
+      "step": 34250
+    },
+    {
+      "epoch": 0.6589063424062775,
+      "grad_norm": 1605.3487600450017,
+      "learning_rate": 5.307829256974173e-06,
+      "loss": 235.8148,
+      "step": 34260
+    },
+    {
+      "epoch": 0.6590986676667581,
+      "grad_norm": 1535.623240763071,
+      "learning_rate": 5.302440550565739e-06,
+      "loss": 230.1232,
+      "step": 34270
+    },
+    {
+      "epoch": 0.6592909929272386,
+      "grad_norm": 1586.308704450949,
+      "learning_rate": 5.297053593920618e-06,
+      "loss": 232.7758,
+      "step": 34280
+    },
+    {
+      "epoch": 0.6594833181877191,
+      "grad_norm": 1455.5176649254452,
+      "learning_rate": 5.291668389045362e-06,
+      "loss": 238.7155,
+      "step": 34290
+    },
+    {
+      "epoch": 0.6596756434481996,
+      "grad_norm": 1566.9981409277634,
+      "learning_rate": 5.286284937945866e-06,
+      "loss": 241.0195,
+      "step": 34300
+    },
+    {
+      "epoch": 0.6598679687086801,
+      "grad_norm": 1591.0879362389355,
+      "learning_rate": 5.280903242627384e-06,
+      "loss": 244.9684,
+      "step": 34310
+    },
+    {
+      "epoch": 0.6600602939691607,
+      "grad_norm": 3271.3285393791975,
+      "learning_rate": 5.2755233050945076e-06,
+      "loss": 245.5639,
+      "step": 34320
+    },
+    {
+      "epoch": 0.6602526192296412,
+      "grad_norm": 1682.0916508380708,
+      "learning_rate": 5.270145127351168e-06,
+      "loss": 247.4215,
+      "step": 34330
+    },
+    {
+      "epoch": 0.6604449444901217,
+      "grad_norm": 1789.31363571357,
+      "learning_rate": 5.264768711400656e-06,
+      "loss": 241.0164,
+      "step": 34340
+    },
+    {
+      "epoch": 0.6606372697506022,
+      "grad_norm": 1719.6523963782524,
+      "learning_rate": 5.25939405924559e-06,
+      "loss": 242.2833,
+      "step": 34350
+    },
+    {
+      "epoch": 0.6608295950110827,
+      "grad_norm": 1601.5005012523118,
+      "learning_rate": 5.254021172887947e-06,
+      "loss": 247.5436,
+      "step": 34360
+    },
+    {
+      "epoch": 0.6610219202715633,
+      "grad_norm": 1612.383745768123,
+      "learning_rate": 5.248650054329032e-06,
+      "loss": 247.6403,
+      "step": 34370
+    },
+    {
+      "epoch": 0.6612142455320438,
+      "grad_norm": 1623.4026519042443,
+      "learning_rate": 5.2432807055695035e-06,
+      "loss": 247.4154,
+      "step": 34380
+    },
+    {
+      "epoch": 0.6614065707925243,
+      "grad_norm": 1579.6288849252182,
+      "learning_rate": 5.237913128609352e-06,
+      "loss": 243.2603,
+      "step": 34390
+    },
+    {
+      "epoch": 0.6615988960530048,
+      "grad_norm": 1455.270238227104,
+      "learning_rate": 5.232547325447908e-06,
+      "loss": 250.4422,
+      "step": 34400
+    },
+    {
+      "epoch": 0.6617912213134853,
+      "grad_norm": 1630.009850769735,
+      "learning_rate": 5.227183298083854e-06,
+      "loss": 239.656,
+      "step": 34410
+    },
+    {
+      "epoch": 0.6619835465739659,
+      "grad_norm": 1751.6551799830158,
+      "learning_rate": 5.221821048515193e-06,
+      "loss": 242.0374,
+      "step": 34420
+    },
+    {
+      "epoch": 0.6621758718344464,
+      "grad_norm": 1514.6136013261407,
+      "learning_rate": 5.216460578739278e-06,
+      "loss": 236.949,
+      "step": 34430
+    },
+    {
+      "epoch": 0.6623681970949269,
+      "grad_norm": 1515.4167284499051,
+      "learning_rate": 5.211101890752792e-06,
+      "loss": 236.9444,
+      "step": 34440
+    },
+    {
+      "epoch": 0.6625605223554074,
+      "grad_norm": 1554.3141051753453,
+      "learning_rate": 5.205744986551763e-06,
+      "loss": 240.2105,
+      "step": 34450
+    },
+    {
+      "epoch": 0.662752847615888,
+      "grad_norm": 1677.5415540828621,
+      "learning_rate": 5.200389868131547e-06,
+      "loss": 248.5601,
+      "step": 34460
+    },
+    {
+      "epoch": 0.6629451728763686,
+      "grad_norm": 1559.733496865752,
+      "learning_rate": 5.195036537486833e-06,
+      "loss": 247.374,
+      "step": 34470
+    },
+    {
+      "epoch": 0.6631374981368491,
+      "grad_norm": 1550.7657091462142,
+      "learning_rate": 5.189684996611657e-06,
+      "loss": 246.8185,
+      "step": 34480
+    },
+    {
+      "epoch": 0.6633298233973296,
+      "grad_norm": 1428.9433311814219,
+      "learning_rate": 5.18433524749937e-06,
+      "loss": 235.8814,
+      "step": 34490
+    },
+    {
+      "epoch": 0.6635221486578101,
+      "grad_norm": 1799.0210163625984,
+      "learning_rate": 5.178987292142674e-06,
+      "loss": 252.3958,
+      "step": 34500
+    },
+    {
+      "epoch": 0.6637144739182906,
+      "grad_norm": 1443.1877247379198,
+      "learning_rate": 5.173641132533586e-06,
+      "loss": 240.2568,
+      "step": 34510
+    },
+    {
+      "epoch": 0.6639067991787712,
+      "grad_norm": 1635.034512203725,
+      "learning_rate": 5.16829677066347e-06,
+      "loss": 247.8541,
+      "step": 34520
+    },
+    {
+      "epoch": 0.6640991244392517,
+      "grad_norm": 1674.1642518045078,
+      "learning_rate": 5.16295420852301e-06,
+      "loss": 238.8359,
+      "step": 34530
+    },
+    {
+      "epoch": 0.6642914496997322,
+      "grad_norm": 1542.1724582160432,
+      "learning_rate": 5.157613448102219e-06,
+      "loss": 243.5318,
+      "step": 34540
+    },
+    {
+      "epoch": 0.6644837749602127,
+      "grad_norm": 1359.6359175561622,
+      "learning_rate": 5.152274491390441e-06,
+      "loss": 241.1148,
+      "step": 34550
+    },
+    {
+      "epoch": 0.6646761002206932,
+      "grad_norm": 1633.8622693225025,
+      "learning_rate": 5.1469373403763555e-06,
+      "loss": 243.5518,
+      "step": 34560
+    },
+    {
+      "epoch": 0.6648684254811738,
+      "grad_norm": 1550.9344748639865,
+      "learning_rate": 5.14160199704796e-06,
+      "loss": 238.6767,
+      "step": 34570
+    },
+    {
+      "epoch": 0.6650607507416543,
+      "grad_norm": 1499.9267239020364,
+      "learning_rate": 5.136268463392578e-06,
+      "loss": 237.2557,
+      "step": 34580
+    },
+    {
+      "epoch": 0.6652530760021348,
+      "grad_norm": 1665.5736182747498,
+      "learning_rate": 5.13093674139687e-06,
+      "loss": 239.5459,
+      "step": 34590
+    },
+    {
+      "epoch": 0.6654454012626153,
+      "grad_norm": 1485.474032002765,
+      "learning_rate": 5.12560683304681e-06,
+      "loss": 235.1012,
+      "step": 34600
+    },
+    {
+      "epoch": 0.6656377265230958,
+      "grad_norm": 1669.3176931646353,
+      "learning_rate": 5.120278740327702e-06,
+      "loss": 238.2485,
+      "step": 34610
+    },
+    {
+      "epoch": 0.6658300517835763,
+      "grad_norm": 1429.0662090165415,
+      "learning_rate": 5.114952465224168e-06,
+      "loss": 240.7864,
+      "step": 34620
+    },
+    {
+      "epoch": 0.6660223770440569,
+      "grad_norm": 1574.384026635983,
+      "learning_rate": 5.109628009720162e-06,
+      "loss": 247.8457,
+      "step": 34630
+    },
+    {
+      "epoch": 0.6662147023045374,
+      "grad_norm": 1974.0278155291112,
+      "learning_rate": 5.104305375798958e-06,
+      "loss": 242.7548,
+      "step": 34640
+    },
+    {
+      "epoch": 0.6664070275650179,
+      "grad_norm": 1865.3875014337664,
+      "learning_rate": 5.098984565443148e-06,
+      "loss": 244.976,
+      "step": 34650
+    },
+    {
+      "epoch": 0.6665993528254985,
+      "grad_norm": 1799.3689359589089,
+      "learning_rate": 5.093665580634639e-06,
+      "loss": 236.0072,
+      "step": 34660
+    },
+    {
+      "epoch": 0.6667916780859791,
+      "grad_norm": 1586.830044361748,
+      "learning_rate": 5.088348423354674e-06,
+      "loss": 236.3255,
+      "step": 34670
+    },
+    {
+      "epoch": 0.6669840033464596,
+      "grad_norm": 1477.1281301542635,
+      "learning_rate": 5.0830330955838045e-06,
+      "loss": 244.4166,
+      "step": 34680
+    },
+    {
+      "epoch": 0.6671763286069401,
+      "grad_norm": 1453.389875571511,
+      "learning_rate": 5.077719599301895e-06,
+      "loss": 241.1291,
+      "step": 34690
+    },
+    {
+      "epoch": 0.6673686538674206,
+      "grad_norm": 1741.008433516878,
+      "learning_rate": 5.072407936488145e-06,
+      "loss": 254.7444,
+      "step": 34700
+    },
+    {
+      "epoch": 0.6675609791279011,
+      "grad_norm": 1450.1987424526549,
+      "learning_rate": 5.067098109121058e-06,
+      "loss": 240.3621,
+      "step": 34710
+    },
+    {
+      "epoch": 0.6677533043883817,
+      "grad_norm": 1656.220009503437,
+      "learning_rate": 5.0617901191784536e-06,
+      "loss": 240.9367,
+      "step": 34720
+    },
+    {
+      "epoch": 0.6679456296488622,
+      "grad_norm": 1482.6727507809046,
+      "learning_rate": 5.05648396863747e-06,
+      "loss": 240.2776,
+      "step": 34730
+    },
+    {
+      "epoch": 0.6681379549093427,
+      "grad_norm": 1811.330848051615,
+      "learning_rate": 5.051179659474568e-06,
+      "loss": 242.9302,
+      "step": 34740
+    },
+    {
+      "epoch": 0.6683302801698232,
+      "grad_norm": 1593.0864925162757,
+      "learning_rate": 5.045877193665508e-06,
+      "loss": 243.2492,
+      "step": 34750
+    },
+    {
+      "epoch": 0.6685226054303037,
+      "grad_norm": 1649.0097673598555,
+      "learning_rate": 5.040576573185372e-06,
+      "loss": 240.529,
+      "step": 34760
+    },
+    {
+      "epoch": 0.6687149306907842,
+      "grad_norm": 1683.9299016641,
+      "learning_rate": 5.0352778000085565e-06,
+      "loss": 238.4384,
+      "step": 34770
+    },
+    {
+      "epoch": 0.6689072559512648,
+      "grad_norm": 1576.440869984676,
+      "learning_rate": 5.029980876108762e-06,
+      "loss": 242.4429,
+      "step": 34780
+    },
+    {
+      "epoch": 0.6690995812117453,
+      "grad_norm": 1422.2642081678534,
+      "learning_rate": 5.024685803459013e-06,
+      "loss": 234.8015,
+      "step": 34790
+    },
+    {
+      "epoch": 0.6692919064722258,
+      "grad_norm": 1499.545862262694,
+      "learning_rate": 5.019392584031628e-06,
+      "loss": 232.897,
+      "step": 34800
+    },
+    {
+      "epoch": 0.6694842317327063,
+      "grad_norm": 1664.45033227328,
+      "learning_rate": 5.0141012197982534e-06,
+      "loss": 248.728,
+      "step": 34810
+    },
+    {
+      "epoch": 0.6696765569931868,
+      "grad_norm": 1612.1865810594797,
+      "learning_rate": 5.0088117127298285e-06,
+      "loss": 237.6479,
+      "step": 34820
+    },
+    {
+      "epoch": 0.6698688822536674,
+      "grad_norm": 1462.8324785069049,
+      "learning_rate": 5.003524064796608e-06,
+      "loss": 240.4195,
+      "step": 34830
+    },
+    {
+      "epoch": 0.6700612075141479,
+      "grad_norm": 1560.2940839790592,
+      "learning_rate": 4.998238277968149e-06,
+      "loss": 242.2532,
+      "step": 34840
+    },
+    {
+      "epoch": 0.6702535327746284,
+      "grad_norm": 1653.3102655389425,
+      "learning_rate": 4.992954354213331e-06,
+      "loss": 236.3106,
+      "step": 34850
+    },
+    {
+      "epoch": 0.6704458580351089,
+      "grad_norm": 1717.3012319069778,
+      "learning_rate": 4.98767229550032e-06,
+      "loss": 245.5809,
+      "step": 34860
+    },
+    {
+      "epoch": 0.6706381832955896,
+      "grad_norm": 1649.3152282098256,
+      "learning_rate": 4.982392103796595e-06,
+      "loss": 236.3104,
+      "step": 34870
+    },
+    {
+      "epoch": 0.6708305085560701,
+      "grad_norm": 1592.2440026780055,
+      "learning_rate": 4.977113781068945e-06,
+      "loss": 245.7599,
+      "step": 34880
+    },
+    {
+      "epoch": 0.6710228338165506,
+      "grad_norm": 1477.387494514049,
+      "learning_rate": 4.971837329283458e-06,
+      "loss": 243.1606,
+      "step": 34890
+    },
+    {
+      "epoch": 0.6712151590770311,
+      "grad_norm": 1510.9082515170664,
+      "learning_rate": 4.966562750405517e-06,
+      "loss": 249.1312,
+      "step": 34900
+    },
+    {
+      "epoch": 0.6714074843375116,
+      "grad_norm": 1667.1164304552096,
+      "learning_rate": 4.9612900463998274e-06,
+      "loss": 239.2116,
+      "step": 34910
+    },
+    {
+      "epoch": 0.6715998095979921,
+      "grad_norm": 1596.9511155995442,
+      "learning_rate": 4.9560192192303735e-06,
+      "loss": 250.2845,
+      "step": 34920
+    },
+    {
+      "epoch": 0.6717921348584727,
+      "grad_norm": 1638.6961634633678,
+      "learning_rate": 4.95075027086046e-06,
+      "loss": 236.8568,
+      "step": 34930
+    },
+    {
+      "epoch": 0.6719844601189532,
+      "grad_norm": 1707.5169579675846,
+      "learning_rate": 4.9454832032526755e-06,
+      "loss": 247.0217,
+      "step": 34940
+    },
+    {
+      "epoch": 0.6721767853794337,
+      "grad_norm": 1534.7959342305119,
+      "learning_rate": 4.940218018368924e-06,
+      "loss": 240.9309,
+      "step": 34950
+    },
+    {
+      "epoch": 0.6723691106399142,
+      "grad_norm": 1477.2685584523297,
+      "learning_rate": 4.934954718170396e-06,
+      "loss": 238.7597,
+      "step": 34960
+    },
+    {
+      "epoch": 0.6725614359003947,
+      "grad_norm": 1573.6164713650567,
+      "learning_rate": 4.9296933046175834e-06,
+      "loss": 234.4981,
+      "step": 34970
+    },
+    {
+      "epoch": 0.6727537611608753,
+      "grad_norm": 1660.8753066222212,
+      "learning_rate": 4.924433779670271e-06,
+      "loss": 234.8812,
+      "step": 34980
+    },
+    {
+      "epoch": 0.6729460864213558,
+      "grad_norm": 1472.2716227435594,
+      "learning_rate": 4.9191761452875554e-06,
+      "loss": 248.6342,
+      "step": 34990
+    },
+    {
+      "epoch": 0.6731384116818363,
+      "grad_norm": 1549.4541004053892,
+      "learning_rate": 4.913920403427812e-06,
+      "loss": 238.4336,
+      "step": 35000
+    },
+    {
+      "epoch": 0.6733307369423168,
+      "grad_norm": 1985.7517243036866,
+      "learning_rate": 4.908666556048719e-06,
+      "loss": 236.0099,
+      "step": 35010
+    },
+    {
+      "epoch": 0.6735230622027973,
+      "grad_norm": 1546.281868327463,
+      "learning_rate": 4.903414605107244e-06,
+      "loss": 237.7881,
+      "step": 35020
+    },
+    {
+      "epoch": 0.6737153874632779,
+      "grad_norm": 1556.804223861094,
+      "learning_rate": 4.89816455255966e-06,
+      "loss": 239.3844,
+      "step": 35030
+    },
+    {
+      "epoch": 0.6739077127237584,
+      "grad_norm": 1660.4441269248118,
+      "learning_rate": 4.892916400361516e-06,
+      "loss": 241.0597,
+      "step": 35040
+    },
+    {
+      "epoch": 0.6741000379842389,
+      "grad_norm": 1537.1815918548596,
+      "learning_rate": 4.887670150467671e-06,
+      "loss": 240.1067,
+      "step": 35050
+    },
+    {
+      "epoch": 0.6742923632447194,
+      "grad_norm": 1374.8726925474718,
+      "learning_rate": 4.882425804832258e-06,
+      "loss": 239.3176,
+      "step": 35060
+    },
+    {
+      "epoch": 0.6744846885052,
+      "grad_norm": 13431.556008525862,
+      "learning_rate": 4.8771833654087165e-06,
+      "loss": 241.3402,
+      "step": 35070
+    },
+    {
+      "epoch": 0.6746770137656806,
+      "grad_norm": 1681.119131350578,
+      "learning_rate": 4.8719428341497665e-06,
+      "loss": 240.1362,
+      "step": 35080
+    },
+    {
+      "epoch": 0.6748693390261611,
+      "grad_norm": 1565.9377754503635,
+      "learning_rate": 4.866704213007413e-06,
+      "loss": 243.6768,
+      "step": 35090
+    },
+    {
+      "epoch": 0.6750616642866416,
+      "grad_norm": 1595.1903327164287,
+      "learning_rate": 4.861467503932965e-06,
+      "loss": 242.1631,
+      "step": 35100
+    },
+    {
+      "epoch": 0.6752539895471221,
+      "grad_norm": 1430.9472111458415,
+      "learning_rate": 4.856232708877007e-06,
+      "loss": 234.3875,
+      "step": 35110
+    },
+    {
+      "epoch": 0.6754463148076026,
+      "grad_norm": 1757.7523260708806,
+      "learning_rate": 4.8509998297894075e-06,
+      "loss": 245.6894,
+      "step": 35120
+    },
+    {
+      "epoch": 0.6756386400680832,
+      "grad_norm": 1709.9388251090809,
+      "learning_rate": 4.845768868619338e-06,
+      "loss": 239.7329,
+      "step": 35130
+    },
+    {
+      "epoch": 0.6758309653285637,
+      "grad_norm": 1790.13059566027,
+      "learning_rate": 4.8405398273152405e-06,
+      "loss": 260.7546,
+      "step": 35140
+    },
+    {
+      "epoch": 0.6760232905890442,
+      "grad_norm": 1695.99320182758,
+      "learning_rate": 4.8353127078248455e-06,
+      "loss": 250.915,
+      "step": 35150
+    },
+    {
+      "epoch": 0.6762156158495247,
+      "grad_norm": 1654.7286934425988,
+      "learning_rate": 4.830087512095164e-06,
+      "loss": 236.3527,
+      "step": 35160
+    },
+    {
+      "epoch": 0.6764079411100052,
+      "grad_norm": 1518.8809802475955,
+      "learning_rate": 4.824864242072506e-06,
+      "loss": 236.217,
+      "step": 35170
+    },
+    {
+      "epoch": 0.6766002663704858,
+      "grad_norm": 1676.1492450616627,
+      "learning_rate": 4.819642899702447e-06,
+      "loss": 239.2257,
+      "step": 35180
+    },
+    {
+      "epoch": 0.6767925916309663,
+      "grad_norm": 1604.8433463634062,
+      "learning_rate": 4.814423486929846e-06,
+      "loss": 238.3168,
+      "step": 35190
+    },
+    {
+      "epoch": 0.6769849168914468,
+      "grad_norm": 1555.6284985922198,
+      "learning_rate": 4.809206005698856e-06,
+      "loss": 236.2358,
+      "step": 35200
+    },
+    {
+      "epoch": 0.6771772421519273,
+      "grad_norm": 1501.771519891142,
+      "learning_rate": 4.803990457952903e-06,
+      "loss": 250.7805,
+      "step": 35210
+    },
+    {
+      "epoch": 0.6773695674124078,
+      "grad_norm": 1598.6267349997,
+      "learning_rate": 4.798776845634688e-06,
+      "loss": 235.7079,
+      "step": 35220
+    },
+    {
+      "epoch": 0.6775618926728884,
+      "grad_norm": 1610.666340493888,
+      "learning_rate": 4.793565170686196e-06,
+      "loss": 242.6579,
+      "step": 35230
+    },
+    {
+      "epoch": 0.6777542179333689,
+      "grad_norm": 1514.145068771424,
+      "learning_rate": 4.7883554350486925e-06,
+      "loss": 246.1071,
+      "step": 35240
+    },
+    {
+      "epoch": 0.6779465431938494,
+      "grad_norm": 1569.7645123417265,
+      "learning_rate": 4.783147640662716e-06,
+      "loss": 234.9182,
+      "step": 35250
+    },
+    {
+      "epoch": 0.6781388684543299,
+      "grad_norm": 1843.8223095579276,
+      "learning_rate": 4.777941789468086e-06,
+      "loss": 238.5698,
+      "step": 35260
+    },
+    {
+      "epoch": 0.6783311937148104,
+      "grad_norm": 1702.281904489595,
+      "learning_rate": 4.772737883403889e-06,
+      "loss": 247.3601,
+      "step": 35270
+    },
+    {
+      "epoch": 0.6785235189752911,
+      "grad_norm": 1628.4594832928703,
+      "learning_rate": 4.767535924408504e-06,
+      "loss": 242.4072,
+      "step": 35280
+    },
+    {
+      "epoch": 0.6787158442357716,
+      "grad_norm": 1548.370651253724,
+      "learning_rate": 4.76233591441957e-06,
+      "loss": 241.6772,
+      "step": 35290
+    },
+    {
+      "epoch": 0.6789081694962521,
+      "grad_norm": 1793.823798929674,
+      "learning_rate": 4.757137855374001e-06,
+      "loss": 236.6288,
+      "step": 35300
+    },
+    {
+      "epoch": 0.6791004947567326,
+      "grad_norm": 1642.3865718885472,
+      "learning_rate": 4.751941749207996e-06,
+      "loss": 242.9599,
+      "step": 35310
+    },
+    {
+      "epoch": 0.6792928200172131,
+      "grad_norm": 1465.3930417500717,
+      "learning_rate": 4.746747597857014e-06,
+      "loss": 238.4086,
+      "step": 35320
+    },
+    {
+      "epoch": 0.6794851452776937,
+      "grad_norm": 1535.2361090234504,
+      "learning_rate": 4.74155540325579e-06,
+      "loss": 241.5043,
+      "step": 35330
+    },
+    {
+      "epoch": 0.6796774705381742,
+      "grad_norm": 1792.7497882019732,
+      "learning_rate": 4.736365167338333e-06,
+      "loss": 247.4346,
+      "step": 35340
+    },
+    {
+      "epoch": 0.6798697957986547,
+      "grad_norm": 1669.936825913557,
+      "learning_rate": 4.731176892037918e-06,
+      "loss": 235.9599,
+      "step": 35350
+    },
+    {
+      "epoch": 0.6800621210591352,
+      "grad_norm": 1473.0361533499533,
+      "learning_rate": 4.725990579287097e-06,
+      "loss": 233.579,
+      "step": 35360
+    },
+    {
+      "epoch": 0.6802544463196157,
+      "grad_norm": 2305.0392250606014,
+      "learning_rate": 4.720806231017676e-06,
+      "loss": 242.3682,
+      "step": 35370
+    },
+    {
+      "epoch": 0.6804467715800963,
+      "grad_norm": 1739.121755322293,
+      "learning_rate": 4.7156238491607506e-06,
+      "loss": 241.138,
+      "step": 35380
+    },
+    {
+      "epoch": 0.6806390968405768,
+      "grad_norm": 1638.4784515221852,
+      "learning_rate": 4.710443435646666e-06,
+      "loss": 237.0229,
+      "step": 35390
+    },
+    {
+      "epoch": 0.6808314221010573,
+      "grad_norm": 1481.2497186672954,
+      "learning_rate": 4.705264992405043e-06,
+      "loss": 236.662,
+      "step": 35400
+    },
+    {
+      "epoch": 0.6810237473615378,
+      "grad_norm": 1684.8178681389418,
+      "learning_rate": 4.700088521364761e-06,
+      "loss": 236.6188,
+      "step": 35410
+    },
+    {
+      "epoch": 0.6812160726220183,
+      "grad_norm": 1562.625843975324,
+      "learning_rate": 4.694914024453977e-06,
+      "loss": 241.3342,
+      "step": 35420
+    },
+    {
+      "epoch": 0.6814083978824989,
+      "grad_norm": 1727.2125847675445,
+      "learning_rate": 4.689741503600103e-06,
+      "loss": 240.6263,
+      "step": 35430
+    },
+    {
+      "epoch": 0.6816007231429794,
+      "grad_norm": 1902.1493212950902,
+      "learning_rate": 4.684570960729818e-06,
+      "loss": 240.4563,
+      "step": 35440
+    },
+    {
+      "epoch": 0.6817930484034599,
+      "grad_norm": 1510.626143137195,
+      "learning_rate": 4.679402397769057e-06,
+      "loss": 234.0074,
+      "step": 35450
+    },
+    {
+      "epoch": 0.6819853736639404,
+      "grad_norm": 1834.008118031445,
+      "learning_rate": 4.674235816643035e-06,
+      "loss": 251.336,
+      "step": 35460
+    },
+    {
+      "epoch": 0.6821776989244209,
+      "grad_norm": 1391.0513980372607,
+      "learning_rate": 4.66907121927621e-06,
+      "loss": 237.735,
+      "step": 35470
+    },
+    {
+      "epoch": 0.6823700241849016,
+      "grad_norm": 1525.4357475208928,
+      "learning_rate": 4.663908607592316e-06,
+      "loss": 235.0554,
+      "step": 35480
+    },
+    {
+      "epoch": 0.6825623494453821,
+      "grad_norm": 1417.3135561199801,
+      "learning_rate": 4.658747983514334e-06,
+      "loss": 229.8492,
+      "step": 35490
+    },
+    {
+      "epoch": 0.6827546747058626,
+      "grad_norm": 1576.8775379958165,
+      "learning_rate": 4.653589348964517e-06,
+      "loss": 238.249,
+      "step": 35500
+    },
+    {
+      "epoch": 0.6829469999663431,
+      "grad_norm": 1602.8624964794892,
+      "learning_rate": 4.648432705864369e-06,
+      "loss": 259.1102,
+      "step": 35510
+    },
+    {
+      "epoch": 0.6831393252268236,
+      "grad_norm": 1517.6239958002438,
+      "learning_rate": 4.64327805613465e-06,
+      "loss": 246.4283,
+      "step": 35520
+    },
+    {
+      "epoch": 0.6833316504873042,
+      "grad_norm": 1534.8169871231958,
+      "learning_rate": 4.638125401695391e-06,
+      "loss": 236.2565,
+      "step": 35530
+    },
+    {
+      "epoch": 0.6835239757477847,
+      "grad_norm": 1584.166090396211,
+      "learning_rate": 4.632974744465865e-06,
+      "loss": 244.4494,
+      "step": 35540
+    },
+    {
+      "epoch": 0.6837163010082652,
+      "grad_norm": 1473.9590014564478,
+      "learning_rate": 4.627826086364603e-06,
+      "loss": 246.8307,
+      "step": 35550
+    },
+    {
+      "epoch": 0.6839086262687457,
+      "grad_norm": 2086.6172954712015,
+      "learning_rate": 4.622679429309404e-06,
+      "loss": 240.9162,
+      "step": 35560
+    },
+    {
+      "epoch": 0.6841009515292262,
+      "grad_norm": 1533.2570787807176,
+      "learning_rate": 4.617534775217307e-06,
+      "loss": 229.8946,
+      "step": 35570
+    },
+    {
+      "epoch": 0.6842932767897068,
+      "grad_norm": 1578.873240031328,
+      "learning_rate": 4.6123921260046135e-06,
+      "loss": 234.8505,
+      "step": 35580
+    },
+    {
+      "epoch": 0.6844856020501873,
+      "grad_norm": 1577.3051897353623,
+      "learning_rate": 4.607251483586869e-06,
+      "loss": 244.1948,
+      "step": 35590
+    },
+    {
+      "epoch": 0.6846779273106678,
+      "grad_norm": 1467.8752814913016,
+      "learning_rate": 4.6021128498788855e-06,
+      "loss": 231.7717,
+      "step": 35600
+    },
+    {
+      "epoch": 0.6848702525711483,
+      "grad_norm": 1424.6796093895823,
+      "learning_rate": 4.596976226794718e-06,
+      "loss": 232.1213,
+      "step": 35610
+    },
+    {
+      "epoch": 0.6850625778316288,
+      "grad_norm": 1571.2834264316803,
+      "learning_rate": 4.591841616247669e-06,
+      "loss": 245.3532,
+      "step": 35620
+    },
+    {
+      "epoch": 0.6852549030921093,
+      "grad_norm": 1676.7221917131023,
+      "learning_rate": 4.586709020150297e-06,
+      "loss": 236.3314,
+      "step": 35630
+    },
+    {
+      "epoch": 0.6854472283525899,
+      "grad_norm": 1692.7840442568893,
+      "learning_rate": 4.581578440414417e-06,
+      "loss": 235.2596,
+      "step": 35640
+    },
+    {
+      "epoch": 0.6856395536130704,
+      "grad_norm": 1527.4220045571374,
+      "learning_rate": 4.576449878951079e-06,
+      "loss": 241.827,
+      "step": 35650
+    },
+    {
+      "epoch": 0.6858318788735509,
+      "grad_norm": 1589.0963502807458,
+      "learning_rate": 4.571323337670584e-06,
+      "loss": 238.6202,
+      "step": 35660
+    },
+    {
+      "epoch": 0.6860242041340314,
+      "grad_norm": 1476.2895786090012,
+      "learning_rate": 4.566198818482494e-06,
+      "loss": 240.941,
+      "step": 35670
+    },
+    {
+      "epoch": 0.686216529394512,
+      "grad_norm": 1486.6620378889293,
+      "learning_rate": 4.5610763232956e-06,
+      "loss": 241.7036,
+      "step": 35680
+    },
+    {
+      "epoch": 0.6864088546549926,
+      "grad_norm": 1613.6380638201672,
+      "learning_rate": 4.55595585401795e-06,
+      "loss": 248.1098,
+      "step": 35690
+    },
+    {
+      "epoch": 0.6866011799154731,
+      "grad_norm": 1677.7797662336263,
+      "learning_rate": 4.550837412556828e-06,
+      "loss": 249.7874,
+      "step": 35700
+    },
+    {
+      "epoch": 0.6867935051759536,
+      "grad_norm": 1636.632598513363,
+      "learning_rate": 4.545721000818778e-06,
+      "loss": 247.96,
+      "step": 35710
+    },
+    {
+      "epoch": 0.6869858304364341,
+      "grad_norm": 1561.3114680973345,
+      "learning_rate": 4.540606620709575e-06,
+      "loss": 255.0879,
+      "step": 35720
+    },
+    {
+      "epoch": 0.6871781556969146,
+      "grad_norm": 1508.6229173893246,
+      "learning_rate": 4.535494274134236e-06,
+      "loss": 245.9509,
+      "step": 35730
+    },
+    {
+      "epoch": 0.6873704809573952,
+      "grad_norm": 1597.9553692072259,
+      "learning_rate": 4.530383962997036e-06,
+      "loss": 240.9111,
+      "step": 35740
+    },
+    {
+      "epoch": 0.6875628062178757,
+      "grad_norm": 1715.5095172633003,
+      "learning_rate": 4.525275689201476e-06,
+      "loss": 250.077,
+      "step": 35750
+    },
+    {
+      "epoch": 0.6877551314783562,
+      "grad_norm": 1799.9672307505061,
+      "learning_rate": 4.520169454650299e-06,
+      "loss": 242.9612,
+      "step": 35760
+    },
+    {
+      "epoch": 0.6879474567388367,
+      "grad_norm": 1682.473233411806,
+      "learning_rate": 4.515065261245498e-06,
+      "loss": 244.1876,
+      "step": 35770
+    },
+    {
+      "epoch": 0.6881397819993172,
+      "grad_norm": 1641.163603961195,
+      "learning_rate": 4.509963110888305e-06,
+      "loss": 239.5544,
+      "step": 35780
+    },
+    {
+      "epoch": 0.6883321072597978,
+      "grad_norm": 1595.9848110961755,
+      "learning_rate": 4.504863005479182e-06,
+      "loss": 234.0018,
+      "step": 35790
+    },
+    {
+      "epoch": 0.6885244325202783,
+      "grad_norm": 1766.9205349420956,
+      "learning_rate": 4.499764946917834e-06,
+      "loss": 243.1572,
+      "step": 35800
+    },
+    {
+      "epoch": 0.6887167577807588,
+      "grad_norm": 1565.7807598365366,
+      "learning_rate": 4.494668937103201e-06,
+      "loss": 240.1978,
+      "step": 35810
+    },
+    {
+      "epoch": 0.6889090830412393,
+      "grad_norm": 1578.7619336539742,
+      "learning_rate": 4.489574977933467e-06,
+      "loss": 235.9885,
+      "step": 35820
+    },
+    {
+      "epoch": 0.6891014083017198,
+      "grad_norm": 1560.615096373814,
+      "learning_rate": 4.484483071306048e-06,
+      "loss": 246.358,
+      "step": 35830
+    },
+    {
+      "epoch": 0.6892937335622004,
+      "grad_norm": 1433.1483361624723,
+      "learning_rate": 4.479393219117588e-06,
+      "loss": 235.0705,
+      "step": 35840
+    },
+    {
+      "epoch": 0.6894860588226809,
+      "grad_norm": 1432.4922794531635,
+      "learning_rate": 4.474305423263984e-06,
+      "loss": 236.9843,
+      "step": 35850
+    },
+    {
+      "epoch": 0.6896783840831614,
+      "grad_norm": 1669.7486936920873,
+      "learning_rate": 4.469219685640348e-06,
+      "loss": 232.8931,
+      "step": 35860
+    },
+    {
+      "epoch": 0.6898707093436419,
+      "grad_norm": 1670.3213881608501,
+      "learning_rate": 4.4641360081410356e-06,
+      "loss": 239.7855,
+      "step": 35870
+    },
+    {
+      "epoch": 0.6900630346041224,
+      "grad_norm": 1769.9006754087213,
+      "learning_rate": 4.4590543926596285e-06,
+      "loss": 243.5827,
+      "step": 35880
+    },
+    {
+      "epoch": 0.6902553598646031,
+      "grad_norm": 1546.2195805647068,
+      "learning_rate": 4.453974841088953e-06,
+      "loss": 244.6879,
+      "step": 35890
+    },
+    {
+      "epoch": 0.6904476851250836,
+      "grad_norm": 1529.8960950734172,
+      "learning_rate": 4.448897355321049e-06,
+      "loss": 238.6065,
+      "step": 35900
+    },
+    {
+      "epoch": 0.6906400103855641,
+      "grad_norm": 1515.8719221340143,
+      "learning_rate": 4.443821937247205e-06,
+      "loss": 237.8188,
+      "step": 35910
+    },
+    {
+      "epoch": 0.6908323356460446,
+      "grad_norm": 1620.5687062007185,
+      "learning_rate": 4.438748588757922e-06,
+      "loss": 238.8739,
+      "step": 35920
+    },
+    {
+      "epoch": 0.6910246609065251,
+      "grad_norm": 1553.720992753021,
+      "learning_rate": 4.433677311742945e-06,
+      "loss": 236.6553,
+      "step": 35930
+    },
+    {
+      "epoch": 0.6912169861670057,
+      "grad_norm": 1592.7902512353996,
+      "learning_rate": 4.428608108091241e-06,
+      "loss": 236.2994,
+      "step": 35940
+    },
+    {
+      "epoch": 0.6914093114274862,
+      "grad_norm": 1512.5315467280025,
+      "learning_rate": 4.4235409796909965e-06,
+      "loss": 237.001,
+      "step": 35950
+    },
+    {
+      "epoch": 0.6916016366879667,
+      "grad_norm": 1592.8641353374699,
+      "learning_rate": 4.418475928429644e-06,
+      "loss": 237.5281,
+      "step": 35960
+    },
+    {
+      "epoch": 0.6917939619484472,
+      "grad_norm": 1673.5055085300598,
+      "learning_rate": 4.413412956193826e-06,
+      "loss": 247.5375,
+      "step": 35970
+    },
+    {
+      "epoch": 0.6919862872089277,
+      "grad_norm": 1658.7893733343092,
+      "learning_rate": 4.4083520648694156e-06,
+      "loss": 230.0056,
+      "step": 35980
+    },
+    {
+      "epoch": 0.6921786124694083,
+      "grad_norm": 1564.9546440812223,
+      "learning_rate": 4.4032932563415075e-06,
+      "loss": 236.3262,
+      "step": 35990
+    },
+    {
+      "epoch": 0.6923709377298888,
+      "grad_norm": 1537.2178303272024,
+      "learning_rate": 4.398236532494434e-06,
+      "loss": 237.2992,
+      "step": 36000
+    },
+    {
+      "epoch": 0.6925632629903693,
+      "grad_norm": 1657.592525169498,
+      "learning_rate": 4.393181895211735e-06,
+      "loss": 248.6368,
+      "step": 36010
+    },
+    {
+      "epoch": 0.6927555882508498,
+      "grad_norm": 1578.9140177466093,
+      "learning_rate": 4.388129346376177e-06,
+      "loss": 236.5844,
+      "step": 36020
+    },
+    {
+      "epoch": 0.6929479135113303,
+      "grad_norm": 1603.4710988208183,
+      "learning_rate": 4.383078887869759e-06,
+      "loss": 234.7896,
+      "step": 36030
+    },
+    {
+      "epoch": 0.6931402387718109,
+      "grad_norm": 1451.0021624587962,
+      "learning_rate": 4.378030521573683e-06,
+      "loss": 233.4693,
+      "step": 36040
+    },
+    {
+      "epoch": 0.6933325640322914,
+      "grad_norm": 1572.0228397647154,
+      "learning_rate": 4.372984249368393e-06,
+      "loss": 237.8678,
+      "step": 36050
+    },
+    {
+      "epoch": 0.6935248892927719,
+      "grad_norm": 1603.752642575329,
+      "learning_rate": 4.367940073133533e-06,
+      "loss": 235.3691,
+      "step": 36060
+    },
+    {
+      "epoch": 0.6937172145532524,
+      "grad_norm": 1522.568548592612,
+      "learning_rate": 4.362897994747982e-06,
+      "loss": 238.5645,
+      "step": 36070
+    },
+    {
+      "epoch": 0.6939095398137329,
+      "grad_norm": 1468.6477682731438,
+      "learning_rate": 4.35785801608983e-06,
+      "loss": 230.4714,
+      "step": 36080
+    },
+    {
+      "epoch": 0.6941018650742136,
+      "grad_norm": 1834.847687325395,
+      "learning_rate": 4.352820139036379e-06,
+      "loss": 238.1109,
+      "step": 36090
+    },
+    {
+      "epoch": 0.6942941903346941,
+      "grad_norm": 1462.0342093144627,
+      "learning_rate": 4.347784365464163e-06,
+      "loss": 237.521,
+      "step": 36100
+    },
+    {
+      "epoch": 0.6944865155951746,
+      "grad_norm": 1418.0825557405747,
+      "learning_rate": 4.342750697248922e-06,
+      "loss": 240.2118,
+      "step": 36110
+    },
+    {
+      "epoch": 0.6946788408556551,
+      "grad_norm": 1559.7967771585688,
+      "learning_rate": 4.337719136265614e-06,
+      "loss": 230.327,
+      "step": 36120
+    },
+    {
+      "epoch": 0.6948711661161356,
+      "grad_norm": 1417.2288115778736,
+      "learning_rate": 4.332689684388408e-06,
+      "loss": 233.4392,
+      "step": 36130
+    },
+    {
+      "epoch": 0.6950634913766162,
+      "grad_norm": 1523.724144053179,
+      "learning_rate": 4.327662343490701e-06,
+      "loss": 237.2281,
+      "step": 36140
+    },
+    {
+      "epoch": 0.6952558166370967,
+      "grad_norm": 1581.0526726482626,
+      "learning_rate": 4.322637115445088e-06,
+      "loss": 244.6629,
+      "step": 36150
+    },
+    {
+      "epoch": 0.6954481418975772,
+      "grad_norm": 1550.2562505480962,
+      "learning_rate": 4.3176140021233845e-06,
+      "loss": 236.3199,
+      "step": 36160
+    },
+    {
+      "epoch": 0.6956404671580577,
+      "grad_norm": 1579.3722451659926,
+      "learning_rate": 4.312593005396615e-06,
+      "loss": 239.1987,
+      "step": 36170
+    },
+    {
+      "epoch": 0.6958327924185382,
+      "grad_norm": 1474.3929174394623,
+      "learning_rate": 4.307574127135022e-06,
+      "loss": 233.4023,
+      "step": 36180
+    },
+    {
+      "epoch": 0.6960251176790188,
+      "grad_norm": 1748.2313424856814,
+      "learning_rate": 4.3025573692080516e-06,
+      "loss": 239.5724,
+      "step": 36190
+    },
+    {
+      "epoch": 0.6962174429394993,
+      "grad_norm": 1668.808792585717,
+      "learning_rate": 4.297542733484364e-06,
+      "loss": 234.4606,
+      "step": 36200
+    },
+    {
+      "epoch": 0.6964097681999798,
+      "grad_norm": 1461.0731753006285,
+      "learning_rate": 4.292530221831832e-06,
+      "loss": 240.304,
+      "step": 36210
+    },
+    {
+      "epoch": 0.6966020934604603,
+      "grad_norm": 1615.1909924772508,
+      "learning_rate": 4.2875198361175305e-06,
+      "loss": 237.9068,
+      "step": 36220
+    },
+    {
+      "epoch": 0.6967944187209408,
+      "grad_norm": 1383.8775109463547,
+      "learning_rate": 4.282511578207746e-06,
+      "loss": 234.1359,
+      "step": 36230
+    },
+    {
+      "epoch": 0.6969867439814214,
+      "grad_norm": 1466.7274005129375,
+      "learning_rate": 4.277505449967967e-06,
+      "loss": 240.9818,
+      "step": 36240
+    },
+    {
+      "epoch": 0.6971790692419019,
+      "grad_norm": 1392.4392078728338,
+      "learning_rate": 4.2725014532629015e-06,
+      "loss": 239.9398,
+      "step": 36250
+    },
+    {
+      "epoch": 0.6973713945023824,
+      "grad_norm": 1472.6089944240452,
+      "learning_rate": 4.267499589956453e-06,
+      "loss": 237.7319,
+      "step": 36260
+    },
+    {
+      "epoch": 0.6975637197628629,
+      "grad_norm": 1523.7424218507942,
+      "learning_rate": 4.262499861911727e-06,
+      "loss": 243.8434,
+      "step": 36270
+    },
+    {
+      "epoch": 0.6977560450233434,
+      "grad_norm": 1652.6107732906125,
+      "learning_rate": 4.257502270991048e-06,
+      "loss": 239.0709,
+      "step": 36280
+    },
+    {
+      "epoch": 0.697948370283824,
+      "grad_norm": 1577.8605099156312,
+      "learning_rate": 4.252506819055934e-06,
+      "loss": 242.0587,
+      "step": 36290
+    },
+    {
+      "epoch": 0.6981406955443046,
+      "grad_norm": 1610.7692035032255,
+      "learning_rate": 4.2475135079671045e-06,
+      "loss": 237.3633,
+      "step": 36300
+    },
+    {
+      "epoch": 0.6983330208047851,
+      "grad_norm": 1413.6707835153559,
+      "learning_rate": 4.242522339584486e-06,
+      "loss": 233.5362,
+      "step": 36310
+    },
+    {
+      "epoch": 0.6985253460652656,
+      "grad_norm": 1373.494526053818,
+      "learning_rate": 4.2375333157672114e-06,
+      "loss": 238.8478,
+      "step": 36320
+    },
+    {
+      "epoch": 0.6987176713257461,
+      "grad_norm": 1474.978860781716,
+      "learning_rate": 4.232546438373604e-06,
+      "loss": 237.1909,
+      "step": 36330
+    },
+    {
+      "epoch": 0.6989099965862267,
+      "grad_norm": 1685.7446852498474,
+      "learning_rate": 4.227561709261198e-06,
+      "loss": 230.6656,
+      "step": 36340
+    },
+    {
+      "epoch": 0.6991023218467072,
+      "grad_norm": 1444.990274603899,
+      "learning_rate": 4.222579130286716e-06,
+      "loss": 239.9439,
+      "step": 36350
+    },
+    {
+      "epoch": 0.6992946471071877,
+      "grad_norm": 1440.9177629702247,
+      "learning_rate": 4.217598703306095e-06,
+      "loss": 242.8563,
+      "step": 36360
+    },
+    {
+      "epoch": 0.6994869723676682,
+      "grad_norm": 1482.7561204465965,
+      "learning_rate": 4.212620430174457e-06,
+      "loss": 242.8523,
+      "step": 36370
+    },
+    {
+      "epoch": 0.6996792976281487,
+      "grad_norm": 1684.3337713894173,
+      "learning_rate": 4.207644312746124e-06,
+      "loss": 236.7045,
+      "step": 36380
+    },
+    {
+      "epoch": 0.6998716228886293,
+      "grad_norm": 1570.0048378372696,
+      "learning_rate": 4.202670352874625e-06,
+      "loss": 234.627,
+      "step": 36390
+    },
+    {
+      "epoch": 0.7000639481491098,
+      "grad_norm": 1551.606464965818,
+      "learning_rate": 4.197698552412672e-06,
+      "loss": 245.2033,
+      "step": 36400
+    },
+    {
+      "epoch": 0.7002562734095903,
+      "grad_norm": 1571.8202653053443,
+      "learning_rate": 4.192728913212181e-06,
+      "loss": 239.5426,
+      "step": 36410
+    },
+    {
+      "epoch": 0.7004485986700708,
+      "grad_norm": 1429.5459380760037,
+      "learning_rate": 4.187761437124256e-06,
+      "loss": 230.2199,
+      "step": 36420
+    },
+    {
+      "epoch": 0.7006409239305513,
+      "grad_norm": 1998.7112756520833,
+      "learning_rate": 4.182796125999207e-06,
+      "loss": 242.5665,
+      "step": 36430
+    },
+    {
+      "epoch": 0.7008332491910318,
+      "grad_norm": 1859.073618336908,
+      "learning_rate": 4.177832981686526e-06,
+      "loss": 242.1965,
+      "step": 36440
+    },
+    {
+      "epoch": 0.7010255744515124,
+      "grad_norm": 1619.6290495473081,
+      "learning_rate": 4.172872006034899e-06,
+      "loss": 238.2478,
+      "step": 36450
+    },
+    {
+      "epoch": 0.7012178997119929,
+      "grad_norm": 1620.2636076833003,
+      "learning_rate": 4.167913200892217e-06,
+      "loss": 233.8886,
+      "step": 36460
+    },
+    {
+      "epoch": 0.7014102249724734,
+      "grad_norm": 1422.7864452436954,
+      "learning_rate": 4.162956568105543e-06,
+      "loss": 245.792,
+      "step": 36470
+    },
+    {
+      "epoch": 0.7016025502329539,
