{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999896050976601,
  "eval_steps": 500,
  "global_step": 48100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 5.09375,
      "learning_rate": 1.2474012474012475e-08,
      "loss": 2.414,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.890625,
      "learning_rate": 1.2474012474012475e-07,
      "loss": 2.316,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.96875,
      "learning_rate": 2.494802494802495e-07,
      "loss": 2.3409,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.5,
      "learning_rate": 3.7422037422037426e-07,
      "loss": 2.3086,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.6875,
      "learning_rate": 4.98960498960499e-07,
      "loss": 2.3287,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.375,
      "learning_rate": 6.237006237006237e-07,
      "loss": 2.3292,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.09375,
      "learning_rate": 7.484407484407485e-07,
      "loss": 2.3661,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.875,
      "learning_rate": 8.731808731808732e-07,
      "loss": 2.3403,
      "step": 70
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.0625,
      "learning_rate": 9.97920997920998e-07,
      "loss": 2.3216,
      "step": 80
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.3125,
      "learning_rate": 1.1226611226611228e-06,
      "loss": 2.3108,
      "step": 90
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.34375,
      "learning_rate": 1.2474012474012474e-06,
      "loss": 2.3116,
      "step": 100
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.578125,
      "learning_rate": 1.3721413721413722e-06,
      "loss": 2.3412,
      "step": 110
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.109375,
      "learning_rate": 1.496881496881497e-06,
      "loss": 2.3043,
      "step": 120
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.15625,
      "learning_rate": 1.6216216216216219e-06,
      "loss": 2.2906,
      "step": 130
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8671875,
      "learning_rate": 1.7463617463617464e-06,
      "loss": 2.295,
      "step": 140
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3671875,
      "learning_rate": 1.8711018711018713e-06,
      "loss": 2.2948,
      "step": 150
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.65625,
      "learning_rate": 1.995841995841996e-06,
      "loss": 2.282,
      "step": 160
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.390625,
      "learning_rate": 2.120582120582121e-06,
      "loss": 2.3002,
      "step": 170
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.98828125,
      "learning_rate": 2.2453222453222457e-06,
      "loss": 2.2917,
      "step": 180
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.03125,
      "learning_rate": 2.37006237006237e-06,
      "loss": 2.2711,
      "step": 190
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.73046875,
      "learning_rate": 2.494802494802495e-06,
      "loss": 2.29,
      "step": 200
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.83203125,
      "learning_rate": 2.6195426195426197e-06,
      "loss": 2.2714,
      "step": 210
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.015625,
      "learning_rate": 2.7442827442827445e-06,
      "loss": 2.2805,
      "step": 220
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.67578125,
      "learning_rate": 2.8690228690228693e-06,
      "loss": 2.2621,
      "step": 230
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.62890625,
      "learning_rate": 2.993762993762994e-06,
      "loss": 2.2599,
      "step": 240
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5625,
      "learning_rate": 3.118503118503119e-06,
      "loss": 2.2501,
      "step": 250
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.67578125,
      "learning_rate": 3.2432432432432437e-06,
      "loss": 2.255,
      "step": 260
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.65234375,
      "learning_rate": 3.367983367983368e-06,
      "loss": 2.2622,
      "step": 270
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.58203125,
      "learning_rate": 3.492723492723493e-06,
      "loss": 2.302,
      "step": 280
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.65234375,
      "learning_rate": 3.6174636174636177e-06,
      "loss": 2.2871,
      "step": 290
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.50390625,
      "learning_rate": 3.7422037422037425e-06,
      "loss": 2.2389,
      "step": 300
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.51953125,
      "learning_rate": 3.866943866943867e-06,
      "loss": 2.251,
      "step": 310
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5234375,
      "learning_rate": 3.991683991683992e-06,
      "loss": 2.263,
      "step": 320
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.50390625,
      "learning_rate": 4.116424116424117e-06,
      "loss": 2.2557,
      "step": 330
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.53515625,
      "learning_rate": 4.241164241164242e-06,
      "loss": 2.2275,
      "step": 340
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.59765625,
      "learning_rate": 4.3659043659043665e-06,
      "loss": 2.2436,
      "step": 350
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.55078125,
      "learning_rate": 4.490644490644491e-06,
      "loss": 2.2556,
      "step": 360
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5,
      "learning_rate": 4.615384615384616e-06,
      "loss": 2.2675,
      "step": 370
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.515625,
      "learning_rate": 4.74012474012474e-06,
      "loss": 2.2425,
      "step": 380
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.59375,
      "learning_rate": 4.864864864864865e-06,
      "loss": 2.2649,
      "step": 390
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.55078125,
      "learning_rate": 4.98960498960499e-06,
      "loss": 2.2611,
      "step": 400
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.61328125,
      "learning_rate": 5.1143451143451145e-06,
      "loss": 2.2612,
      "step": 410
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.498046875,
      "learning_rate": 5.239085239085239e-06,
      "loss": 2.2397,
      "step": 420
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.65625,
      "learning_rate": 5.363825363825364e-06,
      "loss": 2.2806,
      "step": 430
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.53515625,
      "learning_rate": 5.488565488565489e-06,
      "loss": 2.25,
      "step": 440
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.474609375,
      "learning_rate": 5.613305613305614e-06,
      "loss": 2.2517,
      "step": 450
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.51953125,
      "learning_rate": 5.7380457380457386e-06,
      "loss": 2.2445,
      "step": 460
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.52734375,
      "learning_rate": 5.862785862785863e-06,
      "loss": 2.2713,
      "step": 470
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.478515625,
      "learning_rate": 5.987525987525988e-06,
      "loss": 2.253,
      "step": 480
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.51171875,
      "learning_rate": 6.112266112266113e-06,
      "loss": 2.2655,
      "step": 490
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.54296875,
      "learning_rate": 6.237006237006238e-06,
      "loss": 2.2432,
      "step": 500
    },
    {
      "epoch": 0.01,
      "eval_accuracy": 0.5358239483476676,
      "eval_loss": 2.1239066123962402,
      "eval_runtime": 16.4503,
      "eval_samples_per_second": 36.17,
      "eval_steps_per_second": 1.155,
      "step": 500
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.47265625,
      "learning_rate": 6.361746361746363e-06,
      "loss": 2.2518,
      "step": 510
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.50390625,
      "learning_rate": 6.486486486486487e-06,
      "loss": 2.2418,
      "step": 520
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5,
      "learning_rate": 6.611226611226611e-06,
      "loss": 2.2301,
      "step": 530
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.46484375,
      "learning_rate": 6.735966735966736e-06,
      "loss": 2.2528,
      "step": 540
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.470703125,
      "learning_rate": 6.860706860706861e-06,
      "loss": 2.277,
      "step": 550
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.48046875,
      "learning_rate": 6.985446985446986e-06,
      "loss": 2.2413,
      "step": 560
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.462890625,
      "learning_rate": 7.110187110187111e-06,
      "loss": 2.2362,
      "step": 570
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.50390625,
      "learning_rate": 7.234927234927235e-06,
      "loss": 2.2555,
      "step": 580
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.482421875,
      "learning_rate": 7.35966735966736e-06,
      "loss": 2.2618,
      "step": 590
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.48046875,
      "learning_rate": 7.484407484407485e-06,
      "loss": 2.2621,
      "step": 600
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.515625,
      "learning_rate": 7.60914760914761e-06,
      "loss": 2.2627,
      "step": 610
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5390625,
      "learning_rate": 7.733887733887735e-06,
      "loss": 2.2253,
      "step": 620
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.46875,
      "learning_rate": 7.85862785862786e-06,
      "loss": 2.2166,
      "step": 630
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5703125,
      "learning_rate": 7.983367983367984e-06,
      "loss": 2.2339,
      "step": 640
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.51953125,
      "learning_rate": 8.108108108108109e-06,
      "loss": 2.2368,
      "step": 650
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.4765625,
      "learning_rate": 8.232848232848234e-06,
      "loss": 2.2122,
      "step": 660
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.4609375,
      "learning_rate": 8.357588357588359e-06,
      "loss": 2.2463,
      "step": 670
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.455078125,
      "learning_rate": 8.482328482328483e-06,
      "loss": 2.1995,
      "step": 680
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.50390625,
      "learning_rate": 8.607068607068608e-06,
      "loss": 2.2342,
      "step": 690
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.45703125,
      "learning_rate": 8.731808731808733e-06,
      "loss": 2.2459,
      "step": 700
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5078125,
      "learning_rate": 8.856548856548858e-06,
      "loss": 2.2487,
      "step": 710
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6171875,
      "learning_rate": 8.981288981288983e-06,
      "loss": 2.2721,
      "step": 720
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.478515625,
      "learning_rate": 9.106029106029107e-06,
      "loss": 2.2112,
      "step": 730
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.46484375,
      "learning_rate": 9.230769230769232e-06,
      "loss": 2.2318,
      "step": 740
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.447265625,
      "learning_rate": 9.355509355509355e-06,
      "loss": 2.2131,
      "step": 750
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4609375,
      "learning_rate": 9.48024948024948e-06,
      "loss": 2.2346,
      "step": 760
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.455078125,
      "learning_rate": 9.604989604989605e-06,
      "loss": 2.2356,
      "step": 770
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.50390625,
      "learning_rate": 9.72972972972973e-06,
      "loss": 2.2433,
      "step": 780
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.52734375,
      "learning_rate": 9.854469854469855e-06,
      "loss": 2.2118,
      "step": 790
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.48046875,
      "learning_rate": 9.97920997920998e-06,
      "loss": 2.2434,
      "step": 800
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.50390625,