+      "grad_norm": 1540.3041101411256,
+      "learning_rate": 4.158002109521149e-06,
+      "loss": 233.4258,
+      "step": 36480
+    },
+    {
+      "epoch": 0.7017948754934344,
+      "grad_norm": 1590.403755164191,
+      "learning_rate": 4.153049826984482e-06,
+      "loss": 238.6875,
+      "step": 36490
+    },
+    {
+      "epoch": 0.7019872007539151,
+      "grad_norm": 1486.442548055403,
+      "learning_rate": 4.148099722340192e-06,
+      "loss": 245.4797,
+      "step": 36500
+    },
+    {
+      "epoch": 0.7021795260143956,
+      "grad_norm": 1573.1200138846857,
+      "learning_rate": 4.143151797432109e-06,
+      "loss": 236.4589,
+      "step": 36510
+    },
+    {
+      "epoch": 0.7023718512748761,
+      "grad_norm": 1465.6549467487025,
+      "learning_rate": 4.1382060541032505e-06,
+      "loss": 239.0211,
+      "step": 36520
+    },
+    {
+      "epoch": 0.7025641765353566,
+      "grad_norm": 1486.1555173082197,
+      "learning_rate": 4.133262494195824e-06,
+      "loss": 238.1677,
+      "step": 36530
+    },
+    {
+      "epoch": 0.7027565017958372,
+      "grad_norm": 1468.431071874715,
+      "learning_rate": 4.12832111955123e-06,
+      "loss": 228.7636,
+      "step": 36540
+    },
+    {
+      "epoch": 0.7029488270563177,
+      "grad_norm": 1635.3959797153484,
+      "learning_rate": 4.123381932010044e-06,
+      "loss": 240.603,
+      "step": 36550
+    },
+    {
+      "epoch": 0.7031411523167982,
+      "grad_norm": 1513.0296072196882,
+      "learning_rate": 4.11844493341203e-06,
+      "loss": 238.3765,
+      "step": 36560
+    },
+    {
+      "epoch": 0.7033334775772787,
+      "grad_norm": 1420.5802294866357,
+      "learning_rate": 4.113510125596145e-06,
+      "loss": 240.6861,
+      "step": 36570
+    },
+    {
+      "epoch": 0.7035258028377592,
+      "grad_norm": 1433.1778599123843,
+      "learning_rate": 4.1085775104005186e-06,
+      "loss": 229.3261,
+      "step": 36580
+    },
+    {
+      "epoch": 0.7037181280982397,
+      "grad_norm": 1581.9472138163,
+      "learning_rate": 4.10364708966247e-06,
+      "loss": 239.3117,
+      "step": 36590
+    },
+    {
+      "epoch": 0.7039104533587203,
+      "grad_norm": 1431.5859957338066,
+      "learning_rate": 4.098718865218496e-06,
+      "loss": 229.8606,
+      "step": 36600
+    },
+    {
+      "epoch": 0.7041027786192008,
+      "grad_norm": 1592.5707742720695,
+      "learning_rate": 4.0937928389042815e-06,
+      "loss": 245.4869,
+      "step": 36610
+    },
+    {
+      "epoch": 0.7042951038796813,
+      "grad_norm": 1541.539229887895,
+      "learning_rate": 4.088869012554694e-06,
+      "loss": 232.0533,
+      "step": 36620
+    },
+    {
+      "epoch": 0.7044874291401618,
+      "grad_norm": 1655.9729420262363,
+      "learning_rate": 4.08394738800377e-06,
+      "loss": 236.9838,
+      "step": 36630
+    },
+    {
+      "epoch": 0.7046797544006423,
+      "grad_norm": 1548.9981635393065,
+      "learning_rate": 4.07902796708474e-06,
+      "loss": 231.4997,
+      "step": 36640
+    },
+    {
+      "epoch": 0.7048720796611229,
+      "grad_norm": 1533.6145908623128,
+      "learning_rate": 4.074110751630005e-06,
+      "loss": 243.34,
+      "step": 36650
+    },
+    {
+      "epoch": 0.7050644049216034,
+      "grad_norm": 1534.2479666720096,
+      "learning_rate": 4.0691957434711446e-06,
+      "loss": 233.8075,
+      "step": 36660
+    },
+    {
+      "epoch": 0.7052567301820839,
+      "grad_norm": 1466.9178573390122,
+      "learning_rate": 4.0642829444389165e-06,
+      "loss": 235.0018,
+      "step": 36670
+    },
+    {
+      "epoch": 0.7054490554425644,
+      "grad_norm": 1371.9886259276007,
+      "learning_rate": 4.059372356363263e-06,
+      "loss": 244.473,
+      "step": 36680
+    },
+    {
+      "epoch": 0.7056413807030449,
+      "grad_norm": 1395.573012847938,
+      "learning_rate": 4.054463981073296e-06,
+      "loss": 236.107,
+      "step": 36690
+    },
+    {
+      "epoch": 0.7058337059635255,
+      "grad_norm": 1460.9548441821817,
+      "learning_rate": 4.049557820397297e-06,
+      "loss": 238.7327,
+      "step": 36700
+    },
+    {
+      "epoch": 0.7060260312240061,
+      "grad_norm": 1556.3728261556753,
+      "learning_rate": 4.044653876162738e-06,
+      "loss": 228.712,
+      "step": 36710
+    },
+    {
+      "epoch": 0.7062183564844866,
+      "grad_norm": 1532.8862849101301,
+      "learning_rate": 4.039752150196257e-06,
+      "loss": 243.1279,
+      "step": 36720
+    },
+    {
+      "epoch": 0.7064106817449671,
+      "grad_norm": 1510.087979493412,
+      "learning_rate": 4.034852644323661e-06,
+      "loss": 240.6138,
+      "step": 36730
+    },
+    {
+      "epoch": 0.7066030070054476,
+      "grad_norm": 1488.1642748194001,
+      "learning_rate": 4.029955360369935e-06,
+      "loss": 231.6899,
+      "step": 36740
+    },
+    {
+      "epoch": 0.7067953322659282,
+      "grad_norm": 1668.262096692243,
+      "learning_rate": 4.0250603001592416e-06,
+      "loss": 228.8245,
+      "step": 36750
+    },
+    {
+      "epoch": 0.7069876575264087,
+      "grad_norm": 1449.0467594894396,
+      "learning_rate": 4.020167465514903e-06,
+      "loss": 232.0522,
+      "step": 36760
+    },
+    {
+      "epoch": 0.7071799827868892,
+      "grad_norm": 1732.14284558293,
+      "learning_rate": 4.015276858259427e-06,
+      "loss": 236.3973,
+      "step": 36770
+    },
+    {
+      "epoch": 0.7073723080473697,
+      "grad_norm": 1620.0365120255449,
+      "learning_rate": 4.0103884802144775e-06,
+      "loss": 237.6175,
+      "step": 36780
+    },
+    {
+      "epoch": 0.7075646333078502,
+      "grad_norm": 1435.9687837395886,
+      "learning_rate": 4.0055023332009e-06,
+      "loss": 237.5677,
+      "step": 36790
+    },
+    {
+      "epoch": 0.7077569585683308,
+      "grad_norm": 1430.424433405911,
+      "learning_rate": 4.000618419038702e-06,
+      "loss": 224.1724,
+      "step": 36800
+    },
+    {
+      "epoch": 0.7079492838288113,
+      "grad_norm": 1539.9769017012263,
+      "learning_rate": 3.9957367395470555e-06,
+      "loss": 240.6433,
+      "step": 36810
+    },
+    {
+      "epoch": 0.7081416090892918,
+      "grad_norm": 1497.8846945501607,
+      "learning_rate": 3.990857296544315e-06,
+      "loss": 229.3633,
+      "step": 36820
+    },
+    {
+      "epoch": 0.7083339343497723,
+      "grad_norm": 1478.890507936883,
+      "learning_rate": 3.985980091847985e-06,
+      "loss": 232.3452,
+      "step": 36830
+    },
+    {
+      "epoch": 0.7085262596102528,
+      "grad_norm": 1664.6373127632035,
+      "learning_rate": 3.981105127274748e-06,
+      "loss": 237.4508,
+      "step": 36840
+    },
+    {
+      "epoch": 0.7087185848707334,
+      "grad_norm": 1746.158692369701,
+      "learning_rate": 3.976232404640441e-06,
+      "loss": 227.9658,
+      "step": 36850
+    },
+    {
+      "epoch": 0.7089109101312139,
+      "grad_norm": 1567.4040581206111,
+      "learning_rate": 3.971361925760081e-06,
+      "loss": 236.1024,
+      "step": 36860
+    },
+    {
+      "epoch": 0.7091032353916944,
+      "grad_norm": 1722.1054415455667,
+      "learning_rate": 3.966493692447838e-06,
+      "loss": 237.7352,
+      "step": 36870
+    },
+    {
+      "epoch": 0.7092955606521749,
+      "grad_norm": 1566.9968636591109,
+      "learning_rate": 3.961627706517044e-06,
+      "loss": 233.6227,
+      "step": 36880
+    },
+    {
+      "epoch": 0.7094878859126554,
+      "grad_norm": 1365.377334290341,
+      "learning_rate": 3.956763969780206e-06,
+      "loss": 234.3301,
+      "step": 36890
+    },
+    {
+      "epoch": 0.709680211173136,
+      "grad_norm": 1445.511445977611,
+      "learning_rate": 3.951902484048978e-06,
+      "loss": 231.3383,
+      "step": 36900
+    },
+    {
+      "epoch": 0.7098725364336166,
+      "grad_norm": 1663.3247427441008,
+      "learning_rate": 3.94704325113419e-06,
+      "loss": 232.236,
+      "step": 36910
+    },
+    {
+      "epoch": 0.7100648616940971,
+      "grad_norm": 1400.0997349840698,
+      "learning_rate": 3.942186272845821e-06,
+      "loss": 232.9099,
+      "step": 36920
+    },
+    {
+      "epoch": 0.7102571869545776,
+      "grad_norm": 1363.7355472352858,
+      "learning_rate": 3.937331550993021e-06,
+      "loss": 235.6557,
+      "step": 36930
+    },
+    {
+      "epoch": 0.7104495122150581,
+      "grad_norm": 1670.4160699662211,
+      "learning_rate": 3.932479087384089e-06,
+      "loss": 240.1717,
+      "step": 36940
+    },
+    {
+      "epoch": 0.7106418374755387,
+      "grad_norm": 1505.6961227861493,
+      "learning_rate": 3.927628883826488e-06,
+      "loss": 234.9185,
+      "step": 36950
+    },
+    {
+      "epoch": 0.7108341627360192,
+      "grad_norm": 1435.1502078133735,
+      "learning_rate": 3.922780942126837e-06,
+      "loss": 234.7055,
+      "step": 36960
+    },
+    {
+      "epoch": 0.7110264879964997,
+      "grad_norm": 1376.20604381166,
+      "learning_rate": 3.91793526409092e-06,
+      "loss": 235.231,
+      "step": 36970
+    },
+    {
+      "epoch": 0.7112188132569802,
+      "grad_norm": 1573.3670881740359,
+      "learning_rate": 3.913091851523667e-06,
+      "loss": 236.8275,
+      "step": 36980
+    },
+    {
+      "epoch": 0.7114111385174607,
+      "grad_norm": 1453.3263390209486,
+      "learning_rate": 3.908250706229168e-06,
+      "loss": 230.027,
+      "step": 36990
+    },
+    {
+      "epoch": 0.7116034637779413,
+      "grad_norm": 1485.769996440751,
+      "learning_rate": 3.903411830010676e-06,
+      "loss": 242.0607,
+      "step": 37000
+    },
+    {
+      "epoch": 0.7117957890384218,
+      "grad_norm": 1532.2970997159234,
+      "learning_rate": 3.8985752246705885e-06,
+      "loss": 234.7675,
+      "step": 37010
+    },
+    {
+      "epoch": 0.7119881142989023,
+      "grad_norm": 1547.6265843219417,
+      "learning_rate": 3.893740892010463e-06,
+      "loss": 236.3145,
+      "step": 37020
+    },
+    {
+      "epoch": 0.7121804395593828,
+      "grad_norm": 1561.630207264149,
+      "learning_rate": 3.888908833831002e-06,
+      "loss": 239.4003,
+      "step": 37030
+    },
+    {
+      "epoch": 0.7123727648198633,
+      "grad_norm": 1510.605240861088,
+      "learning_rate": 3.884079051932073e-06,
+      "loss": 236.1792,
+      "step": 37040
+    },
+    {
+      "epoch": 0.7125650900803439,
+      "grad_norm": 1534.3632009540188,
+      "learning_rate": 3.879251548112692e-06,
+      "loss": 241.3086,
+      "step": 37050
+    },
+    {
+      "epoch": 0.7127574153408244,
+      "grad_norm": 1806.6796395736437,
+      "learning_rate": 3.874426324171019e-06,
+      "loss": 249.0332,
+      "step": 37060
+    },
+    {
+      "epoch": 0.7129497406013049,
+      "grad_norm": 2016.3750466670258,
+      "learning_rate": 3.869603381904377e-06,
+      "loss": 236.7132,
+      "step": 37070
+    },
+    {
+      "epoch": 0.7131420658617854,
+      "grad_norm": 1566.106222686981,
+      "learning_rate": 3.864782723109227e-06,
+      "loss": 231.7482,
+      "step": 37080
+    },
+    {
+      "epoch": 0.7133343911222659,
+      "grad_norm": 1560.2987007657275,
+      "learning_rate": 3.859964349581187e-06,
+      "loss": 233.1552,
+      "step": 37090
+    },
+    {
+      "epoch": 0.7135267163827465,
+      "grad_norm": 1769.2260892394934,
+      "learning_rate": 3.855148263115017e-06,
+      "loss": 234.646,
+      "step": 37100
+    },
+    {
+      "epoch": 0.713719041643227,
+      "grad_norm": 1698.0319735982089,
+      "learning_rate": 3.850334465504637e-06,
+      "loss": 236.5427,
+      "step": 37110
+    },
+    {
+      "epoch": 0.7139113669037076,
+      "grad_norm": 1684.4199983771625,
+      "learning_rate": 3.845522958543104e-06,
+      "loss": 241.887,
+      "step": 37120
+    },
+    {
+      "epoch": 0.7141036921641881,
+      "grad_norm": 1844.9808840557555,
+      "learning_rate": 3.840713744022624e-06,
+      "loss": 236.2804,
+      "step": 37130
+    },
+    {
+      "epoch": 0.7142960174246686,
+      "grad_norm": 1541.3606644378585,
+      "learning_rate": 3.835906823734548e-06,
+      "loss": 240.3451,
+      "step": 37140
+    },
+    {
+      "epoch": 0.7144883426851492,
+      "grad_norm": 1627.873940031945,
+      "learning_rate": 3.831102199469379e-06,
+      "loss": 249.0218,
+      "step": 37150
+    },
+    {
+      "epoch": 0.7146806679456297,
+      "grad_norm": 1448.4906422092638,
+      "learning_rate": 3.826299873016758e-06,
+      "loss": 238.2984,
+      "step": 37160
+    },
+    {
+      "epoch": 0.7148729932061102,
+      "grad_norm": 1481.0101010630947,
+      "learning_rate": 3.821499846165468e-06,
+      "loss": 227.4023,
+      "step": 37170
+    },
+    {
+      "epoch": 0.7150653184665907,
+      "grad_norm": 1511.9525245861967,
+      "learning_rate": 3.816702120703449e-06,
+      "loss": 234.5009,
+      "step": 37180
+    },
+    {
+      "epoch": 0.7152576437270712,
+      "grad_norm": 1434.3699061079233,
+      "learning_rate": 3.8119066984177654e-06,
+      "loss": 231.3417,
+      "step": 37190
+    },
+    {
+      "epoch": 0.7154499689875518,
+      "grad_norm": 1602.5048808493707,
+      "learning_rate": 3.8071135810946415e-06,
+      "loss": 237.2059,
+      "step": 37200
+    },
+    {
+      "epoch": 0.7156422942480323,
+      "grad_norm": 1609.5907307870593,
+      "learning_rate": 3.802322770519424e-06,
+      "loss": 230.2667,
+      "step": 37210
+    },
+    {
+      "epoch": 0.7158346195085128,
+      "grad_norm": 1462.1144772868236,
+      "learning_rate": 3.7975342684766215e-06,
+      "loss": 229.9174,
+      "step": 37220
+    },
+    {
+      "epoch": 0.7160269447689933,
+      "grad_norm": 1556.4489530243522,
+      "learning_rate": 3.792748076749867e-06,
+      "loss": 236.9995,
+      "step": 37230
+    },
+    {
+      "epoch": 0.7162192700294738,
+      "grad_norm": 1606.6375549844663,
+      "learning_rate": 3.787964197121934e-06,
+      "loss": 237.1684,
+      "step": 37240
+    },
+    {
+      "epoch": 0.7164115952899544,
+      "grad_norm": 1461.1124908639335,
+      "learning_rate": 3.7831826313747454e-06,
+      "loss": 228.4254,
+      "step": 37250
+    },
+    {
+      "epoch": 0.7166039205504349,
+      "grad_norm": 1624.4875748797022,
+      "learning_rate": 3.778403381289353e-06,
+      "loss": 229.4599,
+      "step": 37260
+    },
+    {
+      "epoch": 0.7167962458109154,
+      "grad_norm": 1461.1178928515196,
+      "learning_rate": 3.7736264486459486e-06,
+      "loss": 231.6607,
+      "step": 37270
+    },
+    {
+      "epoch": 0.7169885710713959,
+      "grad_norm": 1557.728353339117,
+      "learning_rate": 3.7688518352238555e-06,
+      "loss": 232.5229,
+      "step": 37280
+    },
+    {
+      "epoch": 0.7171808963318764,
+      "grad_norm": 1741.0367463683694,
+      "learning_rate": 3.7640795428015462e-06,
+      "loss": 237.2978,
+      "step": 37290
+    },
+    {
+      "epoch": 0.717373221592357,
+      "grad_norm": 1444.1362216994314,
+      "learning_rate": 3.7593095731566186e-06,
+      "loss": 232.3923,
+      "step": 37300
+    },
+    {
+      "epoch": 0.7175655468528375,
+      "grad_norm": 1559.3365992915112,
+      "learning_rate": 3.7545419280658025e-06,
+      "loss": 235.3234,
+      "step": 37310
+    },
+    {
+      "epoch": 0.7177578721133181,
+      "grad_norm": 1553.1545499996446,
+      "learning_rate": 3.749776609304975e-06,
+      "loss": 234.207,
+      "step": 37320
+    },
+    {
+      "epoch": 0.7179501973737986,
+      "grad_norm": 1575.052944165723,
+      "learning_rate": 3.7450136186491315e-06,
+      "loss": 237.8557,
+      "step": 37330
+    },
+    {
+      "epoch": 0.7181425226342791,
+      "grad_norm": 1544.7978492163777,
+      "learning_rate": 3.7402529578724134e-06,
+      "loss": 234.4243,
+      "step": 37340
+    },
+    {
+      "epoch": 0.7183348478947597,
+      "grad_norm": 1504.0280172511,
+      "learning_rate": 3.735494628748082e-06,
+      "loss": 234.0796,
+      "step": 37350
+    },
+    {
+      "epoch": 0.7185271731552402,
+      "grad_norm": 1542.6179104680355,
+      "learning_rate": 3.730738633048543e-06,
+      "loss": 230.7367,
+      "step": 37360
+    },
+    {
+      "epoch": 0.7187194984157207,
+      "grad_norm": 1503.465130204258,
+      "learning_rate": 3.7259849725453225e-06,
+      "loss": 238.3747,
+      "step": 37370
+    },
+    {
+      "epoch": 0.7189118236762012,
+      "grad_norm": 1560.8548193230586,
+      "learning_rate": 3.7212336490090815e-06,
+      "loss": 231.5233,
+      "step": 37380
+    },
+    {
+      "epoch": 0.7191041489366817,
+      "grad_norm": 1625.0760171848426,
+      "learning_rate": 3.7164846642096053e-06,
+      "loss": 235.499,
+      "step": 37390
+    },
+    {
+      "epoch": 0.7192964741971622,
+      "grad_norm": 1611.7693898449481,
+      "learning_rate": 3.7117380199158204e-06,
+      "loss": 241.0182,
+      "step": 37400
+    },
+    {
+      "epoch": 0.7194887994576428,
+      "grad_norm": 1525.2995827557488,
+      "learning_rate": 3.706993717895768e-06,
+      "loss": 231.1189,
+      "step": 37410
+    },
+    {
+      "epoch": 0.7196811247181233,
+      "grad_norm": 1662.461814996555,
+      "learning_rate": 3.70225175991662e-06,
+      "loss": 233.8253,
+      "step": 37420
+    },
+    {
+      "epoch": 0.7198734499786038,
+      "grad_norm": 1627.8829357689497,
+      "learning_rate": 3.697512147744684e-06,
+      "loss": 239.7928,
+      "step": 37430
+    },
+    {
+      "epoch": 0.7200657752390843,
+      "grad_norm": 1580.84535546553,
+      "learning_rate": 3.6927748831453835e-06,
+      "loss": 229.8675,
+      "step": 37440
+    },
+    {
+      "epoch": 0.7202581004995648,
+      "grad_norm": 1691.9899522622936,
+      "learning_rate": 3.688039967883269e-06,
+      "loss": 233.9928,
+      "step": 37450
+    },
+    {
+      "epoch": 0.7204504257600454,
+      "grad_norm": 1646.4956034451407,
+      "learning_rate": 3.683307403722025e-06,
+      "loss": 231.9454,
+      "step": 37460
+    },
+    {
+      "epoch": 0.7206427510205259,
+      "grad_norm": 1452.9406305908544,
+      "learning_rate": 3.678577192424445e-06,
+      "loss": 230.8371,
+      "step": 37470
+    },
+    {
+      "epoch": 0.7208350762810064,
+      "grad_norm": 1909.9190232044855,
+      "learning_rate": 3.6738493357524628e-06,
+      "loss": 231.7844,
+      "step": 37480
+    },
+    {
+      "epoch": 0.7210274015414869,
+      "grad_norm": 1427.9012348817284,
+      "learning_rate": 3.6691238354671233e-06,
+      "loss": 234.9695,
+      "step": 37490
+    },
+    {
+      "epoch": 0.7212197268019674,
+      "grad_norm": 1784.1848292523393,
+      "learning_rate": 3.664400693328595e-06,
+      "loss": 235.5823,
+      "step": 37500
+    },
+    {
+      "epoch": 0.721412052062448,
+      "grad_norm": 1416.093029582277,
+      "learning_rate": 3.6596799110961746e-06,
+      "loss": 228.8743,
+      "step": 37510
+    },
+    {
+      "epoch": 0.7216043773229286,
+      "grad_norm": 1659.8501820012216,
+      "learning_rate": 3.6549614905282724e-06,
+      "loss": 239.8036,
+      "step": 37520
+    },
+    {
+      "epoch": 0.7217967025834091,
+      "grad_norm": 1456.5211880780953,
+      "learning_rate": 3.6502454333824224e-06,
+      "loss": 236.1289,
+      "step": 37530
+    },
+    {
+      "epoch": 0.7219890278438896,
+      "grad_norm": 1852.734474841436,
+      "learning_rate": 3.6455317414152803e-06,
+      "loss": 234.7422,
+      "step": 37540
+    },
+    {
+      "epoch": 0.7221813531043701,
+      "grad_norm": 1547.2671143298355,
+      "learning_rate": 3.640820416382618e-06,
+      "loss": 235.2092,
+      "step": 37550
+    },
+    {
+      "epoch": 0.7223736783648507,
+      "grad_norm": 1453.7809208511683,
+      "learning_rate": 3.6361114600393242e-06,
+      "loss": 225.6464,
+      "step": 37560
+    },
+    {
+      "epoch": 0.7225660036253312,
+      "grad_norm": 1542.9814103822732,
+      "learning_rate": 3.6314048741394057e-06,
+      "loss": 237.9018,
+      "step": 37570
+    },
+    {
+      "epoch": 0.7227583288858117,
+      "grad_norm": 1491.2889064386443,
+      "learning_rate": 3.6267006604359943e-06,
+      "loss": 236.3528,
+      "step": 37580
+    },
+    {
+      "epoch": 0.7229506541462922,
+      "grad_norm": 1678.0753334877468,
+      "learning_rate": 3.6219988206813285e-06,
+      "loss": 240.7572,
+      "step": 37590
+    },
+    {
+      "epoch": 0.7231429794067727,
+      "grad_norm": 1511.7530309298397,
+      "learning_rate": 3.6172993566267623e-06,
+      "loss": 229.9834,
+      "step": 37600
+    },
+    {
+      "epoch": 0.7233353046672533,
+      "grad_norm": 1706.6628385502697,
+      "learning_rate": 3.6126022700227715e-06,
+      "loss": 235.563,
+      "step": 37610
+    },
+    {
+      "epoch": 0.7235276299277338,
+      "grad_norm": 1621.9463661496866,
+      "learning_rate": 3.6079075626189476e-06,
+      "loss": 235.9628,
+      "step": 37620
+    },
+    {
+      "epoch": 0.7237199551882143,
+      "grad_norm": 1541.8512749723611,
+      "learning_rate": 3.603215236163987e-06,
+      "loss": 228.3153,
+      "step": 37630
+    },
+    {
+      "epoch": 0.7239122804486948,
+      "grad_norm": 1430.7869807013765,
+      "learning_rate": 3.5985252924057023e-06,
+      "loss": 229.3261,
+      "step": 37640
+    },
+    {
+      "epoch": 0.7241046057091753,
+      "grad_norm": 1416.9922622696242,
+      "learning_rate": 3.5938377330910245e-06,
+      "loss": 232.6925,
+      "step": 37650
+    },
+    {
+      "epoch": 0.7242969309696559,
+      "grad_norm": 1488.252286117843,
+      "learning_rate": 3.5891525599659905e-06,
+      "loss": 233.1205,
+      "step": 37660
+    },
+    {
+      "epoch": 0.7244892562301364,
+      "grad_norm": 1617.1437070836214,
+      "learning_rate": 3.5844697747757496e-06,
+      "loss": 229.868,
+      "step": 37670
+    },
+    {
+      "epoch": 0.7246815814906169,
+      "grad_norm": 1574.1878686033965,
+      "learning_rate": 3.5797893792645577e-06,
+      "loss": 234.2606,
+      "step": 37680
+    },
+    {
+      "epoch": 0.7248739067510974,
+      "grad_norm": 1689.432210523899,
+      "learning_rate": 3.5751113751757925e-06,
+      "loss": 236.0605,
+      "step": 37690
+    },
+    {
+      "epoch": 0.7250662320115779,
+      "grad_norm": 1479.4623468117168,
+      "learning_rate": 3.5704357642519295e-06,
+      "loss": 230.437,
+      "step": 37700
+    },
+    {
+      "epoch": 0.7252585572720585,
+      "grad_norm": 1619.7816764119484,
+      "learning_rate": 3.5657625482345526e-06,
+      "loss": 231.8185,
+      "step": 37710
+    },
+    {
+      "epoch": 0.725450882532539,
+      "grad_norm": 1693.3841506888232,
+      "learning_rate": 3.5610917288643655e-06,
+      "loss": 232.6709,
+      "step": 37720
+    },
+    {
+      "epoch": 0.7256432077930196,
+      "grad_norm": 1801.9910353301125,
+      "learning_rate": 3.556423307881167e-06,
+      "loss": 235.0963,
+      "step": 37730
+    },
+    {
+      "epoch": 0.7258355330535001,
+      "grad_norm": 1668.7726223905188,
+      "learning_rate": 3.551757287023865e-06,
+      "loss": 235.075,
+      "step": 37740
+    },
+    {
+      "epoch": 0.7260278583139806,
+      "grad_norm": 1388.884080230714,
+      "learning_rate": 3.547093668030479e-06,
+      "loss": 235.9934,
+      "step": 37750
+    },
+    {
+      "epoch": 0.7262201835744612,
+      "grad_norm": 1677.8991278417686,
+      "learning_rate": 3.542432452638126e-06,
+      "loss": 234.0942,
+      "step": 37760
+    },
+    {
+      "epoch": 0.7264125088349417,
+      "grad_norm": 1544.1650085195756,
+      "learning_rate": 3.5377736425830366e-06,
+      "loss": 239.614,
+      "step": 37770
+    },
+    {
+      "epoch": 0.7266048340954222,
+      "grad_norm": 1812.3429776276469,
+      "learning_rate": 3.5331172396005354e-06,
+      "loss": 229.4545,
+      "step": 37780
+    },
+    {
+      "epoch": 0.7267971593559027,
+      "grad_norm": 1603.1230736647542,
+      "learning_rate": 3.528463245425062e-06,
+      "loss": 240.1801,
+      "step": 37790
+    },
+    {
+      "epoch": 0.7269894846163832,
+      "grad_norm": 1443.434335743417,
+      "learning_rate": 3.5238116617901486e-06,
+      "loss": 232.0958,
+      "step": 37800
+    },
+    {
+      "epoch": 0.7271818098768638,
+      "grad_norm": 1605.3742844790415,
+      "learning_rate": 3.519162490428433e-06,
+      "loss": 239.055,
+      "step": 37810
+    },
+    {
+      "epoch": 0.7273741351373443,
+      "grad_norm": 1527.088012355998,
+      "learning_rate": 3.5145157330716516e-06,
+      "loss": 229.7837,
+      "step": 37820
+    },
+    {
+      "epoch": 0.7275664603978248,
+      "grad_norm": 1480.5399344921368,
+      "learning_rate": 3.509871391450652e-06,
+      "loss": 241.1656,
+      "step": 37830
+    },
+    {
+      "epoch": 0.7277587856583053,
+      "grad_norm": 1386.8784457822,
+      "learning_rate": 3.505229467295371e-06,
+      "loss": 231.1997,
+      "step": 37840
+    },
+    {
+      "epoch": 0.7279511109187858,
+      "grad_norm": 1498.2244943293506,
+      "learning_rate": 3.5005899623348493e-06,
+      "loss": 244.6487,
+      "step": 37850
+    },
+    {
+      "epoch": 0.7281434361792664,
+      "grad_norm": 1469.4204061317337,
+      "learning_rate": 3.495952878297221e-06,
+      "loss": 235.9334,
+      "step": 37860
+    },
+    {
+      "epoch": 0.7283357614397469,
+      "grad_norm": 1351.2079624777139,
+      "learning_rate": 3.4913182169097315e-06,
+      "loss": 230.4485,
+      "step": 37870
+    },
+    {
+      "epoch": 0.7285280867002274,
+      "grad_norm": 1502.1192196783863,
+      "learning_rate": 3.4866859798987084e-06,
+      "loss": 231.663,
+      "step": 37880
+    },
+    {
+      "epoch": 0.7287204119607079,
+      "grad_norm": 1759.390405141497,
+      "learning_rate": 3.4820561689895906e-06,
+      "loss": 235.0598,
+      "step": 37890
+    },
+    {
+      "epoch": 0.7289127372211884,
+      "grad_norm": 1577.3634302237792,
+      "learning_rate": 3.4774287859068988e-06,
+      "loss": 229.306,
+      "step": 37900
+    },
+    {
+      "epoch": 0.729105062481669,
+      "grad_norm": 9423.990269975693,
+      "learning_rate": 3.472803832374263e-06,
+      "loss": 226.7137,
+      "step": 37910
+    },
+    {
+      "epoch": 0.7292973877421495,
+      "grad_norm": 1492.2728316103485,
+      "learning_rate": 3.4681813101144e-06,
+      "loss": 236.0989,
+      "step": 37920
+    },
+    {
+      "epoch": 0.7294897130026301,
+      "grad_norm": 1497.4115488474138,
+      "learning_rate": 3.4635612208491197e-06,
+      "loss": 237.5444,
+      "step": 37930
+    },
+    {
+      "epoch": 0.7296820382631106,
+      "grad_norm": 1424.5715929030805,
+      "learning_rate": 3.458943566299334e-06,
+      "loss": 224.4871,
+      "step": 37940
+    },
+    {
+      "epoch": 0.7298743635235911,
+      "grad_norm": 1602.2442483422503,
+      "learning_rate": 3.454328348185042e-06,
+      "loss": 226.0011,
+      "step": 37950
+    },
+    {
+      "epoch": 0.7300666887840717,
+      "grad_norm": 1625.1770808516565,
+      "learning_rate": 3.4497155682253314e-06,
+      "loss": 240.1945,
+      "step": 37960
+    },
+    {
+      "epoch": 0.7302590140445522,
+      "grad_norm": 1535.8323367872435,
+      "learning_rate": 3.4451052281383922e-06,
+      "loss": 227.0235,
+      "step": 37970
+    },
+    {
+      "epoch": 0.7304513393050327,
+      "grad_norm": 1579.472889542994,
+      "learning_rate": 3.440497329641499e-06,
+      "loss": 236.794,
+      "step": 37980
+    },
+    {
+      "epoch": 0.7306436645655132,
+      "grad_norm": 1580.954188423651,
+      "learning_rate": 3.435891874451017e-06,
+      "loss": 230.2277,
+      "step": 37990
+    },
+    {
+      "epoch": 0.7308359898259937,
+      "grad_norm": 1469.7377768417768,
+      "learning_rate": 3.431288864282398e-06,
+      "loss": 230.226,
+      "step": 38000
+    },
+    {
+      "epoch": 0.7310283150864743,
+      "grad_norm": 1421.2548757450052,
+      "learning_rate": 3.4266883008501937e-06,
+      "loss": 228.2719,
+      "step": 38010
+    },
+    {
+      "epoch": 0.7312206403469548,
+      "grad_norm": 1538.2825198599721,
+      "learning_rate": 3.4220901858680365e-06,
+      "loss": 238.3946,
+      "step": 38020
+    },
+    {
+      "epoch": 0.7314129656074353,
+      "grad_norm": 1521.3598967828464,
+      "learning_rate": 3.4174945210486445e-06,
+      "loss": 235.642,
+      "step": 38030
+    },
+    {
+      "epoch": 0.7316052908679158,
+      "grad_norm": 1366.2607252731727,
+      "learning_rate": 3.4129013081038285e-06,
+      "loss": 235.8387,
+      "step": 38040
+    },
+    {
+      "epoch": 0.7317976161283963,
+      "grad_norm": 1469.340021318117,
+      "learning_rate": 3.40831054874449e-06,
+      "loss": 230.4746,
+      "step": 38050
+    },
+    {
+      "epoch": 0.7319899413888769,
+      "grad_norm": 1518.1402450302116,
+      "learning_rate": 3.403722244680606e-06,
+      "loss": 230.0023,
+      "step": 38060
+    },
+    {
+      "epoch": 0.7321822666493574,
+      "grad_norm": 1588.654904140331,
+      "learning_rate": 3.3991363976212423e-06,
+      "loss": 241.2521,
+      "step": 38070
+    },
+    {
+      "epoch": 0.7323745919098379,
+      "grad_norm": 1495.2090854258015,
+      "learning_rate": 3.394553009274556e-06,
+      "loss": 227.6466,
+      "step": 38080
+    },
+    {
+      "epoch": 0.7325669171703184,
+      "grad_norm": 1490.171408351072,
+      "learning_rate": 3.389972081347782e-06,
+      "loss": 229.939,
+      "step": 38090
+    },
+    {
+      "epoch": 0.7327592424307989,
+      "grad_norm": 1565.1137289771389,
+      "learning_rate": 3.385393615547239e-06,
+      "loss": 239.5606,
+      "step": 38100
+    },
+    {
+      "epoch": 0.7329515676912794,
+      "grad_norm": 1555.2400904353592,
+      "learning_rate": 3.3808176135783276e-06,
+      "loss": 245.8193,
+      "step": 38110
+    },
+    {
+      "epoch": 0.73314389295176,
+      "grad_norm": 1563.5939510084056,
+      "learning_rate": 3.3762440771455386e-06,
+      "loss": 236.9324,
+      "step": 38120
+    },
+    {
+      "epoch": 0.7333362182122405,
+      "grad_norm": 1372.7472298470218,
+      "learning_rate": 3.371673007952435e-06,
+      "loss": 229.9706,
+      "step": 38130
+    },
+    {
+      "epoch": 0.7335285434727211,
+      "grad_norm": 1989.4626982753248,
+      "learning_rate": 3.3671044077016634e-06,
+      "loss": 244.1696,
+      "step": 38140
+    },
+    {
+      "epoch": 0.7337208687332016,
+      "grad_norm": 1382.315746654657,
+      "learning_rate": 3.3625382780949576e-06,
+      "loss": 236.8861,
+      "step": 38150
+    },
+    {
+      "epoch": 0.7339131939936822,
+      "grad_norm": 2013.7536739456355,
+      "learning_rate": 3.357974620833121e-06,
+      "loss": 236.7099,
+      "step": 38160
+    },
+    {
+      "epoch": 0.7341055192541627,
+      "grad_norm": 1480.908347081604,
+      "learning_rate": 3.353413437616039e-06,
+      "loss": 237.05,
+      "step": 38170
+    },
+    {
+      "epoch": 0.7342978445146432,
+      "grad_norm": 1609.1857888718864,
+      "learning_rate": 3.3488547301426786e-06,
+      "loss": 221.6165,
+      "step": 38180
+    },
+    {
+      "epoch": 0.7344901697751237,
+      "grad_norm": 1616.1171745614897,
+      "learning_rate": 3.344298500111087e-06,
+      "loss": 233.9652,
+      "step": 38190
+    },
+    {
+      "epoch": 0.7346824950356042,
+      "grad_norm": 1548.548677522442,
+      "learning_rate": 3.3397447492183833e-06,
+      "loss": 232.3867,
+      "step": 38200
+    },
+    {
+      "epoch": 0.7348748202960848,
+      "grad_norm": 1720.5928973720063,
+      "learning_rate": 3.3351934791607576e-06,
+      "loss": 234.0296,
+      "step": 38210
+    },
+    {
+      "epoch": 0.7350671455565653,
+      "grad_norm": 1396.6493745825644,
+      "learning_rate": 3.330644691633492e-06,
+      "loss": 231.8759,
+      "step": 38220
+    },
+    {
+      "epoch": 0.7352594708170458,
+      "grad_norm": 1605.4773960245195,
+      "learning_rate": 3.3260983883309306e-06,
+      "loss": 240.6725,
+      "step": 38230
+    },
+    {
+      "epoch": 0.7354517960775263,
+      "grad_norm": 1500.612684089531,
+      "learning_rate": 3.321554570946497e-06,
+      "loss": 228.6631,
+      "step": 38240
+    },
+    {
+      "epoch": 0.7356441213380068,
+      "grad_norm": 1553.948986128039,
+      "learning_rate": 3.317013241172684e-06,
+      "loss": 233.0377,
+      "step": 38250
+    },
+    {
+      "epoch": 0.7358364465984873,
+      "grad_norm": 1547.010115429101,
+      "learning_rate": 3.3124744007010688e-06,
+      "loss": 230.8883,
+      "step": 38260
+    },
+    {
+      "epoch": 0.7360287718589679,
+      "grad_norm": 1664.7985164728855,
+      "learning_rate": 3.3079380512222904e-06,
+      "loss": 233.1191,
+      "step": 38270
+    },
+    {
+      "epoch": 0.7362210971194484,
+      "grad_norm": 1515.2440430941901,
+      "learning_rate": 3.3034041944260654e-06,
+      "loss": 234.5917,
+      "step": 38280
+    },
+    {
+      "epoch": 0.7364134223799289,
+      "grad_norm": 2032.3442517176109,
+      "learning_rate": 3.2988728320011774e-06,
+      "loss": 231.3146,
+      "step": 38290
+    },
+    {
+      "epoch": 0.7366057476404094,
+      "grad_norm": 1527.9186762694744,
+      "learning_rate": 3.294343965635489e-06,
+      "loss": 232.4708,
+      "step": 38300
+    },
+    {
+      "epoch": 0.7367980729008899,
+      "grad_norm": 1376.744789862261,
+      "learning_rate": 3.289817597015923e-06,
+      "loss": 224.7349,
+      "step": 38310
+    },
+    {
+      "epoch": 0.7369903981613705,
+      "grad_norm": 1724.8007360933777,
+      "learning_rate": 3.2852937278284837e-06,
+      "loss": 236.0009,
+      "step": 38320
+    },
+    {
+      "epoch": 0.737182723421851,
+      "grad_norm": 1583.083827873174,
+      "learning_rate": 3.28077235975823e-06,
+      "loss": 230.9427,
+      "step": 38330
+    },
+    {
+      "epoch": 0.7373750486823316,
+      "grad_norm": 1382.2351877769281,
+      "learning_rate": 3.2762534944893033e-06,
+      "loss": 235.0521,
+      "step": 38340
+    },
+    {
+      "epoch": 0.7375673739428121,
+      "grad_norm": 1465.7386454187583,
+      "learning_rate": 3.271737133704904e-06,
+      "loss": 227.5069,
+      "step": 38350
+    },
+    {
+      "epoch": 0.7377596992032927,
+      "grad_norm": 1548.1191185165078,
+      "learning_rate": 3.2672232790872983e-06,
+      "loss": 232.8356,
+      "step": 38360
+    },
+    {
+      "epoch": 0.7379520244637732,
+      "grad_norm": 1532.9614833623862,
+      "learning_rate": 3.262711932317828e-06,
+      "loss": 234.831,
+      "step": 38370
+    },
+    {
+      "epoch": 0.7381443497242537,
+      "grad_norm": 1546.0956541191586,
+      "learning_rate": 3.258203095076894e-06,
+      "loss": 231.3097,
+      "step": 38380
+    },
+    {
+      "epoch": 0.7383366749847342,
+      "grad_norm": 1593.369081047998,
+      "learning_rate": 3.2536967690439592e-06,
+      "loss": 241.5989,
+      "step": 38390
+    },
+    {
+      "epoch": 0.7385290002452147,
+      "grad_norm": 1391.2124942627154,
+      "learning_rate": 3.249192955897562e-06,
+      "loss": 232.1474,
+      "step": 38400
+    },
+    {
+      "epoch": 0.7387213255056952,
+      "grad_norm": 1770.44050331133,
+      "learning_rate": 3.2446916573152955e-06,
+      "loss": 233.9994,
+      "step": 38410
+    },
+    {
+      "epoch": 0.7389136507661758,
+      "grad_norm": 1363.0946982386781,
+      "learning_rate": 3.24019287497382e-06,
+      "loss": 227.8752,
+      "step": 38420
+    },
+    {
+      "epoch": 0.7391059760266563,
+      "grad_norm": 1524.4093907439187,
+      "learning_rate": 3.235696610548852e-06,
+      "loss": 225.0379,
+      "step": 38430
+    },
+    {
+      "epoch": 0.7392983012871368,
+      "grad_norm": 1511.6932896158028,
+      "learning_rate": 3.231202865715184e-06,
+      "loss": 234.9341,
+      "step": 38440
+    },
+    {
+      "epoch": 0.7394906265476173,
+      "grad_norm": 1431.2248540907865,
+      "learning_rate": 3.226711642146655e-06,
+      "loss": 240.6318,
+      "step": 38450
+    },
+    {
+      "epoch": 0.7396829518080978,
+      "grad_norm": 1439.2073383165941,
+      "learning_rate": 3.222222941516179e-06,
+      "loss": 230.4061,
+      "step": 38460
+    },
+    {
+      "epoch": 0.7398752770685784,
+      "grad_norm": 1403.5401126611653,
+      "learning_rate": 3.2177367654957137e-06,
+      "loss": 223.8518,
+      "step": 38470
+    },
+    {
+      "epoch": 0.7400676023290589,
+      "grad_norm": 1559.0925481611444,
+      "learning_rate": 3.213253115756295e-06,
+      "loss": 232.6755,
+      "step": 38480
+    },
+    {
+      "epoch": 0.7402599275895394,
+      "grad_norm": 1480.5121393484953,
+      "learning_rate": 3.208771993968003e-06,
+      "loss": 232.3182,
+      "step": 38490
+    },
+    {
+      "epoch": 0.7404522528500199,
+      "grad_norm": 1539.960055255188,
+      "learning_rate": 3.2042934017999795e-06,
+      "loss": 228.5095,
+      "step": 38500
+    },
+    {
+      "epoch": 0.7406445781105004,
+      "grad_norm": 1354.0956620780714,
+      "learning_rate": 3.1998173409204326e-06,
+      "loss": 229.5166,
+      "step": 38510
+    },
+    {
+      "epoch": 0.740836903370981,
+      "grad_norm": 1508.161123468857,
+      "learning_rate": 3.1953438129966175e-06,
+      "loss": 241.1044,
+      "step": 38520
+    },
+    {
+      "epoch": 0.7410292286314615,
+      "grad_norm": 1475.1107350472876,
+      "learning_rate": 3.190872819694849e-06,
+      "loss": 239.4382,
+      "step": 38530
+    },
+    {
+      "epoch": 0.741221553891942,
+      "grad_norm": 1391.2570241695958,
+      "learning_rate": 3.1864043626804953e-06,
+      "loss": 225.6426,
+      "step": 38540
+    },
+    {
+      "epoch": 0.7414138791524226,
+      "grad_norm": 1414.0084050629603,
+      "learning_rate": 3.18193844361799e-06,
+      "loss": 224.4675,
+      "step": 38550
+    },
+    {
+      "epoch": 0.7416062044129031,
+      "grad_norm": 1607.8175152907313,
+      "learning_rate": 3.1774750641708095e-06,
+      "loss": 228.3828,
+      "step": 38560
+    },
+    {
+      "epoch": 0.7417985296733837,
+      "grad_norm": 1577.781339626044,
+      "learning_rate": 3.1730142260014875e-06,
+      "loss": 231.7078,
+      "step": 38570
+    },
+    {
+      "epoch": 0.7419908549338642,
+      "grad_norm": 1523.5309648075292,
+      "learning_rate": 3.1685559307716187e-06,
+      "loss": 229.4399,
+      "step": 38580
+    },
+    {
+      "epoch": 0.7421831801943447,
+      "grad_norm": 1486.2068497748805,
+      "learning_rate": 3.164100180141839e-06,
+      "loss": 231.492,
+      "step": 38590
+    },
+    {
+      "epoch": 0.7423755054548252,
+      "grad_norm": 1519.7079700238107,
+      "learning_rate": 3.159646975771842e-06,
+      "loss": 235.6418,
+      "step": 38600
+    },
+    {
+      "epoch": 0.7425678307153057,
+      "grad_norm": 1722.8987081581117,
+      "learning_rate": 3.155196319320374e-06,
+      "loss": 237.3484,
+      "step": 38610
+    },
+    {
+      "epoch": 0.7427601559757863,
+      "grad_norm": 1659.9162008450585,
+      "learning_rate": 3.1507482124452337e-06,
+      "loss": 236.7098,
+      "step": 38620
+    },
+    {
+      "epoch": 0.7429524812362668,
+      "grad_norm": 1494.1360038720754,
+      "learning_rate": 3.146302656803266e-06,
+      "loss": 228.4177,
+      "step": 38630
+    },
+    {
+      "epoch": 0.7431448064967473,
+      "grad_norm": 1514.5651028015404,
+      "learning_rate": 3.1418596540503653e-06,
+      "loss": 228.4572,
+      "step": 38640
+    },
+    {
+      "epoch": 0.7433371317572278,
+      "grad_norm": 1577.7417549788806,
+      "learning_rate": 3.1374192058414755e-06,
+      "loss": 240.3857,
+      "step": 38650
+    },
+    {
+      "epoch": 0.7435294570177083,
+      "grad_norm": 1483.4734957474009,
+      "learning_rate": 3.1329813138305944e-06,
+      "loss": 235.6175,
+      "step": 38660
+    },
+    {
+      "epoch": 0.7437217822781889,
+      "grad_norm": 1517.817491423927,
+      "learning_rate": 3.128545979670762e-06,
+      "loss": 224.0461,
+      "step": 38670
+    },
+    {
+      "epoch": 0.7439141075386694,
+      "grad_norm": 1472.852040987415,
+      "learning_rate": 3.124113205014063e-06,
+      "loss": 227.5622,
+      "step": 38680
+    },
+    {
+      "epoch": 0.7441064327991499,
+      "grad_norm": 1647.6381240449646,
+      "learning_rate": 3.119682991511639e-06,
+      "loss": 233.6358,
+      "step": 38690
+    },
+    {
+      "epoch": 0.7442987580596304,
+      "grad_norm": 1636.7000026118264,