      "learning_rate": 1.0103950103950104e-05,
      "loss": 2.2281,
      "step": 810
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5546875,
      "learning_rate": 1.0228690228690229e-05,
      "loss": 2.2246,
      "step": 820
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4921875,
      "learning_rate": 1.0353430353430354e-05,
      "loss": 2.2342,
      "step": 830
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.458984375,
      "learning_rate": 1.0478170478170479e-05,
      "loss": 2.2294,
      "step": 840
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.482421875,
      "learning_rate": 1.0602910602910603e-05,
      "loss": 2.2337,
      "step": 850
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.484375,
      "learning_rate": 1.0727650727650728e-05,
      "loss": 2.2471,
      "step": 860
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.46484375,
      "learning_rate": 1.0852390852390853e-05,
      "loss": 2.2424,
      "step": 870
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4375,
      "learning_rate": 1.0977130977130978e-05,
      "loss": 2.2182,
      "step": 880
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.46484375,
      "learning_rate": 1.1101871101871103e-05,
      "loss": 2.2314,
      "step": 890
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.46484375,
      "learning_rate": 1.1226611226611228e-05,
      "loss": 2.2541,
      "step": 900
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.451171875,
      "learning_rate": 1.1351351351351352e-05,
      "loss": 2.2618,
      "step": 910
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.451171875,
      "learning_rate": 1.1476091476091477e-05,
      "loss": 2.2213,
      "step": 920
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.46875,
      "learning_rate": 1.1600831600831602e-05,
      "loss": 2.2102,
      "step": 930
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.49609375,
      "learning_rate": 1.1725571725571727e-05,
      "loss": 2.2121,
      "step": 940
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.484375,
      "learning_rate": 1.1850311850311852e-05,
      "loss": 2.2016,
      "step": 950
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.50390625,
      "learning_rate": 1.1975051975051976e-05,
      "loss": 2.2033,
      "step": 960
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.515625,
      "learning_rate": 1.2099792099792101e-05,
      "loss": 2.2137,
      "step": 970
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.54296875,
      "learning_rate": 1.2224532224532226e-05,
      "loss": 2.1972,
      "step": 980
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.455078125,
      "learning_rate": 1.234927234927235e-05,
      "loss": 2.2055,
      "step": 990
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4609375,
      "learning_rate": 1.2474012474012476e-05,
      "loss": 2.209,
      "step": 1000
    },
    {
      "epoch": 0.02,
      "eval_accuracy": 0.5404174573055028,
      "eval_loss": 2.0921764373779297,
      "eval_runtime": 16.4273,
      "eval_samples_per_second": 36.22,
      "eval_steps_per_second": 1.157,
      "step": 1000
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.462890625,
      "learning_rate": 1.25987525987526e-05,
      "loss": 2.2232,
      "step": 1010
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.484375,
      "learning_rate": 1.2723492723492725e-05,
      "loss": 2.2188,
      "step": 1020
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4765625,
      "learning_rate": 1.284823284823285e-05,
      "loss": 2.2393,
      "step": 1030
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.498046875,
      "learning_rate": 1.2972972972972975e-05,
      "loss": 2.2115,
      "step": 1040
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.462890625,
      "learning_rate": 1.3097713097713098e-05,
      "loss": 2.2207,
      "step": 1050
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.466796875,
      "learning_rate": 1.3222453222453223e-05,
      "loss": 2.2665,
      "step": 1060
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5234375,
      "learning_rate": 1.3347193347193348e-05,
      "loss": 2.2192,
      "step": 1070
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.546875,
      "learning_rate": 1.3471933471933472e-05,
      "loss": 2.246,
      "step": 1080
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.4453125,
      "learning_rate": 1.3596673596673597e-05,
      "loss": 2.2309,
      "step": 1090
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.48828125,
      "learning_rate": 1.3721413721413722e-05,
      "loss": 2.2237,
      "step": 1100
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.44921875,
      "learning_rate": 1.3846153846153847e-05,
      "loss": 2.2247,
      "step": 1110
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.61328125,
      "learning_rate": 1.3970893970893972e-05,
      "loss": 2.2323,
      "step": 1120
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.470703125,
      "learning_rate": 1.4095634095634096e-05,
      "loss": 2.2141,
      "step": 1130
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.466796875,
      "learning_rate": 1.4220374220374221e-05,
      "loss": 2.239,
      "step": 1140
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.48046875,
      "learning_rate": 1.4345114345114346e-05,
      "loss": 2.2332,
      "step": 1150
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.470703125,
      "learning_rate": 1.446985446985447e-05,
      "loss": 2.1918,
      "step": 1160
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.50390625,
      "learning_rate": 1.4594594594594596e-05,
      "loss": 2.1926,
      "step": 1170
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.49609375,
      "learning_rate": 1.471933471933472e-05,
      "loss": 2.2116,
      "step": 1180
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6328125,
      "learning_rate": 1.4844074844074845e-05,
      "loss": 2.2029,
      "step": 1190
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.470703125,