+      "learning_rate": 3.1152553408136686e-06,
+      "loss": 229.2395,
+      "step": 38700
+    },
+    {
+      "epoch": 0.7444910833201109,
+      "grad_norm": 1626.5114430746166,
+      "learning_rate": 3.110830254569378e-06,
+      "loss": 233.4884,
+      "step": 38710
+    },
+    {
+      "epoch": 0.7446834085805915,
+      "grad_norm": 1588.9076680875623,
+      "learning_rate": 3.106407734427037e-06,
+      "loss": 233.4497,
+      "step": 38720
+    },
+    {
+      "epoch": 0.744875733841072,
+      "grad_norm": 1546.1459562253833,
+      "learning_rate": 3.101987782033966e-06,
+      "loss": 234.1466,
+      "step": 38730
+    },
+    {
+      "epoch": 0.7450680591015525,
+      "grad_norm": 1536.6085840720723,
+      "learning_rate": 3.097570399036519e-06,
+      "loss": 232.2633,
+      "step": 38740
+    },
+    {
+      "epoch": 0.7452603843620331,
+      "grad_norm": 1477.7490614590422,
+      "learning_rate": 3.0931555870801033e-06,
+      "loss": 224.3291,
+      "step": 38750
+    },
+    {
+      "epoch": 0.7454527096225136,
+      "grad_norm": 1354.5010656531415,
+      "learning_rate": 3.0887433478091587e-06,
+      "loss": 228.2884,
+      "step": 38760
+    },
+    {
+      "epoch": 0.7456450348829942,
+      "grad_norm": 1465.9569984578495,
+      "learning_rate": 3.0843336828671765e-06,
+      "loss": 229.9976,
+      "step": 38770
+    },
+    {
+      "epoch": 0.7458373601434747,
+      "grad_norm": 1622.1769824796472,
+      "learning_rate": 3.079926593896683e-06,
+      "loss": 229.3872,
+      "step": 38780
+    },
+    {
+      "epoch": 0.7460296854039552,
+      "grad_norm": 1532.9826618181792,
+      "learning_rate": 3.0755220825392397e-06,
+      "loss": 226.9077,
+      "step": 38790
+    },
+    {
+      "epoch": 0.7462220106644357,
+      "grad_norm": 1627.3066234340752,
+      "learning_rate": 3.0711201504354628e-06,
+      "loss": 230.8235,
+      "step": 38800
+    },
+    {
+      "epoch": 0.7464143359249162,
+      "grad_norm": 1375.838055736071,
+      "learning_rate": 3.0667207992249948e-06,
+      "loss": 233.8211,
+      "step": 38810
+    },
+    {
+      "epoch": 0.7466066611853968,
+      "grad_norm": 1417.7960527685434,
+      "learning_rate": 3.062324030546523e-06,
+      "loss": 232.4826,
+      "step": 38820
+    },
+    {
+      "epoch": 0.7467989864458773,
+      "grad_norm": 1503.0315132252572,
+      "learning_rate": 3.057929846037767e-06,
+      "loss": 227.4745,
+      "step": 38830
+    },
+    {
+      "epoch": 0.7469913117063578,
+      "grad_norm": 1918.148383930101,
+      "learning_rate": 3.0535382473354945e-06,
+      "loss": 232.5834,
+      "step": 38840
+    },
+    {
+      "epoch": 0.7471836369668383,
+      "grad_norm": 1541.8218319319228,
+      "learning_rate": 3.0491492360755003e-06,
+      "loss": 232.8663,
+      "step": 38850
+    },
+    {
+      "epoch": 0.7473759622273188,
+      "grad_norm": 1619.755062191748,
+      "learning_rate": 3.0447628138926153e-06,
+      "loss": 223.3882,
+      "step": 38860
+    },
+    {
+      "epoch": 0.7475682874877994,
+      "grad_norm": 1552.9096540805451,
+      "learning_rate": 3.0403789824207165e-06,
+      "loss": 224.4841,
+      "step": 38870
+    },
+    {
+      "epoch": 0.7477606127482799,
+      "grad_norm": 1606.2692061301036,
+      "learning_rate": 3.0359977432927013e-06,
+      "loss": 228.7691,
+      "step": 38880
+    },
+    {
+      "epoch": 0.7479529380087604,
+      "grad_norm": 1493.6444073551415,
+      "learning_rate": 3.0316190981405147e-06,
+      "loss": 228.6588,
+      "step": 38890
+    },
+    {
+      "epoch": 0.7481452632692409,
+      "grad_norm": 1621.4724822994094,
+      "learning_rate": 3.0272430485951244e-06,
+      "loss": 226.9325,
+      "step": 38900
+    },
+    {
+      "epoch": 0.7483375885297214,
+      "grad_norm": 1529.2698704758027,
+      "learning_rate": 3.0228695962865438e-06,
+      "loss": 232.5227,
+      "step": 38910
+    },
+    {
+      "epoch": 0.748529913790202,
+      "grad_norm": 1611.6355212872631,
+      "learning_rate": 3.018498742843806e-06,
+      "loss": 233.6747,
+      "step": 38920
+    },
+    {
+      "epoch": 0.7487222390506825,
+      "grad_norm": 1534.9337549557347,
+      "learning_rate": 3.014130489894982e-06,
+      "loss": 230.3002,
+      "step": 38930
+    },
+    {
+      "epoch": 0.748914564311163,
+      "grad_norm": 1485.581125702763,
+      "learning_rate": 3.0097648390671765e-06,
+      "loss": 223.6374,
+      "step": 38940
+    },
+    {
+      "epoch": 0.7491068895716435,
+      "grad_norm": 1618.7800635046221,
+      "learning_rate": 3.005401791986522e-06,
+      "loss": 236.3089,
+      "step": 38950
+    },
+    {
+      "epoch": 0.7492992148321241,
+      "grad_norm": 1600.9724572405305,
+      "learning_rate": 3.00104135027818e-06,
+      "loss": 235.8808,
+      "step": 38960
+    },
+    {
+      "epoch": 0.7494915400926047,
+      "grad_norm": 1632.6882586007207,
+      "learning_rate": 2.99668351556634e-06,
+      "loss": 233.8207,
+      "step": 38970
+    },
+    {
+      "epoch": 0.7496838653530852,
+      "grad_norm": 1742.7987924364313,
+      "learning_rate": 2.99232828947423e-06,
+      "loss": 234.3095,
+      "step": 38980
+    },
+    {
+      "epoch": 0.7498761906135657,
+      "grad_norm": 1560.8204002034142,
+      "learning_rate": 2.987975673624096e-06,
+      "loss": 226.3867,
+      "step": 38990
+    },
+    {
+      "epoch": 0.7500685158740462,
+      "grad_norm": 1559.7950189364128,
+      "learning_rate": 2.9836256696372178e-06,
+      "loss": 236.0132,
+      "step": 39000
+    },
+    {
+      "epoch": 0.7502608411345267,
+      "grad_norm": 1461.7826027381584,
+      "learning_rate": 2.9792782791338936e-06,
+      "loss": 232.3441,
+      "step": 39010
+    },
+    {
+      "epoch": 0.7504531663950073,
+      "grad_norm": 1540.1667159474748,
+      "learning_rate": 2.9749335037334604e-06,
+      "loss": 233.2916,
+      "step": 39020
+    },
+    {
+      "epoch": 0.7506454916554878,
+      "grad_norm": 1493.4633613370054,
+      "learning_rate": 2.9705913450542777e-06,
+      "loss": 225.4259,
+      "step": 39030
+    },
+    {
+      "epoch": 0.7508378169159683,
+      "grad_norm": 1681.9987349814191,
+      "learning_rate": 2.9662518047137214e-06,
+      "loss": 227.9373,
+      "step": 39040
+    },
+    {
+      "epoch": 0.7510301421764488,
+      "grad_norm": 1551.8812699037337,
+      "learning_rate": 2.961914884328203e-06,
+      "loss": 232.0307,
+      "step": 39050
+    },
+    {
+      "epoch": 0.7512224674369293,
+      "grad_norm": 1577.1057744004038,
+      "learning_rate": 2.9575805855131546e-06,
+      "loss": 230.7319,
+      "step": 39060
+    },
+    {
+      "epoch": 0.7514147926974099,
+      "grad_norm": 1387.5241059839784,
+      "learning_rate": 2.9532489098830274e-06,
+      "loss": 222.7067,
+      "step": 39070
+    },
+    {
+      "epoch": 0.7516071179578904,
+      "grad_norm": 1618.4106613229749,
+      "learning_rate": 2.9489198590512967e-06,
+      "loss": 229.0409,
+      "step": 39080
+    },
+    {
+      "epoch": 0.7517994432183709,
+      "grad_norm": 1545.6290078684374,
+      "learning_rate": 2.9445934346304706e-06,
+      "loss": 231.7582,
+      "step": 39090
+    },
+    {
+      "epoch": 0.7519917684788514,
+      "grad_norm": 1498.1802867621911,
+      "learning_rate": 2.940269638232065e-06,
+      "loss": 229.0374,
+      "step": 39100
+    },
+    {
+      "epoch": 0.7521840937393319,
+      "grad_norm": 1484.7504843406407,
+      "learning_rate": 2.935948471466622e-06,
+      "loss": 227.3486,
+      "step": 39110
+    },
+    {
+      "epoch": 0.7523764189998124,
+      "grad_norm": 1406.5905899381798,
+      "learning_rate": 2.9316299359437085e-06,
+      "loss": 228.8529,
+      "step": 39120
+    },
+    {
+      "epoch": 0.752568744260293,
+      "grad_norm": 1468.5098236518395,
+      "learning_rate": 2.9273140332719064e-06,
+      "loss": 227.1665,
+      "step": 39130
+    },
+    {
+      "epoch": 0.7527610695207735,
+      "grad_norm": 1480.4347493309658,
+      "learning_rate": 2.923000765058818e-06,
+      "loss": 229.7315,
+      "step": 39140
+    },
+    {
+      "epoch": 0.752953394781254,
+      "grad_norm": 1459.8983937371859,
+      "learning_rate": 2.9186901329110605e-06,
+      "loss": 223.861,
+      "step": 39150
+    },
+    {
+      "epoch": 0.7531457200417346,
+      "grad_norm": 1630.1424199653043,
+      "learning_rate": 2.9143821384342808e-06,
+      "loss": 226.2624,
+      "step": 39160
+    },
+    {
+      "epoch": 0.7533380453022152,
+      "grad_norm": 1560.6245436710506,
+      "learning_rate": 2.9100767832331277e-06,
+      "loss": 227.4073,
+      "step": 39170
+    },
+    {
+      "epoch": 0.7535303705626957,
+      "grad_norm": 1501.8897167206867,
+      "learning_rate": 2.9057740689112822e-06,
+      "loss": 235.3969,
+      "step": 39180
+    },
+    {
+      "epoch": 0.7537226958231762,
+      "grad_norm": 1522.0067963644365,
+      "learning_rate": 2.901473997071428e-06,
+      "loss": 229.2319,
+      "step": 39190
+    },
+    {
+      "epoch": 0.7539150210836567,
+      "grad_norm": 1431.8775443390805,
+      "learning_rate": 2.8971765693152767e-06,
+      "loss": 228.2309,
+      "step": 39200
+    },
+    {
+      "epoch": 0.7541073463441372,
+      "grad_norm": 1341.6012616189569,
+      "learning_rate": 2.8928817872435465e-06,
+      "loss": 227.5759,
+      "step": 39210
+    },
+    {
+      "epoch": 0.7542996716046177,
+      "grad_norm": 1492.1599328719174,
+      "learning_rate": 2.8885896524559696e-06,
+      "loss": 232.1361,
+      "step": 39220
+    },
+    {
+      "epoch": 0.7544919968650983,
+      "grad_norm": 1633.6914536316206,
+      "learning_rate": 2.8843001665513016e-06,
+      "loss": 229.4284,
+      "step": 39230
+    },
+    {
+      "epoch": 0.7546843221255788,
+      "grad_norm": 1533.30097259094,
+      "learning_rate": 2.8800133311273016e-06,
+      "loss": 235.565,
+      "step": 39240
+    },
+    {
+      "epoch": 0.7548766473860593,
+      "grad_norm": 1737.6545061013549,
+      "learning_rate": 2.875729147780745e-06,
+      "loss": 233.1697,
+      "step": 39250
+    },
+    {
+      "epoch": 0.7550689726465398,
+      "grad_norm": 1712.2424769954955,
+      "learning_rate": 2.871447618107417e-06,
+      "loss": 229.7837,
+      "step": 39260
+    },
+    {
+      "epoch": 0.7552612979070203,
+      "grad_norm": 1560.7419996991025,
+      "learning_rate": 2.867168743702122e-06,
+      "loss": 224.7095,
+      "step": 39270
+    },
+    {
+      "epoch": 0.7554536231675009,
+      "grad_norm": 1702.4049797609039,
+      "learning_rate": 2.8628925261586683e-06,
+      "loss": 234.5142,
+      "step": 39280
+    },
+    {
+      "epoch": 0.7556459484279814,
+      "grad_norm": 2075.1287585968976,
+      "learning_rate": 2.8586189670698717e-06,
+      "loss": 231.4869,
+      "step": 39290
+    },
+    {
+      "epoch": 0.7558382736884619,
+      "grad_norm": 1428.5769655978036,
+      "learning_rate": 2.854348068027568e-06,
+      "loss": 227.7387,
+      "step": 39300
+    },
+    {
+      "epoch": 0.7560305989489424,
+      "grad_norm": 1553.079126384359,
+      "learning_rate": 2.850079830622593e-06,
+      "loss": 234.0741,
+      "step": 39310
+    },
+    {
+      "epoch": 0.7562229242094229,
+      "grad_norm": 1556.8255438387275,
+      "learning_rate": 2.845814256444799e-06,
+      "loss": 239.6045,
+      "step": 39320
+    },
+    {
+      "epoch": 0.7564152494699035,
+      "grad_norm": 1432.1019826302088,
+      "learning_rate": 2.8415513470830357e-06,
+      "loss": 236.7049,
+      "step": 39330
+    },
+    {
+      "epoch": 0.756607574730384,
+      "grad_norm": 1556.9228249657856,
+      "learning_rate": 2.837291104125174e-06,
+      "loss": 234.3567,
+      "step": 39340
+    },
+    {
+      "epoch": 0.7567998999908645,
+      "grad_norm": 1507.964162233076,
+      "learning_rate": 2.833033529158079e-06,
+      "loss": 226.55,
+      "step": 39350
+    },
+    {
+      "epoch": 0.7569922252513451,
+      "grad_norm": 1578.3682066255808,
+      "learning_rate": 2.8287786237676253e-06,
+      "loss": 238.0503,
+      "step": 39360
+    },
+    {
+      "epoch": 0.7571845505118256,
+      "grad_norm": 1609.993364757536,
+      "learning_rate": 2.824526389538701e-06,
+      "loss": 236.4321,
+      "step": 39370
+    },
+    {
+      "epoch": 0.7573768757723062,
+      "grad_norm": 1565.811264480137,
+      "learning_rate": 2.8202768280551894e-06,
+      "loss": 227.1677,
+      "step": 39380
+    },
+    {
+      "epoch": 0.7575692010327867,
+      "grad_norm": 1522.7060655441767,
+      "learning_rate": 2.8160299408999827e-06,
+      "loss": 238.6643,
+      "step": 39390
+    },
+    {
+      "epoch": 0.7577615262932672,
+      "grad_norm": 1525.0299837914583,
+      "learning_rate": 2.811785729654972e-06,
+      "loss": 232.3769,
+      "step": 39400
+    },
+    {
+      "epoch": 0.7579538515537477,
+      "grad_norm": 1830.8695918774251,
+      "learning_rate": 2.8075441959010628e-06,
+      "loss": 227.8009,
+      "step": 39410
+    },
+    {
+      "epoch": 0.7581461768142282,
+      "grad_norm": 1425.616184175756,
+      "learning_rate": 2.8033053412181543e-06,
+      "loss": 227.7011,
+      "step": 39420
+    },
+    {
+      "epoch": 0.7583385020747088,
+      "grad_norm": 1318.8439040850305,
+      "learning_rate": 2.799069167185148e-06,
+      "loss": 217.5075,
+      "step": 39430
+    },
+    {
+      "epoch": 0.7585308273351893,
+      "grad_norm": 1429.719327200252,
+      "learning_rate": 2.7948356753799466e-06,
+      "loss": 238.4496,
+      "step": 39440
+    },
+    {
+      "epoch": 0.7587231525956698,
+      "grad_norm": 1469.4435178137517,
+      "learning_rate": 2.7906048673794593e-06,
+      "loss": 227.6901,
+      "step": 39450
+    },
+    {
+      "epoch": 0.7589154778561503,
+      "grad_norm": 1493.9387265662892,
+      "learning_rate": 2.7863767447595946e-06,
+      "loss": 233.9194,
+      "step": 39460
+    },
+    {
+      "epoch": 0.7591078031166308,
+      "grad_norm": 1373.719482542066,
+      "learning_rate": 2.7821513090952523e-06,
+      "loss": 226.7143,
+      "step": 39470
+    },
+    {
+      "epoch": 0.7593001283771114,
+      "grad_norm": 1488.3352043271668,
+      "learning_rate": 2.7779285619603446e-06,
+      "loss": 229.5452,
+      "step": 39480
+    },
+    {
+      "epoch": 0.7594924536375919,
+      "grad_norm": 1457.1325948435813,
+      "learning_rate": 2.77370850492777e-06,
+      "loss": 241.8746,
+      "step": 39490
+    },
+    {
+      "epoch": 0.7596847788980724,
+      "grad_norm": 1524.312368437958,
+      "learning_rate": 2.7694911395694324e-06,
+      "loss": 230.2175,
+      "step": 39500
+    },
+    {
+      "epoch": 0.7598771041585529,
+      "grad_norm": 1497.1561124764387,
+      "learning_rate": 2.765276467456225e-06,
+      "loss": 233.6009,
+      "step": 39510
+    },
+    {
+      "epoch": 0.7600694294190334,
+      "grad_norm": 1840.2446602403843,
+      "learning_rate": 2.761064490158052e-06,
+      "loss": 231.5688,
+      "step": 39520
+    },
+    {
+      "epoch": 0.760261754679514,
+      "grad_norm": 1493.0030869830234,
+      "learning_rate": 2.7568552092438018e-06,
+      "loss": 229.5692,
+      "step": 39530
+    },
+    {
+      "epoch": 0.7604540799399945,
+      "grad_norm": 1458.218326324335,
+      "learning_rate": 2.7526486262813578e-06,
+      "loss": 227.9972,
+      "step": 39540
+    },
+    {
+      "epoch": 0.760646405200475,
+      "grad_norm": 1519.3327051377792,
+      "learning_rate": 2.7484447428376094e-06,
+      "loss": 223.1645,
+      "step": 39550
+    },
+    {
+      "epoch": 0.7608387304609555,
+      "grad_norm": 1544.6972057422195,
+      "learning_rate": 2.7442435604784313e-06,
+      "loss": 231.1909,
+      "step": 39560
+    },
+    {
+      "epoch": 0.7610310557214361,
+      "grad_norm": 1734.2695560302896,
+      "learning_rate": 2.740045080768694e-06,
+      "loss": 230.289,
+      "step": 39570
+    },
+    {
+      "epoch": 0.7612233809819167,
+      "grad_norm": 1345.318831638213,
+      "learning_rate": 2.7358493052722603e-06,
+      "loss": 227.446,
+      "step": 39580
+    },
+    {
+      "epoch": 0.7614157062423972,
+      "grad_norm": 1768.76847183003,
+      "learning_rate": 2.7316562355519904e-06,
+      "loss": 228.2119,
+      "step": 39590
+    },
+    {
+      "epoch": 0.7616080315028777,
+      "grad_norm": 1615.1567394760389,
+      "learning_rate": 2.727465873169729e-06,
+      "loss": 230.1732,
+      "step": 39600
+    },
+    {
+      "epoch": 0.7618003567633582,
+      "grad_norm": 1475.6891761370462,
+      "learning_rate": 2.723278219686324e-06,
+      "loss": 230.1224,
+      "step": 39610
+    },
+    {
+      "epoch": 0.7619926820238387,
+      "grad_norm": 1623.8386817986877,
+      "learning_rate": 2.7190932766615998e-06,
+      "loss": 230.9577,
+      "step": 39620
+    },
+    {
+      "epoch": 0.7621850072843193,
+      "grad_norm": 1456.9462614488823,
+      "learning_rate": 2.714911045654385e-06,
+      "loss": 224.0631,
+      "step": 39630
+    },
+    {
+      "epoch": 0.7623773325447998,
+      "grad_norm": 1709.329324322735,
+      "learning_rate": 2.7107315282224878e-06,
+      "loss": 226.8969,
+      "step": 39640
+    },
+    {
+      "epoch": 0.7625696578052803,
+      "grad_norm": 1550.3853116982405,
+      "learning_rate": 2.7065547259227078e-06,
+      "loss": 231.5462,
+      "step": 39650
+    },
+    {
+      "epoch": 0.7627619830657608,
+      "grad_norm": 1538.7643760089115,
+      "learning_rate": 2.7023806403108397e-06,
+      "loss": 233.9908,
+      "step": 39660
+    },
+    {
+      "epoch": 0.7629543083262413,
+      "grad_norm": 1502.0380791874352,
+      "learning_rate": 2.698209272941659e-06,
+      "loss": 223.9186,
+      "step": 39670
+    },
+    {
+      "epoch": 0.7631466335867219,
+      "grad_norm": 1770.0076136099817,
+      "learning_rate": 2.694040625368931e-06,
+      "loss": 240.11,
+      "step": 39680
+    },
+    {
+      "epoch": 0.7633389588472024,
+      "grad_norm": 1476.1021015682165,
+      "learning_rate": 2.689874699145405e-06,
+      "loss": 226.428,
+      "step": 39690
+    },
+    {
+      "epoch": 0.7635312841076829,
+      "grad_norm": 1479.3706646795874,
+      "learning_rate": 2.685711495822827e-06,
+      "loss": 226.8066,
+      "step": 39700
+    },
+    {
+      "epoch": 0.7637236093681634,
+      "grad_norm": 1485.3689278382758,
+      "learning_rate": 2.6815510169519164e-06,
+      "loss": 225.0812,
+      "step": 39710
+    },
+    {
+      "epoch": 0.7639159346286439,
+      "grad_norm": 1465.0505655110665,
+      "learning_rate": 2.677393264082381e-06,
+      "loss": 239.3931,
+      "step": 39720
+    },
+    {
+      "epoch": 0.7641082598891245,
+      "grad_norm": 1356.5680751083773,
+      "learning_rate": 2.673238238762921e-06,
+      "loss": 233.6438,
+      "step": 39730
+    },
+    {
+      "epoch": 0.764300585149605,
+      "grad_norm": 1532.2535572660252,
+      "learning_rate": 2.6690859425412075e-06,
+      "loss": 229.7484,
+      "step": 39740
+    },
+    {
+      "epoch": 0.7644929104100855,
+      "grad_norm": 1826.663160506281,
+      "learning_rate": 2.6649363769639103e-06,
+      "loss": 240.9609,
+      "step": 39750
+    },
+    {
+      "epoch": 0.764685235670566,
+      "grad_norm": 1619.1353528456075,
+      "learning_rate": 2.660789543576667e-06,
+      "loss": 230.513,
+      "step": 39760
+    },
+    {
+      "epoch": 0.7648775609310466,
+      "grad_norm": 1451.953162359556,
+      "learning_rate": 2.6566454439241107e-06,
+      "loss": 232.0023,
+      "step": 39770
+    },
+    {
+      "epoch": 0.7650698861915272,
+      "grad_norm": 1608.5655613210492,
+      "learning_rate": 2.652504079549848e-06,
+      "loss": 236.1966,
+      "step": 39780
+    },
+    {
+      "epoch": 0.7652622114520077,
+      "grad_norm": 1425.7565820048073,
+      "learning_rate": 2.648365451996466e-06,
+      "loss": 231.6341,
+      "step": 39790
+    },
+    {
+      "epoch": 0.7654545367124882,
+      "grad_norm": 1359.7013947533796,
+      "learning_rate": 2.6442295628055346e-06,
+      "loss": 222.9614,
+      "step": 39800
+    },
+    {
+      "epoch": 0.7656468619729687,
+      "grad_norm": 1571.4898704025986,
+      "learning_rate": 2.64009641351761e-06,
+      "loss": 226.5332,
+      "step": 39810
+    },
+    {
+      "epoch": 0.7658391872334492,
+      "grad_norm": 1475.7123169859588,
+      "learning_rate": 2.635966005672218e-06,
+      "loss": 231.0349,
+      "step": 39820
+    },
+    {
+      "epoch": 0.7660315124939298,
+      "grad_norm": 1564.6585313295045,
+      "learning_rate": 2.631838340807865e-06,
+      "loss": 232.6851,
+      "step": 39830
+    },
+    {
+      "epoch": 0.7662238377544103,
+      "grad_norm": 1456.3971795270434,
+      "learning_rate": 2.6277134204620436e-06,
+      "loss": 229.4299,
+      "step": 39840
+    },
+    {
+      "epoch": 0.7664161630148908,
+      "grad_norm": 1654.790642375442,
+      "learning_rate": 2.6235912461712167e-06,
+      "loss": 237.5063,
+      "step": 39850
+    },
+    {
+      "epoch": 0.7666084882753713,
+      "grad_norm": 1622.1438050190582,
+      "learning_rate": 2.619471819470821e-06,
+      "loss": 227.2596,
+      "step": 39860
+    },
+    {
+      "epoch": 0.7668008135358518,
+      "grad_norm": 1445.822420320407,
+      "learning_rate": 2.6153551418952827e-06,
+      "loss": 223.3597,
+      "step": 39870
+    },
+    {
+      "epoch": 0.7669931387963324,
+      "grad_norm": 1468.4502517005567,
+      "learning_rate": 2.6112412149779888e-06,
+      "loss": 231.4466,
+      "step": 39880
+    },
+    {
+      "epoch": 0.7671854640568129,
+      "grad_norm": 1624.240968786334,
+      "learning_rate": 2.6071300402513165e-06,
+      "loss": 232.0156,
+      "step": 39890
+    },
+    {
+      "epoch": 0.7673777893172934,
+      "grad_norm": 1592.8445640755456,
+      "learning_rate": 2.603021619246604e-06,
+      "loss": 232.5667,
+      "step": 39900
+    },
+    {
+      "epoch": 0.7675701145777739,
+      "grad_norm": 1364.0891676198787,
+      "learning_rate": 2.5989159534941768e-06,
+      "loss": 228.7056,
+      "step": 39910
+    },
+    {
+      "epoch": 0.7677624398382544,
+      "grad_norm": 1362.1823548520736,
+      "learning_rate": 2.5948130445233232e-06,
+      "loss": 228.5754,
+      "step": 39920
+    },
+    {
+      "epoch": 0.767954765098735,
+      "grad_norm": 1540.7993054253168,
+      "learning_rate": 2.5907128938623093e-06,
+      "loss": 230.0194,
+      "step": 39930
+    },
+    {
+      "epoch": 0.7681470903592155,
+      "grad_norm": 1729.8661181939376,
+      "learning_rate": 2.5866155030383722e-06,
+      "loss": 236.0014,
+      "step": 39940
+    },
+    {
+      "epoch": 0.768339415619696,
+      "grad_norm": 1680.22327267366,
+      "learning_rate": 2.582520873577726e-06,
+      "loss": 224.4291,
+      "step": 39950
+    },
+    {
+      "epoch": 0.7685317408801765,
+      "grad_norm": 1355.6570509368657,
+      "learning_rate": 2.578429007005552e-06,
+      "loss": 227.8073,
+      "step": 39960
+    },
+    {
+      "epoch": 0.768724066140657,
+      "grad_norm": 1517.5289082127358,
+      "learning_rate": 2.5743399048460004e-06,
+      "loss": 233.4162,
+      "step": 39970
+    },
+    {
+      "epoch": 0.7689163914011377,
+      "grad_norm": 1529.6247759705575,
+      "learning_rate": 2.570253568622193e-06,
+      "loss": 230.2485,
+      "step": 39980
+    },
+    {
+      "epoch": 0.7691087166616182,
+      "grad_norm": 1521.9041550375655,
+      "learning_rate": 2.5661699998562286e-06,
+      "loss": 229.9751,
+      "step": 39990
+    },
+    {
+      "epoch": 0.7693010419220987,
+      "grad_norm": 1592.7690722009593,
+      "learning_rate": 2.5620892000691643e-06,
+      "loss": 239.7344,
+      "step": 40000
+    },
+    {
+      "epoch": 0.7694933671825792,
+      "grad_norm": 1536.1139050369718,
+      "learning_rate": 2.5580111707810296e-06,
+      "loss": 223.93,
+      "step": 40010
+    },
+    {
+      "epoch": 0.7696856924430597,
+      "grad_norm": 1701.5655218724514,
+      "learning_rate": 2.5539359135108244e-06,
+      "loss": 230.5768,
+      "step": 40020
+    },
+    {
+      "epoch": 0.7698780177035403,
+      "grad_norm": 1373.3462386369006,
+      "learning_rate": 2.549863429776519e-06,
+      "loss": 225.1608,
+      "step": 40030
+    },
+    {
+      "epoch": 0.7700703429640208,
+      "grad_norm": 1556.2910638255069,
+      "learning_rate": 2.5457937210950433e-06,
+      "loss": 229.5541,
+      "step": 40040
+    },
+    {
+      "epoch": 0.7702626682245013,
+      "grad_norm": 1579.768718214408,
+      "learning_rate": 2.541726788982294e-06,
+      "loss": 227.8932,
+      "step": 40050
+    },
+    {
+      "epoch": 0.7704549934849818,
+      "grad_norm": 1434.5027060967066,
+      "learning_rate": 2.5376626349531395e-06,
+      "loss": 229.1317,
+      "step": 40060
+    },
+    {
+      "epoch": 0.7706473187454623,
+      "grad_norm": 1517.7401641889667,
+      "learning_rate": 2.53360126052141e-06,
+      "loss": 233.8522,
+      "step": 40070
+    },
+    {
+      "epoch": 0.7708396440059428,
+      "grad_norm": 1594.181864715632,
+      "learning_rate": 2.529542667199896e-06,
+      "loss": 233.9767,
+      "step": 40080
+    },
+    {
+      "epoch": 0.7710319692664234,
+      "grad_norm": 1532.3078184249018,
+      "learning_rate": 2.525486856500363e-06,
+      "loss": 225.3485,
+      "step": 40090
+    },
+    {
+      "epoch": 0.7712242945269039,
+      "grad_norm": 1350.3336512430171,
+      "learning_rate": 2.5214338299335306e-06,
+      "loss": 229.533,
+      "step": 40100
+    },
+    {
+      "epoch": 0.7714166197873844,
+      "grad_norm": 1488.1248068203115,
+      "learning_rate": 2.5173835890090826e-06,
+      "loss": 232.382,
+      "step": 40110
+    },
+    {
+      "epoch": 0.7716089450478649,
+      "grad_norm": 1620.2971711654454,
+      "learning_rate": 2.5133361352356666e-06,
+      "loss": 228.0891,
+      "step": 40120
+    },
+    {
+      "epoch": 0.7718012703083454,
+      "grad_norm": 1413.8491979953592,
+      "learning_rate": 2.5092914701208958e-06,
+      "loss": 221.1879,
+      "step": 40130
+    },
+    {
+      "epoch": 0.771993595568826,
+      "grad_norm": 1386.9575119558153,
+      "learning_rate": 2.5052495951713406e-06,
+      "loss": 229.0098,
+      "step": 40140
+    },
+    {
+      "epoch": 0.7721859208293065,
+      "grad_norm": 1411.5234756516775,
+      "learning_rate": 2.5012105118925267e-06,
+      "loss": 226.465,
+      "step": 40150
+    },
+    {
+      "epoch": 0.772378246089787,
+      "grad_norm": 1566.5085338791948,
+      "learning_rate": 2.497174221788955e-06,
+      "loss": 226.8515,
+      "step": 40160
+    },
+    {
+      "epoch": 0.7725705713502675,
+      "grad_norm": 1554.7575223229185,
+      "learning_rate": 2.4931407263640683e-06,
+      "loss": 228.7184,
+      "step": 40170
+    },
+    {
+      "epoch": 0.7727628966107482,
+      "grad_norm": 1498.1724529363096,
+      "learning_rate": 2.489110027120285e-06,
+      "loss": 231.9207,
+      "step": 40180
+    },
+    {
+      "epoch": 0.7729552218712287,
+      "grad_norm": 1369.6770834399115,
+      "learning_rate": 2.4850821255589664e-06,
+      "loss": 226.0766,
+      "step": 40190
+    },
+    {
+      "epoch": 0.7731475471317092,
+      "grad_norm": 1381.2827277417787,
+      "learning_rate": 2.4810570231804463e-06,
+      "loss": 235.2267,
+      "step": 40200
+    },
+    {
+      "epoch": 0.7733398723921897,
+      "grad_norm": 1441.4432966750805,
+      "learning_rate": 2.4770347214840063e-06,
+      "loss": 225.7035,
+      "step": 40210
+    },
+    {
+      "epoch": 0.7735321976526702,
+      "grad_norm": 1443.7836043588495,
+      "learning_rate": 2.473015221967886e-06,
+      "loss": 225.2694,
+      "step": 40220
+    },
+    {
+      "epoch": 0.7737245229131507,
+      "grad_norm": 1808.5225238707985,
+      "learning_rate": 2.4689985261292805e-06,
+      "loss": 226.0416,
+      "step": 40230
+    },
+    {
+      "epoch": 0.7739168481736313,
+      "grad_norm": 1478.8505085391268,
+      "learning_rate": 2.464984635464348e-06,
+      "loss": 226.1488,
+      "step": 40240
+    },
+    {
+      "epoch": 0.7741091734341118,
+      "grad_norm": 1355.2658061098073,
+      "learning_rate": 2.460973551468194e-06,
+      "loss": 215.6184,
+      "step": 40250
+    },
+    {
+      "epoch": 0.7743014986945923,
+      "grad_norm": 1484.8008634830073,
+      "learning_rate": 2.456965275634878e-06,
+      "loss": 232.3315,
+      "step": 40260
+    },
+    {
+      "epoch": 0.7744938239550728,
+      "grad_norm": 1617.43936160178,
+      "learning_rate": 2.4529598094574226e-06,
+      "loss": 231.874,
+      "step": 40270
+    },
+    {
+      "epoch": 0.7746861492155533,
+      "grad_norm": 1566.1125054785784,
+      "learning_rate": 2.4489571544277944e-06,
+      "loss": 225.6771,
+      "step": 40280
+    },
+    {
+      "epoch": 0.7748784744760339,
+      "grad_norm": 1418.8194346181263,
+      "learning_rate": 2.444957312036914e-06,
+      "loss": 235.0988,
+      "step": 40290
+    },
+    {
+      "epoch": 0.7750707997365144,
+      "grad_norm": 1379.16405744577,
+      "learning_rate": 2.4409602837746625e-06,
+      "loss": 223.6478,
+      "step": 40300
+    },
+    {
+      "epoch": 0.7752631249969949,
+      "grad_norm": 1341.2557067983473,
+      "learning_rate": 2.4369660711298603e-06,
+      "loss": 224.6993,
+      "step": 40310
+    },
+    {
+      "epoch": 0.7754554502574754,
+      "grad_norm": 1531.9208625184538,
+      "learning_rate": 2.4329746755902917e-06,
+      "loss": 235.4249,
+      "step": 40320
+    },
+    {
+      "epoch": 0.7756477755179559,
+      "grad_norm": 1540.8569156599892,
+      "learning_rate": 2.428986098642684e-06,
+      "loss": 224.8458,
+      "step": 40330
+    },
+    {
+      "epoch": 0.7758401007784365,
+      "grad_norm": 1434.9122551682985,
+      "learning_rate": 2.425000341772711e-06,
+      "loss": 228.2546,
+      "step": 40340
+    },
+    {
+      "epoch": 0.776032426038917,
+      "grad_norm": 1533.5765462473894,
+      "learning_rate": 2.4210174064650084e-06,
+      "loss": 221.0433,
+      "step": 40350
+    },
+    {
+      "epoch": 0.7762247512993975,
+      "grad_norm": 1519.58865568952,
+      "learning_rate": 2.4170372942031506e-06,
+      "loss": 227.4079,
+      "step": 40360
+    },
+    {
+      "epoch": 0.776417076559878,
+      "grad_norm": 1603.639785063924,
+      "learning_rate": 2.4130600064696618e-06,
+      "loss": 227.0689,
+      "step": 40370
+    },
+    {
+      "epoch": 0.7766094018203585,
+      "grad_norm": 1572.8152371465953,
+      "learning_rate": 2.4090855447460205e-06,
+      "loss": 231.0838,
+      "step": 40380
+    },
+    {
+      "epoch": 0.7768017270808392,
+      "grad_norm": 1516.5582203496074,
+      "learning_rate": 2.4051139105126463e-06,
+      "loss": 228.535,
+      "step": 40390
+    },
+    {
+      "epoch": 0.7769940523413197,
+      "grad_norm": 1600.033725151574,
+      "learning_rate": 2.4011451052489064e-06,
+      "loss": 233.3731,
+      "step": 40400
+    },
+    {
+      "epoch": 0.7771863776018002,
+      "grad_norm": 1543.3682251301354,
+      "learning_rate": 2.3971791304331125e-06,
+      "loss": 223.3933,
+      "step": 40410
+    },
+    {
+      "epoch": 0.7773787028622807,
+      "grad_norm": 1401.8375212049893,
+      "learning_rate": 2.393215987542531e-06,
+      "loss": 227.7518,
+      "step": 40420
+    },
+    {
+      "epoch": 0.7775710281227612,
+      "grad_norm": 1542.0076337547246,
+      "learning_rate": 2.3892556780533606e-06,
+      "loss": 242.05,
+      "step": 40430
+    },
+    {
+      "epoch": 0.7777633533832418,
+      "grad_norm": 1670.9464137610357,
+      "learning_rate": 2.385298203440758e-06,
+      "loss": 226.5708,
+      "step": 40440
+    },
+    {
+      "epoch": 0.7779556786437223,
+      "grad_norm": 1379.6058889042256,
+      "learning_rate": 2.3813435651788107e-06,
+      "loss": 226.8058,
+      "step": 40450
+    },
+    {
+      "epoch": 0.7781480039042028,
+      "grad_norm": 1411.6719599942162,
+      "learning_rate": 2.377391764740562e-06,
+      "loss": 223.2353,
+      "step": 40460
+    },
+    {
+      "epoch": 0.7783403291646833,
+      "grad_norm": 1503.4557370807229,
+      "learning_rate": 2.3734428035979883e-06,
+      "loss": 220.4783,
+      "step": 40470
+    },
+    {
+      "epoch": 0.7785326544251638,
+      "grad_norm": 1357.084009449347,
+      "learning_rate": 2.3694966832220123e-06,
+      "loss": 226.8251,
+      "step": 40480
+    },
+    {
+      "epoch": 0.7787249796856444,
+      "grad_norm": 1489.3730747509462,
+      "learning_rate": 2.365553405082501e-06,
+      "loss": 230.1492,
+      "step": 40490
+    },
+    {
+      "epoch": 0.7789173049461249,
+      "grad_norm": 1338.5946968965673,
+      "learning_rate": 2.3616129706482604e-06,
+      "loss": 229.6824,
+      "step": 40500
+    },
+    {
+      "epoch": 0.7791096302066054,
+      "grad_norm": 1552.2410985614727,
+      "learning_rate": 2.357675381387036e-06,
+      "loss": 229.3827,
+      "step": 40510
+    },
+    {
+      "epoch": 0.7793019554670859,
+      "grad_norm": 1508.2537415895647,
+      "learning_rate": 2.3537406387655114e-06,
+      "loss": 224.9,
+      "step": 40520
+    },
+    {
+      "epoch": 0.7794942807275664,
+      "grad_norm": 1526.085897457577,
+      "learning_rate": 2.349808744249321e-06,
+      "loss": 227.9886,
+      "step": 40530
+    },
+    {
+      "epoch": 0.779686605988047,
+      "grad_norm": 1323.7807537402425,
+      "learning_rate": 2.345879699303025e-06,
+      "loss": 227.4332,
+      "step": 40540
+    },
+    {
+      "epoch": 0.7798789312485275,
+      "grad_norm": 1533.2274887748467,
+      "learning_rate": 2.3419535053901264e-06,
+      "loss": 226.85,
+      "step": 40550
+    },
+    {
+      "epoch": 0.780071256509008,
+      "grad_norm": 1892.5537707451906,
+      "learning_rate": 2.338030163973073e-06,
+      "loss": 241.7282,
+      "step": 40560
+    },
+    {
+      "epoch": 0.7802635817694885,
+      "grad_norm": 2098.48051973417,
+      "learning_rate": 2.334109676513242e-06,
+      "loss": 227.7498,
+      "step": 40570
+    },
+    {
+      "epoch": 0.780455907029969,
+      "grad_norm": 1402.0562953995811,
+      "learning_rate": 2.330192044470948e-06,
+      "loss": 224.9712,
+      "step": 40580
+    },
+    {
+      "epoch": 0.7806482322904497,
+      "grad_norm": 1523.3168543907718,
+      "learning_rate": 2.3262772693054457e-06,
+      "loss": 234.5564,
+      "step": 40590
+    },
+    {
+      "epoch": 0.7808405575509302,
+      "grad_norm": 1526.1610008678756,
+      "learning_rate": 2.322365352474928e-06,
+      "loss": 226.7791,
+      "step": 40600
+    },
+    {
+      "epoch": 0.7810328828114107,
+      "grad_norm": 1627.713212806367,
+      "learning_rate": 2.3184562954365153e-06,
+      "loss": 233.5697,
+      "step": 40610
+    },
+    {
+      "epoch": 0.7812252080718912,
+      "grad_norm": 1536.8872791126494,
+      "learning_rate": 2.3145500996462656e-06,
+      "loss": 233.8028,
+      "step": 40620
+    },
+    {
+      "epoch": 0.7814175333323717,
+      "grad_norm": 1544.0005262375464,
+      "learning_rate": 2.310646766559177e-06,
+      "loss": 236.3372,
+      "step": 40630
+    },
+    {
+      "epoch": 0.7816098585928523,
+      "grad_norm": 1355.3349946567957,
+      "learning_rate": 2.3067462976291744e-06,
+      "loss": 239.5528,
+      "step": 40640
+    },
+    {
+      "epoch": 0.7818021838533328,
+      "grad_norm": 1393.7196373154275,
+      "learning_rate": 2.302848694309118e-06,
+      "loss": 232.6877,
+      "step": 40650
+    },
+    {
+      "epoch": 0.7819945091138133,
+      "grad_norm": 1521.018926923358,
+      "learning_rate": 2.2989539580507957e-06,
+      "loss": 229.9074,
+      "step": 40660
+    },
+    {
+      "epoch": 0.7821868343742938,
+      "grad_norm": 1393.9483678876109,
+      "learning_rate": 2.2950620903049414e-06,
+      "loss": 231.9409,
+      "step": 40670
+    },
+    {
+      "epoch": 0.7823791596347743,
+      "grad_norm": 1449.210680831741,
+      "learning_rate": 2.2911730925212073e-06,
+      "loss": 231.3504,
+      "step": 40680
+    },
+    {
+      "epoch": 0.7825714848952549,
+      "grad_norm": 1644.9880550694936,
+      "learning_rate": 2.2872869661481766e-06,
+      "loss": 222.984,
+      "step": 40690
+    },
+    {
+      "epoch": 0.7827638101557354,
+      "grad_norm": 1471.3705734924645,
+      "learning_rate": 2.283403712633375e-06,
+      "loss": 223.9134,
+      "step": 40700
+    },
+    {
+      "epoch": 0.7829561354162159,
+      "grad_norm": 1535.5573364192828,
+      "learning_rate": 2.279523333423247e-06,
+      "loss": 235.2469,
+      "step": 40710
+    },
+    {
+      "epoch": 0.7831484606766964,
+      "grad_norm": 1510.987186257855,
+      "learning_rate": 2.2756458299631667e-06,
+      "loss": 220.6051,
+      "step": 40720
+    },
+    {
+      "epoch": 0.7833407859371769,
+      "grad_norm": 1562.9839076985177,
+      "learning_rate": 2.271771203697445e-06,
+      "loss": 225.6615,
+      "step": 40730
+    },
+    {
+      "epoch": 0.7835331111976575,
+      "grad_norm": 1642.8036388046785,
+      "learning_rate": 2.267899456069311e-06,
+      "loss": 229.096,
+      "step": 40740
+    },
+    {
+      "epoch": 0.783725436458138,
+      "grad_norm": 1514.0328472822666,
+      "learning_rate": 2.2640305885209336e-06,
+      "loss": 233.7914,
+      "step": 40750
+    },
+    {
+      "epoch": 0.7839177617186185,
+      "grad_norm": 1863.5990877946347,
+      "learning_rate": 2.2601646024933976e-06,
+      "loss": 235.72,
+      "step": 40760
+    },
+    {
+      "epoch": 0.784110086979099,
+      "grad_norm": 1425.6558101053422,
+      "learning_rate": 2.256301499426716e-06,
+      "loss": 226.4633,
+      "step": 40770
+    },
+    {
+      "epoch": 0.7843024122395795,
+      "grad_norm": 1472.4617980259707,
+      "learning_rate": 2.252441280759838e-06,
+      "loss": 226.7071,
+      "step": 40780
+    },
+    {
+      "epoch": 0.78449473750006,
+      "grad_norm": 1528.3722409226293,
+      "learning_rate": 2.248583947930628e-06,
+      "loss": 221.2189,
+      "step": 40790
+    },
+    {
+      "epoch": 0.7846870627605407,
+      "grad_norm": 1374.7888939546492,
+      "learning_rate": 2.2447295023758755e-06,
+      "loss": 225.7388,
+      "step": 40800
+    },
+    {
+      "epoch": 0.7848793880210212,
+      "grad_norm": 1620.8582955241884,
+      "learning_rate": 2.2408779455313035e-06,
+      "loss": 228.7956,
+      "step": 40810
+    },
+    {
+      "epoch": 0.7850717132815017,
+      "grad_norm": 1490.526283775315,
+      "learning_rate": 2.2370292788315505e-06,
+      "loss": 229.4782,
+      "step": 40820
+    },
+    {
+      "epoch": 0.7852640385419822,
+      "grad_norm": 1460.5091473901052,
+      "learning_rate": 2.2331835037101825e-06,
+      "loss": 220.2695,
+      "step": 40830
+    },
+    {
+      "epoch": 0.7854563638024628,
+      "grad_norm": 1465.1085778814581,
+      "learning_rate": 2.2293406215996814e-06,
+      "loss": 226.2115,
+      "step": 40840
+    },
+    {
+      "epoch": 0.7856486890629433,
+      "grad_norm": 1412.0950540608337,
+      "learning_rate": 2.2255006339314667e-06,
+      "loss": 227.4971,
+      "step": 40850
+    },
+    {
+      "epoch": 0.7858410143234238,
+      "grad_norm": 1447.6390826221968,
+      "learning_rate": 2.2216635421358623e-06,
+      "loss": 222.4478,
+      "step": 40860
+    },
+    {
+      "epoch": 0.7860333395839043,
+      "grad_norm": 1461.774153868207,
+      "learning_rate": 2.2178293476421276e-06,
+      "loss": 223.5899,
+      "step": 40870
+    },
+    {
+      "epoch": 0.7862256648443848,
+      "grad_norm": 1848.7914057486973,
+      "learning_rate": 2.213998051878431e-06,
+      "loss": 224.4984,
+      "step": 40880
+    },
+    {
+      "epoch": 0.7864179901048653,
+      "grad_norm": 1575.747483253397,
+      "learning_rate": 2.2101696562718735e-06,
+      "loss": 220.7876,
+      "step": 40890
+    },
+    {
+      "epoch": 0.7866103153653459,
+      "grad_norm": 1645.0734697340743,
+      "learning_rate": 2.206344162248466e-06,
+      "loss": 234.4414,
+      "step": 40900
+    },
+    {
+      "epoch": 0.7868026406258264,
+      "grad_norm": 1441.4290721070868,
+      "learning_rate": 2.2025215712331383e-06,
+      "loss": 227.5106,
+      "step": 40910
+    },
+    {
+      "epoch": 0.7869949658863069,
+      "grad_norm": 1643.0677555218856,