      "learning_rate": 1.496881496881497e-05,
      "loss": 2.2405,
      "step": 1200
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.439453125,
      "learning_rate": 1.5093555093555095e-05,
      "loss": 2.2376,
      "step": 1210
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.466796875,
      "learning_rate": 1.521829521829522e-05,
      "loss": 2.2264,
      "step": 1220
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47265625,
      "learning_rate": 1.5343035343035344e-05,
      "loss": 2.213,
      "step": 1230
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.439453125,
      "learning_rate": 1.546777546777547e-05,
      "loss": 2.2164,
      "step": 1240
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.439453125,
      "learning_rate": 1.5592515592515594e-05,
      "loss": 2.2008,
      "step": 1250
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4921875,
      "learning_rate": 1.571725571725572e-05,
      "loss": 2.2186,
      "step": 1260
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4609375,
      "learning_rate": 1.5841995841995844e-05,
      "loss": 2.1956,
      "step": 1270
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.462890625,
      "learning_rate": 1.596673596673597e-05,
      "loss": 2.2043,
      "step": 1280
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.498046875,
      "learning_rate": 1.6091476091476093e-05,
      "loss": 2.2153,
      "step": 1290
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.482421875,
      "learning_rate": 1.6216216216216218e-05,
      "loss": 2.2334,
      "step": 1300
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.484375,
      "learning_rate": 1.6340956340956343e-05,
      "loss": 2.224,
      "step": 1310
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4609375,
      "learning_rate": 1.6465696465696468e-05,
      "loss": 2.1958,
      "step": 1320
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.474609375,
      "learning_rate": 1.6590436590436593e-05,
      "loss": 2.1913,
      "step": 1330
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.47265625,
      "learning_rate": 1.6715176715176717e-05,
      "loss": 2.2135,
      "step": 1340
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.478515625,
      "learning_rate": 1.6839916839916842e-05,
      "loss": 2.1942,
      "step": 1350
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.462890625,
      "learning_rate": 1.6964656964656967e-05,
      "loss": 2.1935,
      "step": 1360
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.470703125,
      "learning_rate": 1.7089397089397092e-05,
      "loss": 2.196,
      "step": 1370
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4375,
      "learning_rate": 1.7214137214137217e-05,
      "loss": 2.2106,
      "step": 1380
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.478515625,
      "learning_rate": 1.733887733887734e-05,
      "loss": 2.191,
      "step": 1390
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4453125,
      "learning_rate": 1.7463617463617466e-05,
      "loss": 2.2065,
      "step": 1400
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46875,
      "learning_rate": 1.758835758835759e-05,
      "loss": 2.1872,
      "step": 1410
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.48046875,
      "learning_rate": 1.7713097713097716e-05,
      "loss": 2.2071,
      "step": 1420
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.455078125,
      "learning_rate": 1.783783783783784e-05,
      "loss": 2.1725,
      "step": 1430
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.44140625,
      "learning_rate": 1.7962577962577965e-05,
      "loss": 2.203,
      "step": 1440
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.546875,
      "learning_rate": 1.808731808731809e-05,
      "loss": 2.2122,
      "step": 1450
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.474609375,
      "learning_rate": 1.8212058212058215e-05,
      "loss": 2.1848,
      "step": 1460
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.45703125,
      "learning_rate": 1.833679833679834e-05,
      "loss": 2.1927,
      "step": 1470
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.498046875,
      "learning_rate": 1.8461538461538465e-05,
      "loss": 2.2109,
      "step": 1480
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.490234375,
      "learning_rate": 1.858627858627859e-05,
      "loss": 2.2122,
      "step": 1490
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.466796875,
      "learning_rate": 1.871101871101871e-05,
      "loss": 2.1988,
      "step": 1500
    },
    {
      "epoch": 0.03,
      "eval_accuracy": 0.5435980843950483,
      "eval_loss": 2.074204683303833,
      "eval_runtime": 16.4328,
      "eval_samples_per_second": 36.208,
      "eval_steps_per_second": 1.156,
      "step": 1500
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5078125,
      "learning_rate": 1.8835758835758836e-05,
      "loss": 2.1792,
      "step": 1510
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.494140625,
      "learning_rate": 1.896049896049896e-05,
      "loss": 2.1899,
      "step": 1520
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.458984375,
      "learning_rate": 1.9085239085239085e-05,
      "loss": 2.2182,
      "step": 1530
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.462890625,
      "learning_rate": 1.920997920997921e-05,
      "loss": 2.2034,
      "step": 1540
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.466796875,
      "learning_rate": 1.9334719334719335e-05,
      "loss": 2.2021,
      "step": 1550
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.462890625,
      "learning_rate": 1.945945945945946e-05,
      "loss": 2.2105,
      "step": 1560
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46484375,
      "learning_rate": 1.9584199584199584e-05,
      "loss": 2.2134,
      "step": 1570
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.470703125,
      "learning_rate": 1.970893970893971e-05,
      "loss": 2.237,
      "step": 1580
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.455078125,
      "learning_rate": 1.9833679833679834e-05,
      "loss": 2.1847,
      "step": 1590
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.484375,
      "learning_rate": 1.995841995841996e-05,
      "loss": 2.2396,
      "step": 1600
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.451171875,
      "learning_rate": 2.0083160083160084e-05,
      "loss": 2.2503,
      "step": 1610
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.447265625,
      "learning_rate": 2.020790020790021e-05,
      "loss": 2.185,
      "step": 1620
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46484375,
      "learning_rate": 2.0332640332640333e-05,
      "loss": 2.2091,
      "step": 1630
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.462890625,
      "learning_rate": 2.0457380457380458e-05,
      "loss": 2.1969,
      "step": 1640
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46875,
      "learning_rate": 2.0582120582120583e-05,
      "loss": 2.2218,
      "step": 1650
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.447265625,
      "learning_rate": 2.0706860706860708e-05,
      "loss": 2.1999,
      "step": 1660
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.486328125,
      "learning_rate": 2.0831600831600833e-05,
      "loss": 2.213,
      "step": 1670
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.46484375,
      "learning_rate": 2.0956340956340957e-05,
      "loss": 2.207,
      "step": 1680
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4453125,
      "learning_rate": 2.1081081081081082e-05,
      "loss": 2.1633,
      "step": 1690
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.453125,
      "learning_rate": 2.1205821205821207e-05,
      "loss": 2.2196,
      "step": 1700
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5,
      "learning_rate": 2.1330561330561332e-05,
      "loss": 2.2069,
      "step": 1710
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.447265625,
      "learning_rate": 2.1455301455301457e-05,
      "loss": 2.2024,
      "step": 1720
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4609375,
      "learning_rate": 2.158004158004158e-05,
      "loss": 2.2111,
      "step": 1730
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.474609375,
      "learning_rate": 2.1704781704781706e-05,
      "loss": 2.2228,
      "step": 1740
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4609375,
      "learning_rate": 2.182952182952183e-05,
      "loss": 2.2005,
      "step": 1750
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.48046875,
      "learning_rate": 2.1954261954261956e-05,
      "loss": 2.1753,
      "step": 1760
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.439453125,
      "learning_rate": 2.207900207900208e-05,
      "loss": 2.1739,
      "step": 1770
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4453125,
      "learning_rate": 2.2203742203742205e-05,
      "loss": 2.2136,
      "step": 1780
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.453125,
      "learning_rate": 2.232848232848233e-05,
      "loss": 2.1799,
      "step": 1790
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.462890625,
      "learning_rate": 2.2453222453222455e-05,
      "loss": 2.1874,
      "step": 1800
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5,
      "learning_rate": 2.257796257796258e-05,
      "loss": 2.1957,
      "step": 1810
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4609375,
      "learning_rate": 2.2702702702702705e-05,
      "loss": 2.19,
      "step": 1820
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.462890625,
      "learning_rate": 2.282744282744283e-05,
      "loss": 2.2404,
      "step": 1830
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44140625,
      "learning_rate": 2.2952182952182954e-05,
      "loss": 2.2362,
      "step": 1840
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.482421875,
      "learning_rate": 2.307692307692308e-05,
      "loss": 2.173,
      "step": 1850
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.451171875,
      "learning_rate": 2.3201663201663204e-05,
      "loss": 2.1802,
      "step": 1860
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.466796875,
      "learning_rate": 2.332640332640333e-05,
      "loss": 2.2011,
      "step": 1870
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.451171875,
      "learning_rate": 2.3451143451143453e-05,
      "loss": 2.1798,
      "step": 1880
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44921875,
      "learning_rate": 2.3575883575883578e-05,
      "loss": 2.2108,
      "step": 1890
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46875,
      "learning_rate": 2.3700623700623703e-05,
      "loss": 2.1934,
      "step": 1900
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4375,
      "learning_rate": 2.3825363825363828e-05,
      "loss": 2.1969,
      "step": 1910
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0546875,
      "learning_rate": 2.3950103950103953e-05,
      "loss": 2.194,
      "step": 1920
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45703125,
      "learning_rate": 2.4074844074844078e-05,
      "loss": 2.2106,
      "step": 1930
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.439453125,
      "learning_rate": 2.4199584199584202e-05,
      "loss": 2.2011,
      "step": 1940
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4453125,
      "learning_rate": 2.4324324324324327e-05,
      "loss": 2.2109,
      "step": 1950
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45703125,
      "learning_rate": 2.4449064449064452e-05,
      "loss": 2.1533,
      "step": 1960
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.53515625,
      "learning_rate": 2.4573804573804577e-05,
      "loss": 2.2166,
      "step": 1970
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45703125,
      "learning_rate": 2.46985446985447e-05,
      "loss": 2.216,
      "step": 1980
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.443359375,
      "learning_rate": 2.4823284823284826e-05,
      "loss": 2.165,
      "step": 1990
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.451171875,