+      "learning_rate": 2.1987018846497487e-06,
+      "loss": 223.9698,
+      "step": 40920
+    },
+    {
+      "epoch": 0.7871872911467874,
+      "grad_norm": 1439.6237334511682,
+      "learning_rate": 2.194885103921064e-06,
+      "loss": 228.259,
+      "step": 40930
+    },
+    {
+      "epoch": 0.787379616407268,
+      "grad_norm": 1467.4516329863982,
+      "learning_rate": 2.191071230468772e-06,
+      "loss": 230.5067,
+      "step": 40940
+    },
+    {
+      "epoch": 0.7875719416677485,
+      "grad_norm": 1759.973294282919,
+      "learning_rate": 2.1872602657134757e-06,
+      "loss": 231.1099,
+      "step": 40950
+    },
+    {
+      "epoch": 0.787764266928229,
+      "grad_norm": 1485.6441736705547,
+      "learning_rate": 2.1834522110747014e-06,
+      "loss": 227.9294,
+      "step": 40960
+    },
+    {
+      "epoch": 0.7879565921887095,
+      "grad_norm": 1450.771930508663,
+      "learning_rate": 2.179647067970885e-06,
+      "loss": 225.9228,
+      "step": 40970
+    },
+    {
+      "epoch": 0.78814891744919,
+      "grad_norm": 1638.076217006254,
+      "learning_rate": 2.1758448378193743e-06,
+      "loss": 232.0011,
+      "step": 40980
+    },
+    {
+      "epoch": 0.7883412427096705,
+      "grad_norm": 1523.1982805396844,
+      "learning_rate": 2.1720455220364443e-06,
+      "loss": 225.8148,
+      "step": 40990
+    },
+    {
+      "epoch": 0.7885335679701512,
+      "grad_norm": 1652.9155537154872,
+      "learning_rate": 2.168249122037275e-06,
+      "loss": 223.9595,
+      "step": 41000
+    },
+    {
+      "epoch": 0.7887258932306317,
+      "grad_norm": 1451.996061054277,
+      "learning_rate": 2.1644556392359583e-06,
+      "loss": 225.9462,
+      "step": 41010
+    },
+    {
+      "epoch": 0.7889182184911122,
+      "grad_norm": 1420.226376785837,
+      "learning_rate": 2.160665075045508e-06,
+      "loss": 217.4813,
+      "step": 41020
+    },
+    {
+      "epoch": 0.7891105437515927,
+      "grad_norm": 1481.5532658731017,
+      "learning_rate": 2.1568774308778494e-06,
+      "loss": 229.8827,
+      "step": 41030
+    },
+    {
+      "epoch": 0.7893028690120732,
+      "grad_norm": 1565.4603306776348,
+      "learning_rate": 2.1530927081438148e-06,
+      "loss": 239.1232,
+      "step": 41040
+    },
+    {
+      "epoch": 0.7894951942725538,
+      "grad_norm": 1486.631631394588,
+      "learning_rate": 2.1493109082531473e-06,
+      "loss": 217.7836,
+      "step": 41050
+    },
+    {
+      "epoch": 0.7896875195330343,
+      "grad_norm": 1412.8829691386582,
+      "learning_rate": 2.1455320326145103e-06,
+      "loss": 221.3021,
+      "step": 41060
+    },
+    {
+      "epoch": 0.7898798447935148,
+      "grad_norm": 1387.1226392915962,
+      "learning_rate": 2.141756082635471e-06,
+      "loss": 224.7436,
+      "step": 41070
+    },
+    {
+      "epoch": 0.7900721700539953,
+      "grad_norm": 1453.4667795473927,
+      "learning_rate": 2.137983059722507e-06,
+      "loss": 218.7273,
+      "step": 41080
+    },
+    {
+      "epoch": 0.7902644953144758,
+      "grad_norm": 1596.264832146378,
+      "learning_rate": 2.1342129652810063e-06,
+      "loss": 225.6246,
+      "step": 41090
+    },
+    {
+      "epoch": 0.7904568205749564,
+      "grad_norm": 1658.0471103267905,
+      "learning_rate": 2.1304458007152694e-06,
+      "loss": 222.6364,
+      "step": 41100
+    },
+    {
+      "epoch": 0.7906491458354369,
+      "grad_norm": 1593.536466513159,
+      "learning_rate": 2.1266815674285026e-06,
+      "loss": 234.3179,
+      "step": 41110
+    },
+    {
+      "epoch": 0.7908414710959174,
+      "grad_norm": 1875.4041549400936,
+      "learning_rate": 2.1229202668228197e-06,
+      "loss": 237.6631,
+      "step": 41120
+    },
+    {
+      "epoch": 0.7910337963563979,
+      "grad_norm": 1421.251489425682,
+      "learning_rate": 2.1191619002992405e-06,
+      "loss": 218.0321,
+      "step": 41130
+    },
+    {
+      "epoch": 0.7912261216168784,
+      "grad_norm": 1855.742910843341,
+      "learning_rate": 2.1154064692577e-06,
+      "loss": 225.4099,
+      "step": 41140
+    },
+    {
+      "epoch": 0.791418446877359,
+      "grad_norm": 1457.0500531867021,
+      "learning_rate": 2.111653975097029e-06,
+      "loss": 223.2658,
+      "step": 41150
+    },
+    {
+      "epoch": 0.7916107721378395,
+      "grad_norm": 1588.1258031117427,
+      "learning_rate": 2.1079044192149713e-06,
+      "loss": 226.5197,
+      "step": 41160
+    },
+    {
+      "epoch": 0.79180309739832,
+      "grad_norm": 1478.0282946570778,
+      "learning_rate": 2.1041578030081777e-06,
+      "loss": 225.7318,
+      "step": 41170
+    },
+    {
+      "epoch": 0.7919954226588005,
+      "grad_norm": 1437.370431691596,
+      "learning_rate": 2.100414127872198e-06,
+      "loss": 221.009,
+      "step": 41180
+    },
+    {
+      "epoch": 0.792187747919281,
+      "grad_norm": 1615.315421894461,
+      "learning_rate": 2.0966733952014904e-06,
+      "loss": 233.6552,
+      "step": 41190
+    },
+    {
+      "epoch": 0.7923800731797617,
+      "grad_norm": 1333.4924681749596,
+      "learning_rate": 2.0929356063894125e-06,
+      "loss": 225.8451,
+      "step": 41200
+    },
+    {
+      "epoch": 0.7925723984402422,
+      "grad_norm": 1664.8868194383929,
+      "learning_rate": 2.089200762828234e-06,
+      "loss": 227.6047,
+      "step": 41210
+    },
+    {
+      "epoch": 0.7927647237007227,
+      "grad_norm": 1423.8189205683134,
+      "learning_rate": 2.0854688659091203e-06,
+      "loss": 228.7122,
+      "step": 41220
+    },
+    {
+      "epoch": 0.7929570489612032,
+      "grad_norm": 1473.3704616277328,
+      "learning_rate": 2.08173991702214e-06,
+      "loss": 218.8115,
+      "step": 41230
+    },
+    {
+      "epoch": 0.7931493742216837,
+      "grad_norm": 1610.98361990669,
+      "learning_rate": 2.0780139175562675e-06,
+      "loss": 228.3723,
+      "step": 41240
+    },
+    {
+      "epoch": 0.7933416994821643,
+      "grad_norm": 1597.2059162794196,
+      "learning_rate": 2.0742908688993746e-06,
+      "loss": 222.0698,
+      "step": 41250
+    },
+    {
+      "epoch": 0.7935340247426448,
+      "grad_norm": 1464.8980865615158,
+      "learning_rate": 2.070570772438236e-06,
+      "loss": 220.8634,
+      "step": 41260
+    },
+    {
+      "epoch": 0.7937263500031253,
+      "grad_norm": 1585.9093915142003,
+      "learning_rate": 2.066853629558524e-06,
+      "loss": 228.2862,
+      "step": 41270
+    },
+    {
+      "epoch": 0.7939186752636058,
+      "grad_norm": 1400.350859921243,
+      "learning_rate": 2.0631394416448157e-06,
+      "loss": 217.2972,
+      "step": 41280
+    },
+    {
+      "epoch": 0.7941110005240863,
+      "grad_norm": 1381.219744142666,
+      "learning_rate": 2.059428210080583e-06,
+      "loss": 219.7499,
+      "step": 41290
+    },
+    {
+      "epoch": 0.7943033257845669,
+      "grad_norm": 1631.2974577290759,
+      "learning_rate": 2.0557199362482005e-06,
+      "loss": 228.6004,
+      "step": 41300
+    },
+    {
+      "epoch": 0.7944956510450474,
+      "grad_norm": 1523.135496361386,
+      "learning_rate": 2.052014621528935e-06,
+      "loss": 225.3294,
+      "step": 41310
+    },
+    {
+      "epoch": 0.7946879763055279,
+      "grad_norm": 1443.709965686497,
+      "learning_rate": 2.048312267302961e-06,
+      "loss": 223.9784,
+      "step": 41320
+    },
+    {
+      "epoch": 0.7948803015660084,
+      "grad_norm": 1299.031510551774,
+      "learning_rate": 2.044612874949341e-06,
+      "loss": 221.0394,
+      "step": 41330
+    },
+    {
+      "epoch": 0.7950726268264889,
+      "grad_norm": 1524.2265805064194,
+      "learning_rate": 2.040916445846034e-06,
+      "loss": 232.1858,
+      "step": 41340
+    },
+    {
+      "epoch": 0.7952649520869695,
+      "grad_norm": 1603.2730902821102,
+      "learning_rate": 2.037222981369905e-06,
+      "loss": 222.1017,
+      "step": 41350
+    },
+    {
+      "epoch": 0.79545727734745,
+      "grad_norm": 1587.3299425360408,
+      "learning_rate": 2.033532482896707e-06,
+      "loss": 223.766,
+      "step": 41360
+    },
+    {
+      "epoch": 0.7956496026079305,
+      "grad_norm": 1795.0011121604073,
+      "learning_rate": 2.0298449518010875e-06,
+      "loss": 224.8888,
+      "step": 41370
+    },
+    {
+      "epoch": 0.795841927868411,
+      "grad_norm": 1633.8645199689108,
+      "learning_rate": 2.0261603894565897e-06,
+      "loss": 224.6307,
+      "step": 41380
+    },
+    {
+      "epoch": 0.7960342531288915,
+      "grad_norm": 1460.093598687193,
+      "learning_rate": 2.0224787972356574e-06,
+      "loss": 228.3823,
+      "step": 41390
+    },
+    {
+      "epoch": 0.796226578389372,
+      "grad_norm": 1368.5075483467213,
+      "learning_rate": 2.0188001765096198e-06,
+      "loss": 229.9522,
+      "step": 41400
+    },
+    {
+      "epoch": 0.7964189036498527,
+      "grad_norm": 1474.3032963938929,
+      "learning_rate": 2.0151245286486998e-06,
+      "loss": 225.8562,
+      "step": 41410
+    },
+    {
+      "epoch": 0.7966112289103332,
+      "grad_norm": 1457.0164609325093,
+      "learning_rate": 2.011451855022021e-06,
+      "loss": 229.4697,
+      "step": 41420
+    },
+    {
+      "epoch": 0.7968035541708137,
+      "grad_norm": 1436.3632642150014,
+      "learning_rate": 2.0077821569975885e-06,
+      "loss": 225.8121,
+      "step": 41430
+    },
+    {
+      "epoch": 0.7969958794312942,
+      "grad_norm": 1513.7343517483034,
+      "learning_rate": 2.0041154359423087e-06,
+      "loss": 232.8264,
+      "step": 41440
+    },
+    {
+      "epoch": 0.7971882046917748,
+      "grad_norm": 1610.768583100523,
+      "learning_rate": 2.000451693221971e-06,
+      "loss": 232.3558,
+      "step": 41450
+    },
+    {
+      "epoch": 0.7973805299522553,
+      "grad_norm": 1437.5389649037047,
+      "learning_rate": 1.9967909302012635e-06,
+      "loss": 223.8995,
+      "step": 41460
+    },
+    {
+      "epoch": 0.7975728552127358,
+      "grad_norm": 1443.5420866599811,
+      "learning_rate": 1.9931331482437553e-06,
+      "loss": 226.6861,
+      "step": 41470
+    },
+    {
+      "epoch": 0.7977651804732163,
+      "grad_norm": 1414.388069525892,
+      "learning_rate": 1.989478348711913e-06,
+      "loss": 223.2291,
+      "step": 41480
+    },
+    {
+      "epoch": 0.7979575057336968,
+      "grad_norm": 1537.3167023340775,
+      "learning_rate": 1.9858265329670844e-06,
+      "loss": 223.2962,
+      "step": 41490
+    },
+    {
+      "epoch": 0.7981498309941774,
+      "grad_norm": 1459.426850248904,
+      "learning_rate": 1.9821777023695178e-06,
+      "loss": 224.2818,
+      "step": 41500
+    },
+    {
+      "epoch": 0.7983421562546579,
+      "grad_norm": 1378.8985638760314,
+      "learning_rate": 1.9785318582783375e-06,
+      "loss": 223.5442,
+      "step": 41510
+    },
+    {
+      "epoch": 0.7985344815151384,
+      "grad_norm": 1365.9414691439122,
+      "learning_rate": 1.9748890020515577e-06,
+      "loss": 226.3647,
+      "step": 41520
+    },
+    {
+      "epoch": 0.7987268067756189,
+      "grad_norm": 1406.3202830468524,
+      "learning_rate": 1.9712491350460895e-06,
+      "loss": 220.562,
+      "step": 41530
+    },
+    {
+      "epoch": 0.7989191320360994,
+      "grad_norm": 1353.166381529778,
+      "learning_rate": 1.967612258617718e-06,
+      "loss": 221.5289,
+      "step": 41540
+    },
+    {
+      "epoch": 0.79911145729658,
+      "grad_norm": 1626.867499781207,
+      "learning_rate": 1.9639783741211218e-06,
+      "loss": 226.8763,
+      "step": 41550
+    },
+    {
+      "epoch": 0.7993037825570605,
+      "grad_norm": 1480.278065129493,
+      "learning_rate": 1.960347482909859e-06,
+      "loss": 226.7404,
+      "step": 41560
+    },
+    {
+      "epoch": 0.799496107817541,
+      "grad_norm": 1572.5062905441623,
+      "learning_rate": 1.956719586336382e-06,
+      "loss": 230.2421,
+      "step": 41570
+    },
+    {
+      "epoch": 0.7996884330780215,
+      "grad_norm": 1474.517313489486,
+      "learning_rate": 1.953094685752017e-06,
+      "loss": 226.2306,
+      "step": 41580
+    },
+    {
+      "epoch": 0.799880758338502,
+      "grad_norm": 1400.2555283665315,
+      "learning_rate": 1.949472782506984e-06,
+      "loss": 228.999,
+      "step": 41590
+    },
+    {
+      "epoch": 0.8000730835989825,
+      "grad_norm": 1734.7104279328498,
+      "learning_rate": 1.945853877950382e-06,
+      "loss": 236.8713,
+      "step": 41600
+    },
+    {
+      "epoch": 0.8002654088594632,
+      "grad_norm": 1705.9170945908565,
+      "learning_rate": 1.942237973430192e-06,
+      "loss": 231.3811,
+      "step": 41610
+    },
+    {
+      "epoch": 0.8004577341199437,
+      "grad_norm": 1619.689816450686,
+      "learning_rate": 1.9386250702932784e-06,
+      "loss": 228.5134,
+      "step": 41620
+    },
+    {
+      "epoch": 0.8006500593804242,
+      "grad_norm": 1423.624254683205,
+      "learning_rate": 1.9350151698853857e-06,
+      "loss": 228.8275,
+      "step": 41630
+    },
+    {
+      "epoch": 0.8008423846409047,
+      "grad_norm": 1381.19426376922,
+      "learning_rate": 1.9314082735511475e-06,
+      "loss": 225.8395,
+      "step": 41640
+    },
+    {
+      "epoch": 0.8010347099013853,
+      "grad_norm": 1518.1720519766825,
+      "learning_rate": 1.92780438263407e-06,
+      "loss": 221.9989,
+      "step": 41650
+    },
+    {
+      "epoch": 0.8012270351618658,
+      "grad_norm": 1322.3398325351816,
+      "learning_rate": 1.9242034984765436e-06,
+      "loss": 226.5761,
+      "step": 41660
+    },
+    {
+      "epoch": 0.8014193604223463,
+      "grad_norm": 1410.095477432377,
+      "learning_rate": 1.9206056224198346e-06,
+      "loss": 220.5587,
+      "step": 41670
+    },
+    {
+      "epoch": 0.8016116856828268,
+      "grad_norm": 1541.3107475864845,
+      "learning_rate": 1.9170107558040983e-06,
+      "loss": 233.8152,
+      "step": 41680
+    },
+    {
+      "epoch": 0.8018040109433073,
+      "grad_norm": 1450.8107893911642,
+      "learning_rate": 1.9134188999683613e-06,
+      "loss": 219.7217,
+      "step": 41690
+    },
+    {
+      "epoch": 0.8019963362037879,
+      "grad_norm": 1512.6253332264887,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 224.997,
+      "step": 41700
+    },
+    {
+      "epoch": 0.8021886614642684,
+      "grad_norm": 1432.0950359231251,
+      "learning_rate": 1.9062442259873847e-06,
+      "loss": 228.6781,
+      "step": 41710
+    },
+    {
+      "epoch": 0.8023809867247489,
+      "grad_norm": 1612.6722181101088,
+      "learning_rate": 1.9026614105145935e-06,
+      "loss": 233.8999,
+      "step": 41720
+    },
+    {
+      "epoch": 0.8025733119852294,
+      "grad_norm": 1468.310800618677,
+      "learning_rate": 1.8990816111666976e-06,
+      "loss": 225.3823,
+      "step": 41730
+    },
+    {
+      "epoch": 0.8027656372457099,
+      "grad_norm": 1398.6367937041161,
+      "learning_rate": 1.8955048292771083e-06,
+      "loss": 222.4204,
+      "step": 41740
+    },
+    {
+      "epoch": 0.8029579625061904,
+      "grad_norm": 1540.886255575622,
+      "learning_rate": 1.891931066178122e-06,
+      "loss": 217.259,
+      "step": 41750
+    },
+    {
+      "epoch": 0.803150287766671,
+      "grad_norm": 1517.518704529034,
+      "learning_rate": 1.888360323200904e-06,
+      "loss": 216.4562,
+      "step": 41760
+    },
+    {
+      "epoch": 0.8033426130271515,
+      "grad_norm": 2003.9853512557702,
+      "learning_rate": 1.8847926016754947e-06,
+      "loss": 223.3393,
+      "step": 41770
+    },
+    {
+      "epoch": 0.803534938287632,
+      "grad_norm": 1663.3804137696186,
+      "learning_rate": 1.8812279029308177e-06,
+      "loss": 223.0596,
+      "step": 41780
+    },
+    {
+      "epoch": 0.8037272635481125,
+      "grad_norm": 1407.2459427677816,
+      "learning_rate": 1.87766622829466e-06,
+      "loss": 226.0139,
+      "step": 41790
+    },
+    {
+      "epoch": 0.803919588808593,
+      "grad_norm": 1568.0152663864164,
+      "learning_rate": 1.874107579093688e-06,
+      "loss": 236.8645,
+      "step": 41800
+    },
+    {
+      "epoch": 0.8041119140690736,
+      "grad_norm": 1468.331395695032,
+      "learning_rate": 1.870551956653437e-06,
+      "loss": 230.0562,
+      "step": 41810
+    },
+    {
+      "epoch": 0.8043042393295542,
+      "grad_norm": 1485.1079855934788,
+      "learning_rate": 1.8669993622983217e-06,
+      "loss": 230.1099,
+      "step": 41820
+    },
+    {
+      "epoch": 0.8044965645900347,
+      "grad_norm": 1353.9816137716862,
+      "learning_rate": 1.863449797351624e-06,
+      "loss": 224.6421,
+      "step": 41830
+    },
+    {
+      "epoch": 0.8046888898505152,
+      "grad_norm": 1438.6240927843733,
+      "learning_rate": 1.8599032631354963e-06,
+      "loss": 227.7561,
+      "step": 41840
+    },
+    {
+      "epoch": 0.8048812151109958,
+      "grad_norm": 1842.5025861498002,
+      "learning_rate": 1.8563597609709626e-06,
+      "loss": 234.0146,
+      "step": 41850
+    },
+    {
+      "epoch": 0.8050735403714763,
+      "grad_norm": 1443.658980293271,
+      "learning_rate": 1.852819292177922e-06,
+      "loss": 227.0709,
+      "step": 41860
+    },
+    {
+      "epoch": 0.8052658656319568,
+      "grad_norm": 1760.242506004807,
+      "learning_rate": 1.8492818580751414e-06,
+      "loss": 236.386,
+      "step": 41870
+    },
+    {
+      "epoch": 0.8054581908924373,
+      "grad_norm": 1478.5638971645387,
+      "learning_rate": 1.8457474599802527e-06,
+      "loss": 220.4513,
+      "step": 41880
+    },
+    {
+      "epoch": 0.8056505161529178,
+      "grad_norm": 1570.9939045590684,
+      "learning_rate": 1.842216099209767e-06,
+      "loss": 230.289,
+      "step": 41890
+    },
+    {
+      "epoch": 0.8058428414133983,
+      "grad_norm": 1355.8225959178471,
+      "learning_rate": 1.8386877770790524e-06,
+      "loss": 223.0868,
+      "step": 41900
+    },
+    {
+      "epoch": 0.8060351666738789,
+      "grad_norm": 1718.523060136808,
+      "learning_rate": 1.8351624949023539e-06,
+      "loss": 232.3824,
+      "step": 41910
+    },
+    {
+      "epoch": 0.8062274919343594,
+      "grad_norm": 1451.5035113607748,
+      "learning_rate": 1.8316402539927757e-06,
+      "loss": 227.675,
+      "step": 41920
+    },
+    {
+      "epoch": 0.8064198171948399,
+      "grad_norm": 1538.859997449665,
+      "learning_rate": 1.8281210556623007e-06,
+      "loss": 220.1817,
+      "step": 41930
+    },
+    {
+      "epoch": 0.8066121424553204,
+      "grad_norm": 1464.86440166237,
+      "learning_rate": 1.8246049012217693e-06,
+      "loss": 221.5458,
+      "step": 41940
+    },
+    {
+      "epoch": 0.8068044677158009,
+      "grad_norm": 1512.595131305657,
+      "learning_rate": 1.8210917919808891e-06,
+      "loss": 230.8413,
+      "step": 41950
+    },
+    {
+      "epoch": 0.8069967929762815,
+      "grad_norm": 1482.5182904132826,
+      "learning_rate": 1.817581729248239e-06,
+      "loss": 225.2418,
+      "step": 41960
+    },
+    {
+      "epoch": 0.807189118236762,
+      "grad_norm": 1725.5288490985367,
+      "learning_rate": 1.8140747143312588e-06,
+      "loss": 224.3757,
+      "step": 41970
+    },
+    {
+      "epoch": 0.8073814434972425,
+      "grad_norm": 1571.8246060725655,
+      "learning_rate": 1.8105707485362511e-06,
+      "loss": 217.8115,
+      "step": 41980
+    },
+    {
+      "epoch": 0.807573768757723,
+      "grad_norm": 1374.5560541387367,
+      "learning_rate": 1.8070698331683844e-06,
+      "loss": 225.0413,
+      "step": 41990
+    },
+    {
+      "epoch": 0.8077660940182035,
+      "grad_norm": 1430.9684476880302,
+      "learning_rate": 1.8035719695316955e-06,
+      "loss": 233.4534,
+      "step": 42000
+    },
+    {
+      "epoch": 0.8079584192786841,
+      "grad_norm": 1565.0280665823886,
+      "learning_rate": 1.800077158929081e-06,
+      "loss": 225.7904,
+      "step": 42010
+    },
+    {
+      "epoch": 0.8081507445391647,
+      "grad_norm": 1473.0317998942007,
+      "learning_rate": 1.7965854026622953e-06,
+      "loss": 222.2656,
+      "step": 42020
+    },
+    {
+      "epoch": 0.8083430697996452,
+      "grad_norm": 1615.1995173455418,
+      "learning_rate": 1.7930967020319667e-06,
+      "loss": 228.0199,
+      "step": 42030
+    },
+    {
+      "epoch": 0.8085353950601257,
+      "grad_norm": 1425.7883677433506,
+      "learning_rate": 1.7896110583375747e-06,
+      "loss": 216.8364,
+      "step": 42040
+    },
+    {
+      "epoch": 0.8087277203206062,
+      "grad_norm": 1339.312100319773,
+      "learning_rate": 1.7861284728774652e-06,
+      "loss": 224.1174,
+      "step": 42050
+    },
+    {
+      "epoch": 0.8089200455810868,
+      "grad_norm": 1388.5394112114393,
+      "learning_rate": 1.7826489469488395e-06,
+      "loss": 221.5267,
+      "step": 42060
+    },
+    {
+      "epoch": 0.8091123708415673,
+      "grad_norm": 1595.4362923310557,
+      "learning_rate": 1.7791724818477708e-06,
+      "loss": 222.603,
+      "step": 42070
+    },
+    {
+      "epoch": 0.8093046961020478,
+      "grad_norm": 1597.9080246972799,
+      "learning_rate": 1.7756990788691797e-06,
+      "loss": 220.4012,
+      "step": 42080
+    },
+    {
+      "epoch": 0.8094970213625283,
+      "grad_norm": 1392.3267254806283,
+      "learning_rate": 1.772228739306854e-06,
+      "loss": 224.5479,
+      "step": 42090
+    },
+    {
+      "epoch": 0.8096893466230088,
+      "grad_norm": 1491.138464024459,
+      "learning_rate": 1.7687614644534333e-06,
+      "loss": 221.3646,
+      "step": 42100
+    },
+    {
+      "epoch": 0.8098816718834894,
+      "grad_norm": 1676.4568754630063,
+      "learning_rate": 1.7652972556004267e-06,
+      "loss": 230.1084,
+      "step": 42110
+    },
+    {
+      "epoch": 0.8100739971439699,
+      "grad_norm": 1533.1927379000833,
+      "learning_rate": 1.7618361140381922e-06,
+      "loss": 234.9796,
+      "step": 42120
+    },
+    {
+      "epoch": 0.8102663224044504,
+      "grad_norm": 1694.2491070808348,
+      "learning_rate": 1.7583780410559449e-06,
+      "loss": 226.4873,
+      "step": 42130
+    },
+    {
+      "epoch": 0.8104586476649309,
+      "grad_norm": 1452.8672302265786,
+      "learning_rate": 1.7549230379417636e-06,
+      "loss": 226.541,
+      "step": 42140
+    },
+    {
+      "epoch": 0.8106509729254114,
+      "grad_norm": 1499.1505710043314,
+      "learning_rate": 1.7514711059825773e-06,
+      "loss": 220.3394,
+      "step": 42150
+    },
+    {
+      "epoch": 0.810843298185892,
+      "grad_norm": 1533.378060626979,
+      "learning_rate": 1.7480222464641783e-06,
+      "loss": 236.7391,
+      "step": 42160
+    },
+    {
+      "epoch": 0.8110356234463725,
+      "grad_norm": 1482.6842334290056,
+      "learning_rate": 1.7445764606712024e-06,
+      "loss": 221.3659,
+      "step": 42170
+    },
+    {
+      "epoch": 0.811227948706853,
+      "grad_norm": 1559.034946778375,
+      "learning_rate": 1.7411337498871561e-06,
+      "loss": 222.6689,
+      "step": 42180
+    },
+    {
+      "epoch": 0.8114202739673335,
+      "grad_norm": 1637.9939160450422,
+      "learning_rate": 1.737694115394387e-06,
+      "loss": 228.9204,
+      "step": 42190
+    },
+    {
+      "epoch": 0.811612599227814,
+      "grad_norm": 1509.8790838070415,
+      "learning_rate": 1.7342575584741018e-06,
+      "loss": 223.104,
+      "step": 42200
+    },
+    {
+      "epoch": 0.8118049244882946,
+      "grad_norm": 1635.9237807301313,
+      "learning_rate": 1.7308240804063648e-06,
+      "loss": 223.4572,
+      "step": 42210
+    },
+    {
+      "epoch": 0.8119972497487751,
+      "grad_norm": 1379.2380085877755,
+      "learning_rate": 1.7273936824700888e-06,
+      "loss": 230.1087,
+      "step": 42220
+    },
+    {
+      "epoch": 0.8121895750092557,
+      "grad_norm": 1305.2883850681742,
+      "learning_rate": 1.7239663659430384e-06,
+      "loss": 225.0166,
+      "step": 42230
+    },
+    {
+      "epoch": 0.8123819002697362,
+      "grad_norm": 1556.707919190861,
+      "learning_rate": 1.7205421321018312e-06,
+      "loss": 230.2891,
+      "step": 42240
+    },
+    {
+      "epoch": 0.8125742255302167,
+      "grad_norm": 1424.9442666222474,
+      "learning_rate": 1.7171209822219427e-06,
+      "loss": 232.6741,
+      "step": 42250
+    },
+    {
+      "epoch": 0.8127665507906973,
+      "grad_norm": 1508.260282414269,
+      "learning_rate": 1.713702917577692e-06,
+      "loss": 227.0938,
+      "step": 42260
+    },
+    {
+      "epoch": 0.8129588760511778,
+      "grad_norm": 1428.3730558534426,
+      "learning_rate": 1.71028793944225e-06,
+      "loss": 223.3841,
+      "step": 42270
+    },
+    {
+      "epoch": 0.8131512013116583,
+      "grad_norm": 1429.5273679361626,
+      "learning_rate": 1.7068760490876425e-06,
+      "loss": 223.1546,
+      "step": 42280
+    },
+    {
+      "epoch": 0.8133435265721388,
+      "grad_norm": 1482.4694123804115,
+      "learning_rate": 1.7034672477847402e-06,
+      "loss": 225.6837,
+      "step": 42290
+    },
+    {
+      "epoch": 0.8135358518326193,
+      "grad_norm": 1380.4252499273334,
+      "learning_rate": 1.700061536803268e-06,
+      "loss": 221.5166,
+      "step": 42300
+    },
+    {
+      "epoch": 0.8137281770930999,
+      "grad_norm": 1901.3136030811497,
+      "learning_rate": 1.696658917411793e-06,
+      "loss": 222.6794,
+      "step": 42310
+    },
+    {
+      "epoch": 0.8139205023535804,
+      "grad_norm": 1365.8335175244717,
+      "learning_rate": 1.6932593908777394e-06,
+      "loss": 227.1507,
+      "step": 42320
+    },
+    {
+      "epoch": 0.8141128276140609,
+      "grad_norm": 1375.7556180125393,
+      "learning_rate": 1.689862958467372e-06,
+      "loss": 221.4263,
+      "step": 42330
+    },
+    {
+      "epoch": 0.8143051528745414,
+      "grad_norm": 1626.2539559401287,
+      "learning_rate": 1.6864696214458065e-06,
+      "loss": 225.4705,
+      "step": 42340
+    },
+    {
+      "epoch": 0.8144974781350219,
+      "grad_norm": 1434.6336067280931,
+      "learning_rate": 1.683079381077003e-06,
+      "loss": 229.2176,
+      "step": 42350
+    },
+    {
+      "epoch": 0.8146898033955025,
+      "grad_norm": 1724.9467242078645,
+      "learning_rate": 1.6796922386237724e-06,
+      "loss": 231.7839,
+      "step": 42360
+    },
+    {
+      "epoch": 0.814882128655983,
+      "grad_norm": 1568.501378833988,
+      "learning_rate": 1.67630819534777e-06,
+      "loss": 226.6076,
+      "step": 42370
+    },
+    {
+      "epoch": 0.8150744539164635,
+      "grad_norm": 1670.451033621797,
+      "learning_rate": 1.6729272525094908e-06,
+      "loss": 226.68,
+      "step": 42380
+    },
+    {
+      "epoch": 0.815266779176944,
+      "grad_norm": 1510.4298470406395,
+      "learning_rate": 1.6695494113682874e-06,
+      "loss": 228.1382,
+      "step": 42390
+    },
+    {
+      "epoch": 0.8154591044374245,
+      "grad_norm": 1544.1046478193641,
+      "learning_rate": 1.6661746731823458e-06,
+      "loss": 221.8056,
+      "step": 42400
+    },
+    {
+      "epoch": 0.815651429697905,
+      "grad_norm": 1409.7897076184765,
+      "learning_rate": 1.6628030392087001e-06,
+      "loss": 227.2639,
+      "step": 42410
+    },
+    {
+      "epoch": 0.8158437549583856,
+      "grad_norm": 1441.3276676767996,
+      "learning_rate": 1.6594345107032273e-06,
+      "loss": 230.0636,
+      "step": 42420
+    },
+    {
+      "epoch": 0.8160360802188662,
+      "grad_norm": 1467.6132957286763,
+      "learning_rate": 1.6560690889206499e-06,
+      "loss": 229.1264,
+      "step": 42430
+    },
+    {
+      "epoch": 0.8162284054793467,
+      "grad_norm": 1403.3290535279677,
+      "learning_rate": 1.6527067751145354e-06,
+      "loss": 218.5342,
+      "step": 42440
+    },
+    {
+      "epoch": 0.8164207307398272,
+      "grad_norm": 1554.803857859893,
+      "learning_rate": 1.6493475705372863e-06,
+      "loss": 224.7972,
+      "step": 42450
+    },
+    {
+      "epoch": 0.8166130560003078,
+      "grad_norm": 1568.6354465635306,
+      "learning_rate": 1.6459914764401497e-06,
+      "loss": 230.3151,
+      "step": 42460
+    },
+    {
+      "epoch": 0.8168053812607883,
+      "grad_norm": 1645.2967112572787,
+      "learning_rate": 1.64263849407322e-06,
+      "loss": 225.4138,
+      "step": 42470
+    },
+    {
+      "epoch": 0.8169977065212688,
+      "grad_norm": 1417.7049701690194,
+      "learning_rate": 1.6392886246854234e-06,
+      "loss": 213.743,
+      "step": 42480
+    },
+    {
+      "epoch": 0.8171900317817493,
+      "grad_norm": 1582.4693283726072,
+      "learning_rate": 1.6359418695245311e-06,
+      "loss": 221.4301,
+      "step": 42490
+    },
+    {
+      "epoch": 0.8173823570422298,
+      "grad_norm": 1482.2087313161098,
+      "learning_rate": 1.632598229837158e-06,
+      "loss": 225.2374,
+      "step": 42500
+    },
+    {
+      "epoch": 0.8175746823027104,
+      "grad_norm": 1442.3268427688743,
+      "learning_rate": 1.629257706868751e-06,
+      "loss": 223.0,
+      "step": 42510
+    },
+    {
+      "epoch": 0.8177670075631909,
+      "grad_norm": 1407.8046978948455,
+      "learning_rate": 1.6259203018636016e-06,
+      "loss": 222.0701,
+      "step": 42520
+    },
+    {
+      "epoch": 0.8179593328236714,
+      "grad_norm": 1520.653430167389,
+      "learning_rate": 1.6225860160648343e-06,
+      "loss": 231.2354,
+      "step": 42530
+    },
+    {
+      "epoch": 0.8181516580841519,
+      "grad_norm": 1449.6154322147395,
+      "learning_rate": 1.6192548507144213e-06,
+      "loss": 227.8629,
+      "step": 42540
+    },
+    {
+      "epoch": 0.8183439833446324,
+      "grad_norm": 1433.257553441374,
+      "learning_rate": 1.6159268070531642e-06,
+      "loss": 219.5006,
+      "step": 42550
+    },
+    {
+      "epoch": 0.818536308605113,
+      "grad_norm": 1654.1740475795948,
+      "learning_rate": 1.6126018863207005e-06,
+      "loss": 232.3285,
+      "step": 42560
+    },
+    {
+      "epoch": 0.8187286338655935,
+      "grad_norm": 1476.4215860476145,
+      "learning_rate": 1.609280089755515e-06,
+      "loss": 226.2384,
+      "step": 42570
+    },
+    {
+      "epoch": 0.818920959126074,
+      "grad_norm": 1614.6135528003551,
+      "learning_rate": 1.6059614185949157e-06,
+      "loss": 226.9501,
+      "step": 42580
+    },
+    {
+      "epoch": 0.8191132843865545,
+      "grad_norm": 1451.2994149600986,
+      "learning_rate": 1.6026458740750584e-06,
+      "loss": 214.5616,
+      "step": 42590
+    },
+    {
+      "epoch": 0.819305609647035,
+      "grad_norm": 1410.1708830708812,
+      "learning_rate": 1.5993334574309238e-06,
+      "loss": 228.6564,
+      "step": 42600
+    },
+    {
+      "epoch": 0.8194979349075155,
+      "grad_norm": 1427.8235152665707,
+      "learning_rate": 1.5960241698963374e-06,
+      "loss": 218.9107,
+      "step": 42610
+    },
+    {
+      "epoch": 0.8196902601679961,
+      "grad_norm": 1438.13525418863,
+      "learning_rate": 1.592718012703951e-06,
+      "loss": 223.5326,
+      "step": 42620
+    },
+    {
+      "epoch": 0.8198825854284766,
+      "grad_norm": 1436.9970690690038,
+      "learning_rate": 1.589414987085255e-06,
+      "loss": 216.0942,
+      "step": 42630
+    },
+    {
+      "epoch": 0.8200749106889572,
+      "grad_norm": 1440.1952059084329,
+      "learning_rate": 1.5861150942705672e-06,
+      "loss": 223.2584,
+      "step": 42640
+    },
+    {
+      "epoch": 0.8202672359494377,
+      "grad_norm": 1310.5762976050596,
+      "learning_rate": 1.5828183354890504e-06,
+      "loss": 216.1386,
+      "step": 42650
+    },
+    {
+      "epoch": 0.8204595612099183,
+      "grad_norm": 1314.2962857885525,
+      "learning_rate": 1.5795247119686885e-06,
+      "loss": 217.2636,
+      "step": 42660
+    },
+    {
+      "epoch": 0.8206518864703988,
+      "grad_norm": 1648.8155214634844,
+      "learning_rate": 1.5762342249363006e-06,
+      "loss": 229.1493,
+      "step": 42670
+    },
+    {
+      "epoch": 0.8208442117308793,
+      "grad_norm": 1380.3027020197358,
+      "learning_rate": 1.5729468756175426e-06,
+      "loss": 215.5619,
+      "step": 42680
+    },
+    {
+      "epoch": 0.8210365369913598,
+      "grad_norm": 1497.7760674089052,
+      "learning_rate": 1.5696626652368973e-06,
+      "loss": 220.6415,
+      "step": 42690
+    },
+    {
+      "epoch": 0.8212288622518403,
+      "grad_norm": 1531.6529123032683,
+      "learning_rate": 1.5663815950176742e-06,
+      "loss": 225.0178,
+      "step": 42700
+    },
+    {
+      "epoch": 0.8214211875123208,
+      "grad_norm": 1496.7894727302607,
+      "learning_rate": 1.5631036661820232e-06,
+      "loss": 229.5158,
+      "step": 42710
+    },
+    {
+      "epoch": 0.8216135127728014,
+      "grad_norm": 1310.4907574438366,
+      "learning_rate": 1.5598288799509153e-06,
+      "loss": 220.1782,
+      "step": 42720
+    },
+    {
+      "epoch": 0.8218058380332819,
+      "grad_norm": 1465.0493818727255,
+      "learning_rate": 1.5565572375441573e-06,
+      "loss": 222.9248,
+      "step": 42730
+    },
+    {
+      "epoch": 0.8219981632937624,
+      "grad_norm": 1699.6861573890478,
+      "learning_rate": 1.5532887401803787e-06,
+      "loss": 225.4941,
+      "step": 42740
+    },
+    {
+      "epoch": 0.8221904885542429,
+      "grad_norm": 1474.509262915218,
+      "learning_rate": 1.5500233890770434e-06,
+      "loss": 223.9475,
+      "step": 42750
+    },
+    {
+      "epoch": 0.8223828138147234,
+      "grad_norm": 1479.9237445482188,
+      "learning_rate": 1.5467611854504406e-06,
+      "loss": 218.3563,
+      "step": 42760
+    },
+    {
+      "epoch": 0.822575139075204,
+      "grad_norm": 1478.6754869003344,
+      "learning_rate": 1.5435021305156862e-06,
+      "loss": 225.0581,
+      "step": 42770
+    },
+    {
+      "epoch": 0.8227674643356845,
+      "grad_norm": 1434.4670315392,
+      "learning_rate": 1.5402462254867222e-06,
+      "loss": 230.8776,
+      "step": 42780
+    },
+    {
+      "epoch": 0.822959789596165,
+      "grad_norm": 1617.2008651536803,
+      "learning_rate": 1.5369934715763235e-06,
+      "loss": 225.4059,
+      "step": 42790
+    },
+    {
+      "epoch": 0.8231521148566455,
+      "grad_norm": 1342.971304790718,
+      "learning_rate": 1.533743869996086e-06,
+      "loss": 219.8375,
+      "step": 42800
+    },
+    {
+      "epoch": 0.823344440117126,
+      "grad_norm": 1564.6699496237663,
+      "learning_rate": 1.5304974219564318e-06,
+      "loss": 222.778,
+      "step": 42810
+    },
+    {
+      "epoch": 0.8235367653776066,
+      "grad_norm": 1449.2396650691437,
+      "learning_rate": 1.5272541286666075e-06,
+      "loss": 219.3594,
+      "step": 42820
+    },
+    {
+      "epoch": 0.8237290906380871,
+      "grad_norm": 1502.664272263605,
+      "learning_rate": 1.5240139913346906e-06,
+      "loss": 224.089,
+      "step": 42830
+    },
+    {
+      "epoch": 0.8239214158985677,
+      "grad_norm": 1689.1833276265363,
+      "learning_rate": 1.5207770111675735e-06,
+      "loss": 230.1453,
+      "step": 42840
+    },
+    {
+      "epoch": 0.8241137411590482,
+      "grad_norm": 1334.5046971311374,
+      "learning_rate": 1.5175431893709836e-06,
+      "loss": 225.4314,
+      "step": 42850
+    },
+    {
+      "epoch": 0.8243060664195287,
+      "grad_norm": 1523.1610938396905,
+      "learning_rate": 1.5143125271494607e-06,
+      "loss": 221.144,
+      "step": 42860
+    },
+    {
+      "epoch": 0.8244983916800093,
+      "grad_norm": 1497.6574033801644,
+      "learning_rate": 1.511085025706378e-06,
+      "loss": 221.9221,
+      "step": 42870
+    },
+    {
+      "epoch": 0.8246907169404898,
+      "grad_norm": 1438.2043595609148,
+      "learning_rate": 1.5078606862439248e-06,
+      "loss": 220.46,
+      "step": 42880
+    },
+    {
+      "epoch": 0.8248830422009703,
+      "grad_norm": 1564.0477266607927,
+      "learning_rate": 1.5046395099631106e-06,
+      "loss": 224.4029,
+      "step": 42890
+    },
+    {
+      "epoch": 0.8250753674614508,
+      "grad_norm": 1635.0379462651622,
+      "learning_rate": 1.5014214980637754e-06,
+      "loss": 221.3462,
+      "step": 42900
+    },
+    {
+      "epoch": 0.8252676927219313,
+      "grad_norm": 1551.5149649277223,
+      "learning_rate": 1.4982066517445748e-06,
+      "loss": 219.9687,
+      "step": 42910
+    },
+    {
+      "epoch": 0.8254600179824119,
+      "grad_norm": 1707.4302422145995,
+      "learning_rate": 1.4949949722029811e-06,
+      "loss": 222.5139,
+      "step": 42920
+    },
+    {
+      "epoch": 0.8256523432428924,
+      "grad_norm": 1445.9578331543678,
+      "learning_rate": 1.4917864606352983e-06,
+      "loss": 233.4687,
+      "step": 42930
+    },
+    {
+      "epoch": 0.8258446685033729,
+      "grad_norm": 1599.6250489434442,
+      "learning_rate": 1.4885811182366406e-06,
+      "loss": 228.2906,
+      "step": 42940
+    },
+    {
+      "epoch": 0.8260369937638534,
+      "grad_norm": 1409.1313919489892,
+      "learning_rate": 1.485378946200946e-06,
+      "loss": 223.3477,
+      "step": 42950
+    },
+    {
+      "epoch": 0.8262293190243339,
+      "grad_norm": 1337.6787003616673,
+      "learning_rate": 1.4821799457209684e-06,
+      "loss": 220.3701,
+      "step": 42960
+    },
+    {
+      "epoch": 0.8264216442848145,
+      "grad_norm": 1462.1516407910956,
+      "learning_rate": 1.478984117988287e-06,
+      "loss": 224.9028,
+      "step": 42970
+    },
+    {
+      "epoch": 0.826613969545295,
+      "grad_norm": 1461.5549226889477,
+      "learning_rate": 1.4757914641932924e-06,
+      "loss": 219.3262,
+      "step": 42980
+    },
+    {
+      "epoch": 0.8268062948057755,
+      "grad_norm": 1452.2664113251153,
+      "learning_rate": 1.4726019855251928e-06,
+      "loss": 224.7554,
+      "step": 42990
+    },
+    {
+      "epoch": 0.826998620066256,
+      "grad_norm": 1406.4120536350338,
+      "learning_rate": 1.4694156831720185e-06,
+      "loss": 221.8885,
+      "step": 43000
+    },
+    {
+      "epoch": 0.8271909453267365,
+      "grad_norm": 1441.9756662111256,
+      "learning_rate": 1.4662325583206172e-06,
+      "loss": 228.0354,
+      "step": 43010
+    },
+    {
+      "epoch": 0.8273832705872171,
+      "grad_norm": 1452.276030064802,
+      "learning_rate": 1.463052612156649e-06,
+      "loss": 226.7396,
+      "step": 43020
+    },
+    {
+      "epoch": 0.8275755958476976,
+      "grad_norm": 1589.3922746547805,
+      "learning_rate": 1.4598758458645878e-06,
+      "loss": 226.6761,
+      "step": 43030
+    },
+    {
+      "epoch": 0.8277679211081782,
+      "grad_norm": 1606.813383999375,
+      "learning_rate": 1.4567022606277314e-06,
+      "loss": 222.9822,
+      "step": 43040
+    },
+    {
+      "epoch": 0.8279602463686587,
+      "grad_norm": 1482.623935353459,
+      "learning_rate": 1.4535318576281854e-06,
+      "loss": 219.5229,
+      "step": 43050
+    },
+    {
+      "epoch": 0.8281525716291392,
+      "grad_norm": 1470.459030106924,
+      "learning_rate": 1.4503646380468729e-06,
+      "loss": 219.3171,
+      "step": 43060
+    },
+    {
+      "epoch": 0.8283448968896198,
+      "grad_norm": 1678.8660847012534,
+      "learning_rate": 1.4472006030635288e-06,
+      "loss": 225.4163,
+      "step": 43070
+    },
+    {
+      "epoch": 0.8285372221501003,
+      "grad_norm": 1715.8514913998072,
+      "learning_rate": 1.4440397538567086e-06,
+      "loss": 227.607,
+      "step": 43080
+    },
+    {
+      "epoch": 0.8287295474105808,
+      "grad_norm": 1472.8729075439112,
+      "learning_rate": 1.4408820916037735e-06,
+      "loss": 222.1429,
+      "step": 43090
+    },
+    {
+      "epoch": 0.8289218726710613,
+      "grad_norm": 1623.528660479627,
+      "learning_rate": 1.4377276174808984e-06,
+      "loss": 221.4274,
+      "step": 43100
+    },
+    {
+      "epoch": 0.8291141979315418,
+      "grad_norm": 1559.8267466951684,
+      "learning_rate": 1.4345763326630768e-06,
+      "loss": 223.9511,
+      "step": 43110
+    },
+    {
+      "epoch": 0.8293065231920224,
+      "grad_norm": 1526.2159550879214,
+      "learning_rate": 1.4314282383241097e-06,
+      "loss": 221.1235,
+      "step": 43120
+    },
+    {
+      "epoch": 0.8294988484525029,
+      "grad_norm": 1578.8018533864104,
+      "learning_rate": 1.4282833356366066e-06,
+      "loss": 222.0095,
+      "step": 43130
+    },
+    {
+      "epoch": 0.8296911737129834,
+      "grad_norm": 1300.7080183334708,