      "learning_rate": 2.494802494802495e-05,
      "loss": 2.1877,
      "step": 2000
    },
    {
      "epoch": 0.04,
      "eval_accuracy": 0.5462529879987186,
      "eval_loss": 2.0614967346191406,
      "eval_runtime": 16.4441,
      "eval_samples_per_second": 36.183,
      "eval_steps_per_second": 1.155,
      "step": 2000
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43359375,
      "learning_rate": 2.5072765072765076e-05,
      "loss": 2.1612,
      "step": 2010
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5,
      "learning_rate": 2.51975051975052e-05,
      "loss": 2.1958,
      "step": 2020
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.439453125,
      "learning_rate": 2.5322245322245326e-05,
      "loss": 2.1965,
      "step": 2030
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.458984375,
      "learning_rate": 2.544698544698545e-05,
      "loss": 2.1974,
      "step": 2040
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4453125,
      "learning_rate": 2.5571725571725575e-05,
      "loss": 2.1879,
      "step": 2050
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.451171875,
      "learning_rate": 2.56964656964657e-05,
      "loss": 2.1704,
      "step": 2060
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45703125,
      "learning_rate": 2.5821205821205825e-05,
      "loss": 2.2034,
      "step": 2070
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44140625,
      "learning_rate": 2.594594594594595e-05,
      "loss": 2.1839,
      "step": 2080
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.474609375,
      "learning_rate": 2.6070686070686074e-05,
      "loss": 2.1888,
      "step": 2090
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44140625,
      "learning_rate": 2.6195426195426196e-05,
      "loss": 2.1959,
      "step": 2100
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.439453125,
      "learning_rate": 2.632016632016632e-05,
      "loss": 2.1911,
      "step": 2110
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.453125,
      "learning_rate": 2.6444906444906445e-05,
      "loss": 2.1735,
      "step": 2120
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.51171875,
      "learning_rate": 2.656964656964657e-05,
      "loss": 2.2117,
      "step": 2130
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4765625,
      "learning_rate": 2.6694386694386695e-05,
      "loss": 2.1977,
      "step": 2140
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.51171875,
      "learning_rate": 2.681912681912682e-05,
      "loss": 2.1918,
      "step": 2150
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4453125,
      "learning_rate": 2.6943866943866945e-05,
      "loss": 2.2017,
      "step": 2160
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.439453125,
      "learning_rate": 2.706860706860707e-05,
      "loss": 2.1722,
      "step": 2170
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.447265625,
      "learning_rate": 2.7193347193347194e-05,
      "loss": 2.2026,
      "step": 2180
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4375,
      "learning_rate": 2.731808731808732e-05,
      "loss": 2.2064,
      "step": 2190
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.59765625,
      "learning_rate": 2.7442827442827444e-05,
      "loss": 2.1722,
      "step": 2200
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46875,
      "learning_rate": 2.756756756756757e-05,
      "loss": 2.2206,
      "step": 2210
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.451171875,
      "learning_rate": 2.7692307692307694e-05,
      "loss": 2.2084,
      "step": 2220
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4765625,
      "learning_rate": 2.781704781704782e-05,
      "loss": 2.1849,
      "step": 2230
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.427734375,
      "learning_rate": 2.7941787941787943e-05,
      "loss": 2.1716,
      "step": 2240
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44140625,
      "learning_rate": 2.8066528066528068e-05,
      "loss": 2.1911,
      "step": 2250
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44140625,
      "learning_rate": 2.8191268191268193e-05,
      "loss": 2.2008,
      "step": 2260
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4375,
      "learning_rate": 2.8316008316008318e-05,
      "loss": 2.1709,
      "step": 2270
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.427734375,
      "learning_rate": 2.8440748440748442e-05,
      "loss": 2.1745,
      "step": 2280
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44140625,
      "learning_rate": 2.8565488565488567e-05,
      "loss": 2.1732,
      "step": 2290
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46484375,
      "learning_rate": 2.8690228690228692e-05,
      "loss": 2.1823,
      "step": 2300
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4375,
      "learning_rate": 2.8814968814968817e-05,
      "loss": 2.1953,
      "step": 2310
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.482421875,
      "learning_rate": 2.893970893970894e-05,
      "loss": 2.1624,
      "step": 2320
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9064449064449066e-05,
      "loss": 2.2022,
      "step": 2330
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.63671875,
      "learning_rate": 2.918918918918919e-05,
      "loss": 2.1941,
      "step": 2340
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45703125,
      "learning_rate": 2.9313929313929316e-05,
      "loss": 2.1828,
      "step": 2350
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44921875,
      "learning_rate": 2.943866943866944e-05,
      "loss": 2.1982,
      "step": 2360
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9563409563409566e-05,
      "loss": 2.1808,
      "step": 2370
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4375,
      "learning_rate": 2.968814968814969e-05,
      "loss": 2.1894,
      "step": 2380
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.88671875,
      "learning_rate": 2.9812889812889815e-05,
      "loss": 2.175,
      "step": 2390