+      "learning_rate": 1.4251416257719962e-06,
+      "loss": 219.9324,
+      "step": 43140
+    },
+    {
+      "epoch": 0.8298834989734639,
+      "grad_norm": 1772.3218074526933,
+      "learning_rate": 1.4220031099005094e-06,
+      "loss": 223.9319,
+      "step": 43150
+    },
+    {
+      "epoch": 0.8300758242339444,
+      "grad_norm": 1409.969836447749,
+      "learning_rate": 1.4188677891911961e-06,
+      "loss": 215.4104,
+      "step": 43160
+    },
+    {
+      "epoch": 0.830268149494425,
+      "grad_norm": 1550.5193590378076,
+      "learning_rate": 1.4157356648119103e-06,
+      "loss": 218.971,
+      "step": 43170
+    },
+    {
+      "epoch": 0.8304604747549055,
+      "grad_norm": 1563.572617345743,
+      "learning_rate": 1.412606737929313e-06,
+      "loss": 227.1725,
+      "step": 43180
+    },
+    {
+      "epoch": 0.830652800015386,
+      "grad_norm": 1311.857591919677,
+      "learning_rate": 1.4094810097088817e-06,
+      "loss": 224.0366,
+      "step": 43190
+    },
+    {
+      "epoch": 0.8308451252758665,
+      "grad_norm": 1407.0813132596777,
+      "learning_rate": 1.4063584813148979e-06,
+      "loss": 220.8986,
+      "step": 43200
+    },
+    {
+      "epoch": 0.831037450536347,
+      "grad_norm": 1509.6331052617006,
+      "learning_rate": 1.4032391539104484e-06,
+      "loss": 225.7195,
+      "step": 43210
+    },
+    {
+      "epoch": 0.8312297757968276,
+      "grad_norm": 1436.965555732776,
+      "learning_rate": 1.4001230286574363e-06,
+      "loss": 224.4622,
+      "step": 43220
+    },
+    {
+      "epoch": 0.8314221010573081,
+      "grad_norm": 1604.1059247403773,
+      "learning_rate": 1.3970101067165642e-06,
+      "loss": 222.9508,
+      "step": 43230
+    },
+    {
+      "epoch": 0.8316144263177886,
+      "grad_norm": 1426.6876703718547,
+      "learning_rate": 1.3939003892473446e-06,
+      "loss": 219.5681,
+      "step": 43240
+    },
+    {
+      "epoch": 0.8318067515782692,
+      "grad_norm": 1951.9408808587825,
+      "learning_rate": 1.390793877408093e-06,
+      "loss": 225.5036,
+      "step": 43250
+    },
+    {
+      "epoch": 0.8319990768387497,
+      "grad_norm": 1360.2465553514594,
+      "learning_rate": 1.3876905723559397e-06,
+      "loss": 219.0112,
+      "step": 43260
+    },
+    {
+      "epoch": 0.8321914020992303,
+      "grad_norm": 1403.5484396162897,
+      "learning_rate": 1.3845904752468075e-06,
+      "loss": 225.2093,
+      "step": 43270
+    },
+    {
+      "epoch": 0.8323837273597108,
+      "grad_norm": 1591.7749258548881,
+      "learning_rate": 1.3814935872354385e-06,
+      "loss": 220.1429,
+      "step": 43280
+    },
+    {
+      "epoch": 0.8325760526201913,
+      "grad_norm": 1395.5504830242419,
+      "learning_rate": 1.3783999094753653e-06,
+      "loss": 222.5725,
+      "step": 43290
+    },
+    {
+      "epoch": 0.8327683778806718,
+      "grad_norm": 1478.6462610159524,
+      "learning_rate": 1.3753094431189385e-06,
+      "loss": 227.9779,
+      "step": 43300
+    },
+    {
+      "epoch": 0.8329607031411523,
+      "grad_norm": 1328.720582165635,
+      "learning_rate": 1.3722221893173027e-06,
+      "loss": 215.7067,
+      "step": 43310
+    },
+    {
+      "epoch": 0.8331530284016329,
+      "grad_norm": 1331.731828943892,
+      "learning_rate": 1.369138149220407e-06,
+      "loss": 222.5609,
+      "step": 43320
+    },
+    {
+      "epoch": 0.8333453536621134,
+      "grad_norm": 1525.8811052669473,
+      "learning_rate": 1.3660573239770091e-06,
+      "loss": 226.8789,
+      "step": 43330
+    },
+    {
+      "epoch": 0.8335376789225939,
+      "grad_norm": 1507.0222285506968,
+      "learning_rate": 1.3629797147346635e-06,
+      "loss": 231.5131,
+      "step": 43340
+    },
+    {
+      "epoch": 0.8337300041830744,
+      "grad_norm": 1518.3584506104048,
+      "learning_rate": 1.3599053226397275e-06,
+      "loss": 226.9579,
+      "step": 43350
+    },
+    {
+      "epoch": 0.8339223294435549,
+      "grad_norm": 1421.9651108724152,
+      "learning_rate": 1.3568341488373637e-06,
+      "loss": 225.9569,
+      "step": 43360
+    },
+    {
+      "epoch": 0.8341146547040355,
+      "grad_norm": 1473.318928649225,
+      "learning_rate": 1.3537661944715342e-06,
+      "loss": 221.6633,
+      "step": 43370
+    },
+    {
+      "epoch": 0.834306979964516,
+      "grad_norm": 1378.9753465096528,
+      "learning_rate": 1.350701460685e-06,
+      "loss": 227.3245,
+      "step": 43380
+    },
+    {
+      "epoch": 0.8344993052249965,
+      "grad_norm": 1551.987471309156,
+      "learning_rate": 1.3476399486193214e-06,
+      "loss": 219.946,
+      "step": 43390
+    },
+    {
+      "epoch": 0.834691630485477,
+      "grad_norm": 1680.3512826887784,
+      "learning_rate": 1.3445816594148654e-06,
+      "loss": 231.3844,
+      "step": 43400
+    },
+    {
+      "epoch": 0.8348839557459575,
+      "grad_norm": 1737.9875027238609,
+      "learning_rate": 1.3415265942107925e-06,
+      "loss": 227.8542,
+      "step": 43410
+    },
+    {
+      "epoch": 0.835076281006438,
+      "grad_norm": 1318.0451602571711,
+      "learning_rate": 1.3384747541450615e-06,
+      "loss": 216.5979,
+      "step": 43420
+    },
+    {
+      "epoch": 0.8352686062669186,
+      "grad_norm": 1640.0894669784952,
+      "learning_rate": 1.3354261403544345e-06,
+      "loss": 224.2853,
+      "step": 43430
+    },
+    {
+      "epoch": 0.8354609315273991,
+      "grad_norm": 1575.223275261,
+      "learning_rate": 1.3323807539744726e-06,
+      "loss": 232.1573,
+      "step": 43440
+    },
+    {
+      "epoch": 0.8356532567878797,
+      "grad_norm": 1544.904791717601,
+      "learning_rate": 1.32933859613953e-06,
+      "loss": 224.5349,
+      "step": 43450
+    },
+    {
+      "epoch": 0.8358455820483602,
+      "grad_norm": 1606.8584481255227,
+      "learning_rate": 1.3262996679827567e-06,
+      "loss": 225.4846,
+      "step": 43460
+    },
+    {
+      "epoch": 0.8360379073088408,
+      "grad_norm": 1495.2814562081387,
+      "learning_rate": 1.3232639706361083e-06,
+      "loss": 224.0928,
+      "step": 43470
+    },
+    {
+      "epoch": 0.8362302325693213,
+      "grad_norm": 1749.0174161938844,
+      "learning_rate": 1.3202315052303304e-06,
+      "loss": 214.18,
+      "step": 43480
+    },
+    {
+      "epoch": 0.8364225578298018,
+      "grad_norm": 1457.572530900232,
+      "learning_rate": 1.3172022728949651e-06,
+      "loss": 223.0677,
+      "step": 43490
+    },
+    {
+      "epoch": 0.8366148830902823,
+      "grad_norm": 1659.302804104909,
+      "learning_rate": 1.3141762747583498e-06,
+      "loss": 216.7371,
+      "step": 43500
+    },
+    {
+      "epoch": 0.8368072083507628,
+      "grad_norm": 1420.9136059960424,
+      "learning_rate": 1.3111535119476237e-06,
+      "loss": 223.184,
+      "step": 43510
+    },
+    {
+      "epoch": 0.8369995336112434,
+      "grad_norm": 1676.0494682498827,
+      "learning_rate": 1.3081339855887133e-06,
+      "loss": 229.2944,
+      "step": 43520
+    },
+    {
+      "epoch": 0.8371918588717239,
+      "grad_norm": 1477.8419388152613,
+      "learning_rate": 1.3051176968063407e-06,
+      "loss": 226.9572,
+      "step": 43530
+    },
+    {
+      "epoch": 0.8373841841322044,
+      "grad_norm": 1455.6740985715503,
+      "learning_rate": 1.302104646724026e-06,
+      "loss": 222.4934,
+      "step": 43540
+    },
+    {
+      "epoch": 0.8375765093926849,
+      "grad_norm": 1473.3636505248817,
+      "learning_rate": 1.299094836464081e-06,
+      "loss": 225.0963,
+      "step": 43550
+    },
+    {
+      "epoch": 0.8377688346531654,
+      "grad_norm": 1497.7140480845171,
+      "learning_rate": 1.2960882671476062e-06,
+      "loss": 220.6281,
+      "step": 43560
+    },
+    {
+      "epoch": 0.837961159913646,
+      "grad_norm": 1291.8235591520413,
+      "learning_rate": 1.2930849398945033e-06,
+      "loss": 219.9122,
+      "step": 43570
+    },
+    {
+      "epoch": 0.8381534851741265,
+      "grad_norm": 1472.0839395415944,
+      "learning_rate": 1.2900848558234625e-06,
+      "loss": 224.1345,
+      "step": 43580
+    },
+    {
+      "epoch": 0.838345810434607,
+      "grad_norm": 1393.9235174101336,
+      "learning_rate": 1.2870880160519628e-06,
+      "loss": 223.6116,
+      "step": 43590
+    },
+    {
+      "epoch": 0.8385381356950875,
+      "grad_norm": 1556.480565862197,
+      "learning_rate": 1.2840944216962802e-06,
+      "loss": 226.3156,
+      "step": 43600
+    },
+    {
+      "epoch": 0.838730460955568,
+      "grad_norm": 1393.7299081988726,
+      "learning_rate": 1.2811040738714742e-06,
+      "loss": 220.957,
+      "step": 43610
+    },
+    {
+      "epoch": 0.8389227862160485,
+      "grad_norm": 1391.969938940345,
+      "learning_rate": 1.2781169736914067e-06,
+      "loss": 219.2455,
+      "step": 43620
+    },
+    {
+      "epoch": 0.8391151114765291,
+      "grad_norm": 1467.4454476909652,
+      "learning_rate": 1.275133122268719e-06,
+      "loss": 216.2487,
+      "step": 43630
+    },
+    {
+      "epoch": 0.8393074367370096,
+      "grad_norm": 1434.4400913722538,
+      "learning_rate": 1.2721525207148456e-06,
+      "loss": 221.2937,
+      "step": 43640
+    },
+    {
+      "epoch": 0.8394997619974901,
+      "grad_norm": 1527.7518616830482,
+      "learning_rate": 1.2691751701400145e-06,
+      "loss": 223.8823,
+      "step": 43650
+    },
+    {
+      "epoch": 0.8396920872579707,
+      "grad_norm": 1519.8938594863816,
+      "learning_rate": 1.2662010716532392e-06,
+      "loss": 223.0174,
+      "step": 43660
+    },
+    {
+      "epoch": 0.8398844125184513,
+      "grad_norm": 1515.1833066857318,
+      "learning_rate": 1.2632302263623198e-06,
+      "loss": 217.628,
+      "step": 43670
+    },
+    {
+      "epoch": 0.8400767377789318,
+      "grad_norm": 1410.9299564052328,
+      "learning_rate": 1.2602626353738479e-06,
+      "loss": 223.9453,
+      "step": 43680
+    },
+    {
+      "epoch": 0.8402690630394123,
+      "grad_norm": 1620.277058779955,
+      "learning_rate": 1.2572982997932037e-06,
+      "loss": 227.3114,
+      "step": 43690
+    },
+    {
+      "epoch": 0.8404613882998928,
+      "grad_norm": 1533.6290807670493,
+      "learning_rate": 1.2543372207245508e-06,
+      "loss": 222.9592,
+      "step": 43700
+    },
+    {
+      "epoch": 0.8406537135603733,
+      "grad_norm": 1402.3982433221956,
+      "learning_rate": 1.2513793992708467e-06,
+      "loss": 226.6808,
+      "step": 43710
+    },
+    {
+      "epoch": 0.8408460388208538,
+      "grad_norm": 1416.0522248472248,
+      "learning_rate": 1.2484248365338248e-06,
+      "loss": 222.8583,
+      "step": 43720
+    },
+    {
+      "epoch": 0.8410383640813344,
+      "grad_norm": 1411.279093768394,
+      "learning_rate": 1.2454735336140167e-06,
+      "loss": 224.7821,
+      "step": 43730
+    },
+    {
+      "epoch": 0.8412306893418149,
+      "grad_norm": 1430.420101239886,
+      "learning_rate": 1.2425254916107321e-06,
+      "loss": 222.7541,
+      "step": 43740
+    },
+    {
+      "epoch": 0.8414230146022954,
+      "grad_norm": 1434.1488896863368,
+      "learning_rate": 1.2395807116220648e-06,
+      "loss": 225.4853,
+      "step": 43750
+    },
+    {
+      "epoch": 0.8416153398627759,
+      "grad_norm": 1496.9988491196166,
+      "learning_rate": 1.236639194744902e-06,
+      "loss": 227.2553,
+      "step": 43760
+    },
+    {
+      "epoch": 0.8418076651232564,
+      "grad_norm": 1587.7390295015002,
+      "learning_rate": 1.233700942074907e-06,
+      "loss": 222.028,
+      "step": 43770
+    },
+    {
+      "epoch": 0.841999990383737,
+      "grad_norm": 1231.4497036693028,
+      "learning_rate": 1.230765954706531e-06,
+      "loss": 219.2602,
+      "step": 43780
+    },
+    {
+      "epoch": 0.8421923156442175,
+      "grad_norm": 1630.3885380414208,
+      "learning_rate": 1.227834233733005e-06,
+      "loss": 226.5649,
+      "step": 43790
+    },
+    {
+      "epoch": 0.842384640904698,
+      "grad_norm": 1494.4716992968256,
+      "learning_rate": 1.2249057802463527e-06,
+      "loss": 222.6803,
+      "step": 43800
+    },
+    {
+      "epoch": 0.8425769661651785,
+      "grad_norm": 1302.6174037147368,
+      "learning_rate": 1.221980595337372e-06,
+      "loss": 222.1433,
+      "step": 43810
+    },
+    {
+      "epoch": 0.842769291425659,
+      "grad_norm": 1445.7922549214225,
+      "learning_rate": 1.219058680095644e-06,
+      "loss": 224.5352,
+      "step": 43820
+    },
+    {
+      "epoch": 0.8429616166861396,
+      "grad_norm": 1415.5083646614253,
+      "learning_rate": 1.2161400356095376e-06,
+      "loss": 216.6406,
+      "step": 43830
+    },
+    {
+      "epoch": 0.8431539419466201,
+      "grad_norm": 1361.6642000422942,
+      "learning_rate": 1.2132246629661948e-06,
+      "loss": 221.9346,
+      "step": 43840
+    },
+    {
+      "epoch": 0.8433462672071006,
+      "grad_norm": 1357.4578283708697,
+      "learning_rate": 1.21031256325155e-06,
+      "loss": 222.678,
+      "step": 43850
+    },
+    {
+      "epoch": 0.8435385924675812,
+      "grad_norm": 1920.1957526131248,
+      "learning_rate": 1.2074037375503056e-06,
+      "loss": 214.6385,
+      "step": 43860
+    },
+    {
+      "epoch": 0.8437309177280617,
+      "grad_norm": 1392.2492299323367,
+      "learning_rate": 1.2044981869459571e-06,
+      "loss": 227.5373,
+      "step": 43870
+    },
+    {
+      "epoch": 0.8439232429885423,
+      "grad_norm": 1423.496928685203,
+      "learning_rate": 1.201595912520771e-06,
+      "loss": 225.3438,
+      "step": 43880
+    },
+    {
+      "epoch": 0.8441155682490228,
+      "grad_norm": 1565.7378212014708,
+      "learning_rate": 1.198696915355796e-06,
+      "loss": 232.4191,
+      "step": 43890
+    },
+    {
+      "epoch": 0.8443078935095033,
+      "grad_norm": 1586.833360537692,
+      "learning_rate": 1.1958011965308624e-06,
+      "loss": 228.0277,
+      "step": 43900
+    },
+    {
+      "epoch": 0.8445002187699838,
+      "grad_norm": 1599.4084633087273,
+      "learning_rate": 1.192908757124578e-06,
+      "loss": 224.7376,
+      "step": 43910
+    },
+    {
+      "epoch": 0.8446925440304643,
+      "grad_norm": 1355.9518693533403,
+      "learning_rate": 1.190019598214327e-06,
+      "loss": 224.9384,
+      "step": 43920
+    },
+    {
+      "epoch": 0.8448848692909449,
+      "grad_norm": 1757.2523577222307,
+      "learning_rate": 1.1871337208762723e-06,
+      "loss": 218.5743,
+      "step": 43930
+    },
+    {
+      "epoch": 0.8450771945514254,
+      "grad_norm": 1610.8442993721146,
+      "learning_rate": 1.1842511261853596e-06,
+      "loss": 228.0625,
+      "step": 43940
+    },
+    {
+      "epoch": 0.8452695198119059,
+      "grad_norm": 1406.8465327967401,
+      "learning_rate": 1.1813718152153054e-06,
+      "loss": 219.3295,
+      "step": 43950
+    },
+    {
+      "epoch": 0.8454618450723864,
+      "grad_norm": 1600.0839402379015,
+      "learning_rate": 1.1784957890386051e-06,
+      "loss": 219.9695,
+      "step": 43960
+    },
+    {
+      "epoch": 0.8456541703328669,
+      "grad_norm": 1600.5809616542167,
+      "learning_rate": 1.1756230487265296e-06,
+      "loss": 220.5824,
+      "step": 43970
+    },
+    {
+      "epoch": 0.8458464955933475,
+      "grad_norm": 1487.417388175908,
+      "learning_rate": 1.1727535953491308e-06,
+      "loss": 221.6486,
+      "step": 43980
+    },
+    {
+      "epoch": 0.846038820853828,
+      "grad_norm": 1334.7323064862499,
+      "learning_rate": 1.1698874299752293e-06,
+      "loss": 216.1754,
+      "step": 43990
+    },
+    {
+      "epoch": 0.8462311461143085,
+      "grad_norm": 1419.6007140269585,
+      "learning_rate": 1.1670245536724267e-06,
+      "loss": 223.4285,
+      "step": 44000
+    },
+    {
+      "epoch": 0.846423471374789,
+      "grad_norm": 1431.1831841195954,
+      "learning_rate": 1.1641649675070975e-06,
+      "loss": 224.9286,
+      "step": 44010
+    },
+    {
+      "epoch": 0.8466157966352695,
+      "grad_norm": 1701.9355507971266,
+      "learning_rate": 1.161308672544389e-06,
+      "loss": 237.1768,
+      "step": 44020
+    },
+    {
+      "epoch": 0.84680812189575,
+      "grad_norm": 1536.6224091129468,
+      "learning_rate": 1.1584556698482252e-06,
+      "loss": 220.6659,
+      "step": 44030
+    },
+    {
+      "epoch": 0.8470004471562306,
+      "grad_norm": 1439.5508660709074,
+      "learning_rate": 1.1556059604812985e-06,
+      "loss": 220.4054,
+      "step": 44040
+    },
+    {
+      "epoch": 0.8471927724167111,
+      "grad_norm": 1449.787752541528,
+      "learning_rate": 1.1527595455050844e-06,
+      "loss": 223.6956,
+      "step": 44050
+    },
+    {
+      "epoch": 0.8473850976771916,
+      "grad_norm": 1941.3541577436552,
+      "learning_rate": 1.1499164259798223e-06,
+      "loss": 223.6243,
+      "step": 44060
+    },
+    {
+      "epoch": 0.8475774229376722,
+      "grad_norm": 1329.2756865018018,
+      "learning_rate": 1.1470766029645253e-06,
+      "loss": 226.6673,
+      "step": 44070
+    },
+    {
+      "epoch": 0.8477697481981528,
+      "grad_norm": 1629.470305184223,
+      "learning_rate": 1.1442400775169849e-06,
+      "loss": 225.2041,
+      "step": 44080
+    },
+    {
+      "epoch": 0.8479620734586333,
+      "grad_norm": 1342.6852915218171,
+      "learning_rate": 1.141406850693757e-06,
+      "loss": 215.5427,
+      "step": 44090
+    },
+    {
+      "epoch": 0.8481543987191138,
+      "grad_norm": 1437.5793214251187,
+      "learning_rate": 1.1385769235501742e-06,
+      "loss": 226.4937,
+      "step": 44100
+    },
+    {
+      "epoch": 0.8483467239795943,
+      "grad_norm": 1616.9790363480909,
+      "learning_rate": 1.1357502971403335e-06,
+      "loss": 225.5648,
+      "step": 44110
+    },
+    {
+      "epoch": 0.8485390492400748,
+      "grad_norm": 1506.5620085123712,
+      "learning_rate": 1.132926972517111e-06,
+      "loss": 220.107,
+      "step": 44120
+    },
+    {
+      "epoch": 0.8487313745005554,
+      "grad_norm": 1889.0140679439746,
+      "learning_rate": 1.130106950732145e-06,
+      "loss": 217.7576,
+      "step": 44130
+    },
+    {
+      "epoch": 0.8489236997610359,
+      "grad_norm": 1456.0511442976265,
+      "learning_rate": 1.1272902328358514e-06,
+      "loss": 222.2518,
+      "step": 44140
+    },
+    {
+      "epoch": 0.8491160250215164,
+      "grad_norm": 1573.3631460039203,
+      "learning_rate": 1.1244768198774047e-06,
+      "loss": 229.7807,
+      "step": 44150
+    },
+    {
+      "epoch": 0.8493083502819969,
+      "grad_norm": 1680.161434159948,
+      "learning_rate": 1.121666712904762e-06,
+      "loss": 240.4775,
+      "step": 44160
+    },
+    {
+      "epoch": 0.8495006755424774,
+      "grad_norm": 1376.2295667532603,
+      "learning_rate": 1.1188599129646382e-06,
+      "loss": 222.535,
+      "step": 44170
+    },
+    {
+      "epoch": 0.849693000802958,
+      "grad_norm": 1611.2210517756535,
+      "learning_rate": 1.116056421102517e-06,
+      "loss": 230.8724,
+      "step": 44180
+    },
+    {
+      "epoch": 0.8498853260634385,
+      "grad_norm": 1445.7080513888532,
+      "learning_rate": 1.113256238362659e-06,
+      "loss": 227.2497,
+      "step": 44190
+    },
+    {
+      "epoch": 0.850077651323919,
+      "grad_norm": 1727.3940019080158,
+      "learning_rate": 1.1104593657880812e-06,
+      "loss": 221.5471,
+      "step": 44200
+    },
+    {
+      "epoch": 0.8502699765843995,
+      "grad_norm": 1576.8126251934868,
+      "learning_rate": 1.1076658044205746e-06,
+      "loss": 229.829,
+      "step": 44210
+    },
+    {
+      "epoch": 0.85046230184488,
+      "grad_norm": 1510.919988458017,
+      "learning_rate": 1.1048755553006928e-06,
+      "loss": 227.9921,
+      "step": 44220
+    },
+    {
+      "epoch": 0.8506546271053606,
+      "grad_norm": 1407.745251374931,
+      "learning_rate": 1.1020886194677605e-06,
+      "loss": 217.1494,
+      "step": 44230
+    },
+    {
+      "epoch": 0.8508469523658411,
+      "grad_norm": 1453.471618245384,
+      "learning_rate": 1.0993049979598635e-06,
+      "loss": 224.0534,
+      "step": 44240
+    },
+    {
+      "epoch": 0.8510392776263216,
+      "grad_norm": 1558.454782872489,
+      "learning_rate": 1.0965246918138529e-06,
+      "loss": 230.5848,
+      "step": 44250
+    },
+    {
+      "epoch": 0.8512316028868021,
+      "grad_norm": 1364.2571575929665,
+      "learning_rate": 1.093747702065351e-06,
+      "loss": 221.4876,
+      "step": 44260
+    },
+    {
+      "epoch": 0.8514239281472827,
+      "grad_norm": 1425.7823028069986,
+      "learning_rate": 1.090974029748736e-06,
+      "loss": 224.7708,
+      "step": 44270
+    },
+    {
+      "epoch": 0.8516162534077633,
+      "grad_norm": 1518.0736735190394,
+      "learning_rate": 1.0882036758971592e-06,
+      "loss": 222.334,
+      "step": 44280
+    },
+    {
+      "epoch": 0.8518085786682438,
+      "grad_norm": 1500.5306460131499,
+      "learning_rate": 1.0854366415425289e-06,
+      "loss": 231.6998,
+      "step": 44290
+    },
+    {
+      "epoch": 0.8520009039287243,
+      "grad_norm": 1259.386747688931,
+      "learning_rate": 1.0826729277155224e-06,
+      "loss": 222.8064,
+      "step": 44300
+    },
+    {
+      "epoch": 0.8521932291892048,
+      "grad_norm": 1519.9163576944618,
+      "learning_rate": 1.0799125354455752e-06,
+      "loss": 231.1701,
+      "step": 44310
+    },
+    {
+      "epoch": 0.8523855544496853,
+      "grad_norm": 1568.6990583958495,
+      "learning_rate": 1.0771554657608896e-06,
+      "loss": 222.5424,
+      "step": 44320
+    },
+    {
+      "epoch": 0.8525778797101659,
+      "grad_norm": 1632.8774978537,
+      "learning_rate": 1.0744017196884248e-06,
+      "loss": 221.433,
+      "step": 44330
+    },
+    {
+      "epoch": 0.8527702049706464,
+      "grad_norm": 1490.702483080145,
+      "learning_rate": 1.0716512982539106e-06,
+      "loss": 218.7103,
+      "step": 44340
+    },
+    {
+      "epoch": 0.8529625302311269,
+      "grad_norm": 1454.0263266258567,
+      "learning_rate": 1.0689042024818307e-06,
+      "loss": 221.9324,
+      "step": 44350
+    },
+    {
+      "epoch": 0.8531548554916074,
+      "grad_norm": 1826.897269366102,
+      "learning_rate": 1.0661604333954312e-06,
+      "loss": 228.8074,
+      "step": 44360
+    },
+    {
+      "epoch": 0.8533471807520879,
+      "grad_norm": 1324.664578344857,
+      "learning_rate": 1.0634199920167255e-06,
+      "loss": 218.7342,
+      "step": 44370
+    },
+    {
+      "epoch": 0.8535395060125684,
+      "grad_norm": 1476.943747636655,
+      "learning_rate": 1.0606828793664804e-06,
+      "loss": 221.6734,
+      "step": 44380
+    },
+    {
+      "epoch": 0.853731831273049,
+      "grad_norm": 1540.1715394889052,
+      "learning_rate": 1.0579490964642247e-06,
+      "loss": 224.3827,
+      "step": 44390
+    },
+    {
+      "epoch": 0.8539241565335295,
+      "grad_norm": 1664.3088120174123,
+      "learning_rate": 1.0552186443282464e-06,
+      "loss": 223.0133,
+      "step": 44400
+    },
+    {
+      "epoch": 0.85411648179401,
+      "grad_norm": 1376.1979941534355,
+      "learning_rate": 1.0524915239755939e-06,
+      "loss": 217.2271,
+      "step": 44410
+    },
+    {
+      "epoch": 0.8543088070544905,
+      "grad_norm": 1730.3667765054035,
+      "learning_rate": 1.0497677364220792e-06,
+      "loss": 222.5058,
+      "step": 44420
+    },
+    {
+      "epoch": 0.854501132314971,
+      "grad_norm": 1318.9617982524283,
+      "learning_rate": 1.047047282682262e-06,
+      "loss": 226.7995,
+      "step": 44430
+    },
+    {
+      "epoch": 0.8546934575754516,
+      "grad_norm": 1494.160955604607,
+      "learning_rate": 1.0443301637694713e-06,
+      "loss": 229.824,
+      "step": 44440
+    },
+    {
+      "epoch": 0.8548857828359321,
+      "grad_norm": 1616.6699796428238,
+      "learning_rate": 1.0416163806957857e-06,
+      "loss": 213.0148,
+      "step": 44450
+    },
+    {
+      "epoch": 0.8550781080964126,
+      "grad_norm": 1439.2446420159963,
+      "learning_rate": 1.0389059344720475e-06,
+      "loss": 222.4147,
+      "step": 44460
+    },
+    {
+      "epoch": 0.8552704333568931,
+      "grad_norm": 1338.2216121922509,
+      "learning_rate": 1.0361988261078482e-06,
+      "loss": 218.9765,
+      "step": 44470
+    },
+    {
+      "epoch": 0.8554627586173738,
+      "grad_norm": 1446.8531489569643,
+      "learning_rate": 1.0334950566115466e-06,
+      "loss": 216.0462,
+      "step": 44480
+    },
+    {
+      "epoch": 0.8556550838778543,
+      "grad_norm": 1478.387854339477,
+      "learning_rate": 1.0307946269902492e-06,
+      "loss": 223.0573,
+      "step": 44490
+    },
+    {
+      "epoch": 0.8558474091383348,
+      "grad_norm": 1408.6418333847616,
+      "learning_rate": 1.0280975382498225e-06,
+      "loss": 223.774,
+      "step": 44500
+    },
+    {
+      "epoch": 0.8560397343988153,
+      "grad_norm": 1431.47655305968,
+      "learning_rate": 1.0254037913948845e-06,
+      "loss": 211.6654,
+      "step": 44510
+    },
+    {
+      "epoch": 0.8562320596592958,
+      "grad_norm": 1550.89382220905,
+      "learning_rate": 1.0227133874288152e-06,
+      "loss": 227.8492,
+      "step": 44520
+    },
+    {
+      "epoch": 0.8564243849197763,
+      "grad_norm": 1383.1472593550031,
+      "learning_rate": 1.0200263273537458e-06,
+      "loss": 222.0676,
+      "step": 44530
+    },
+    {
+      "epoch": 0.8566167101802569,
+      "grad_norm": 1663.3412184091596,
+      "learning_rate": 1.0173426121705577e-06,
+      "loss": 228.0787,
+      "step": 44540
+    },
+    {
+      "epoch": 0.8568090354407374,
+      "grad_norm": 1449.0111952355635,
+      "learning_rate": 1.0146622428788943e-06,
+      "loss": 223.2425,
+      "step": 44550
+    },
+    {
+      "epoch": 0.8570013607012179,
+      "grad_norm": 1623.2614043960539,
+      "learning_rate": 1.0119852204771463e-06,
+      "loss": 230.3152,
+      "step": 44560
+    },
+    {
+      "epoch": 0.8571936859616984,
+      "grad_norm": 1375.9670550186127,
+      "learning_rate": 1.0093115459624637e-06,
+      "loss": 220.7529,
+      "step": 44570
+    },
+    {
+      "epoch": 0.857386011222179,
+      "grad_norm": 1395.1593052107223,
+      "learning_rate": 1.0066412203307419e-06,
+      "loss": 227.525,
+      "step": 44580
+    },
+    {
+      "epoch": 0.8575783364826595,
+      "grad_norm": 1527.8896744503206,
+      "learning_rate": 1.0039742445766376e-06,
+      "loss": 220.6923,
+      "step": 44590
+    },
+    {
+      "epoch": 0.85777066174314,
+      "grad_norm": 1505.368331046335,
+      "learning_rate": 1.0013106196935528e-06,
+      "loss": 226.6611,
+      "step": 44600
+    },
+    {
+      "epoch": 0.8579629870036205,
+      "grad_norm": 1602.0880846873388,
+      "learning_rate": 9.986503466736419e-07,
+      "loss": 218.8952,
+      "step": 44610
+    },
+    {
+      "epoch": 0.858155312264101,
+      "grad_norm": 1471.3230673162595,
+      "learning_rate": 9.959934265078176e-07,
+      "loss": 230.1545,
+      "step": 44620
+    },
+    {
+      "epoch": 0.8583476375245815,
+      "grad_norm": 1905.7754469448398,
+      "learning_rate": 9.933398601857347e-07,
+      "loss": 219.7658,
+      "step": 44630
+    },
+    {
+      "epoch": 0.8585399627850621,
+      "grad_norm": 1416.584642647331,
+      "learning_rate": 9.90689648695804e-07,
+      "loss": 219.2939,
+      "step": 44640
+    },
+    {
+      "epoch": 0.8587322880455426,
+      "grad_norm": 1490.6522672373076,
+      "learning_rate": 9.880427930251834e-07,
+      "loss": 219.2778,
+      "step": 44650
+    },
+    {
+      "epoch": 0.8589246133060231,
+      "grad_norm": 1387.4361209300773,
+      "learning_rate": 9.853992941597878e-07,
+      "loss": 219.1722,
+      "step": 44660
+    },
+    {
+      "epoch": 0.8591169385665036,
+      "grad_norm": 1330.4902070161509,
+      "learning_rate": 9.827591530842729e-07,
+      "loss": 217.3535,
+      "step": 44670
+    },
+    {
+      "epoch": 0.8593092638269842,
+      "grad_norm": 1289.5976318761966,
+      "learning_rate": 9.801223707820484e-07,
+      "loss": 218.8888,
+      "step": 44680
+    },
+    {
+      "epoch": 0.8595015890874648,
+      "grad_norm": 1396.0786162588342,
+      "learning_rate": 9.774889482352735e-07,
+      "loss": 216.9593,
+      "step": 44690
+    },
+    {
+      "epoch": 0.8596939143479453,
+      "grad_norm": 1314.511252984968,
+      "learning_rate": 9.74858886424852e-07,
+      "loss": 218.7266,
+      "step": 44700
+    },
+    {
+      "epoch": 0.8598862396084258,
+      "grad_norm": 1467.4830446146557,
+      "learning_rate": 9.722321863304418e-07,
+      "loss": 224.7169,
+      "step": 44710
+    },
+    {
+      "epoch": 0.8600785648689063,
+      "grad_norm": 1651.1710817944315,
+      "learning_rate": 9.696088489304412e-07,
+      "loss": 225.7997,
+      "step": 44720
+    },
+    {
+      "epoch": 0.8602708901293868,
+      "grad_norm": 1472.1879795849393,
+      "learning_rate": 9.669888752020061e-07,
+      "loss": 221.0106,
+      "step": 44730
+    },
+    {
+      "epoch": 0.8604632153898674,
+      "grad_norm": 1379.900325401597,
+      "learning_rate": 9.643722661210285e-07,
+      "loss": 213.6883,
+      "step": 44740
+    },
+    {
+      "epoch": 0.8606555406503479,
+      "grad_norm": 1403.5933765281186,
+      "learning_rate": 9.617590226621543e-07,
+      "loss": 218.9994,
+      "step": 44750
+    },
+    {
+      "epoch": 0.8608478659108284,
+      "grad_norm": 1375.9082875043493,
+      "learning_rate": 9.5914914579877e-07,
+      "loss": 216.9004,
+      "step": 44760
+    },
+    {
+      "epoch": 0.8610401911713089,
+      "grad_norm": 1412.0687070540819,
+      "learning_rate": 9.565426365030172e-07,
+      "loss": 222.487,
+      "step": 44770
+    },
+    {
+      "epoch": 0.8612325164317894,
+      "grad_norm": 1341.036516872241,
+      "learning_rate": 9.539394957457737e-07,
+      "loss": 229.0115,
+      "step": 44780
+    },
+    {
+      "epoch": 0.86142484169227,
+      "grad_norm": 1292.2400960749824,
+      "learning_rate": 9.51339724496666e-07,
+      "loss": 220.5234,
+      "step": 44790
+    },
+    {
+      "epoch": 0.8616171669527505,
+      "grad_norm": 1258.388564414747,
+      "learning_rate": 9.487433237240695e-07,
+      "loss": 216.5741,
+      "step": 44800
+    },
+    {
+      "epoch": 0.861809492213231,
+      "grad_norm": 1546.7115017649805,
+      "learning_rate": 9.461502943950973e-07,
+      "loss": 227.9115,
+      "step": 44810
+    },
+    {
+      "epoch": 0.8620018174737115,
+      "grad_norm": 1233.3411836550538,
+      "learning_rate": 9.435606374756123e-07,
+      "loss": 213.2369,
+      "step": 44820
+    },
+    {
+      "epoch": 0.862194142734192,
+      "grad_norm": 1451.3451103240286,
+      "learning_rate": 9.409743539302152e-07,
+      "loss": 214.0367,
+      "step": 44830
+    },
+    {
+      "epoch": 0.8623864679946726,
+      "grad_norm": 1606.773925143552,
+      "learning_rate": 9.383914447222576e-07,
+      "loss": 222.4851,
+      "step": 44840
+    },
+    {
+      "epoch": 0.8625787932551531,
+      "grad_norm": 1467.028494450233,
+      "learning_rate": 9.358119108138309e-07,
+      "loss": 223.8904,
+      "step": 44850
+    },
+    {
+      "epoch": 0.8627711185156336,
+      "grad_norm": 1395.128933470903,
+      "learning_rate": 9.332357531657644e-07,
+      "loss": 214.0299,
+      "step": 44860
+    },
+    {
+      "epoch": 0.8629634437761141,
+      "grad_norm": 1427.7804928539213,
+      "learning_rate": 9.306629727376404e-07,
+      "loss": 220.1048,
+      "step": 44870
+    },
+    {
+      "epoch": 0.8631557690365947,
+      "grad_norm": 1489.3891766195459,
+      "learning_rate": 9.280935704877736e-07,
+      "loss": 216.5844,
+      "step": 44880
+    },
+    {
+      "epoch": 0.8633480942970753,
+      "grad_norm": 1596.6850039793194,
+      "learning_rate": 9.255275473732239e-07,
+      "loss": 223.8956,
+      "step": 44890
+    },
+    {
+      "epoch": 0.8635404195575558,
+      "grad_norm": 1552.2301596822097,
+      "learning_rate": 9.229649043497924e-07,
+      "loss": 223.7186,
+      "step": 44900
+    },
+    {
+      "epoch": 0.8637327448180363,
+      "grad_norm": 1658.1151447156915,
+      "learning_rate": 9.20405642372022e-07,
+      "loss": 226.0265,
+      "step": 44910
+    },
+    {
+      "epoch": 0.8639250700785168,
+      "grad_norm": 1654.2692392789286,
+      "learning_rate": 9.178497623931959e-07,
+      "loss": 222.4252,
+      "step": 44920
+    },
+    {
+      "epoch": 0.8641173953389973,
+      "grad_norm": 1458.4345958966583,
+      "learning_rate": 9.152972653653369e-07,
+      "loss": 218.5969,
+      "step": 44930
+    },
+    {
+      "epoch": 0.8643097205994779,
+      "grad_norm": 1486.6553104634793,
+      "learning_rate": 9.127481522392068e-07,
+      "loss": 222.7092,
+      "step": 44940
+    },
+    {
+      "epoch": 0.8645020458599584,
+      "grad_norm": 1472.2827187767418,
+      "learning_rate": 9.102024239643092e-07,
+      "loss": 220.9687,
+      "step": 44950
+    },
+    {
+      "epoch": 0.8646943711204389,
+      "grad_norm": 1632.8645940264787,
+      "learning_rate": 9.076600814888869e-07,
+      "loss": 223.1505,
+      "step": 44960
+    },
+    {
+      "epoch": 0.8648866963809194,
+      "grad_norm": 1430.1459777674636,
+      "learning_rate": 9.051211257599169e-07,
+      "loss": 223.5466,
+      "step": 44970
+    },
+    {
+      "epoch": 0.8650790216413999,
+      "grad_norm": 1289.12944328112,
+      "learning_rate": 9.025855577231224e-07,
+      "loss": 223.2009,
+      "step": 44980
+    },
+    {
+      "epoch": 0.8652713469018805,
+      "grad_norm": 1509.108472176515,
+      "learning_rate": 9.000533783229581e-07,
+      "loss": 224.9003,
+      "step": 44990
+    },
+    {
+      "epoch": 0.865463672162361,
+      "grad_norm": 1355.3375270870872,
+      "learning_rate": 8.975245885026207e-07,
+      "loss": 226.8134,
+      "step": 45000
+    },
+    {
+      "epoch": 0.8656559974228415,
+      "grad_norm": 1628.4568008727547,
+      "learning_rate": 8.949991892040399e-07,
+      "loss": 224.4073,
+      "step": 45010
+    },
+    {
+      "epoch": 0.865848322683322,
+      "grad_norm": 1581.0882427538825,
+      "learning_rate": 8.92477181367889e-07,
+      "loss": 218.0373,
+      "step": 45020
+    },
+    {
+      "epoch": 0.8660406479438025,
+      "grad_norm": 1430.9257248548472,
+      "learning_rate": 8.899585659335719e-07,
+      "loss": 219.8096,
+      "step": 45030
+    },
+    {
+      "epoch": 0.866232973204283,
+      "grad_norm": 1634.061489982721,
+      "learning_rate": 8.874433438392305e-07,
+      "loss": 237.2581,
+      "step": 45040
+    },
+    {
+      "epoch": 0.8664252984647636,
+      "grad_norm": 1455.2076029075677,
+      "learning_rate": 8.849315160217465e-07,
+      "loss": 222.0419,
+      "step": 45050
+    },
+    {
+      "epoch": 0.8666176237252441,
+      "grad_norm": 1568.5322390886718,
+      "learning_rate": 8.824230834167325e-07,
+      "loss": 221.4173,
+      "step": 45060
+    },
+    {
+      "epoch": 0.8668099489857246,
+      "grad_norm": 1378.8242173660622,
+      "learning_rate": 8.799180469585378e-07,
+      "loss": 233.1714,
+      "step": 45070
+    },
+    {
+      "epoch": 0.8670022742462051,
+      "grad_norm": 1381.1798836559813,
+      "learning_rate": 8.77416407580246e-07,
+      "loss": 220.9373,
+      "step": 45080
+    },
+    {
+      "epoch": 0.8671945995066858,
+      "grad_norm": 1524.988751031165,
+      "learning_rate": 8.749181662136785e-07,
+      "loss": 227.4413,
+      "step": 45090
+    },
+    {
+      "epoch": 0.8673869247671663,
+      "grad_norm": 1581.5874654034042,
+      "learning_rate": 8.724233237893897e-07,
+      "loss": 219.9931,
+      "step": 45100
+    },
+    {
+      "epoch": 0.8675792500276468,
+      "grad_norm": 1483.5962658636972,
+      "learning_rate": 8.699318812366641e-07,
+      "loss": 223.0417,
+      "step": 45110
+    },
+    {
+      "epoch": 0.8677715752881273,
+      "grad_norm": 1390.2358468890366,
+      "learning_rate": 8.67443839483526e-07,
+      "loss": 226.2141,
+      "step": 45120
+    },
+    {
+      "epoch": 0.8679639005486078,
+      "grad_norm": 1369.8699902251954,
+      "learning_rate": 8.649591994567275e-07,
+      "loss": 219.5672,
+      "step": 45130
+    },
+    {
+      "epoch": 0.8681562258090884,
+      "grad_norm": 1333.5229238336788,
+      "learning_rate": 8.62477962081758e-07,
+      "loss": 221.5544,
+      "step": 45140
+    },
+    {
+      "epoch": 0.8683485510695689,
+      "grad_norm": 1616.3378021343551,
+      "learning_rate": 8.600001282828341e-07,
+      "loss": 229.0536,
+      "step": 45150
+    },
+    {
+      "epoch": 0.8685408763300494,
+      "grad_norm": 1475.6091182030611,
+      "learning_rate": 8.57525698982914e-07,
+      "loss": 224.3053,
+      "step": 45160
+    },
+    {
+      "epoch": 0.8687332015905299,
+      "grad_norm": 1461.103468956903,
+      "learning_rate": 8.550546751036759e-07,
+      "loss": 227.872,
+      "step": 45170
+    },
+    {
+      "epoch": 0.8689255268510104,
+      "grad_norm": 1622.1520427962919,
+      "learning_rate": 8.525870575655393e-07,
+      "loss": 221.3598,
+      "step": 45180
+    },
+    {
+      "epoch": 0.869117852111491,
+      "grad_norm": 1466.3661645519253,
+      "learning_rate": 8.501228472876466e-07,
+      "loss": 225.4994,
+      "step": 45190
+    },
+    {
+      "epoch": 0.8693101773719715,
+      "grad_norm": 1515.9959590644955,
+      "learning_rate": 8.476620451878803e-07,
+      "loss": 219.3026,
+      "step": 45200
+    },
+    {
+      "epoch": 0.869502502632452,
+      "grad_norm": 1418.7419482863531,
+      "learning_rate": 8.45204652182846e-07,
+      "loss": 224.0093,
+      "step": 45210
+    },
+    {
+      "epoch": 0.8696948278929325,
+      "grad_norm": 1488.427391426985,
+      "learning_rate": 8.427506691878806e-07,
+      "loss": 223.803,
+      "step": 45220
+    },
+    {
+      "epoch": 0.869887153153413,
+      "grad_norm": 2231.450311309086,
+      "learning_rate": 8.403000971170561e-07,
+      "loss": 218.2955,
+      "step": 45230
+    },
+    {
+      "epoch": 0.8700794784138935,
+      "grad_norm": 1375.2335577883996,
+      "learning_rate": 8.378529368831667e-07,
+      "loss": 221.5526,
+      "step": 45240
+    },
+    {
+      "epoch": 0.8702718036743741,
+      "grad_norm": 1424.9905254612718,
+      "learning_rate": 8.354091893977401e-07,
+      "loss": 218.7038,
+      "step": 45250
+    },
+    {
+      "epoch": 0.8704641289348546,
+      "grad_norm": 1247.697137215622,
+      "learning_rate": 8.329688555710336e-07,
+      "loss": 221.3929,
+      "step": 45260
+    },
+    {
+      "epoch": 0.8706564541953351,
+      "grad_norm": 1351.9565887296799,
+      "learning_rate": 8.305319363120279e-07,
+      "loss": 226.2381,
+      "step": 45270
+    },
+    {
+      "epoch": 0.8708487794558156,
+      "grad_norm": 1512.986668930324,
+      "learning_rate": 8.280984325284392e-07,
+      "loss": 220.8623,
+      "step": 45280
+    },
+    {
+      "epoch": 0.8710411047162963,
+      "grad_norm": 1421.0483552375156,
+      "learning_rate": 8.256683451267044e-07,
+      "loss": 218.3972,
+      "step": 45290
+    },
+    {
+      "epoch": 0.8712334299767768,
+      "grad_norm": 1622.1881608777096,
+      "learning_rate": 8.232416750119921e-07,
+      "loss": 220.7554,
+      "step": 45300
+    },
+    {
+      "epoch": 0.8714257552372573,
+      "grad_norm": 2046.142356125899,
+      "learning_rate": 8.208184230881966e-07,
+      "loss": 220.5351,
+      "step": 45310
+    },
+    {
+      "epoch": 0.8716180804977378,
+      "grad_norm": 1521.4251426531416,
+      "learning_rate": 8.183985902579405e-07,
+      "loss": 218.0601,
+      "step": 45320
+    },
+    {
+      "epoch": 0.8718104057582183,
+      "grad_norm": 1304.96593477745,
+      "learning_rate": 8.159821774225685e-07,
+      "loss": 213.2861,
+      "step": 45330
+    },
+    {
+      "epoch": 0.8720027310186989,
+      "grad_norm": 1541.0009248092513,
+      "learning_rate": 8.13569185482157e-07,
+      "loss": 217.9085,
+      "step": 45340
+    },
+    {
+      "epoch": 0.8721950562791794,
+      "grad_norm": 1342.556267511391,
+      "learning_rate": 8.111596153355061e-07,
+      "loss": 214.8615,
+      "step": 45350
+    },
+    {
+      "epoch": 0.8723873815396599,
+      "grad_norm": 1433.9957221971968,