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.435546875,
      "learning_rate": 2.993762993762994e-05,
      "loss": 2.194,
      "step": 2400
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9999999113734896e-05,
      "loss": 2.2173,
      "step": 2410
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9999992023614688e-05,
      "loss": 2.2032,
      "step": 2420
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.56640625,
      "learning_rate": 2.9999977843377625e-05,
      "loss": 2.1914,
      "step": 2430
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.435546875,
      "learning_rate": 2.999995657303041e-05,
      "loss": 2.1739,
      "step": 2440
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43359375,
      "learning_rate": 2.9999928212583095e-05,
      "loss": 2.2069,
      "step": 2450
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9999892762049083e-05,
      "loss": 2.173,
      "step": 2460
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46875,
      "learning_rate": 2.999985022144514e-05,
      "loss": 2.1694,
      "step": 2470
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45703125,
      "learning_rate": 2.9999800590791363e-05,
      "loss": 2.1733,
      "step": 2480
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.94921875,
      "learning_rate": 2.9999743870111215e-05,
      "loss": 2.191,
      "step": 2490
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4609375,
      "learning_rate": 2.999968005943151e-05,
      "loss": 2.1743,
      "step": 2500
    },
    {
      "epoch": 0.05,
      "eval_accuracy": 0.5478597304024249,
      "eval_loss": 2.0514397621154785,
      "eval_runtime": 16.4515,
      "eval_samples_per_second": 36.167,
      "eval_steps_per_second": 1.155,
      "step": 2500
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9999609158782405e-05,
      "loss": 2.2029,
      "step": 2510
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9999531168197414e-05,
      "loss": 2.1817,
      "step": 2520
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.640625,
      "learning_rate": 2.99994460877134e-05,
      "loss": 2.2111,
      "step": 2530
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9999353917370584e-05,
      "loss": 2.1583,
      "step": 2540
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.427734375,
      "learning_rate": 2.999925465721253e-05,
      "loss": 2.1684,
      "step": 2550
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44140625,
      "learning_rate": 2.999914830728615e-05,
      "loss": 2.1606,
      "step": 2560
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.439453125,
      "learning_rate": 2.999903486764172e-05,
      "loss": 2.1909,
      "step": 2570
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.478515625,
      "learning_rate": 2.9998914338332854e-05,
      "loss": 2.1801,
      "step": 2580
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45703125,
      "learning_rate": 2.9998786719416534e-05,
      "loss": 2.208,
      "step": 2590
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.458984375,
      "learning_rate": 2.9998652010953068e-05,
      "loss": 2.1974,
      "step": 2600
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9998510213006138e-05,
      "loss": 2.1862,
      "step": 2610
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45703125,
      "learning_rate": 2.999836132564277e-05,
      "loss": 2.1778,
      "step": 2620
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.458984375,
      "learning_rate": 2.9998205348933333e-05,
      "loss": 2.1937,
      "step": 2630
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46484375,
      "learning_rate": 2.9998042282951557e-05,
      "loss": 2.1744,
      "step": 2640
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.453125,
      "learning_rate": 2.999787212777452e-05,
      "loss": 2.2019,
      "step": 2650
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5,
      "learning_rate": 2.999769488348264e-05,
      "loss": 2.1894,
      "step": 2660
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9997510550159716e-05,
      "loss": 2.1506,
      "step": 2670
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.51171875,
      "learning_rate": 2.9997319127892858e-05,
      "loss": 2.1915,
      "step": 2680
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.451171875,
      "learning_rate": 2.999712061677256e-05,
      "loss": 2.2194,
      "step": 2690
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48828125,
      "learning_rate": 2.9996915016892644e-05,
      "loss": 2.207,
      "step": 2700
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44921875,
      "learning_rate": 2.999670232835029e-05,
      "loss": 2.1725,
      "step": 2710
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43359375,
      "learning_rate": 2.9996482551246048e-05,
      "loss": 2.1666,
      "step": 2720
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.419921875,
      "learning_rate": 2.999625568568378e-05,
      "loss": 2.186,
      "step": 2730
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45703125,
      "learning_rate": 2.999602173177073e-05,
      "loss": 2.1696,
      "step": 2740
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.015625,
      "learning_rate": 2.9995780689617485e-05,
      "loss": 2.1701,
      "step": 2750
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44921875,
      "learning_rate": 2.999553255933797e-05,
      "loss": 2.1956,
      "step": 2760
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.455078125,
      "learning_rate": 2.999527734104948e-05,
      "loss": 2.1865,
      "step": 2770
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9995015034872642e-05,
      "loss": 2.1913,
      "step": 2780
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.51171875,
      "learning_rate": 2.9994745640931443e-05,
      "loss": 2.2192,
      "step": 2790
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.46484375,
      "learning_rate": 2.999446915935322e-05,
      "loss": 2.1883,
      "step": 2800
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.486328125,