+      "learning_rate": 8.08753467880139e-07,
+      "loss": 223.0772,
+      "step": 45360
+    },
+    {
+      "epoch": 0.8725797068001404,
+      "grad_norm": 1404.9413843291154,
+      "learning_rate": 8.063507440123052e-07,
+      "loss": 216.1271,
+      "step": 45370
+    },
+    {
+      "epoch": 0.8727720320606209,
+      "grad_norm": 1585.5505368463832,
+      "learning_rate": 8.039514446269836e-07,
+      "loss": 225.7973,
+      "step": 45380
+    },
+    {
+      "epoch": 0.8729643573211014,
+      "grad_norm": 1503.9200999930997,
+      "learning_rate": 8.015555706178702e-07,
+      "loss": 220.2815,
+      "step": 45390
+    },
+    {
+      "epoch": 0.873156682581582,
+      "grad_norm": 1564.5730298590745,
+      "learning_rate": 7.991631228773889e-07,
+      "loss": 224.5119,
+      "step": 45400
+    },
+    {
+      "epoch": 0.8733490078420625,
+      "grad_norm": 1470.1508064096213,
+      "learning_rate": 7.967741022966857e-07,
+      "loss": 216.5776,
+      "step": 45410
+    },
+    {
+      "epoch": 0.873541333102543,
+      "grad_norm": 1455.5821755407633,
+      "learning_rate": 7.943885097656356e-07,
+      "loss": 214.1013,
+      "step": 45420
+    },
+    {
+      "epoch": 0.8737336583630235,
+      "grad_norm": 1414.6460342623484,
+      "learning_rate": 7.920063461728311e-07,
+      "loss": 217.3768,
+      "step": 45430
+    },
+    {
+      "epoch": 0.873925983623504,
+      "grad_norm": 1421.9858631033972,
+      "learning_rate": 7.896276124055846e-07,
+      "loss": 220.0106,
+      "step": 45440
+    },
+    {
+      "epoch": 0.8741183088839846,
+      "grad_norm": 1500.0846151527558,
+      "learning_rate": 7.872523093499396e-07,
+      "loss": 223.4208,
+      "step": 45450
+    },
+    {
+      "epoch": 0.8743106341444651,
+      "grad_norm": 1495.365602861149,
+      "learning_rate": 7.848804378906561e-07,
+      "loss": 216.5039,
+      "step": 45460
+    },
+    {
+      "epoch": 0.8745029594049456,
+      "grad_norm": 1285.9684832354624,
+      "learning_rate": 7.825119989112173e-07,
+      "loss": 213.8228,
+      "step": 45470
+    },
+    {
+      "epoch": 0.8746952846654261,
+      "grad_norm": 1457.220260123584,
+      "learning_rate": 7.801469932938255e-07,
+      "loss": 222.097,
+      "step": 45480
+    },
+    {
+      "epoch": 0.8748876099259066,
+      "grad_norm": 1345.7025421902993,
+      "learning_rate": 7.777854219194092e-07,
+      "loss": 215.61,
+      "step": 45490
+    },
+    {
+      "epoch": 0.8750799351863873,
+      "grad_norm": 1358.6142492246643,
+      "learning_rate": 7.754272856676126e-07,
+      "loss": 223.7884,
+      "step": 45500
+    },
+    {
+      "epoch": 0.8752722604468678,
+      "grad_norm": 1661.8384848373844,
+      "learning_rate": 7.73072585416802e-07,
+      "loss": 225.425,
+      "step": 45510
+    },
+    {
+      "epoch": 0.8754645857073483,
+      "grad_norm": 1437.152989007928,
+      "learning_rate": 7.707213220440679e-07,
+      "loss": 214.1261,
+      "step": 45520
+    },
+    {
+      "epoch": 0.8756569109678288,
+      "grad_norm": 1391.568431687959,
+      "learning_rate": 7.683734964252143e-07,
+      "loss": 219.6732,
+      "step": 45530
+    },
+    {
+      "epoch": 0.8758492362283093,
+      "grad_norm": 1767.2067400187598,
+      "learning_rate": 7.66029109434766e-07,
+      "loss": 223.9062,
+      "step": 45540
+    },
+    {
+      "epoch": 0.8760415614887899,
+      "grad_norm": 1676.618011885755,
+      "learning_rate": 7.636881619459724e-07,
+      "loss": 220.9183,
+      "step": 45550
+    },
+    {
+      "epoch": 0.8762338867492704,
+      "grad_norm": 1699.694816764324,
+      "learning_rate": 7.613506548307936e-07,
+      "loss": 222.0517,
+      "step": 45560
+    },
+    {
+      "epoch": 0.8764262120097509,
+      "grad_norm": 1458.4701739727388,
+      "learning_rate": 7.590165889599166e-07,
+      "loss": 219.5811,
+      "step": 45570
+    },
+    {
+      "epoch": 0.8766185372702314,
+      "grad_norm": 1284.517256290324,
+      "learning_rate": 7.566859652027381e-07,
+      "loss": 217.8593,
+      "step": 45580
+    },
+    {
+      "epoch": 0.8768108625307119,
+      "grad_norm": 1589.057420851381,
+      "learning_rate": 7.543587844273814e-07,
+      "loss": 221.2403,
+      "step": 45590
+    },
+    {
+      "epoch": 0.8770031877911925,
+      "grad_norm": 1435.0343680908527,
+      "learning_rate": 7.52035047500681e-07,
+      "loss": 215.9271,
+      "step": 45600
+    },
+    {
+      "epoch": 0.877195513051673,
+      "grad_norm": 1497.3560958758344,
+      "learning_rate": 7.497147552881901e-07,
+      "loss": 222.116,
+      "step": 45610
+    },
+    {
+      "epoch": 0.8773878383121535,
+      "grad_norm": 1523.3861710197339,
+      "learning_rate": 7.473979086541772e-07,
+      "loss": 221.9458,
+      "step": 45620
+    },
+    {
+      "epoch": 0.877580163572634,
+      "grad_norm": 1440.1717469910482,
+      "learning_rate": 7.450845084616332e-07,
+      "loss": 217.7438,
+      "step": 45630
+    },
+    {
+      "epoch": 0.8777724888331145,
+      "grad_norm": 1462.3226228629226,
+      "learning_rate": 7.427745555722598e-07,
+      "loss": 227.7105,
+      "step": 45640
+    },
+    {
+      "epoch": 0.8779648140935951,
+      "grad_norm": 1560.5866363561129,
+      "learning_rate": 7.404680508464767e-07,
+      "loss": 223.216,
+      "step": 45650
+    },
+    {
+      "epoch": 0.8781571393540756,
+      "grad_norm": 1723.0172870025417,
+      "learning_rate": 7.381649951434167e-07,
+      "loss": 222.4714,
+      "step": 45660
+    },
+    {
+      "epoch": 0.8783494646145561,
+      "grad_norm": 1546.0606515429151,
+      "learning_rate": 7.358653893209333e-07,
+      "loss": 228.8557,
+      "step": 45670
+    },
+    {
+      "epoch": 0.8785417898750366,
+      "grad_norm": 1319.804459303906,
+      "learning_rate": 7.335692342355882e-07,
+      "loss": 221.1393,
+      "step": 45680
+    },
+    {
+      "epoch": 0.8787341151355171,
+      "grad_norm": 1265.532708727102,
+      "learning_rate": 7.312765307426662e-07,
+      "loss": 215.6548,
+      "step": 45690
+    },
+    {
+      "epoch": 0.8789264403959978,
+      "grad_norm": 1423.9723752623859,
+      "learning_rate": 7.28987279696155e-07,
+      "loss": 222.775,
+      "step": 45700
+    },
+    {
+      "epoch": 0.8791187656564783,
+      "grad_norm": 1382.3509709150283,
+      "learning_rate": 7.267014819487695e-07,
+      "loss": 226.3133,
+      "step": 45710
+    },
+    {
+      "epoch": 0.8793110909169588,
+      "grad_norm": 1436.0670010700708,
+      "learning_rate": 7.244191383519272e-07,
+      "loss": 219.3337,
+      "step": 45720
+    },
+    {
+      "epoch": 0.8795034161774393,
+      "grad_norm": 1577.488486576761,
+      "learning_rate": 7.221402497557629e-07,
+      "loss": 217.439,
+      "step": 45730
+    },
+    {
+      "epoch": 0.8796957414379198,
+      "grad_norm": 1461.6194812951574,
+      "learning_rate": 7.198648170091294e-07,
+      "loss": 215.2838,
+      "step": 45740
+    },
+    {
+      "epoch": 0.8798880666984004,
+      "grad_norm": 1517.99586248658,
+      "learning_rate": 7.175928409595844e-07,
+      "loss": 220.1899,
+      "step": 45750
+    },
+    {
+      "epoch": 0.8800803919588809,
+      "grad_norm": 1657.9740523783892,
+      "learning_rate": 7.153243224534001e-07,
+      "loss": 224.8396,
+      "step": 45760
+    },
+    {
+      "epoch": 0.8802727172193614,
+      "grad_norm": 1689.1605040515612,
+      "learning_rate": 7.130592623355659e-07,
+      "loss": 217.9619,
+      "step": 45770
+    },
+    {
+      "epoch": 0.8804650424798419,
+      "grad_norm": 1365.0743638550964,
+      "learning_rate": 7.10797661449778e-07,
+      "loss": 218.4567,
+      "step": 45780
+    },
+    {
+      "epoch": 0.8806573677403224,
+      "grad_norm": 1506.705052941599,
+      "learning_rate": 7.085395206384449e-07,
+      "loss": 220.4682,
+      "step": 45790
+    },
+    {
+      "epoch": 0.880849693000803,
+      "grad_norm": 1354.267316484208,
+      "learning_rate": 7.062848407426859e-07,
+      "loss": 217.3667,
+      "step": 45800
+    },
+    {
+      "epoch": 0.8810420182612835,
+      "grad_norm": 1489.8298171483382,
+      "learning_rate": 7.040336226023336e-07,
+      "loss": 215.0995,
+      "step": 45810
+    },
+    {
+      "epoch": 0.881234343521764,
+      "grad_norm": 1376.774603339723,
+      "learning_rate": 7.017858670559274e-07,
+      "loss": 222.7015,
+      "step": 45820
+    },
+    {
+      "epoch": 0.8814266687822445,
+      "grad_norm": 1580.9217395512842,
+      "learning_rate": 6.99541574940722e-07,
+      "loss": 226.3356,
+      "step": 45830
+    },
+    {
+      "epoch": 0.881618994042725,
+      "grad_norm": 1492.3541294544646,
+      "learning_rate": 6.973007470926774e-07,
+      "loss": 220.9657,
+      "step": 45840
+    },
+    {
+      "epoch": 0.8818113193032056,
+      "grad_norm": 1692.0635701177573,
+      "learning_rate": 6.95063384346466e-07,
+      "loss": 221.4249,
+      "step": 45850
+    },
+    {
+      "epoch": 0.8820036445636861,
+      "grad_norm": 1332.0832671726175,
+      "learning_rate": 6.92829487535468e-07,
+      "loss": 223.9194,
+      "step": 45860
+    },
+    {
+      "epoch": 0.8821959698241666,
+      "grad_norm": 1414.4945731135972,
+      "learning_rate": 6.905990574917709e-07,
+      "loss": 222.3355,
+      "step": 45870
+    },
+    {
+      "epoch": 0.8823882950846471,
+      "grad_norm": 1388.3547762166402,
+      "learning_rate": 6.88372095046177e-07,
+      "loss": 215.6123,
+      "step": 45880
+    },
+    {
+      "epoch": 0.8825806203451276,
+      "grad_norm": 1461.3707583105254,
+      "learning_rate": 6.861486010281915e-07,
+      "loss": 226.8827,
+      "step": 45890
+    },
+    {
+      "epoch": 0.8827729456056082,
+      "grad_norm": 1346.9723806785742,
+      "learning_rate": 6.839285762660275e-07,
+      "loss": 213.6974,
+      "step": 45900
+    },
+    {
+      "epoch": 0.8829652708660888,
+      "grad_norm": 1356.4343327787112,
+      "learning_rate": 6.81712021586608e-07,
+      "loss": 220.1619,
+      "step": 45910
+    },
+    {
+      "epoch": 0.8831575961265693,
+      "grad_norm": 1361.4909883178495,
+      "learning_rate": 6.794989378155659e-07,
+      "loss": 223.4412,
+      "step": 45920
+    },
+    {
+      "epoch": 0.8833499213870498,
+      "grad_norm": 1374.9365533472032,
+      "learning_rate": 6.772893257772361e-07,
+      "loss": 225.8385,
+      "step": 45930
+    },
+    {
+      "epoch": 0.8835422466475303,
+      "grad_norm": 1605.6222221516082,
+      "learning_rate": 6.750831862946605e-07,
+      "loss": 230.5761,
+      "step": 45940
+    },
+    {
+      "epoch": 0.8837345719080109,
+      "grad_norm": 1462.61106839348,
+      "learning_rate": 6.728805201895949e-07,
+      "loss": 225.3928,
+      "step": 45950
+    },
+    {
+      "epoch": 0.8839268971684914,
+      "grad_norm": 1541.5948137478513,
+      "learning_rate": 6.70681328282492e-07,
+      "loss": 208.7761,
+      "step": 45960
+    },
+    {
+      "epoch": 0.8841192224289719,
+      "grad_norm": 1423.7975209623678,
+      "learning_rate": 6.684856113925143e-07,
+      "loss": 221.4539,
+      "step": 45970
+    },
+    {
+      "epoch": 0.8843115476894524,
+      "grad_norm": 1520.9614802863646,
+      "learning_rate": 6.662933703375307e-07,
+      "loss": 226.6568,
+      "step": 45980
+    },
+    {
+      "epoch": 0.8845038729499329,
+      "grad_norm": 1476.3241792098945,
+      "learning_rate": 6.641046059341171e-07,
+      "loss": 213.7385,
+      "step": 45990
+    },
+    {
+      "epoch": 0.8846961982104135,
+      "grad_norm": 1618.9427451080335,
+      "learning_rate": 6.619193189975515e-07,
+      "loss": 223.3078,
+      "step": 46000
+    },
+    {
+      "epoch": 0.884888523470894,
+      "grad_norm": 1454.2732669525421,
+      "learning_rate": 6.597375103418135e-07,
+      "loss": 218.7141,
+      "step": 46010
+    },
+    {
+      "epoch": 0.8850808487313745,
+      "grad_norm": 1320.791435154261,
+      "learning_rate": 6.575591807795944e-07,
+      "loss": 212.0283,
+      "step": 46020
+    },
+    {
+      "epoch": 0.885273173991855,
+      "grad_norm": 1412.6693213232759,
+      "learning_rate": 6.553843311222863e-07,
+      "loss": 217.7653,
+      "step": 46030
+    },
+    {
+      "epoch": 0.8854654992523355,
+      "grad_norm": 1293.7639851681747,
+      "learning_rate": 6.532129621799832e-07,
+      "loss": 221.3312,
+      "step": 46040
+    },
+    {
+      "epoch": 0.885657824512816,
+      "grad_norm": 1660.7692875702126,
+      "learning_rate": 6.510450747614816e-07,
+      "loss": 226.6156,
+      "step": 46050
+    },
+    {
+      "epoch": 0.8858501497732966,
+      "grad_norm": 1443.8799605617487,
+      "learning_rate": 6.488806696742889e-07,
+      "loss": 229.9846,
+      "step": 46060
+    },
+    {
+      "epoch": 0.8860424750337771,
+      "grad_norm": 1490.2750334231596,
+      "learning_rate": 6.46719747724609e-07,
+      "loss": 233.9441,
+      "step": 46070
+    },
+    {
+      "epoch": 0.8862348002942576,
+      "grad_norm": 1428.396812622404,
+      "learning_rate": 6.44562309717347e-07,
+      "loss": 227.9069,
+      "step": 46080
+    },
+    {
+      "epoch": 0.8864271255547381,
+      "grad_norm": 1537.7132188985988,
+      "learning_rate": 6.424083564561134e-07,
+      "loss": 224.3031,
+      "step": 46090
+    },
+    {
+      "epoch": 0.8866194508152186,
+      "grad_norm": 1408.4615940504423,
+      "learning_rate": 6.402578887432232e-07,
+      "loss": 216.4268,
+      "step": 46100
+    },
+    {
+      "epoch": 0.8868117760756993,
+      "grad_norm": 1613.007077744771,
+      "learning_rate": 6.381109073796865e-07,
+      "loss": 223.7321,
+      "step": 46110
+    },
+    {
+      "epoch": 0.8870041013361798,
+      "grad_norm": 1448.9604743948537,
+      "learning_rate": 6.359674131652204e-07,
+      "loss": 215.2867,
+      "step": 46120
+    },
+    {
+      "epoch": 0.8871964265966603,
+      "grad_norm": 1595.2968969784808,
+      "learning_rate": 6.338274068982408e-07,
+      "loss": 221.4875,
+      "step": 46130
+    },
+    {
+      "epoch": 0.8873887518571408,
+      "grad_norm": 1490.5522710098103,
+      "learning_rate": 6.316908893758656e-07,
+      "loss": 219.4413,
+      "step": 46140
+    },
+    {
+      "epoch": 0.8875810771176214,
+      "grad_norm": 1480.085861014192,
+      "learning_rate": 6.295578613939113e-07,
+      "loss": 220.9146,
+      "step": 46150
+    },
+    {
+      "epoch": 0.8877734023781019,
+      "grad_norm": 1508.3910262287934,
+      "learning_rate": 6.274283237468948e-07,
+      "loss": 214.3749,
+      "step": 46160
+    },
+    {
+      "epoch": 0.8879657276385824,
+      "grad_norm": 1475.6728959557029,
+      "learning_rate": 6.25302277228036e-07,
+      "loss": 223.789,
+      "step": 46170
+    },
+    {
+      "epoch": 0.8881580528990629,
+      "grad_norm": 1611.933041951094,
+      "learning_rate": 6.231797226292502e-07,
+      "loss": 225.78,
+      "step": 46180
+    },
+    {
+      "epoch": 0.8883503781595434,
+      "grad_norm": 1483.3084428805128,
+      "learning_rate": 6.210606607411529e-07,
+      "loss": 214.7251,
+      "step": 46190
+    },
+    {
+      "epoch": 0.888542703420024,
+      "grad_norm": 1318.8166480175398,
+      "learning_rate": 6.189450923530627e-07,
+      "loss": 216.3322,
+      "step": 46200
+    },
+    {
+      "epoch": 0.8887350286805045,
+      "grad_norm": 1384.306628668819,
+      "learning_rate": 6.168330182529924e-07,
+      "loss": 225.431,
+      "step": 46210
+    },
+    {
+      "epoch": 0.888927353940985,
+      "grad_norm": 1472.0396016767743,
+      "learning_rate": 6.147244392276541e-07,
+      "loss": 220.7442,
+      "step": 46220
+    },
+    {
+      "epoch": 0.8891196792014655,
+      "grad_norm": 1280.2636634885869,
+      "learning_rate": 6.126193560624583e-07,
+      "loss": 223.7494,
+      "step": 46230
+    },
+    {
+      "epoch": 0.889312004461946,
+      "grad_norm": 1479.376389148199,
+      "learning_rate": 6.105177695415165e-07,
+      "loss": 218.8775,
+      "step": 46240
+    },
+    {
+      "epoch": 0.8895043297224265,
+      "grad_norm": 1337.081755758868,
+      "learning_rate": 6.084196804476317e-07,
+      "loss": 217.6594,
+      "step": 46250
+    },
+    {
+      "epoch": 0.8896966549829071,
+      "grad_norm": 1461.7468868415488,
+      "learning_rate": 6.063250895623096e-07,
+      "loss": 216.627,
+      "step": 46260
+    },
+    {
+      "epoch": 0.8898889802433876,
+      "grad_norm": 2250.7741556144065,
+      "learning_rate": 6.042339976657486e-07,
+      "loss": 227.4446,
+      "step": 46270
+    },
+    {
+      "epoch": 0.8900813055038681,
+      "grad_norm": 1431.2070494282514,
+      "learning_rate": 6.021464055368498e-07,
+      "loss": 220.2569,
+      "step": 46280
+    },
+    {
+      "epoch": 0.8902736307643486,
+      "grad_norm": 1326.0558774466288,
+      "learning_rate": 6.000623139532036e-07,
+      "loss": 228.0942,
+      "step": 46290
+    },
+    {
+      "epoch": 0.8904659560248291,
+      "grad_norm": 1294.28514245179,
+      "learning_rate": 5.979817236910979e-07,
+      "loss": 220.3547,
+      "step": 46300
+    },
+    {
+      "epoch": 0.8906582812853097,
+      "grad_norm": 1404.4233102421995,
+      "learning_rate": 5.959046355255238e-07,
+      "loss": 232.473,
+      "step": 46310
+    },
+    {
+      "epoch": 0.8908506065457903,
+      "grad_norm": 1458.577998790589,
+      "learning_rate": 5.93831050230158e-07,
+      "loss": 219.4227,
+      "step": 46320
+    },
+    {
+      "epoch": 0.8910429318062708,
+      "grad_norm": 1484.8663443865212,
+      "learning_rate": 5.917609685773784e-07,
+      "loss": 218.1252,
+      "step": 46330
+    },
+    {
+      "epoch": 0.8912352570667513,
+      "grad_norm": 1482.839545464784,
+      "learning_rate": 5.896943913382547e-07,
+      "loss": 213.3422,
+      "step": 46340
+    },
+    {
+      "epoch": 0.8914275823272318,
+      "grad_norm": 1272.5806136822664,
+      "learning_rate": 5.876313192825544e-07,
+      "loss": 214.6004,
+      "step": 46350
+    },
+    {
+      "epoch": 0.8916199075877124,
+      "grad_norm": 1575.626673681023,
+      "learning_rate": 5.855717531787375e-07,
+      "loss": 227.8473,
+      "step": 46360
+    },
+    {
+      "epoch": 0.8918122328481929,
+      "grad_norm": 1520.8216265388264,
+      "learning_rate": 5.835156937939568e-07,
+      "loss": 226.4484,
+      "step": 46370
+    },
+    {
+      "epoch": 0.8920045581086734,
+      "grad_norm": 1388.38765817661,
+      "learning_rate": 5.814631418940641e-07,
+      "loss": 218.6067,
+      "step": 46380
+    },
+    {
+      "epoch": 0.8921968833691539,
+      "grad_norm": 1391.1677029761859,
+      "learning_rate": 5.794140982435981e-07,
+      "loss": 218.7422,
+      "step": 46390
+    },
+    {
+      "epoch": 0.8923892086296344,
+      "grad_norm": 1477.1739172444923,
+      "learning_rate": 5.773685636057924e-07,
+      "loss": 220.2675,
+      "step": 46400
+    },
+    {
+      "epoch": 0.892581533890115,
+      "grad_norm": 1363.037747944765,
+      "learning_rate": 5.753265387425777e-07,
+      "loss": 229.6712,
+      "step": 46410
+    },
+    {
+      "epoch": 0.8927738591505955,
+      "grad_norm": 1866.725434299131,
+      "learning_rate": 5.732880244145744e-07,
+      "loss": 216.3705,
+      "step": 46420
+    },
+    {
+      "epoch": 0.892966184411076,
+      "grad_norm": 1309.853753693657,
+      "learning_rate": 5.712530213810951e-07,
+      "loss": 221.7039,
+      "step": 46430
+    },
+    {
+      "epoch": 0.8931585096715565,
+      "grad_norm": 1392.555337628686,
+      "learning_rate": 5.692215304001447e-07,
+      "loss": 217.0223,
+      "step": 46440
+    },
+    {
+      "epoch": 0.893350834932037,
+      "grad_norm": 1430.7355999475292,
+      "learning_rate": 5.671935522284177e-07,
+      "loss": 223.6912,
+      "step": 46450
+    },
+    {
+      "epoch": 0.8935431601925176,
+      "grad_norm": 1538.112836502128,
+      "learning_rate": 5.651690876213067e-07,
+      "loss": 226.1131,
+      "step": 46460
+    },
+    {
+      "epoch": 0.8937354854529981,
+      "grad_norm": 1463.9496914215588,
+      "learning_rate": 5.631481373328895e-07,
+      "loss": 227.5667,
+      "step": 46470
+    },
+    {
+      "epoch": 0.8939278107134786,
+      "grad_norm": 1344.7445157577852,
+      "learning_rate": 5.61130702115934e-07,
+      "loss": 224.5837,
+      "step": 46480
+    },
+    {
+      "epoch": 0.8941201359739591,
+      "grad_norm": 1542.225333096471,
+      "learning_rate": 5.591167827219057e-07,
+      "loss": 220.551,
+      "step": 46490
+    },
+    {
+      "epoch": 0.8943124612344396,
+      "grad_norm": 1656.3510143022625,
+      "learning_rate": 5.571063799009546e-07,
+      "loss": 222.3722,
+      "step": 46500
+    },
+    {
+      "epoch": 0.8945047864949202,
+      "grad_norm": 1368.3589018208604,
+      "learning_rate": 5.550994944019216e-07,
+      "loss": 221.5293,
+      "step": 46510
+    },
+    {
+      "epoch": 0.8946971117554008,
+      "grad_norm": 1399.5642335697032,
+      "learning_rate": 5.53096126972339e-07,
+      "loss": 222.8575,
+      "step": 46520
+    },
+    {
+      "epoch": 0.8948894370158813,
+      "grad_norm": 1375.8183121848929,
+      "learning_rate": 5.510962783584295e-07,
+      "loss": 222.4795,
+      "step": 46530
+    },
+    {
+      "epoch": 0.8950817622763618,
+      "grad_norm": 1610.9080169549472,
+      "learning_rate": 5.490999493051008e-07,
+      "loss": 218.747,
+      "step": 46540
+    },
+    {
+      "epoch": 0.8952740875368423,
+      "grad_norm": 1424.7380763758229,
+      "learning_rate": 5.471071405559547e-07,
+      "loss": 220.7228,
+      "step": 46550
+    },
+    {
+      "epoch": 0.8954664127973229,
+      "grad_norm": 1436.3949981178098,
+      "learning_rate": 5.451178528532786e-07,
+      "loss": 222.3391,
+      "step": 46560
+    },
+    {
+      "epoch": 0.8956587380578034,
+      "grad_norm": 1360.3733143259442,
+      "learning_rate": 5.431320869380519e-07,
+      "loss": 210.3986,
+      "step": 46570
+    },
+    {
+      "epoch": 0.8958510633182839,
+      "grad_norm": 1331.1294292070177,
+      "learning_rate": 5.411498435499363e-07,
+      "loss": 226.2973,
+      "step": 46580
+    },
+    {
+      "epoch": 0.8960433885787644,
+      "grad_norm": 1430.736991708031,
+      "learning_rate": 5.391711234272856e-07,
+      "loss": 217.365,
+      "step": 46590
+    },
+    {
+      "epoch": 0.8962357138392449,
+      "grad_norm": 1394.9522651995821,
+      "learning_rate": 5.371959273071414e-07,
+      "loss": 221.7367,
+      "step": 46600
+    },
+    {
+      "epoch": 0.8964280390997255,
+      "grad_norm": 1330.574883087887,
+      "learning_rate": 5.352242559252308e-07,
+      "loss": 212.5495,
+      "step": 46610
+    },
+    {
+      "epoch": 0.896620364360206,
+      "grad_norm": 1339.4661413834856,
+      "learning_rate": 5.332561100159683e-07,
+      "loss": 222.5836,
+      "step": 46620
+    },
+    {
+      "epoch": 0.8968126896206865,
+      "grad_norm": 1403.4451768756446,
+      "learning_rate": 5.312914903124566e-07,
+      "loss": 218.0726,
+      "step": 46630
+    },
+    {
+      "epoch": 0.897005014881167,
+      "grad_norm": 1542.6051427233544,
+      "learning_rate": 5.293303975464836e-07,
+      "loss": 219.1479,
+      "step": 46640
+    },
+    {
+      "epoch": 0.8971973401416475,
+      "grad_norm": 1537.3246365725583,
+      "learning_rate": 5.273728324485261e-07,
+      "loss": 221.9458,
+      "step": 46650
+    },
+    {
+      "epoch": 0.8973896654021281,
+      "grad_norm": 1478.0167750122955,
+      "learning_rate": 5.254187957477397e-07,
+      "loss": 232.0157,
+      "step": 46660
+    },
+    {
+      "epoch": 0.8975819906626086,
+      "grad_norm": 1411.1370098994937,
+      "learning_rate": 5.234682881719766e-07,
+      "loss": 222.2815,
+      "step": 46670
+    },
+    {
+      "epoch": 0.8977743159230891,
+      "grad_norm": 1376.0540284869267,
+      "learning_rate": 5.215213104477645e-07,
+      "loss": 218.5224,
+      "step": 46680
+    },
+    {
+      "epoch": 0.8979666411835696,
+      "grad_norm": 1360.111196179622,
+      "learning_rate": 5.195778633003223e-07,
+      "loss": 214.8784,
+      "step": 46690
+    },
+    {
+      "epoch": 0.8981589664440501,
+      "grad_norm": 1299.634393360051,
+      "learning_rate": 5.176379474535509e-07,
+      "loss": 217.2428,
+      "step": 46700
+    },
+    {
+      "epoch": 0.8983512917045307,
+      "grad_norm": 1606.7830309270275,
+      "learning_rate": 5.1570156363004e-07,
+      "loss": 224.4943,
+      "step": 46710
+    },
+    {
+      "epoch": 0.8985436169650112,
+      "grad_norm": 1408.556361161924,
+      "learning_rate": 5.13768712551057e-07,
+      "loss": 227.2928,
+      "step": 46720
+    },
+    {
+      "epoch": 0.8987359422254918,
+      "grad_norm": 1324.8942740295813,
+      "learning_rate": 5.118393949365574e-07,
+      "loss": 220.0283,
+      "step": 46730
+    },
+    {
+      "epoch": 0.8989282674859723,
+      "grad_norm": 1901.0604238102271,
+      "learning_rate": 5.099136115051829e-07,
+      "loss": 220.363,
+      "step": 46740
+    },
+    {
+      "epoch": 0.8991205927464528,
+      "grad_norm": 1478.726525220824,
+      "learning_rate": 5.079913629742539e-07,
+      "loss": 222.2089,
+      "step": 46750
+    },
+    {
+      "epoch": 0.8993129180069334,
+      "grad_norm": 1505.128030958836,
+      "learning_rate": 5.060726500597768e-07,
+      "loss": 227.9248,
+      "step": 46760
+    },
+    {
+      "epoch": 0.8995052432674139,
+      "grad_norm": 1450.2383454067806,
+      "learning_rate": 5.041574734764376e-07,
+      "loss": 220.9869,
+      "step": 46770
+    },
+    {
+      "epoch": 0.8996975685278944,
+      "grad_norm": 1453.839559140101,
+      "learning_rate": 5.022458339376124e-07,
+      "loss": 218.7223,
+      "step": 46780
+    },
+    {
+      "epoch": 0.8998898937883749,
+      "grad_norm": 1468.469766703546,
+      "learning_rate": 5.003377321553538e-07,
+      "loss": 225.2005,
+      "step": 46790
+    },
+    {
+      "epoch": 0.9000822190488554,
+      "grad_norm": 1355.747466162097,
+      "learning_rate": 4.984331688403976e-07,
+      "loss": 224.5639,
+      "step": 46800
+    },
+    {
+      "epoch": 0.900274544309336,
+      "grad_norm": 1444.184541351923,
+      "learning_rate": 4.96532144702162e-07,
+      "loss": 218.3931,
+      "step": 46810
+    },
+    {
+      "epoch": 0.9004668695698165,
+      "grad_norm": 1431.5097195521855,
+      "learning_rate": 4.946346604487462e-07,
+      "loss": 231.0745,
+      "step": 46820
+    },
+    {
+      "epoch": 0.900659194830297,
+      "grad_norm": 1372.0452974625207,
+      "learning_rate": 4.927407167869346e-07,
+      "loss": 218.4352,
+      "step": 46830
+    },
+    {
+      "epoch": 0.9008515200907775,
+      "grad_norm": 1429.7090153273914,
+      "learning_rate": 4.908503144221877e-07,
+      "loss": 220.1105,
+      "step": 46840
+    },
+    {
+      "epoch": 0.901043845351258,
+      "grad_norm": 1422.8227749181583,
+      "learning_rate": 4.889634540586518e-07,
+      "loss": 220.7422,
+      "step": 46850
+    },
+    {
+      "epoch": 0.9012361706117386,
+      "grad_norm": 1519.5353671623918,
+      "learning_rate": 4.870801363991484e-07,
+      "loss": 227.8265,
+      "step": 46860
+    },
+    {
+      "epoch": 0.9014284958722191,
+      "grad_norm": 1359.3086555893549,
+      "learning_rate": 4.852003621451829e-07,
+      "loss": 216.379,
+      "step": 46870
+    },
+    {
+      "epoch": 0.9016208211326996,
+      "grad_norm": 1595.9811679470838,
+      "learning_rate": 4.833241319969395e-07,
+      "loss": 215.0732,
+      "step": 46880
+    },
+    {
+      "epoch": 0.9018131463931801,
+      "grad_norm": 1329.4766522062168,
+      "learning_rate": 4.814514466532849e-07,
+      "loss": 219.9756,
+      "step": 46890
+    },
+    {
+      "epoch": 0.9020054716536606,
+      "grad_norm": 1519.7936145474603,
+      "learning_rate": 4.795823068117622e-07,
+      "loss": 223.6884,
+      "step": 46900
+    },
+    {
+      "epoch": 0.9021977969141411,
+      "grad_norm": 1441.189960715862,
+      "learning_rate": 4.777167131685945e-07,
+      "loss": 214.417,
+      "step": 46910
+    },
+    {
+      "epoch": 0.9023901221746217,
+      "grad_norm": 1379.65981457216,
+      "learning_rate": 4.7585466641868696e-07,
+      "loss": 226.1294,
+      "step": 46920
+    },
+    {
+      "epoch": 0.9025824474351023,
+      "grad_norm": 1387.7175842725867,
+      "learning_rate": 4.7399616725561925e-07,
+      "loss": 215.7953,
+      "step": 46930
+    },
+    {
+      "epoch": 0.9027747726955828,
+      "grad_norm": 1490.7144996476823,
+      "learning_rate": 4.721412163716521e-07,
+      "loss": 219.4926,
+      "step": 46940
+    },
+    {
+      "epoch": 0.9029670979560633,
+      "grad_norm": 1606.9868485269926,
+      "learning_rate": 4.702898144577228e-07,
+      "loss": 232.0976,
+      "step": 46950
+    },
+    {
+      "epoch": 0.9031594232165439,
+      "grad_norm": 1390.983723362061,
+      "learning_rate": 4.6844196220345086e-07,
+      "loss": 216.0086,
+      "step": 46960
+    },
+    {
+      "epoch": 0.9033517484770244,
+      "grad_norm": 1617.6546655662457,
+      "learning_rate": 4.665976602971278e-07,
+      "loss": 223.763,
+      "step": 46970
+    },
+    {
+      "epoch": 0.9035440737375049,
+      "grad_norm": 1559.5855001690002,
+      "learning_rate": 4.647569094257276e-07,
+      "loss": 227.5162,
+      "step": 46980
+    },
+    {
+      "epoch": 0.9037363989979854,
+      "grad_norm": 1358.3030328722803,
+      "learning_rate": 4.629197102748984e-07,
+      "loss": 214.7696,
+      "step": 46990
+    },
+    {
+      "epoch": 0.9039287242584659,
+      "grad_norm": 1525.3432992106088,
+      "learning_rate": 4.610860635289671e-07,
+      "loss": 221.7334,
+      "step": 47000
+    },
+    {
+      "epoch": 0.9041210495189465,
+      "grad_norm": 1345.8686097189452,
+      "learning_rate": 4.592559698709387e-07,
+      "loss": 219.8919,
+      "step": 47010
+    },
+    {
+      "epoch": 0.904313374779427,
+      "grad_norm": 1331.7024925857322,
+      "learning_rate": 4.5742942998248774e-07,
+      "loss": 220.1635,
+      "step": 47020
+    },
+    {
+      "epoch": 0.9045057000399075,
+      "grad_norm": 1467.937745812986,
+      "learning_rate": 4.5560644454397563e-07,
+      "loss": 220.8538,
+      "step": 47030
+    },
+    {
+      "epoch": 0.904698025300388,
+      "grad_norm": 1359.9598669413365,
+      "learning_rate": 4.537870142344314e-07,
+      "loss": 222.1589,
+      "step": 47040
+    },
+    {
+      "epoch": 0.9048903505608685,
+      "grad_norm": 1480.9876007737616,
+      "learning_rate": 4.5197113973156403e-07,
+      "loss": 218.9359,
+      "step": 47050
+    },
+    {
+      "epoch": 0.905082675821349,
+      "grad_norm": 1470.2394457393145,
+      "learning_rate": 4.5015882171175476e-07,
+      "loss": 225.6769,
+      "step": 47060
+    },
+    {
+      "epoch": 0.9052750010818296,
+      "grad_norm": 1836.6324461425384,
+      "learning_rate": 4.483500608500657e-07,
+      "loss": 218.3168,
+      "step": 47070
+    },
+    {
+      "epoch": 0.9054673263423101,
+      "grad_norm": 1614.6640102453246,
+      "learning_rate": 4.4654485782022697e-07,
+      "loss": 219.166,
+      "step": 47080
+    },
+    {
+      "epoch": 0.9056596516027906,
+      "grad_norm": 1489.5051681586247,
+      "learning_rate": 4.447432132946472e-07,
+      "loss": 224.8024,
+      "step": 47090
+    },
+    {
+      "epoch": 0.9058519768632711,
+      "grad_norm": 1680.262134295633,
+      "learning_rate": 4.429451279444119e-07,
+      "loss": 221.4581,
+      "step": 47100
+    },
+    {
+      "epoch": 0.9060443021237516,
+      "grad_norm": 1444.001840773713,
+      "learning_rate": 4.411506024392753e-07,
+      "loss": 225.6714,
+      "step": 47110
+    },
+    {
+      "epoch": 0.9062366273842322,
+      "grad_norm": 1668.5750334172917,
+      "learning_rate": 4.393596374476705e-07,
+      "loss": 214.7576,
+      "step": 47120
+    },
+    {
+      "epoch": 0.9064289526447128,
+      "grad_norm": 1332.7983133827638,
+      "learning_rate": 4.3757223363670055e-07,
+      "loss": 211.9972,
+      "step": 47130
+    },
+    {
+      "epoch": 0.9066212779051933,
+      "grad_norm": 1314.4716216572058,
+      "learning_rate": 4.3578839167214505e-07,
+      "loss": 224.1075,
+      "step": 47140
+    },
+    {
+      "epoch": 0.9068136031656738,
+      "grad_norm": 1344.0111235957,
+      "learning_rate": 4.3400811221845693e-07,
+      "loss": 223.1015,
+      "step": 47150
+    },
+    {
+      "epoch": 0.9070059284261544,
+      "grad_norm": 1392.1378119485578,
+      "learning_rate": 4.322313959387592e-07,
+      "loss": 220.7298,
+      "step": 47160
+    },
+    {
+      "epoch": 0.9071982536866349,
+      "grad_norm": 1339.662045877732,
+      "learning_rate": 4.304582434948479e-07,
+      "loss": 219.0525,
+      "step": 47170
+    },
+    {
+      "epoch": 0.9073905789471154,
+      "grad_norm": 1237.1396430250047,
+      "learning_rate": 4.2868865554719583e-07,
+      "loss": 220.4794,
+      "step": 47180
+    },
+    {
+      "epoch": 0.9075829042075959,
+      "grad_norm": 1243.689604265467,
+      "learning_rate": 4.269226327549447e-07,
+      "loss": 222.6925,
+      "step": 47190
+    },
+    {
+      "epoch": 0.9077752294680764,
+      "grad_norm": 1525.2456161598323,
+      "learning_rate": 4.251601757759061e-07,
+      "loss": 212.6521,
+      "step": 47200
+    },
+    {
+      "epoch": 0.907967554728557,
+      "grad_norm": 1569.5996124155515,
+      "learning_rate": 4.234012852665703e-07,
+      "loss": 227.1184,
+      "step": 47210
+    },
+    {
+      "epoch": 0.9081598799890375,
+      "grad_norm": 1536.4192739967484,
+      "learning_rate": 4.2164596188209226e-07,
+      "loss": 224.376,
+      "step": 47220
+    },
+    {
+      "epoch": 0.908352205249518,
+      "grad_norm": 1622.905828787272,
+      "learning_rate": 4.198942062763023e-07,
+      "loss": 223.7278,
+      "step": 47230
+    },
+    {
+      "epoch": 0.9085445305099985,
+      "grad_norm": 1513.5467307466727,
+      "learning_rate": 4.181460191016984e-07,
+      "loss": 217.6169,
+      "step": 47240
+    },
+    {
+      "epoch": 0.908736855770479,
+      "grad_norm": 1366.5343375702782,
+      "learning_rate": 4.1640140100945304e-07,
+      "loss": 224.4244,
+      "step": 47250
+    },
+    {
+      "epoch": 0.9089291810309595,
+      "grad_norm": 1605.2650628774873,
+      "learning_rate": 4.146603526494086e-07,
+      "loss": 219.5355,
+      "step": 47260
+    },
+    {
+      "epoch": 0.9091215062914401,
+      "grad_norm": 1553.9891847004822,
+      "learning_rate": 4.129228746700742e-07,
+      "loss": 225.5607,
+      "step": 47270
+    },
+    {
+      "epoch": 0.9093138315519206,
+      "grad_norm": 1734.1490012654092,
+      "learning_rate": 4.111889677186354e-07,
+      "loss": 220.383,
+      "step": 47280
+    },
+    {
+      "epoch": 0.9095061568124011,
+      "grad_norm": 1602.6603601514983,
+      "learning_rate": 4.094586324409411e-07,
+      "loss": 216.7603,
+      "step": 47290
+    },
+    {
+      "epoch": 0.9096984820728816,
+      "grad_norm": 1305.374800293323,
+      "learning_rate": 4.0773186948151246e-07,
+      "loss": 223.2034,
+      "step": 47300
+    },
+    {
+      "epoch": 0.9098908073333621,
+      "grad_norm": 1622.6320761783568,
+      "learning_rate": 4.060086794835405e-07,
+      "loss": 217.493,
+      "step": 47310
+    },
+    {
+      "epoch": 0.9100831325938427,
+      "grad_norm": 1429.1113968411746,
+      "learning_rate": 4.042890630888863e-07,
+      "loss": 217.6178,
+      "step": 47320
+    },
+    {
+      "epoch": 0.9102754578543232,
+      "grad_norm": 1303.8000290881128,
+      "learning_rate": 4.025730209380774e-07,
+      "loss": 212.2856,
+      "step": 47330
+    },
+    {
+      "epoch": 0.9104677831148038,
+      "grad_norm": 1472.4432589699088,
+      "learning_rate": 4.0086055367031027e-07,
+      "loss": 226.2052,
+      "step": 47340
+    },
+    {
+      "epoch": 0.9106601083752843,
+      "grad_norm": 1312.9446923193095,
+      "learning_rate": 3.9915166192345365e-07,
+      "loss": 219.4008,
+      "step": 47350
+    },
+    {
+      "epoch": 0.9108524336357648,
+      "grad_norm": 1476.785470661036,
+      "learning_rate": 3.9744634633403944e-07,
+      "loss": 216.3873,
+      "step": 47360
+    },
+    {
+      "epoch": 0.9110447588962454,
+      "grad_norm": 1370.3625332409579,
+      "learning_rate": 3.957446075372706e-07,
+      "loss": 213.5251,
+      "step": 47370
+    },
+    {
+      "epoch": 0.9112370841567259,
+      "grad_norm": 1398.7029359767273,
+      "learning_rate": 3.940464461670135e-07,
+      "loss": 225.4371,
+      "step": 47380
+    },
+    {
+      "epoch": 0.9114294094172064,
+      "grad_norm": 1478.5102345817131,
+      "learning_rate": 3.923518628558087e-07,
+      "loss": 224.6383,
+      "step": 47390
+    },
+    {
+      "epoch": 0.9116217346776869,
+      "grad_norm": 1344.995625573096,
+      "learning_rate": 3.9066085823485923e-07,
+      "loss": 220.425,
+      "step": 47400
+    },
+    {
+      "epoch": 0.9118140599381674,
+      "grad_norm": 1420.2761277536722,
+      "learning_rate": 3.8897343293403777e-07,
+      "loss": 226.7782,
+      "step": 47410
+    },
+    {
+      "epoch": 0.912006385198648,
+      "grad_norm": 1428.019165995346,
+      "learning_rate": 3.872895875818794e-07,
+      "loss": 219.5789,
+      "step": 47420
+    },
+    {
+      "epoch": 0.9121987104591285,
+      "grad_norm": 1524.2914915650294,
+      "learning_rate": 3.856093228055924e-07,
+      "loss": 216.0496,
+      "step": 47430
+    },
+    {
+      "epoch": 0.912391035719609,
+      "grad_norm": 1436.1415045627953,
+      "learning_rate": 3.83932639231045e-07,
+      "loss": 222.3941,
+      "step": 47440
+    },
+    {
+      "epoch": 0.9125833609800895,
+      "grad_norm": 1503.9586716891579,
+      "learning_rate": 3.822595374827742e-07,
+      "loss": 218.0715,
+      "step": 47450
+    },
+    {
+      "epoch": 0.91277568624057,
+      "grad_norm": 1349.6842768640456,
+      "learning_rate": 3.805900181839839e-07,
+      "loss": 221.7199,
+      "step": 47460
+    },
+    {
+      "epoch": 0.9129680115010506,
+      "grad_norm": 1451.2761476664573,
+      "learning_rate": 3.789240819565432e-07,
+      "loss": 216.8558,
+      "step": 47470
+    },
+    {
+      "epoch": 0.9131603367615311,
+      "grad_norm": 1317.1640497005353,
+      "learning_rate": 3.772617294209835e-07,
+      "loss": 219.8583,
+      "step": 47480
+    },
+    {
+      "epoch": 0.9133526620220116,
+      "grad_norm": 1395.5069540858262,
+      "learning_rate": 3.7560296119650396e-07,
+      "loss": 220.1477,
+      "step": 47490
+    },
+    {
+      "epoch": 0.9135449872824921,
+      "grad_norm": 1593.1648381968544,
+      "learning_rate": 3.739477779009704e-07,
+      "loss": 231.3557,
+      "step": 47500
+    },
+    {
+      "epoch": 0.9137373125429726,
+      "grad_norm": 1328.7719753864867,
+      "learning_rate": 3.7229618015091065e-07,
+      "loss": 222.6315,
+      "step": 47510
+    },
+    {
+      "epoch": 0.9139296378034532,
+      "grad_norm": 1298.0060401238827,
+      "learning_rate": 3.7064816856151484e-07,
+      "loss": 217.4497,
+      "step": 47520
+    },
+    {
+      "epoch": 0.9141219630639337,
+      "grad_norm": 1784.7366103480408,
+      "learning_rate": 3.6900374374664425e-07,
+      "loss": 225.1923,
+      "step": 47530
+    },
+    {
+      "epoch": 0.9143142883244143,
+      "grad_norm": 1406.336524879021,
+      "learning_rate": 3.6736290631881667e-07,
+      "loss": 223.364,
+      "step": 47540
+    },
+    {
+      "epoch": 0.9145066135848948,
+      "grad_norm": 1447.0675423819398,
+      "learning_rate": 3.657256568892187e-07,
+      "loss": 224.4759,
+      "step": 47550
+    },
+    {
+      "epoch": 0.9146989388453753,
+      "grad_norm": 1558.7062176640768,
+      "learning_rate": 3.6409199606769806e-07,
+      "loss": 213.1187,
+      "step": 47560
+    },
+    {
+      "epoch": 0.9148912641058559,
+      "grad_norm": 1439.341310782253,
+      "learning_rate": 3.6246192446276694e-07,
+      "loss": 216.4039,
+      "step": 47570
+    },
+    {
+      "epoch": 0.9150835893663364,
+      "grad_norm": 1471.5174417088792,
+      "learning_rate": 3.6083544268160077e-07,
+      "loss": 222.9255,
+      "step": 47580
+    },
+    {
+      "epoch": 0.9152759146268169,
+      "grad_norm": 1409.9488850591572,
+      "learning_rate": 3.5921255133003483e-07,
+      "loss": 218.6654,
+      "step": 47590
+    },
+    {
+      "epoch": 0.9154682398872974,
+      "grad_norm": 1423.3093621885928,
+      "learning_rate": 3.5759325101257013e-07,
+      "loss": 223.0352,
+      "step": 47600
+    },
+    {
+      "epoch": 0.9156605651477779,
+      "grad_norm": 1505.4007198177992,
+      "learning_rate": 3.559775423323708e-07,
+      "loss": 219.4685,
+      "step": 47610
+    },
+    {
+      "epoch": 0.9158528904082585,
+      "grad_norm": 1388.416041488496,
+      "learning_rate": 3.54365425891261e-07,
+      "loss": 214.4795,
+      "step": 47620
+    },
+    {
+      "epoch": 0.916045215668739,
+      "grad_norm": 1372.0266667616108,
+      "learning_rate": 3.527569022897259e-07,
+      "loss": 224.1236,
+      "step": 47630
+    },
+    {
+      "epoch": 0.9162375409292195,
+      "grad_norm": 1346.3885818848808,
+      "learning_rate": 3.511519721269163e-07,
+      "loss": 222.9773,
+      "step": 47640
+    },
+    {
+      "epoch": 0.9164298661897,
+      "grad_norm": 1451.9466833275196,
+      "learning_rate": 3.4955063600064177e-07,
+      "loss": 215.0467,
+      "step": 47650
+    },
+    {
+      "epoch": 0.9166221914501805,
+      "grad_norm": 1439.0813641269924,
+      "learning_rate": 3.479528945073707e-07,
+      "loss": 218.8649,
+      "step": 47660
+    },
+    {
+      "epoch": 0.916814516710661,
+      "grad_norm": 1374.7710071859926,
+      "learning_rate": 3.4635874824223924e-07,
+      "loss": 221.2719,
+      "step": 47670
+    },
+    {
+      "epoch": 0.9170068419711416,
+      "grad_norm": 4921.765003875099,
+      "learning_rate": 3.4476819779903694e-07,
+      "loss": 221.1381,
+      "step": 47680
+    },
+    {
+      "epoch": 0.9171991672316221,
+      "grad_norm": 1490.0607461887414,
+      "learning_rate": 3.43181243770222e-07,
+      "loss": 217.945,
+      "step": 47690
+    },
+    {
+      "epoch": 0.9173914924921026,
+      "grad_norm": 1404.0460318201362,
+      "learning_rate": 3.4159788674690386e-07,
+      "loss": 215.6464,
+      "step": 47700
+    },
+    {
+      "epoch": 0.9175838177525831,
+      "grad_norm": 1391.8318599368931,
+      "learning_rate": 3.4001812731886077e-07,
+      "loss": 212.6189,
+      "step": 47710
+    },
+    {
+      "epoch": 0.9177761430130636,
+      "grad_norm": 1269.2213887480323,
+      "learning_rate": 3.384419660745253e-07,
+      "loss": 214.7372,
+      "step": 47720
+    },
+    {
+      "epoch": 0.9179684682735442,
+      "grad_norm": 1576.799759515564,
+      "learning_rate": 3.368694036009923e-07,
+      "loss": 223.3609,
+      "step": 47730
+    },
+    {
+      "epoch": 0.9181607935340247,
+      "grad_norm": 1461.9010347646336,
+      "learning_rate": 3.353004404840121e-07,
+      "loss": 211.45,
+      "step": 47740
+    },
+    {
+      "epoch": 0.9183531187945053,
+      "grad_norm": 1559.459309411269,
+      "learning_rate": 3.3373507730800167e-07,
+      "loss": 230.0832,
+      "step": 47750
+    },
+    {
+      "epoch": 0.9185454440549858,
+      "grad_norm": 1600.8985215363884,
+      "learning_rate": 3.321733146560324e-07,
+      "loss": 219.0825,
+      "step": 47760
+    },
+    {
+      "epoch": 0.9187377693154664,
+      "grad_norm": 1585.9395832715118,
+      "learning_rate": 3.306151531098323e-07,
+      "loss": 223.2969,
+      "step": 47770
+    },
+    {
+      "epoch": 0.9189300945759469,
+      "grad_norm": 1438.8339673614244,
+      "learning_rate": 3.2906059324979255e-07,
+      "loss": 219.2484,
+      "step": 47780
+    },
+    {
+      "epoch": 0.9191224198364274,
+      "grad_norm": 1693.7520237797733,
+      "learning_rate": 3.275096356549612e-07,
+      "loss": 224.3883,
+      "step": 47790
+    },
+    {
+      "epoch": 0.9193147450969079,
+      "grad_norm": 1374.1466437365805,
+      "learning_rate": 3.2596228090304496e-07,
+      "loss": 211.2069,
+      "step": 47800
+    },
+    {
+      "epoch": 0.9195070703573884,
+      "grad_norm": 1444.7493813370381,
+      "learning_rate": 3.2441852957040607e-07,
+      "loss": 220.1731,
+      "step": 47810
+    },
+    {
+      "epoch": 0.919699395617869,
+      "grad_norm": 1301.6678714795885,
+      "learning_rate": 3.228783822320669e-07,
+      "loss": 217.9352,
+      "step": 47820
+    },
+    {
+      "epoch": 0.9198917208783495,
+      "grad_norm": 1299.3317831281151,
+      "learning_rate": 3.213418394617085e-07,
+      "loss": 219.5616,
+      "step": 47830
+    },
+    {
+      "epoch": 0.92008404613883,
+      "grad_norm": 1465.8789030895596,
+      "learning_rate": 3.1980890183166633e-07,
+      "loss": 218.4252,
+      "step": 47840
+    },
+    {
+      "epoch": 0.9202763713993105,
+      "grad_norm": 1416.2344950675404,
+      "learning_rate": 3.1827956991293374e-07,
+      "loss": 221.5462,
+      "step": 47850
+    },
+    {
+      "epoch": 0.920468696659791,
+      "grad_norm": 1458.0826619236723,
+      "learning_rate": 3.167538442751639e-07,
+      "loss": 221.894,
+      "step": 47860
+    },
+    {
+      "epoch": 0.9206610219202715,
+      "grad_norm": 1552.4803382268922,
+      "learning_rate": 3.1523172548666215e-07,
+      "loss": 220.0342,
+      "step": 47870
+    },
+    {
+      "epoch": 0.9208533471807521,
+      "grad_norm": 1411.753135894563,
+      "learning_rate": 3.1371321411439284e-07,
+      "loss": 225.0019,
+      "step": 47880
+    },
+    {
+      "epoch": 0.9210456724412326,
+      "grad_norm": 1430.238728284219,
+      "learning_rate": 3.1219831072397787e-07,
+      "loss": 221.1834,
+      "step": 47890
+    },
+    {
+      "epoch": 0.9212379977017131,
+      "grad_norm": 1304.7855657849027,
+      "learning_rate": 3.1068701587969375e-07,
+      "loss": 226.8632,
+      "step": 47900
+    },
+    {
+      "epoch": 0.9214303229621936,
+      "grad_norm": 1654.4673211647885,
+      "learning_rate": 3.091793301444701e-07,
+      "loss": 221.2661,
+      "step": 47910
+    },
+    {
+      "epoch": 0.9216226482226741,
+      "grad_norm": 1394.4288875986913,
+      "learning_rate": 3.076752540798977e-07,
+      "loss": 220.475,
+      "step": 47920
+    },
+    {
+      "epoch": 0.9218149734831547,
+      "grad_norm": 1431.1259954025106,
+      "learning_rate": 3.061747882462185e-07,
+      "loss": 222.9695,
+      "step": 47930
+    },
+    {
+      "epoch": 0.9220072987436352,
+      "grad_norm": 1375.24743677573,
+      "learning_rate": 3.0467793320233306e-07,
+      "loss": 219.8958,
+      "step": 47940
+    },
+    {
+      "epoch": 0.9221996240041158,
+      "grad_norm": 1342.3495925202565,
+      "learning_rate": 3.031846895057922e-07,
+      "loss": 222.7533,
+      "step": 47950
+    },
+    {
+      "epoch": 0.9223919492645963,
+      "grad_norm": 1403.0074302009762,
+      "learning_rate": 3.0169505771280747e-07,
+      "loss": 219.4715,
+      "step": 47960
+    },
+    {
+      "epoch": 0.9225842745250769,
+      "grad_norm": 1406.458761825228,
+      "learning_rate": 3.002090383782408e-07,
+      "loss": 215.6729,
+      "step": 47970
+    },
+    {
+      "epoch": 0.9227765997855574,
+      "grad_norm": 1592.4241206752758,
+      "learning_rate": 2.9872663205561035e-07,
+      "loss": 217.0189,
+      "step": 47980
+    },
+    {
+      "epoch": 0.9229689250460379,
+      "grad_norm": 1593.4532351647367,
+      "learning_rate": 2.972478392970857e-07,
+      "loss": 224.2339,
+      "step": 47990
+    },
+    {
+      "epoch": 0.9231612503065184,
+      "grad_norm": 1519.3356496559938,
+      "learning_rate": 2.9577266065349716e-07,
+      "loss": 217.2127,
+      "step": 48000
+    },
+    {
+      "epoch": 0.9233535755669989,
+      "grad_norm": 1345.652395432279,
+      "learning_rate": 2.9430109667432096e-07,
+      "loss": 219.529,
+      "step": 48010
+    },
+    {
+      "epoch": 0.9235459008274794,
+      "grad_norm": 1671.7656770264464,
+      "learning_rate": 2.9283314790769177e-07,
+      "loss": 226.2386,
+      "step": 48020
+    },
+    {
+      "epoch": 0.92373822608796,
+      "grad_norm": 1520.2096466099476,
+      "learning_rate": 2.913688149003946e-07,
+      "loss": 216.641,
+      "step": 48030
+    },
+    {
+      "epoch": 0.9239305513484405,
+      "grad_norm": 1501.7275251998974,
+      "learning_rate": 2.899080981978719e-07,
+      "loss": 225.2489,
+      "step": 48040
+    },
+    {
+      "epoch": 0.924122876608921,
+      "grad_norm": 1321.0058719172387,
+      "learning_rate": 2.8845099834421517e-07,
+      "loss": 213.8865,
+      "step": 48050
+    },
+    {
+      "epoch": 0.9243152018694015,
+      "grad_norm": 1397.7757444555393,
+      "learning_rate": 2.869975158821681e-07,
+      "loss": 218.6217,
+      "step": 48060
+    },
+    {
+      "epoch": 0.924507527129882,
+      "grad_norm": 1507.939502356167,
+      "learning_rate": 2.8554765135313303e-07,
+      "loss": 226.7254,
+      "step": 48070
+    },
+    {
+      "epoch": 0.9246998523903626,
+      "grad_norm": 1308.9016539988518,
+      "learning_rate": 2.8410140529715803e-07,
+      "loss": 214.2825,
+      "step": 48080
+    },
+    {
+      "epoch": 0.9248921776508431,
+      "grad_norm": 1540.1751435950482,
+      "learning_rate": 2.826587782529444e-07,
+      "loss": 212.4035,
+      "step": 48090
+    },
+    {
+      "epoch": 0.9250845029113236,
+      "grad_norm": 1544.0346071366935,
+      "learning_rate": 2.812197707578501e-07,
+      "loss": 223.1791,
+      "step": 48100
+    },
+    {
+      "epoch": 0.9252768281718041,
+      "grad_norm": 1426.831497021911,
+      "learning_rate": 2.797843833478797e-07,
+      "loss": 219.2706,
+      "step": 48110
+    },
+    {
+      "epoch": 0.9254691534322846,
+      "grad_norm": 1695.5684423480686,
+      "learning_rate": 2.7835261655769217e-07,
+      "loss": 214.1696,
+      "step": 48120
+    },
+    {
+      "epoch": 0.9256614786927652,
+      "grad_norm": 1536.0103384651316,
+      "learning_rate": 2.769244709205976e-07,
+      "loss": 218.4421,
+      "step": 48130
+    },
+    {
+      "epoch": 0.9258538039532457,
+      "grad_norm": 1345.711706299068,
+      "learning_rate": 2.7549994696855376e-07,
+      "loss": 214.9853,
+      "step": 48140
+    },
+    {
+      "epoch": 0.9260461292137262,
+      "grad_norm": 1340.9451295397134,
+      "learning_rate": 2.740790452321751e-07,
+      "loss": 212.6925,
+      "step": 48150
+    },
+    {
+      "epoch": 0.9262384544742068,
+      "grad_norm": 1328.2159913932562,
+      "learning_rate": 2.726617662407238e-07,
+      "loss": 214.3774,
+      "step": 48160
+    },
+    {
+      "epoch": 0.9264307797346873,
+      "grad_norm": 1383.3575376863405,
+      "learning_rate": 2.7124811052211097e-07,
+      "loss": 223.3681,
+      "step": 48170
+    },
+    {
+      "epoch": 0.9266231049951679,
+      "grad_norm": 1462.62844723085,
+      "learning_rate": 2.698380786029031e-07,
+      "loss": 215.6205,
+      "step": 48180
+    },
+    {
+      "epoch": 0.9268154302556484,
+      "grad_norm": 1527.7695244006295,
+      "learning_rate": 2.6843167100831125e-07,
+      "loss": 219.9917,
+      "step": 48190
+    },
+    {
+      "epoch": 0.9270077555161289,
+      "grad_norm": 1418.359165205742,
+      "learning_rate": 2.6702888826219965e-07,
+      "loss": 222.0716,
+      "step": 48200
+    },
+    {
+      "epoch": 0.9272000807766094,
+      "grad_norm": 1405.4897041005295,
+      "learning_rate": 2.6562973088708146e-07,
+      "loss": 218.1003,
+      "step": 48210
+    },
+    {
+      "epoch": 0.92739240603709,
+      "grad_norm": 1372.5701243633775,
+      "learning_rate": 2.6423419940412086e-07,
+      "loss": 215.0814,
+      "step": 48220
+    },
+    {
+      "epoch": 0.9275847312975705,
+      "grad_norm": 1368.4961564221253,
+      "learning_rate": 2.628422943331288e-07,
+      "loss": 223.2442,
+      "step": 48230
+    },
+    {
+      "epoch": 0.927777056558051,
+      "grad_norm": 1347.0173733412985,
+      "learning_rate": 2.614540161925683e-07,
+      "loss": 224.957,
+      "step": 48240
+    },
+    {
+      "epoch": 0.9279693818185315,
+      "grad_norm": 1384.9718141286414,
+      "learning_rate": 2.6006936549954784e-07,
+      "loss": 219.28,
+      "step": 48250
+    },
+    {
+      "epoch": 0.928161707079012,
+      "grad_norm": 1220.9125594825982,
+      "learning_rate": 2.5868834276983057e-07,
+      "loss": 223.2387,
+      "step": 48260
+    },
+    {
+      "epoch": 0.9283540323394925,
+      "grad_norm": 1342.7009983882829,
+      "learning_rate": 2.573109485178216e-07,
+      "loss": 224.9703,
+      "step": 48270
+    },
+    {
+      "epoch": 0.9285463575999731,
+      "grad_norm": 1354.4313956353342,
+      "learning_rate": 2.5593718325657713e-07,
+      "loss": 224.9315,
+      "step": 48280
+    },
+    {
+      "epoch": 0.9287386828604536,
+      "grad_norm": 1663.4861929053834,
+      "learning_rate": 2.545670474978057e-07,
+      "loss": 213.3882,
+      "step": 48290
+    },
+    {
+      "epoch": 0.9289310081209341,
+      "grad_norm": 1443.7121404565762,
+      "learning_rate": 2.532005417518568e-07,
+      "loss": 215.4642,
+      "step": 48300
+    },
+    {
+      "epoch": 0.9291233333814146,
+      "grad_norm": 1366.3725466033384,
+      "learning_rate": 2.5183766652773336e-07,
+      "loss": 216.153,
+      "step": 48310
+    },
+    {
+      "epoch": 0.9293156586418951,
+      "grad_norm": 1431.7955612902772,
+      "learning_rate": 2.504784223330814e-07,
+      "loss": 216.707,
+      "step": 48320
+    },
+    {
+      "epoch": 0.9295079839023757,
+      "grad_norm": 1549.894658463349,
+      "learning_rate": 2.4912280967419934e-07,
+      "loss": 221.1073,
+      "step": 48330
+    },
+    {
+      "epoch": 0.9297003091628562,
+      "grad_norm": 1456.2314518427997,
+      "learning_rate": 2.477708290560299e-07,
+      "loss": 229.2287,
+      "step": 48340
+    },
+    {
+      "epoch": 0.9298926344233367,
+      "grad_norm": 1357.6297752750288,
+      "learning_rate": 2.464224809821614e-07,
+      "loss": 230.4937,
+      "step": 48350
+    },
+    {
+      "epoch": 0.9300849596838173,
+      "grad_norm": 1427.1274384670082,
+      "learning_rate": 2.450777659548353e-07,
+      "loss": 222.4397,
+      "step": 48360
+    },
+    {
+      "epoch": 0.9302772849442978,
+      "grad_norm": 1493.3636922782707,
+      "learning_rate": 2.4373668447493225e-07,
+      "loss": 224.0513,
+      "step": 48370
+    },
+    {
+      "epoch": 0.9304696102047784,
+      "grad_norm": 1228.9819821562182,
+      "learning_rate": 2.4239923704198476e-07,
+      "loss": 216.6722,
+      "step": 48380
+    },
+    {
+      "epoch": 0.9306619354652589,
+      "grad_norm": 1361.5103097302067,
+      "learning_rate": 2.410654241541688e-07,
+      "loss": 215.3536,
+      "step": 48390
+    },
+    {
+      "epoch": 0.9308542607257394,
+      "grad_norm": 1432.3808865103306,
+      "learning_rate": 2.3973524630830804e-07,
+      "loss": 220.539,
+      "step": 48400
+    },
+    {
+      "epoch": 0.9310465859862199,
+      "grad_norm": 1488.1812837252164,
+      "learning_rate": 2.3840870399987283e-07,
+      "loss": 223.2664,
+      "step": 48410
+    },
+    {
+      "epoch": 0.9312389112467004,
+      "grad_norm": 1368.829747338738,
+      "learning_rate": 2.370857977229768e-07,
+      "loss": 223.0433,
+      "step": 48420
+    },
+    {
+      "epoch": 0.931431236507181,
+      "grad_norm": 1328.082648318461,
+      "learning_rate": 2.3576652797038247e-07,
+      "loss": 221.4879,
+      "step": 48430
+    },
+    {
+      "epoch": 0.9316235617676615,
+      "grad_norm": 1418.9465086124178,
+      "learning_rate": 2.344508952334934e-07,
+      "loss": 216.7115,
+      "step": 48440
+    },
+    {
+      "epoch": 0.931815887028142,
+      "grad_norm": 1302.754902764057,
+      "learning_rate": 2.3313890000236316e-07,
+      "loss": 214.921,
+      "step": 48450
+    },
+    {
+      "epoch": 0.9320082122886225,
+      "grad_norm": 1394.5322698155076,
+      "learning_rate": 2.3183054276568752e-07,
+      "loss": 218.6302,
+      "step": 48460
+    },
+    {
+      "epoch": 0.932200537549103,
+      "grad_norm": 1448.7660192805088,
+      "learning_rate": 2.305258240108077e-07,
+      "loss": 215.6557,
+      "step": 48470
+    },
+    {
+      "epoch": 0.9323928628095836,
+      "grad_norm": 1778.5654358557942,
+      "learning_rate": 2.2922474422371166e-07,
+      "loss": 222.8299,
+      "step": 48480
+    },
+    {
+      "epoch": 0.9325851880700641,
+      "grad_norm": 1276.2701348380415,
+      "learning_rate": 2.279273038890273e-07,
+      "loss": 222.6284,
+      "step": 48490
+    },
+    {
+      "epoch": 0.9327775133305446,
+      "grad_norm": 1447.3337888479798,
+      "learning_rate": 2.2663350349003134e-07,
+      "loss": 218.724,
+      "step": 48500
+    },
+    {
+      "epoch": 0.9329698385910251,
+      "grad_norm": 1366.2347467045965,
+      "learning_rate": 2.2534334350864274e-07,
+      "loss": 212.7785,
+      "step": 48510
+    },
+    {
+      "epoch": 0.9331621638515056,
+      "grad_norm": 1581.795152606201,
+      "learning_rate": 2.2405682442542487e-07,
+      "loss": 221.3937,
+      "step": 48520
+    },
+    {
+      "epoch": 0.9333544891119862,
+      "grad_norm": 1326.9763535990462,
+      "learning_rate": 2.2277394671958442e-07,
+      "loss": 220.8175,
+      "step": 48530
+    },
+    {
+      "epoch": 0.9335468143724667,
+      "grad_norm": 1462.1562738443135,
+      "learning_rate": 2.2149471086897355e-07,
+      "loss": 219.5209,
+      "step": 48540
+    },
+    {
+      "epoch": 0.9337391396329472,
+      "grad_norm": 1840.7755545042123,
+      "learning_rate": 2.202191173500845e-07,
+      "loss": 223.9876,
+      "step": 48550
+    },
+    {
+      "epoch": 0.9339314648934277,
+      "grad_norm": 1513.7450983054312,
+      "learning_rate": 2.1894716663805716e-07,
+      "loss": 222.8194,
+      "step": 48560
+    },
+    {
+      "epoch": 0.9341237901539083,
+      "grad_norm": 1512.0765462302688,
+      "learning_rate": 2.176788592066692e-07,
+      "loss": 231.463,
+      "step": 48570
+    },
+    {
+      "epoch": 0.9343161154143889,
+      "grad_norm": 1507.880890171871,
+      "learning_rate": 2.164141955283472e-07,
+      "loss": 213.6017,
+      "step": 48580
+    },
+    {
+      "epoch": 0.9345084406748694,
+      "grad_norm": 1382.4233741772143,
+      "learning_rate": 2.1515317607415654e-07,
+      "loss": 219.7649,
+      "step": 48590
+    },
+    {
+      "epoch": 0.9347007659353499,
+      "grad_norm": 1321.2779752271076,
+      "learning_rate": 2.1389580131380373e-07,
+      "loss": 234.4291,
+      "step": 48600
+    },
+    {
+      "epoch": 0.9348930911958304,
+      "grad_norm": 1604.1237040994413,
+      "learning_rate": 2.126420717156441e-07,
+      "loss": 229.7492,
+      "step": 48610
+    },
+    {
+      "epoch": 0.9350854164563109,
+      "grad_norm": 1507.0068110620718,
+      "learning_rate": 2.113919877466686e-07,
+      "loss": 215.8246,
+      "step": 48620
+    },
+    {
+      "epoch": 0.9352777417167915,
+      "grad_norm": 1894.812129250663,
+      "learning_rate": 2.1014554987251356e-07,
+      "loss": 226.4411,
+      "step": 48630
+    },
+    {
+      "epoch": 0.935470066977272,
+      "grad_norm": 1417.8296026083271,
+      "learning_rate": 2.0890275855745546e-07,
+      "loss": 218.8423,
+      "step": 48640
+    },
+    {
+      "epoch": 0.9356623922377525,
+      "grad_norm": 1351.992081181903,
+      "learning_rate": 2.0766361426441505e-07,
+      "loss": 215.6343,
+      "step": 48650
+    },
+    {
+      "epoch": 0.935854717498233,
+      "grad_norm": 1302.7594731055633,
+      "learning_rate": 2.0642811745495206e-07,
+      "loss": 226.6181,
+      "step": 48660
+    },
+    {
+      "epoch": 0.9360470427587135,
+      "grad_norm": 1498.4650782147276,
+      "learning_rate": 2.0519626858926944e-07,
+      "loss": 219.5578,
+      "step": 48670
+    },
+    {
+      "epoch": 0.936239368019194,
+      "grad_norm": 1413.5460913091172,
+      "learning_rate": 2.0396806812621018e-07,
+      "loss": 220.2387,
+      "step": 48680
+    },
+    {
+      "epoch": 0.9364316932796746,
+      "grad_norm": 1450.5595334943514,
+      "learning_rate": 2.0274351652325942e-07,
+      "loss": 227.4826,
+      "step": 48690
+    },
+    {
+      "epoch": 0.9366240185401551,
+      "grad_norm": 1403.7716562110738,
+      "learning_rate": 2.0152261423654118e-07,
+      "loss": 223.8041,
+      "step": 48700
+    },
+    {
+      "epoch": 0.9368163438006356,
+      "grad_norm": 1331.374860598774,
+      "learning_rate": 2.003053617208217e-07,
+      "loss": 217.9378,
+      "step": 48710
+    },
+    {
+      "epoch": 0.9370086690611161,
+      "grad_norm": 1479.9874435358035,
+      "learning_rate": 1.9909175942950832e-07,
+      "loss": 227.4229,
+      "step": 48720
+    },
+    {
+      "epoch": 0.9372009943215966,
+      "grad_norm": 1659.3378128005795,
+      "learning_rate": 1.9788180781464716e-07,
+      "loss": 222.5422,
+      "step": 48730
+    },
+    {
+      "epoch": 0.9373933195820772,
+      "grad_norm": 1343.347600477696,
+      "learning_rate": 1.9667550732692554e-07,
+      "loss": 222.2111,
+      "step": 48740
+    },
+    {
+      "epoch": 0.9375856448425577,
+      "grad_norm": 1318.6667266753677,
+      "learning_rate": 1.9547285841566843e-07,
+      "loss": 213.8635,
+      "step": 48750
+    },
+    {
+      "epoch": 0.9377779701030382,
+      "grad_norm": 1493.87287592487,
+      "learning_rate": 1.9427386152884643e-07,
+      "loss": 219.1087,
+      "step": 48760
+    },
+    {
+      "epoch": 0.9379702953635188,
+      "grad_norm": 1562.078925959731,
+      "learning_rate": 1.9307851711306336e-07,
+      "loss": 221.5375,
+      "step": 48770
+    },
+    {
+      "epoch": 0.9381626206239994,
+      "grad_norm": 1388.0220825267072,
+      "learning_rate": 1.918868256135653e-07,
+      "loss": 223.972,
+      "step": 48780
+    },
+    {
+      "epoch": 0.9383549458844799,
+      "grad_norm": 1330.1853997904634,
+      "learning_rate": 1.9069878747423943e-07,
+      "loss": 219.3253,
+      "step": 48790
+    },
+    {
+      "epoch": 0.9385472711449604,
+      "grad_norm": 1756.8756966376293,
+      "learning_rate": 1.8951440313760838e-07,
+      "loss": 220.7723,
+      "step": 48800
+    },
+    {
+      "epoch": 0.9387395964054409,
+      "grad_norm": 1356.9913556748725,
+      "learning_rate": 1.8833367304483708e-07,
+      "loss": 223.0061,
+      "step": 48810
+    },
+    {
+      "epoch": 0.9389319216659214,
+      "grad_norm": 1374.9040728679893,
+      "learning_rate": 1.8715659763572703e-07,
+      "loss": 218.9188,
+      "step": 48820
+    },
+    {
+      "epoch": 0.939124246926402,
+      "grad_norm": 1432.1827949199196,
+      "learning_rate": 1.8598317734872086e-07,
+      "loss": 217.1677,
+      "step": 48830
+    },
+    {
+      "epoch": 0.9393165721868825,
+      "grad_norm": 1493.0493848455733,
+      "learning_rate": 1.8481341262089668e-07,
+      "loss": 221.1158,
+      "step": 48840
+    },
+    {
+      "epoch": 0.939508897447363,
+      "grad_norm": 1675.6485118930636,
+      "learning_rate": 1.836473038879727e-07,
+      "loss": 215.9657,
+      "step": 48850
+    },
+    {
+      "epoch": 0.9397012227078435,
+      "grad_norm": 1651.7186047597081,
+      "learning_rate": 1.8248485158430696e-07,
+      "loss": 219.7294,
+      "step": 48860
+    },
+    {
+      "epoch": 0.939893547968324,
+      "grad_norm": 1461.0578275623395,
+      "learning_rate": 1.813260561428909e-07,
+      "loss": 218.0068,
+      "step": 48870
+    },
+    {
+      "epoch": 0.9400858732288045,
+      "grad_norm": 1476.2837872092398,
+      "learning_rate": 1.8017091799535924e-07,
+      "loss": 215.0431,
+      "step": 48880
+    },
+    {
+      "epoch": 0.9402781984892851,
+      "grad_norm": 1309.5088194323287,
+      "learning_rate": 1.7901943757198003e-07,
+      "loss": 214.9583,
+      "step": 48890
+    },
+    {
+      "epoch": 0.9404705237497656,
+      "grad_norm": 1481.5788926876676,
+      "learning_rate": 1.7787161530166242e-07,
+      "loss": 218.8679,
+      "step": 48900
+    },
+    {
+      "epoch": 0.9406628490102461,
+      "grad_norm": 1544.0380454810745,
+      "learning_rate": 1.7672745161194992e-07,
+      "loss": 222.1133,
+      "step": 48910
+    },
+    {
+      "epoch": 0.9408551742707266,
+      "grad_norm": 1321.5615668746314,
+      "learning_rate": 1.75586946929025e-07,
+      "loss": 218.2311,
+      "step": 48920
+    },
+    {
+      "epoch": 0.9410474995312071,
+      "grad_norm": 1384.2284116658982,
+      "learning_rate": 1.7445010167770672e-07,
+      "loss": 222.7567,
+      "step": 48930
+    },
+    {
+      "epoch": 0.9412398247916877,
+      "grad_norm": 1464.750193221155,
+      "learning_rate": 1.7331691628145076e-07,
+      "loss": 222.5026,
+      "step": 48940
+    },
+    {
+      "epoch": 0.9414321500521682,
+      "grad_norm": 1421.0093881818298,
+      "learning_rate": 1.7218739116235061e-07,
+      "loss": 220.2543,
+      "step": 48950
+    },
+    {
+      "epoch": 0.9416244753126487,
+      "grad_norm": 1384.542091524248,
+      "learning_rate": 1.710615267411353e-07,
+      "loss": 214.7182,
+      "step": 48960
+    },
+    {
+      "epoch": 0.9418168005731293,
+      "grad_norm": 1514.8554472681276,
+      "learning_rate": 1.6993932343717158e-07,
+      "loss": 220.3634,
+      "step": 48970
+    },
+    {
+      "epoch": 0.9420091258336099,
+      "grad_norm": 1354.5647171188343,
+      "learning_rate": 1.6882078166846173e-07,
+      "loss": 215.113,
+      "step": 48980
+    },
+    {
+      "epoch": 0.9422014510940904,
+      "grad_norm": 1455.2777205649381,
+      "learning_rate": 1.677059018516425e-07,
+      "loss": 215.3395,
+      "step": 48990
+    },
+    {
+      "epoch": 0.9423937763545709,
+      "grad_norm": 1448.5088923613425,
+      "learning_rate": 1.6659468440198835e-07,
+      "loss": 222.7378,
+      "step": 49000
+    },
+    {
+      "epoch": 0.9425861016150514,
+      "grad_norm": 1250.4333686386178,
+      "learning_rate": 1.6548712973341152e-07,
+      "loss": 230.9148,
+      "step": 49010
+    },
+    {
+      "epoch": 0.9427784268755319,
+      "grad_norm": 1381.3475013362543,
+      "learning_rate": 1.6438323825845647e-07,
+      "loss": 215.9438,
+      "step": 49020
+    },
+    {
+      "epoch": 0.9429707521360124,
+      "grad_norm": 1400.720680680042,
+      "learning_rate": 1.6328301038830429e-07,
+      "loss": 221.7146,
+      "step": 49030
+    },
+    {
+      "epoch": 0.943163077396493,
+      "grad_norm": 1479.3819464434112,
+      "learning_rate": 1.621864465327716e-07,
+      "loss": 212.319,
+      "step": 49040
+    },
+    {
+      "epoch": 0.9433554026569735,
+      "grad_norm": 1666.3186738187007,
+      "learning_rate": 1.6109354710031167e-07,
+      "loss": 223.9077,
+      "step": 49050
+    },
+    {
+      "epoch": 0.943547727917454,
+      "grad_norm": 1287.1865213646581,
+      "learning_rate": 1.6000431249800995e-07,
+      "loss": 209.4681,
+      "step": 49060
+    },
+    {
+      "epoch": 0.9437400531779345,
+      "grad_norm": 1260.8907458725478,
+      "learning_rate": 1.5891874313158862e-07,
+      "loss": 209.7787,
+      "step": 49070
+    },
+    {
+      "epoch": 0.943932378438415,
+      "grad_norm": 1465.5264985262763,
+      "learning_rate": 1.5783683940540528e-07,
+      "loss": 216.3579,
+      "step": 49080
+    },
+    {
+      "epoch": 0.9441247036988956,
+      "grad_norm": 1270.9833342140068,
+      "learning_rate": 1.5675860172244982e-07,
+      "loss": 214.5415,
+      "step": 49090
+    },
+    {
+      "epoch": 0.9443170289593761,
+      "grad_norm": 1376.786440317354,
+      "learning_rate": 1.5568403048434877e-07,
+      "loss": 214.4211,
+      "step": 49100
+    },
+    {
+      "epoch": 0.9445093542198566,
+      "grad_norm": 1270.0883767865007,
+      "learning_rate": 1.5461312609136192e-07,
+      "loss": 216.8254,
+      "step": 49110
+    },
+    {
+      "epoch": 0.9447016794803371,
+      "grad_norm": 1450.5672453293744,
+      "learning_rate": 1.5354588894238465e-07,
+      "loss": 214.8898,
+      "step": 49120
+    },
+    {
+      "epoch": 0.9448940047408176,
+      "grad_norm": 1375.048397461741,
+      "learning_rate": 1.524823194349434e-07,
+      "loss": 222.6178,
+      "step": 49130
+    },
+    {
+      "epoch": 0.9450863300012982,
+      "grad_norm": 1488.461636684655,
+      "learning_rate": 1.514224179652013e-07,
+      "loss": 223.175,
+      "step": 49140
+    },
+    {
+      "epoch": 0.9452786552617787,
+      "grad_norm": 1480.9839638843337,
+      "learning_rate": 1.5036618492795473e-07,
+      "loss": 221.1118,
+      "step": 49150
+    },
+    {
+      "epoch": 0.9454709805222592,
+      "grad_norm": 1382.4697674624388,
+      "learning_rate": 1.4931362071663125e-07,
+      "loss": 218.7365,
+      "step": 49160
+    },
+    {
+      "epoch": 0.9456633057827397,
+      "grad_norm": 1390.1083629149891,
+      "learning_rate": 1.4826472572329498e-07,
+      "loss": 217.7386,
+      "step": 49170
+    },
+    {
+      "epoch": 0.9458556310432203,
+      "grad_norm": 1424.8330751478002,
+      "learning_rate": 1.4721950033864118e-07,
+      "loss": 212.3467,
+      "step": 49180
+    },
+    {
+      "epoch": 0.9460479563037009,
+      "grad_norm": 1383.6156139819386,
+      "learning_rate": 1.4617794495199956e-07,
+      "loss": 221.9907,
+      "step": 49190
+    },
+    {
+      "epoch": 0.9462402815641814,
+      "grad_norm": 1454.9331898579517,
+      "learning_rate": 1.4514005995133197e-07,
+      "loss": 221.3145,
+      "step": 49200
+    },
+    {
+      "epoch": 0.9464326068246619,
+      "grad_norm": 1329.53101841995,
+      "learning_rate": 1.441058457232336e-07,
+      "loss": 215.3924,
+      "step": 49210
+    },
+    {
+      "epoch": 0.9466249320851424,
+      "grad_norm": 1438.784677106811,
+      "learning_rate": 1.43075302652933e-07,
+      "loss": 225.9471,
+      "step": 49220
+    },
+    {
+      "epoch": 0.9468172573456229,
+      "grad_norm": 1392.8093911436142,
+      "learning_rate": 1.4204843112428867e-07,
+      "loss": 220.0533,
+      "step": 49230
+    },
+    {
+      "epoch": 0.9470095826061035,
+      "grad_norm": 1331.6673038411307,
+      "learning_rate": 1.4102523151979574e-07,
+      "loss": 218.5267,
+      "step": 49240
+    },
+    {
+      "epoch": 0.947201907866584,
+      "grad_norm": 1642.9773984773065,
+      "learning_rate": 1.40005704220576e-07,
+      "loss": 222.6884,
+      "step": 49250
+    },
+    {
+      "epoch": 0.9473942331270645,
+      "grad_norm": 1387.9204088687911,
+      "learning_rate": 1.3898984960638907e-07,
+      "loss": 221.0935,
+      "step": 49260
+    },
+    {
+      "epoch": 0.947586558387545,
+      "grad_norm": 1563.2552699163398,
+      "learning_rate": 1.3797766805562328e-07,
+      "loss": 223.8792,
+      "step": 49270
+    },
+    {
+      "epoch": 0.9477788836480255,
+      "grad_norm": 1725.6842547242015,
+      "learning_rate": 1.3696915994530048e-07,
+      "loss": 220.3075,
+      "step": 49280
+    },
+    {
+      "epoch": 0.9479712089085061,
+      "grad_norm": 1351.9204479092296,
+      "learning_rate": 1.359643256510701e-07,
+      "loss": 214.8934,
+      "step": 49290
+    },
+    {
+      "epoch": 0.9481635341689866,
+      "grad_norm": 1372.4787109431147,
+      "learning_rate": 1.3496316554722056e-07,
+      "loss": 217.8872,
+      "step": 49300
+    },
+    {
+      "epoch": 0.9483558594294671,
+      "grad_norm": 1495.7625838423826,
+      "learning_rate": 1.3396568000666467e-07,
+      "loss": 220.4245,
+      "step": 49310
+    },
+    {
+      "epoch": 0.9485481846899476,
+      "grad_norm": 1428.9404245857595,
+      "learning_rate": 1.3297186940094853e-07,
+      "loss": 223.5978,
+      "step": 49320
+    },
+    {
+      "epoch": 0.9487405099504281,
+      "grad_norm": 1797.3066779888977,
+      "learning_rate": 1.3198173410025383e-07,
+      "loss": 216.9476,
+      "step": 49330
+    },
+    {
+      "epoch": 0.9489328352109087,
+      "grad_norm": 1563.3033791239825,
+      "learning_rate": 1.3099527447338668e-07,
+      "loss": 221.7208,
+      "step": 49340
+    },
+    {
+      "epoch": 0.9491251604713892,
+      "grad_norm": 1418.473510444605,
+      "learning_rate": 1.3001249088778756e-07,
+      "loss": 223.9783,
+      "step": 49350
+    },
+    {
+      "epoch": 0.9493174857318697,
+      "grad_norm": 1764.1922197600616,
+      "learning_rate": 1.29033383709527e-07,
+      "loss": 221.6277,
+      "step": 49360
+    },
+    {
+      "epoch": 0.9495098109923502,
+      "grad_norm": 1519.3160579705714,
+      "learning_rate": 1.2805795330330774e-07,
+      "loss": 222.519,
+      "step": 49370
+    },
+    {
+      "epoch": 0.9497021362528308,
+      "grad_norm": 1410.2806834320604,
+      "learning_rate": 1.2708620003245908e-07,
+      "loss": 227.1136,
+      "step": 49380
+    },
+    {
+      "epoch": 0.9498944615133114,
+      "grad_norm": 1384.442499752756,
+      "learning_rate": 1.2611812425894487e-07,
+      "loss": 211.1012,
+      "step": 49390
+    },
+    {
+      "epoch": 0.9500867867737919,
+      "grad_norm": 1449.078012356189,
+      "learning_rate": 1.2515372634335666e-07,
+      "loss": 215.1433,
+      "step": 49400
+    },
+    {
+      "epoch": 0.9502791120342724,
+      "grad_norm": 1525.3608856907586,
+      "learning_rate": 1.2419300664491707e-07,
+      "loss": 231.105,
+      "step": 49410
+    },
+    {
+      "epoch": 0.9504714372947529,
+      "grad_norm": 1274.9084907231727,
+      "learning_rate": 1.2323596552147876e-07,
+      "loss": 220.4786,
+      "step": 49420
+    },
+    {
+      "epoch": 0.9506637625552334,
+      "grad_norm": 1491.2090463219804,
+      "learning_rate": 1.2228260332952213e-07,
+      "loss": 219.7076,
+      "step": 49430
+    },
+    {
+      "epoch": 0.950856087815714,
+      "grad_norm": 1642.0579754967641,
+      "learning_rate": 1.213329204241609e-07,
+      "loss": 215.8749,
+      "step": 49440
+    },
+    {
+      "epoch": 0.9510484130761945,
+      "grad_norm": 1622.2449385890602,
+      "learning_rate": 1.203869171591343e-07,
+      "loss": 214.1462,
+      "step": 49450
+    },
+    {
+      "epoch": 0.951240738336675,
+      "grad_norm": 1639.916115340428,
+      "learning_rate": 1.1944459388681496e-07,
+      "loss": 227.6559,
+      "step": 49460
+    },
+    {
+      "epoch": 0.9514330635971555,
+      "grad_norm": 1531.8280036441413,
+      "learning_rate": 1.1850595095820095e-07,
+      "loss": 221.534,
+      "step": 49470
+    },
+    {
+      "epoch": 0.951625388857636,
+      "grad_norm": 1592.6006498171778,
+      "learning_rate": 1.175709887229215e-07,
+      "loss": 214.6332,
+      "step": 49480
+    },
+    {
+      "epoch": 0.9518177141181166,
+      "grad_norm": 1544.0915759183692,
+      "learning_rate": 1.1663970752923581e-07,
+      "loss": 217.407,
+      "step": 49490
+    },
+    {
+      "epoch": 0.9520100393785971,
+      "grad_norm": 1257.471276929645,
+      "learning_rate": 1.1571210772402975e-07,
+      "loss": 218.6383,
+      "step": 49500
+    },
+    {
+      "epoch": 0.9522023646390776,
+      "grad_norm": 1493.143269834361,
+      "learning_rate": 1.1478818965281912e-07,
+      "loss": 216.8622,
+      "step": 49510
+    },
+    {
+      "epoch": 0.9523946898995581,
+      "grad_norm": 1587.2592223772997,
+      "learning_rate": 1.1386795365974757e-07,
+      "loss": 225.2867,
+      "step": 49520
+    },
+    {
+      "epoch": 0.9525870151600386,
+      "grad_norm": 1436.2136052394492,
+      "learning_rate": 1.1295140008758864e-07,
+      "loss": 218.4989,
+      "step": 49530
+    },
+    {
+      "epoch": 0.9527793404205191,
+      "grad_norm": 1778.5153661587804,
+      "learning_rate": 1.1203852927774372e-07,
+      "loss": 222.3573,
+      "step": 49540
+    },
+    {
+      "epoch": 0.9529716656809997,
+      "grad_norm": 1596.3578970155165,
+      "learning_rate": 1.111293415702408e-07,
+      "loss": 217.9064,
+      "step": 49550
+    },
+    {
+      "epoch": 0.9531639909414802,
+      "grad_norm": 1488.349410158663,
+      "learning_rate": 1.1022383730373897e-07,
+      "loss": 222.3463,
+      "step": 49560
+    },
+    {
+      "epoch": 0.9533563162019607,
+      "grad_norm": 1365.3116597899987,
+      "learning_rate": 1.093220168155218e-07,
+      "loss": 218.0718,
+      "step": 49570
+    },
+    {
+      "epoch": 0.9535486414624412,
+      "grad_norm": 1472.6474533627631,
+      "learning_rate": 1.0842388044150387e-07,
+      "loss": 219.2332,
+      "step": 49580
+    },
+    {
+      "epoch": 0.9537409667229219,
+      "grad_norm": 1419.849759730337,
+      "learning_rate": 1.0752942851622649e-07,
+      "loss": 217.3345,
+      "step": 49590
+    },
+    {
+      "epoch": 0.9539332919834024,
+      "grad_norm": 1414.9693040770362,
+      "learning_rate": 1.066386613728565e-07,
+      "loss": 217.0496,
+      "step": 49600
+    },
+    {
+      "epoch": 0.9541256172438829,
+      "grad_norm": 1628.0844701745593,
+      "learning_rate": 1.0575157934319069e-07,
+      "loss": 223.2177,
+      "step": 49610
+    },
+    {
+      "epoch": 0.9543179425043634,
+      "grad_norm": 1542.2438900909433,
+      "learning_rate": 1.0486818275765364e-07,
+      "loss": 219.1934,
+      "step": 49620
+    },
+    {
+      "epoch": 0.9545102677648439,
+      "grad_norm": 1467.6851503594532,
+      "learning_rate": 1.0398847194529437e-07,
+      "loss": 221.6878,
+      "step": 49630
+    },
+    {
+      "epoch": 0.9547025930253245,
+      "grad_norm": 1500.0813827002046,
+      "learning_rate": 1.0311244723379188e-07,
+      "loss": 219.5877,
+      "step": 49640
+    },
+    {
+      "epoch": 0.954894918285805,
+      "grad_norm": 1397.1826024362097,
+      "learning_rate": 1.0224010894944958e-07,
+      "loss": 225.1783,
+      "step": 49650
+    },
+    {
+      "epoch": 0.9550872435462855,
+      "grad_norm": 1363.445633430055,
+      "learning_rate": 1.0137145741719867e-07,
+      "loss": 222.808,
+      "step": 49660
+    },
+    {
+      "epoch": 0.955279568806766,
+      "grad_norm": 1656.8560138722585,
+      "learning_rate": 1.0050649296060033e-07,
+      "loss": 234.056,
+      "step": 49670
+    },
+    {
+      "epoch": 0.9554718940672465,
+      "grad_norm": 1518.0234312138584,
+      "learning_rate": 9.964521590183684e-08,
+      "loss": 223.1471,
+      "step": 49680
+    },
+    {
+      "epoch": 0.955664219327727,
+      "grad_norm": 1549.3289411961346,
+      "learning_rate": 9.878762656172159e-08,
+      "loss": 219.831,
+      "step": 49690
+    },
+    {
+      "epoch": 0.9558565445882076,
+      "grad_norm": 1392.3647382184045,
+      "learning_rate": 9.793372525969125e-08,
+      "loss": 227.9785,
+      "step": 49700
+    },
+    {
+      "epoch": 0.9560488698486881,
+      "grad_norm": 1521.286342017299,
+      "learning_rate": 9.70835123138103e-08,
+      "loss": 221.397,
+      "step": 49710
+    },
+    {
+      "epoch": 0.9562411951091686,
+      "grad_norm": 1479.5330835399277,
+      "learning_rate": 9.623698804076875e-08,
+      "loss": 222.178,
+      "step": 49720
+    },
+    {
+      "epoch": 0.9564335203696491,
+      "grad_norm": 1476.4323700275932,
+      "learning_rate": 9.539415275588326e-08,
+      "loss": 229.1022,
+      "step": 49730
+    },
+    {
+      "epoch": 0.9566258456301296,
+      "grad_norm": 1321.8535674937316,
+      "learning_rate": 9.455500677309603e-08,
+      "loss": 236.7966,
+      "step": 49740
+    },
+    {
+      "epoch": 0.9568181708906102,
+      "grad_norm": 1387.817208223957,
+      "learning_rate": 9.371955040497371e-08,
+      "loss": 218.5549,
+      "step": 49750
+    },
+    {
+      "epoch": 0.9570104961510907,
+      "grad_norm": 1431.9921146805782,
+      "learning_rate": 9.288778396271292e-08,
+      "loss": 227.8062,
+      "step": 49760
+    },
+    {
+      "epoch": 0.9572028214115712,
+      "grad_norm": 1521.7189171140803,
+      "learning_rate": 9.205970775613027e-08,
+      "loss": 220.0308,
+      "step": 49770
+    },
+    {
+      "epoch": 0.9573951466720517,
+      "grad_norm": 1726.3466369243124,
+      "learning_rate": 9.123532209367237e-08,
+      "loss": 227.9465,
+      "step": 49780
+    },
+    {
+      "epoch": 0.9575874719325324,
+      "grad_norm": 1348.7248495836477,
+      "learning_rate": 9.04146272824069e-08,
+      "loss": 210.339,
+      "step": 49790
+    },
+    {
+      "epoch": 0.9577797971930129,
+      "grad_norm": 1356.720134180842,
+      "learning_rate": 8.959762362803159e-08,
+      "loss": 217.5549,
+      "step": 49800
+    },
+    {
+      "epoch": 0.9579721224534934,
+      "grad_norm": 1299.987739470492,
+      "learning_rate": 8.87843114348652e-08,
+      "loss": 216.9105,
+      "step": 49810
+    },
+    {
+      "epoch": 0.9581644477139739,
+      "grad_norm": 1488.546770358729,
+      "learning_rate": 8.797469100585432e-08,
+      "loss": 224.1143,
+      "step": 49820
+    },
+    {
+      "epoch": 0.9583567729744544,
+      "grad_norm": 1762.9370076163752,
+      "learning_rate": 8.716876264256768e-08,
+      "loss": 232.3192,
+      "step": 49830
+    },
+    {
+      "epoch": 0.958549098234935,
+      "grad_norm": 1372.8568412599043,
+      "learning_rate": 8.636652664520184e-08,
+      "loss": 220.5682,
+      "step": 49840
+    },
+    {
+      "epoch": 0.9587414234954155,
+      "grad_norm": 1458.5762699687893,
+      "learning_rate": 8.556798331257555e-08,
+      "loss": 222.2676,
+      "step": 49850
+    },
+    {
+      "epoch": 0.958933748755896,
+      "grad_norm": 1575.3787980322772,
+      "learning_rate": 8.477313294213307e-08,
+      "loss": 219.2771,
+      "step": 49860
+    },
+    {
+      "epoch": 0.9591260740163765,
+      "grad_norm": 1357.0175353841378,
+      "learning_rate": 8.398197582994316e-08,
+      "loss": 212.6595,
+      "step": 49870
+    },
+    {
+      "epoch": 0.959318399276857,
+      "grad_norm": 1429.2296523014513,
+      "learning_rate": 8.319451227069897e-08,
+      "loss": 220.4824,
+      "step": 49880
+    },
+    {
+      "epoch": 0.9595107245373375,
+      "grad_norm": 1393.5877537069612,
+      "learning_rate": 8.241074255771808e-08,
+      "loss": 215.5422,
+      "step": 49890
+    },
+    {
+      "epoch": 0.9597030497978181,
+      "grad_norm": 1344.6991732241456,
+      "learning_rate": 8.163066698294031e-08,
+      "loss": 220.2936,
+      "step": 49900
+    },
+    {
+      "epoch": 0.9598953750582986,
+      "grad_norm": 1361.6020886917945,
+      "learning_rate": 8.085428583693211e-08,
+      "loss": 217.6353,
+      "step": 49910
+    },
+    {
+      "epoch": 0.9600877003187791,
+      "grad_norm": 1354.6067794420978,
+      "learning_rate": 8.008159940888216e-08,
+      "loss": 218.937,
+      "step": 49920
+    },
+    {
+      "epoch": 0.9602800255792596,
+      "grad_norm": 1398.4695016785956,
+      "learning_rate": 7.931260798660356e-08,
+      "loss": 223.0757,
+      "step": 49930
+    },
+    {
+      "epoch": 0.9604723508397401,
+      "grad_norm": 1348.597500175215,
+      "learning_rate": 7.854731185653386e-08,
+      "loss": 214.6391,
+      "step": 49940
+    },
+    {
+      "epoch": 0.9606646761002207,
+      "grad_norm": 1353.433160214295,
+      "learning_rate": 7.778571130373059e-08,
+      "loss": 218.2911,
+      "step": 49950
+    },
+    {
+      "epoch": 0.9608570013607012,
+      "grad_norm": 1383.1705841048174,
+      "learning_rate": 7.702780661188014e-08,
+      "loss": 222.7571,
+      "step": 49960
+    },
+    {
+      "epoch": 0.9610493266211817,
+      "grad_norm": 1386.0999581259562,
+      "learning_rate": 7.627359806328782e-08,
+      "loss": 221.2229,
+      "step": 49970
+    },
+    {
+      "epoch": 0.9612416518816622,
+      "grad_norm": 1462.5113174928572,
+      "learning_rate": 7.552308593888558e-08,
+      "loss": 220.3561,
+      "step": 49980
+    },
+    {
+      "epoch": 0.9614339771421427,
+      "grad_norm": 1483.0226051717405,
+      "learning_rate": 7.477627051822534e-08,
+      "loss": 224.737,
+      "step": 49990
+    },
+    {
+      "epoch": 0.9616263024026234,
+      "grad_norm": 1592.7711262874652,
+      "learning_rate": 7.403315207948236e-08,
+      "loss": 230.2029,
+      "step": 50000
+    },
+    {
+      "epoch": 0.9618186276631039,
+      "grad_norm": 1581.955196026486,
+      "learning_rate": 7.329373089945968e-08,
+      "loss": 213.5711,
+      "step": 50010
+    },
+    {
+      "epoch": 0.9620109529235844,
+      "grad_norm": 1383.1895906350637,
+      "learning_rate": 7.255800725357586e-08,
+      "loss": 221.3688,
+      "step": 50020
+    },
+    {
+      "epoch": 0.9622032781840649,
+      "grad_norm": 1479.0640633664084,
+      "learning_rate": 7.182598141587838e-08,
+      "loss": 221.1271,
+      "step": 50030
+    },
+    {
+      "epoch": 0.9623956034445454,
+      "grad_norm": 1507.3299417270493,
+      "learning_rate": 7.109765365903243e-08,
+      "loss": 224.5815,
+      "step": 50040
+    },
+    {
+      "epoch": 0.962587928705026,
+      "grad_norm": 1359.2587288594768,
+      "learning_rate": 7.03730242543299e-08,
+      "loss": 221.5924,
+      "step": 50050
+    },
+    {
+      "epoch": 0.9627802539655065,
+      "grad_norm": 1293.2467829465634,
+      "learning_rate": 6.965209347168156e-08,
+      "loss": 215.362,
+      "step": 50060
+    },
+    {
+      "epoch": 0.962972579225987,
+      "grad_norm": 1339.3854791985657,
+      "learning_rate": 6.89348615796237e-08,
+      "loss": 215.3806,
+      "step": 50070
+    },
+    {
+      "epoch": 0.9631649044864675,
+      "grad_norm": 1346.8258537843412,
+      "learning_rate": 6.822132884531373e-08,
+      "loss": 216.8808,
+      "step": 50080
+    },
+    {
+      "epoch": 0.963357229746948,
+      "grad_norm": 1608.656832645142,
+      "learning_rate": 6.751149553452907e-08,
+      "loss": 224.8484,
+      "step": 50090
+    },
+    {
+      "epoch": 0.9635495550074286,
+      "grad_norm": 1347.115068805702,
+      "learning_rate": 6.680536191167263e-08,
+      "loss": 215.7307,
+      "step": 50100
+    },
+    {
+      "epoch": 0.9637418802679091,
+      "grad_norm": 1475.8750918225821,
+      "learning_rate": 6.610292823976628e-08,
+      "loss": 215.2222,
+      "step": 50110
+    },
+    {
+      "epoch": 0.9639342055283896,
+      "grad_norm": 1379.2208797851886,
+      "learning_rate": 6.540419478045623e-08,
+      "loss": 223.4813,
+      "step": 50120
+    },
+    {
+      "epoch": 0.9641265307888701,
+      "grad_norm": 1517.7156384600805,
+      "learning_rate": 6.470916179400765e-08,
+      "loss": 214.4612,
+      "step": 50130
+    },
+    {
+      "epoch": 0.9643188560493506,
+      "grad_norm": 1423.8184394785478,
+      "learning_rate": 6.401782953931013e-08,
+      "loss": 216.5052,
+      "step": 50140
+    },
+    {
+      "epoch": 0.9645111813098312,
+      "grad_norm": 1426.3021935973463,
+      "learning_rate": 6.33301982738721e-08,
+      "loss": 218.0288,
+      "step": 50150
+    },
+    {
+      "epoch": 0.9647035065703117,
+      "grad_norm": 1308.4833197316664,
+      "learning_rate": 6.264626825382647e-08,
+      "loss": 225.292,
+      "step": 50160
+    },
+    {
+      "epoch": 0.9648958318307922,
+      "grad_norm": 1428.450671731036,
+      "learning_rate": 6.196603973392501e-08,
+      "loss": 224.4039,
+      "step": 50170
+    },
+    {
+      "epoch": 0.9650881570912727,
+      "grad_norm": 1692.231727584345,
+      "learning_rate": 6.12895129675406e-08,
+      "loss": 214.8248,
+      "step": 50180
+    },
+    {
+      "epoch": 0.9652804823517532,
+      "grad_norm": 1456.3327960599509,
+      "learning_rate": 6.061668820667055e-08,
+      "loss": 222.6079,
+      "step": 50190
+    },
+    {
+      "epoch": 0.9654728076122339,
+      "grad_norm": 1503.3810416757115,
+      "learning_rate": 5.994756570192994e-08,
+      "loss": 225.3887,
+      "step": 50200
+    },
+    {
+      "epoch": 0.9656651328727144,
+      "grad_norm": 1447.2828900576167,
+      "learning_rate": 5.928214570255497e-08,
+      "loss": 211.5125,
+      "step": 50210
+    },
+    {
+      "epoch": 0.9658574581331949,
+      "grad_norm": 1482.458328963002,
+      "learning_rate": 5.862042845640403e-08,
+      "loss": 217.6564,
+      "step": 50220
+    },
+    {
+      "epoch": 0.9660497833936754,
+      "grad_norm": 1423.4256362289439,
+      "learning_rate": 5.796241420995663e-08,
+      "loss": 211.4614,
+      "step": 50230
+    },
+    {
+      "epoch": 0.9662421086541559,
+      "grad_norm": 1295.2582420324156,
+      "learning_rate": 5.730810320831226e-08,
+      "loss": 219.6026,
+      "step": 50240
+    },
+    {
+      "epoch": 0.9664344339146365,
+      "grad_norm": 1411.0831890260745,
+      "learning_rate": 5.66574956951893e-08,
+      "loss": 221.2823,
+      "step": 50250
+    },
+    {
+      "epoch": 0.966626759175117,
+      "grad_norm": 1963.7417916944628,
+      "learning_rate": 5.6010591912930565e-08,
+      "loss": 225.4074,
+      "step": 50260
+    },
+    {
+      "epoch": 0.9668190844355975,
+      "grad_norm": 1359.6122733778202,
+      "learning_rate": 5.5367392102495534e-08,
+      "loss": 220.1798,
+      "step": 50270
+    },
+    {
+      "epoch": 0.967011409696078,
+      "grad_norm": 1532.095801029173,
+      "learning_rate": 5.472789650346588e-08,
+      "loss": 222.7269,
+      "step": 50280
+    },
+    {
+      "epoch": 0.9672037349565585,
+      "grad_norm": 1508.0119451134963,
+      "learning_rate": 5.4092105354043304e-08,
+      "loss": 221.8856,
+      "step": 50290
+    },
+    {
+      "epoch": 0.967396060217039,
+      "grad_norm": 1378.122736390622,
+      "learning_rate": 5.346001889104946e-08,
+      "loss": 212.4824,
+      "step": 50300
+    },
+    {
+      "epoch": 0.9675883854775196,
+      "grad_norm": 1544.1114348919405,
+      "learning_rate": 5.2831637349926026e-08,
+      "loss": 217.1256,
+      "step": 50310
+    },
+    {
+      "epoch": 0.9677807107380001,
+      "grad_norm": 1522.8427160799283,
+      "learning_rate": 5.220696096473465e-08,
+      "loss": 215.729,
+      "step": 50320
+    },
+    {
+      "epoch": 0.9679730359984806,
+      "grad_norm": 1922.6298739447218,
+      "learning_rate": 5.1585989968157e-08,
+      "loss": 231.2357,
+      "step": 50330
+    },
+    {
+      "epoch": 0.9681653612589611,
+      "grad_norm": 1284.990428705182,
+      "learning_rate": 5.096872459149471e-08,
+      "loss": 218.1397,
+      "step": 50340
+    },
+    {
+      "epoch": 0.9683576865194417,
+      "grad_norm": 1531.5772526680107,
+      "learning_rate": 5.035516506466942e-08,
+      "loss": 229.0436,
+      "step": 50350
+    },
+    {
+      "epoch": 0.9685500117799222,
+      "grad_norm": 1787.2079843825363,
+      "learning_rate": 4.9745311616220535e-08,
+      "loss": 230.0955,
+      "step": 50360
+    },
+    {
+      "epoch": 0.9687423370404027,
+      "grad_norm": 1261.7052608247714,
+      "learning_rate": 4.913916447330858e-08,
+      "loss": 219.1985,
+      "step": 50370
+    },
+    {
+      "epoch": 0.9689346623008832,
+      "grad_norm": 1376.5097051181942,
+      "learning_rate": 4.853672386171515e-08,
+      "loss": 225.4804,
+      "step": 50380
+    },
+    {
+      "epoch": 0.9691269875613637,
+      "grad_norm": 1381.84413928358,
+      "learning_rate": 4.793799000583743e-08,
+      "loss": 218.5802,
+      "step": 50390
+    },
+    {
+      "epoch": 0.9693193128218442,
+      "grad_norm": 1638.0263616664317,
+      "learning_rate": 4.73429631286948e-08,
+      "loss": 218.6352,
+      "step": 50400
+    },
+    {
+      "epoch": 0.9695116380823249,
+      "grad_norm": 1578.8418036853782,
+      "learning_rate": 4.675164345192551e-08,
+      "loss": 212.6262,
+      "step": 50410
+    },
+    {
+      "epoch": 0.9697039633428054,
+      "grad_norm": 1371.1759514757368,
+      "learning_rate": 4.6164031195785606e-08,
+      "loss": 220.3333,
+      "step": 50420
+    },
+    {
+      "epoch": 0.9698962886032859,
+      "grad_norm": 1293.1056215628555,
+      "learning_rate": 4.558012657915112e-08,
+      "loss": 219.4111,
+      "step": 50430
+    },
+    {
+      "epoch": 0.9700886138637664,
+      "grad_norm": 1477.342646008764,
+      "learning_rate": 4.4999929819515844e-08,
+      "loss": 225.7875,
+      "step": 50440
+    },
+    {
+      "epoch": 0.970280939124247,
+      "grad_norm": 1624.6552431312757,
+      "learning_rate": 4.442344113299579e-08,
+      "loss": 220.0965,
+      "step": 50450
+    },
+    {
+      "epoch": 0.9704732643847275,
+      "grad_norm": 1349.1434146453648,
+      "learning_rate": 4.385066073432143e-08,
+      "loss": 213.9362,
+      "step": 50460
+    },
+    {
+      "epoch": 0.970665589645208,
+      "grad_norm": 1553.172914479967,
+      "learning_rate": 4.3281588836844303e-08,
+      "loss": 217.8453,
+      "step": 50470
+    },
+    {
+      "epoch": 0.9708579149056885,
+      "grad_norm": 1374.7270202951709,
+      "learning_rate": 4.271622565253486e-08,
+      "loss": 213.7431,
+      "step": 50480
+    },
+    {
+      "epoch": 0.971050240166169,
+      "grad_norm": 1481.7411706615144,
+      "learning_rate": 4.2154571391982425e-08,
+      "loss": 217.4573,
+      "step": 50490
+    },
+    {
+      "epoch": 0.9712425654266496,
+      "grad_norm": 1487.240944704676,
+      "learning_rate": 4.159662626439187e-08,
+      "loss": 222.2234,
+      "step": 50500
+    },
+    {
+      "epoch": 0.9714348906871301,
+      "grad_norm": 1733.2112032588002,
+      "learning_rate": 4.104239047758918e-08,
+      "loss": 223.9488,
+      "step": 50510
+    },
+    {
+      "epoch": 0.9716272159476106,
+      "grad_norm": 1390.704318018731,
+      "learning_rate": 4.049186423801921e-08,
+      "loss": 212.4791,
+      "step": 50520
+    },
+    {
+      "epoch": 0.9718195412080911,
+      "grad_norm": 1526.7482218638684,
+      "learning_rate": 3.994504775074237e-08,
+      "loss": 213.7319,
+      "step": 50530
+    },
+    {
+      "epoch": 0.9720118664685716,
+      "grad_norm": 1454.2356090929586,
+      "learning_rate": 3.9401941219440186e-08,
+      "loss": 223.8824,
+      "step": 50540
+    },
+    {
+      "epoch": 0.9722041917290521,
+      "grad_norm": 1348.955669625872,
+      "learning_rate": 3.8862544846409725e-08,
+      "loss": 217.2765,
+      "step": 50550
+    },
+    {
+      "epoch": 0.9723965169895327,
+      "grad_norm": 1516.5889832832156,
+      "learning_rate": 3.832685883256915e-08,
+      "loss": 221.4098,
+      "step": 50560
+    },
+    {
+      "epoch": 0.9725888422500132,
+      "grad_norm": 1351.37761180578,
+      "learning_rate": 3.7794883377449966e-08,
+      "loss": 215.5324,
+      "step": 50570
+    },
+    {
+      "epoch": 0.9727811675104937,
+      "grad_norm": 1290.0552578950126,
+      "learning_rate": 3.726661867920478e-08,
+      "loss": 218.6886,
+      "step": 50580
+    },
+    {
+      "epoch": 0.9729734927709742,
+      "grad_norm": 1478.3844432855553,
+      "learning_rate": 3.674206493460508e-08,
+      "loss": 217.3733,
+      "step": 50590
+    },
+    {
+      "epoch": 0.9731658180314547,
+      "grad_norm": 1329.3020218449783,
+      "learning_rate": 3.622122233903791e-08,
+      "loss": 216.0461,
+      "step": 50600
+    },
+    {
+      "epoch": 0.9733581432919354,
+      "grad_norm": 1598.638778495034,
+      "learning_rate": 3.5704091086508076e-08,
+      "loss": 223.5914,
+      "step": 50610
+    },
+    {
+      "epoch": 0.9735504685524159,
+      "grad_norm": 1501.707689531157,
+      "learning_rate": 3.519067136963705e-08,
+      "loss": 217.7369,
+      "step": 50620
+    },
+    {
+      "epoch": 0.9737427938128964,
+      "grad_norm": 1596.2884135282686,
+      "learning_rate": 3.468096337966853e-08,
+      "loss": 219.5405,
+      "step": 50630
+    },
+    {
+      "epoch": 0.9739351190733769,
+      "grad_norm": 1755.822185458902,
+      "learning_rate": 3.417496730645731e-08,
+      "loss": 219.8947,
+      "step": 50640
+    },
+    {
+      "epoch": 0.9741274443338575,
+      "grad_norm": 1368.126446089764,
+      "learning_rate": 3.3672683338480396e-08,
+      "loss": 221.7425,
+      "step": 50650
+    },
+    {
+      "epoch": 0.974319769594338,
+      "grad_norm": 1495.6109987868633,
+      "learning_rate": 3.317411166282813e-08,
+      "loss": 217.1732,
+      "step": 50660
+    },
+    {
+      "epoch": 0.9745120948548185,
+      "grad_norm": 1347.4630626852975,
+      "learning_rate": 3.2679252465213085e-08,
+      "loss": 219.2742,
+      "step": 50670
+    },
+    {
+      "epoch": 0.974704420115299,
+      "grad_norm": 1476.7083655899255,
+      "learning_rate": 3.218810592996113e-08,
+      "loss": 222.0374,
+      "step": 50680
+    },
+    {
+      "epoch": 0.9748967453757795,
+      "grad_norm": 1842.1941916982119,
+      "learning_rate": 3.170067224001483e-08,
+      "loss": 230.6348,
+      "step": 50690
+    },
+    {
+      "epoch": 0.97508907063626,
+      "grad_norm": 1307.7977738124846,
+      "learning_rate": 3.1216951576936714e-08,
+      "loss": 210.7078,
+      "step": 50700
+    },
+    {
+      "epoch": 0.9752813958967406,
+      "grad_norm": 1384.566276610812,
+      "learning_rate": 3.073694412090489e-08,
+      "loss": 216.715,
+      "step": 50710
+    },
+    {
+      "epoch": 0.9754737211572211,
+      "grad_norm": 1530.8551794044608,
+      "learning_rate": 3.026065005071188e-08,
+      "loss": 221.1893,
+      "step": 50720
+    },
+    {
+      "epoch": 0.9756660464177016,
+      "grad_norm": 1486.5955679800163,
+      "learning_rate": 2.9788069543772445e-08,
+      "loss": 224.9625,
+      "step": 50730
+    },
+    {
+      "epoch": 0.9758583716781821,
+      "grad_norm": 1327.569734407385,
+      "learning_rate": 2.9319202776113553e-08,
+      "loss": 216.0199,
+      "step": 50740
+    },
+    {
+      "epoch": 0.9760506969386626,
+      "grad_norm": 1385.0947376121192,
+      "learning_rate": 2.8854049922379946e-08,
+      "loss": 215.7292,
+      "step": 50750
+    },
+    {
+      "epoch": 0.9762430221991432,
+      "grad_norm": 1519.3898920981442,
+      "learning_rate": 2.839261115583303e-08,
+      "loss": 222.5011,
+      "step": 50760
+    },
+    {
+      "epoch": 0.9764353474596237,
+      "grad_norm": 1675.3882091554156,
+      "learning_rate": 2.793488664835309e-08,
+      "loss": 219.7468,
+      "step": 50770
+    },
+    {
+      "epoch": 0.9766276727201042,
+      "grad_norm": 1402.753448994212,
+      "learning_rate": 2.7480876570433746e-08,
+      "loss": 215.0043,
+      "step": 50780
+    },
+    {
+      "epoch": 0.9768199979805847,
+      "grad_norm": 1454.2800896410947,
+      "learning_rate": 2.7030581091186393e-08,
+      "loss": 224.7813,
+      "step": 50790
+    },
+    {
+      "epoch": 0.9770123232410652,
+      "grad_norm": 1436.5471290087116,
+      "learning_rate": 2.658400037833686e-08,
+      "loss": 216.8098,
+      "step": 50800
+    },
+    {
+      "epoch": 0.9772046485015459,
+      "grad_norm": 1354.6174654199986,
+      "learning_rate": 2.6141134598233197e-08,
+      "loss": 217.105,
+      "step": 50810
+    },
+    {
+      "epoch": 0.9773969737620264,
+      "grad_norm": 1810.9408837924473,
+      "learning_rate": 2.5701983915831232e-08,
+      "loss": 223.1486,
+      "step": 50820
+    },
+    {
+      "epoch": 0.9775892990225069,
+      "grad_norm": 1265.015592760325,
+      "learning_rate": 2.5266548494710108e-08,
+      "loss": 219.1568,
+      "step": 50830
+    },
+    {
+      "epoch": 0.9777816242829874,
+      "grad_norm": 1305.251664712247,
+      "learning_rate": 2.4834828497062315e-08,
+      "loss": 214.6504,
+      "step": 50840
+    },
+    {
+      "epoch": 0.977973949543468,
+      "grad_norm": 1358.9980310003082,
+      "learning_rate": 2.4406824083694768e-08,
+      "loss": 222.9667,
+      "step": 50850
+    },
+    {
+      "epoch": 0.9781662748039485,
+      "grad_norm": 1379.5327018798964,
+      "learning_rate": 2.398253541403217e-08,
+      "loss": 226.9672,
+      "step": 50860
+    },
+    {
+      "epoch": 0.978358600064429,
+      "grad_norm": 1369.5129330917473,
+      "learning_rate": 2.3561962646116988e-08,
+      "loss": 218.3629,
+      "step": 50870
+    },
+    {
+      "epoch": 0.9785509253249095,
+      "grad_norm": 1426.930977174834,
+      "learning_rate": 2.3145105936603906e-08,
+      "loss": 211.0509,
+      "step": 50880
+    },
+    {
+      "epoch": 0.97874325058539,
+      "grad_norm": 1644.633872634359,
+      "learning_rate": 2.273196544076539e-08,
+      "loss": 225.7178,
+      "step": 50890
+    },
+    {
+      "epoch": 0.9789355758458705,
+      "grad_norm": 1286.5058961571435,
+      "learning_rate": 2.2322541312490565e-08,
+      "loss": 218.7,
+      "step": 50900
+    },
+    {
+      "epoch": 0.9791279011063511,
+      "grad_norm": 1416.0769614789701,
+      "learning_rate": 2.1916833704281882e-08,
+      "loss": 230.007,
+      "step": 50910
+    },
+    {
+      "epoch": 0.9793202263668316,
+      "grad_norm": 1710.889599421853,
+      "learning_rate": 2.1514842767258458e-08,
+      "loss": 212.4472,
+      "step": 50920
+    },
+    {
+      "epoch": 0.9795125516273121,
+      "grad_norm": 1416.8537592236978,
+      "learning_rate": 2.1116568651156076e-08,
+      "loss": 223.9885,
+      "step": 50930
+    },
+    {
+      "epoch": 0.9797048768877926,
+      "grad_norm": 1442.4390641662933,
+      "learning_rate": 2.0722011504326066e-08,
+      "loss": 218.4397,
+      "step": 50940
+    },
+    {
+      "epoch": 0.9798972021482731,
+      "grad_norm": 1524.4886403365613,
+      "learning_rate": 2.033117147373309e-08,
+      "loss": 214.8394,
+      "step": 50950
+    },
+    {
+      "epoch": 0.9800895274087537,
+      "grad_norm": 1337.31676711825,
+      "learning_rate": 1.9944048704959583e-08,
+      "loss": 219.3454,
+      "step": 50960
+    },
+    {
+      "epoch": 0.9802818526692342,
+      "grad_norm": 1249.9829223873553,
+      "learning_rate": 1.956064334220131e-08,
+      "loss": 214.8916,
+      "step": 50970
+    },
+    {
+      "epoch": 0.9804741779297147,
+      "grad_norm": 1358.3243917619675,
+      "learning_rate": 1.9180955528270706e-08,
+      "loss": 223.45,
+      "step": 50980
+    },
+    {
+      "epoch": 0.9806665031901952,
+      "grad_norm": 1626.3532054361744,
+      "learning_rate": 1.8804985404595743e-08,
+      "loss": 223.7124,
+      "step": 50990
+    },
+    {
+      "epoch": 0.9808588284506757,
+      "grad_norm": 1278.4692814716411,
+      "learning_rate": 1.843273311121885e-08,
+      "loss": 214.2552,
+      "step": 51000
+    },
+    {
+      "epoch": 0.9810511537111563,
+      "grad_norm": 1370.3837110251345,
+      "learning_rate": 1.806419878679799e-08,
+      "loss": 223.5854,
+      "step": 51010
+    },
+    {
+      "epoch": 0.9812434789716369,
+      "grad_norm": 1303.690071878926,
+      "learning_rate": 1.7699382568605595e-08,
+      "loss": 214.8925,
+      "step": 51020
+    },
+    {
+      "epoch": 0.9814358042321174,
+      "grad_norm": 1389.2505052761317,
+      "learning_rate": 1.733828459253073e-08,
+      "loss": 221.8437,
+      "step": 51030
+    },
+    {
+      "epoch": 0.9816281294925979,
+      "grad_norm": 1393.448661121099,
+      "learning_rate": 1.6980904993075808e-08,
+      "loss": 218.3199,
+      "step": 51040
+    },
+    {
+      "epoch": 0.9818204547530784,
+      "grad_norm": 1417.387868844666,
+      "learning_rate": 1.6627243903357682e-08,
+      "loss": 227.3598,
+      "step": 51050
+    },
+    {
+      "epoch": 0.982012780013559,
+      "grad_norm": 1246.5118420206204,
+      "learning_rate": 1.6277301455110972e-08,
+      "loss": 217.5961,
+      "step": 51060
+    },
+    {
+      "epoch": 0.9822051052740395,
+      "grad_norm": 1411.0164990115627,
+      "learning_rate": 1.5931077778682524e-08,
+      "loss": 221.1215,
+      "step": 51070
+    },
+    {
+      "epoch": 0.98239743053452,
+      "grad_norm": 1592.835508119975,
+      "learning_rate": 1.5588573003035847e-08,
+      "loss": 221.5829,
+      "step": 51080
+    },
+    {
+      "epoch": 0.9825897557950005,
+      "grad_norm": 1573.403061390579,
+      "learning_rate": 1.5249787255747774e-08,
+      "loss": 222.5703,
+      "step": 51090
+    },
+    {
+      "epoch": 0.982782081055481,
+      "grad_norm": 1297.3036971801798,
+      "learning_rate": 1.4914720663009585e-08,
+      "loss": 216.8494,
+      "step": 51100
+    },
+    {
+      "epoch": 0.9829744063159616,
+      "grad_norm": 1548.7706593508806,
+      "learning_rate": 1.4583373349629226e-08,
+      "loss": 217.2335,
+      "step": 51110
+    },
+    {
+      "epoch": 0.9831667315764421,
+      "grad_norm": 1374.075364719324,
+      "learning_rate": 1.4255745439027968e-08,
+      "loss": 214.8763,
+      "step": 51120
+    },
+    {
+      "epoch": 0.9833590568369226,
+      "grad_norm": 1451.1855216050324,
+      "learning_rate": 1.3931837053241526e-08,
+      "loss": 222.3662,
+      "step": 51130
+    },
+    {
+      "epoch": 0.9835513820974031,
+      "grad_norm": 1428.063812888039,
+      "learning_rate": 1.3611648312920057e-08,
+      "loss": 223.8568,
+      "step": 51140
+    },
+    {
+      "epoch": 0.9837437073578836,
+      "grad_norm": 1485.2896133925717,
+      "learning_rate": 1.3295179337329267e-08,
+      "loss": 220.7917,
+      "step": 51150
+    },
+    {
+      "epoch": 0.9839360326183642,
+      "grad_norm": 1323.231150432954,
+      "learning_rate": 1.2982430244347088e-08,
+      "loss": 226.1825,
+      "step": 51160
+    },
+    {
+      "epoch": 0.9841283578788447,
+      "grad_norm": 1576.0469204973817,
+      "learning_rate": 1.2673401150470333e-08,
+      "loss": 224.9283,
+      "step": 51170
+    },
+    {
+      "epoch": 0.9843206831393252,
+      "grad_norm": 1489.4402814999119,
+      "learning_rate": 1.2368092170804702e-08,
+      "loss": 219.7486,
+      "step": 51180
+    },
+    {
+      "epoch": 0.9845130083998057,
+      "grad_norm": 1629.4088723652553,
+      "learning_rate": 1.2066503419073672e-08,
+      "loss": 219.009,
+      "step": 51190
+    },
+    {
+      "epoch": 0.9847053336602862,
+      "grad_norm": 1346.2065111979,
+      "learning_rate": 1.1768635007614049e-08,
+      "loss": 219.1161,
+      "step": 51200
+    },
+    {
+      "epoch": 0.9848976589207667,
+      "grad_norm": 1294.1348929093222,
+      "learning_rate": 1.1474487047375971e-08,
+      "loss": 218.8163,
+      "step": 51210
+    },
+    {
+      "epoch": 0.9850899841812474,
+      "grad_norm": 1999.0228280219796,
+      "learning_rate": 1.1184059647926238e-08,
+      "loss": 224.0086,
+      "step": 51220
+    },
+    {
+      "epoch": 0.9852823094417279,
+      "grad_norm": 1565.2317858960037,
+      "learning_rate": 1.0897352917443871e-08,
+      "loss": 224.2722,
+      "step": 51230
+    },
+    {
+      "epoch": 0.9854746347022084,
+      "grad_norm": 1431.285351349309,
+      "learning_rate": 1.0614366962721223e-08,
+      "loss": 222.1215,
+      "step": 51240
+    },
+    {
+      "epoch": 0.9856669599626889,
+      "grad_norm": 1307.262296502763,
+      "learning_rate": 1.033510188916731e-08,
+      "loss": 209.9831,
+      "step": 51250
+    },
+    {
+      "epoch": 0.9858592852231695,
+      "grad_norm": 1478.7529596040304,
+      "learning_rate": 1.0059557800802256e-08,
+      "loss": 218.4157,
+      "step": 51260
+    },
+    {
+      "epoch": 0.98605161048365,
+      "grad_norm": 1342.1128420376638,
+      "learning_rate": 9.78773480026396e-09,
+      "loss": 213.2213,
+      "step": 51270
+    },
+    {
+      "epoch": 0.9862439357441305,
+      "grad_norm": 1550.7535283321577,
+      "learning_rate": 9.519632988800321e-09,
+      "loss": 223.3799,
+      "step": 51280
+    },
+    {
+      "epoch": 0.986436261004611,
+      "grad_norm": 1330.8557496003698,
+      "learning_rate": 9.25525246627479e-09,
+      "loss": 222.1909,
+      "step": 51290
+    },
+    {
+      "epoch": 0.9866285862650915,
+      "grad_norm": 1487.4320419313735,
+      "learning_rate": 8.994593331165257e-09,
+      "loss": 224.6448,
+      "step": 51300
+    },
+    {
+      "epoch": 0.986820911525572,
+      "grad_norm": 1356.2045533333326,
+      "learning_rate": 8.737655680562951e-09,
+      "loss": 213.0719,
+      "step": 51310
+    },
+    {
+      "epoch": 0.9870132367860526,
+      "grad_norm": 1506.8613464237494,
+      "learning_rate": 8.484439610172424e-09,
+      "loss": 219.0308,
+      "step": 51320
+    },
+    {
+      "epoch": 0.9872055620465331,
+      "grad_norm": 1312.2588527689834,
+      "learning_rate": 8.234945214312673e-09,
+      "loss": 221.395,
+      "step": 51330
+    },
+    {
+      "epoch": 0.9873978873070136,
+      "grad_norm": 1349.6400604286332,
+      "learning_rate": 7.989172585917138e-09,
+      "loss": 221.0535,
+      "step": 51340
+    },
+    {
+      "epoch": 0.9875902125674941,
+      "grad_norm": 1431.5555892402901,
+      "learning_rate": 7.747121816530368e-09,
+      "loss": 218.7294,
+      "step": 51350
+    },
+    {
+      "epoch": 0.9877825378279746,
+      "grad_norm": 1389.3445511851914,
+      "learning_rate": 7.508792996313573e-09,
+      "loss": 222.0628,
+      "step": 51360
+    },
+    {
+      "epoch": 0.9879748630884552,
+      "grad_norm": 1359.3494700375438,
+      "learning_rate": 7.274186214040191e-09,
+      "loss": 215.3606,
+      "step": 51370
+    },
+    {
+      "epoch": 0.9881671883489357,
+      "grad_norm": 1401.8634448461696,
+      "learning_rate": 7.043301557096982e-09,
+      "loss": 216.2671,
+      "step": 51380
+    },
+    {
+      "epoch": 0.9883595136094162,
+      "grad_norm": 1405.5532361299563,
+      "learning_rate": 6.816139111484044e-09,
+      "loss": 217.8133,
+      "step": 51390
+    },
+    {
+      "epoch": 0.9885518388698967,
+      "grad_norm": 1344.8531263881591,
+      "learning_rate": 6.592698961818134e-09,
+      "loss": 221.153,
+      "step": 51400
+    },
+    {
+      "epoch": 0.9887441641303772,
+      "grad_norm": 1297.3355723262694,
+      "learning_rate": 6.37298119132379e-09,
+      "loss": 213.2182,
+      "step": 51410
+    },
+    {
+      "epoch": 0.9889364893908578,
+      "grad_norm": 1369.353306862403,
+      "learning_rate": 6.156985881844435e-09,
+      "loss": 216.4058,
+      "step": 51420
+    },
+    {
+      "epoch": 0.9891288146513384,
+      "grad_norm": 1468.7511880833326,
+      "learning_rate": 5.944713113833489e-09,
+      "loss": 218.5407,
+      "step": 51430
+    },
+    {
+      "epoch": 0.9893211399118189,
+      "grad_norm": 1278.996437151188,
+      "learning_rate": 5.736162966359926e-09,
+      "loss": 221.9043,
+      "step": 51440
+    },
+    {
+      "epoch": 0.9895134651722994,
+      "grad_norm": 1375.160320973759,
+      "learning_rate": 5.531335517104941e-09,
+      "loss": 216.5196,
+      "step": 51450
+    },
+    {
+      "epoch": 0.98970579043278,
+      "grad_norm": 1426.4116457371038,
+      "learning_rate": 5.3302308423641704e-09,
+      "loss": 217.575,
+      "step": 51460
+    },
+    {
+      "epoch": 0.9898981156932605,
+      "grad_norm": 1458.9021200607694,
+      "learning_rate": 5.132849017044361e-09,
+      "loss": 222.4083,
+      "step": 51470
+    },
+    {
+      "epoch": 0.990090440953741,
+      "grad_norm": 1694.4203660275061,
+      "learning_rate": 4.939190114666703e-09,
+      "loss": 226.9896,
+      "step": 51480
+    },
+    {
+      "epoch": 0.9902827662142215,
+      "grad_norm": 1391.9248289774064,
+      "learning_rate": 4.749254207367937e-09,
+      "loss": 215.2023,
+      "step": 51490
+    },
+    {
+      "epoch": 0.990475091474702,
+      "grad_norm": 1266.9777005646283,
+      "learning_rate": 4.563041365894805e-09,
+      "loss": 219.9194,
+      "step": 51500
+    },
+    {
+      "epoch": 0.9906674167351825,
+      "grad_norm": 1795.4415090775,
+      "learning_rate": 4.380551659608489e-09,
+      "loss": 221.6285,
+      "step": 51510
+    },
+    {
+      "epoch": 0.9908597419956631,
+      "grad_norm": 1365.928249355028,
+      "learning_rate": 4.2017851564835065e-09,
+      "loss": 218.8447,
+      "step": 51520
+    },
+    {
+      "epoch": 0.9910520672561436,
+      "grad_norm": 1462.1396743521134,
+      "learning_rate": 4.026741923107702e-09,
+      "loss": 214.7567,
+      "step": 51530
+    },
+    {
+      "epoch": 0.9912443925166241,
+      "grad_norm": 1550.7080398876637,
+      "learning_rate": 3.855422024681144e-09,
+      "loss": 224.2405,
+      "step": 51540
+    },
+    {
+      "epoch": 0.9914367177771046,
+      "grad_norm": 1191.5138169966979,
+      "learning_rate": 3.6878255250183406e-09,
+      "loss": 210.5956,
+      "step": 51550
+    },
+    {
+      "epoch": 0.9916290430375851,
+      "grad_norm": 1297.256225222978,
+      "learning_rate": 3.5239524865460227e-09,
+      "loss": 215.2229,
+      "step": 51560
+    },
+    {
+      "epoch": 0.9918213682980657,
+      "grad_norm": 1466.4899683458307,
+      "learning_rate": 3.363802970304253e-09,
+      "loss": 227.8157,
+      "step": 51570
+    },
+    {
+      "epoch": 0.9920136935585462,
+      "grad_norm": 1431.8244867733551,
+      "learning_rate": 3.207377035946424e-09,
+      "loss": 221.4954,
+      "step": 51580
+    },
+    {
+      "epoch": 0.9922060188190267,
+      "grad_norm": 1389.1053902441847,
+      "learning_rate": 3.05467474173704e-09,
+      "loss": 220.8963,
+      "step": 51590
+    },
+    {
+      "epoch": 0.9923983440795072,
+      "grad_norm": 1441.4437514690735,
+      "learning_rate": 2.9056961445572686e-09,
+      "loss": 219.463,
+      "step": 51600
+    },
+    {
+      "epoch": 0.9925906693399877,
+      "grad_norm": 1474.2013426381927,
+      "learning_rate": 2.7604412998982754e-09,
+      "loss": 220.8356,
+      "step": 51610
+    },
+    {
+      "epoch": 0.9927829946004683,
+      "grad_norm": 1449.1238090459485,
+      "learning_rate": 2.61891026186456e-09,
+      "loss": 216.6552,
+      "step": 51620
+    },
+    {
+      "epoch": 0.9929753198609489,
+      "grad_norm": 1408.3734534433443,
+      "learning_rate": 2.4811030831739525e-09,
+      "loss": 229.2871,
+      "step": 51630
+    },
+    {
+      "epoch": 0.9931676451214294,
+      "grad_norm": 1590.7466658228748,
+      "learning_rate": 2.347019815158724e-09,
+      "loss": 223.535,
+      "step": 51640
+    },
+    {
+      "epoch": 0.9933599703819099,
+      "grad_norm": 1562.5799645393317,
+      "learning_rate": 2.216660507762258e-09,
+      "loss": 220.5486,
+      "step": 51650
+    },
+    {
+      "epoch": 0.9935522956423904,
+      "grad_norm": 1454.2733748923456,
+      "learning_rate": 2.0900252095401583e-09,
+      "loss": 215.5121,
+      "step": 51660
+    },
+    {
+      "epoch": 0.993744620902871,
+      "grad_norm": 1603.3997597332047,
+      "learning_rate": 1.9671139676624707e-09,
+      "loss": 219.2345,
+      "step": 51670
+    },
+    {
+      "epoch": 0.9939369461633515,
+      "grad_norm": 1394.3368719356417,
+      "learning_rate": 1.8479268279125717e-09,
+      "loss": 220.9174,
+      "step": 51680
+    },
+    {
+      "epoch": 0.994129271423832,
+      "grad_norm": 1343.380943134161,
+      "learning_rate": 1.73246383468495e-09,
+      "loss": 213.827,
+      "step": 51690
+    },
+    {
+      "epoch": 0.9943215966843125,
+      "grad_norm": 1285.1485220669806,
+      "learning_rate": 1.6207250309874246e-09,
+      "loss": 218.4092,
+      "step": 51700
+    },
+    {
+      "epoch": 0.994513921944793,
+      "grad_norm": 1626.560462404415,
+      "learning_rate": 1.512710458442257e-09,
+      "loss": 223.7305,
+      "step": 51710
+    },
+    {
+      "epoch": 0.9947062472052736,
+      "grad_norm": 1407.782058737943,
+      "learning_rate": 1.408420157280599e-09,
+      "loss": 226.7854,
+      "step": 51720
+    },
+    {
+      "epoch": 0.9948985724657541,
+      "grad_norm": 1400.1296413481173,
+      "learning_rate": 1.3078541663502642e-09,
+      "loss": 222.8753,
+      "step": 51730
+    },
+    {
+      "epoch": 0.9950908977262346,
+      "grad_norm": 1475.3583524220107,
+      "learning_rate": 1.2110125231112879e-09,
+      "loss": 224.1618,
+      "step": 51740
+    },
+    {
+      "epoch": 0.9952832229867151,
+      "grad_norm": 1549.8737218502094,
+      "learning_rate": 1.117895263633706e-09,
+      "loss": 219.9759,
+      "step": 51750
+    },
+    {
+      "epoch": 0.9954755482471956,
+      "grad_norm": 1444.2045857010532,
+      "learning_rate": 1.0285024226042162e-09,
+      "loss": 220.9665,
+      "step": 51760
+    },
+    {
+      "epoch": 0.9956678735076762,
+      "grad_norm": 1298.734984710355,
+      "learning_rate": 9.428340333184072e-10,
+      "loss": 214.748,
+      "step": 51770
+    },
+    {
+      "epoch": 0.9958601987681567,
+      "grad_norm": 1504.4810221591138,
+      "learning_rate": 8.608901276874193e-10,
+      "loss": 229.3391,
+      "step": 51780
+    },
+    {
+      "epoch": 0.9960525240286372,
+      "grad_norm": 1548.7392350156351,
+      "learning_rate": 7.826707362335039e-10,
+      "loss": 224.7887,
+      "step": 51790
+    },
+    {
+      "epoch": 0.9962448492891177,
+      "grad_norm": 1462.720059086555,
+      "learning_rate": 7.081758880911338e-10,
+      "loss": 237.064,
+      "step": 51800
+    },
+    {
+      "epoch": 0.9964371745495982,
+      "grad_norm": 1332.2516227802994,
+      "learning_rate": 6.374056110103332e-10,
+      "loss": 214.0306,
+      "step": 51810
+    },
+    {
+      "epoch": 0.9966294998100788,
+      "grad_norm": 1514.7331971599397,
+      "learning_rate": 5.703599313511277e-10,
+      "loss": 217.9168,
+      "step": 51820
+    },
+    {
+      "epoch": 0.9968218250705593,
+      "grad_norm": 1407.0593314278713,
+      "learning_rate": 5.070388740868737e-10,
+      "loss": 221.0774,
+      "step": 51830
+    },
+    {
+      "epoch": 0.9970141503310399,
+      "grad_norm": 1473.4218750511973,
+      "learning_rate": 4.474424628031493e-10,
+      "loss": 221.3092,
+      "step": 51840
+    },
+    {
+      "epoch": 0.9972064755915204,
+      "grad_norm": 1258.9502542379898,
+      "learning_rate": 3.9157071969997407e-10,
+      "loss": 212.6899,
+      "step": 51850
+    },
+    {
+      "epoch": 0.9973988008520009,
+      "grad_norm": 1571.1567575357894,
+      "learning_rate": 3.394236655873684e-10,
+      "loss": 222.4162,
+      "step": 51860
+    },
+    {
+      "epoch": 0.9975911261124815,
+      "grad_norm": 1349.1337089835238,
+      "learning_rate": 2.910013198886841e-10,
+      "loss": 220.9971,
+      "step": 51870
+    },
+    {
+      "epoch": 0.997783451372962,
+      "grad_norm": 1501.929159723332,
+      "learning_rate": 2.4630370064171463e-10,
+      "loss": 221.988,
+      "step": 51880
+    },
+    {
+      "epoch": 0.9979757766334425,
+      "grad_norm": 1351.6320001798006,
+      "learning_rate": 2.0533082449647467e-10,
+      "loss": 217.8682,
+      "step": 51890
+    },
+    {
+      "epoch": 0.998168101893923,
+      "grad_norm": 1407.9931741017238,
+      "learning_rate": 1.6808270671186954e-10,
+      "loss": 226.0872,
+      "step": 51900
+    },
+    {
+      "epoch": 0.9983604271544035,
+      "grad_norm": 1743.3399384079214,
+      "learning_rate": 1.345593611645768e-10,
+      "loss": 215.9324,
+      "step": 51910
+    },
+    {
+      "epoch": 0.9985527524148841,
+      "grad_norm": 1551.584540557944,
+      "learning_rate": 1.0476080034016456e-10,
+      "loss": 228.4527,
+      "step": 51920
+    },
+    {
+      "epoch": 0.9987450776753646,
+      "grad_norm": 1521.5271051523334,
+      "learning_rate": 7.868703533864264e-11,
+      "loss": 220.933,
+      "step": 51930
+    },
+    {
+      "epoch": 0.9989374029358451,
+      "grad_norm": 1734.0650736528014,
+      "learning_rate": 5.633807587224205e-11,
+      "loss": 222.5149,
+      "step": 51940
+    },
+    {
+      "epoch": 0.9991297281963256,
+      "grad_norm": 1392.847156910018,
+      "learning_rate": 3.771393026541503e-11,
+      "loss": 217.0153,
+      "step": 51950
+    },
+    {
+      "epoch": 0.9993220534568061,
+      "grad_norm": 1324.0422130016884,
+      "learning_rate": 2.281460545594527e-11,
+      "loss": 215.6331,
+      "step": 51960
+    },
+    {
+      "epoch": 0.9995143787172867,
+      "grad_norm": 1394.8856969882577,
+      "learning_rate": 1.164010699272744e-11,
+      "loss": 217.6925,
+      "step": 51970
+    },
+    {
+      "epoch": 0.9997067039777672,
+      "grad_norm": 1348.2654629021827,
+      "learning_rate": 4.190439037987659e-12,
+      "loss": 213.5433,
+      "step": 51980
+    },
+    {
+      "epoch": 0.9998990292382477,
+      "grad_norm": 1423.559718683161,
+      "learning_rate": 4.656043661732668e-13,
+      "loss": 236.7,
+      "step": 51990
+    },
+    {
+      "epoch": 0.999995191868488,
+      "eval_loss": 268.4025573730469,
+      "eval_runtime": 161.3455,
+      "eval_samples_per_second": 4.314,
+      "eval_steps_per_second": 0.136,
+      "step": 51995
+    },
+    {
+      "epoch": 0.999995191868488,
+      "step": 51995,
+      "total_flos": 3.3727656990081024e+16,
+      "train_loss": 478.7030639354971,
+      "train_runtime": 1000248.8491,
+      "train_samples_per_second": 3.327,
+      "train_steps_per_second": 0.052
+    }
+  ],
+  "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": 3.3727656990081024e+16,
+  "train_batch_size": 2,
+  "trial_name": null,
+  "trial_params": null
+}