      "learning_rate": 2.9994185590268666e-05,
      "loss": 2.1691,
      "step": 2810
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.439453125,
      "learning_rate": 2.9993894933811806e-05,
      "loss": 2.1925,
      "step": 2820
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.46484375,
      "learning_rate": 2.999359719012003e-05,
      "loss": 2.1876,
      "step": 2830
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41796875,
      "learning_rate": 2.9993292359334078e-05,
      "loss": 2.1517,
      "step": 2840
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.439453125,
      "learning_rate": 2.9992980441598026e-05,
      "loss": 2.1792,
      "step": 2850
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44140625,
      "learning_rate": 2.999266143705932e-05,
      "loss": 2.181,
      "step": 2860
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9992335345868743e-05,
      "loss": 2.1869,
      "step": 2870
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9992002168180424e-05,
      "loss": 2.1378,
      "step": 2880
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4765625,
      "learning_rate": 2.999166190415185e-05,
      "loss": 2.1861,
      "step": 2890
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45703125,
      "learning_rate": 2.999131455394386e-05,
      "loss": 2.1751,
      "step": 2900
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4765625,
      "learning_rate": 2.9990960117720638e-05,
      "loss": 2.1862,
      "step": 2910
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9990598595649708e-05,
      "loss": 2.1737,
      "step": 2920
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4375,
      "learning_rate": 2.999022998790196e-05,
      "loss": 2.175,
      "step": 2930
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9989854294651622e-05,
      "loss": 2.1782,
      "step": 2940
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.451171875,
      "learning_rate": 2.998947151607628e-05,
      "loss": 2.1847,
      "step": 2950
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.478515625,
      "learning_rate": 2.9989081652356853e-05,
      "loss": 2.177,
      "step": 2960
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.443359375,
      "learning_rate": 2.998868470367763e-05,
      "loss": 2.1798,
      "step": 2970
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.431640625,
      "learning_rate": 2.998828067022624e-05,
      "loss": 2.1934,
      "step": 2980
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42578125,
      "learning_rate": 2.9987869552193653e-05,
      "loss": 2.2019,
      "step": 2990
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.453125,
      "learning_rate": 2.998745134977419e-05,
      "loss": 2.1885,
      "step": 3000
    },
    {
      "epoch": 0.06,
      "eval_accuracy": 0.5495206880406122,
      "eval_loss": 2.0427331924438477,
      "eval_runtime": 16.4333,
      "eval_samples_per_second": 36.207,
      "eval_steps_per_second": 1.156,
      "step": 3000
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9987026063165534e-05,
      "loss": 2.16,
      "step": 3010
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9986593692568706e-05,
      "loss": 2.2021,
      "step": 3020
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.439453125,
      "learning_rate": 2.9986154238188072e-05,
      "loss": 2.1832,
      "step": 3030
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9985707700231352e-05,
      "loss": 2.1827,
      "step": 3040
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.427734375,
      "learning_rate": 2.998525407890962e-05,
      "loss": 2.1906,
      "step": 3050
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9984793374437283e-05,
      "loss": 2.1986,
      "step": 3060
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9984325587032103e-05,
      "loss": 2.1927,
      "step": 3070
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45703125,
      "learning_rate": 2.99838507169152e-05,
      "loss": 2.1772,
      "step": 3080
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45703125,
      "learning_rate": 2.998336876431103e-05,
      "loss": 2.1857,
      "step": 3090
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.466796875,
      "learning_rate": 2.9982879729447396e-05,
      "loss": 2.1608,
      "step": 3100
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45703125,
      "learning_rate": 2.9982383612555453e-05,
      "loss": 2.1536,
      "step": 3110
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9981880413869707e-05,
      "loss": 2.1958,
      "step": 3120
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9981370133628008e-05,
      "loss": 2.196,
      "step": 3130
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.458984375,
      "learning_rate": 2.9980852772071543e-05,
      "loss": 2.1935,
      "step": 3140
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.45703125,
      "learning_rate": 2.9980328329444867e-05,
      "loss": 2.1475,
      "step": 3150
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.451171875,
      "learning_rate": 2.9979796805995864e-05,
      "loss": 2.1844,
      "step": 3160
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9979258201975777e-05,
      "loss": 2.208,
      "step": 3170
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.59375,
      "learning_rate": 2.9978712517639182e-05,
      "loss": 2.1819,
      "step": 3180
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.439453125,
      "learning_rate": 2.9978159753244016e-05,
      "loss": 2.1717,
      "step": 3190
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9977599909051558e-05,
      "loss": 2.1665,
      "step": 3200
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9977032985326433e-05,
      "loss": 2.1821,
      "step": 3210
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.458984375,
      "learning_rate": 2.9976458982336606e-05,
      "loss": 2.1882,
      "step": 3220
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9975877900353396e-05,
      "loss": 2.1881,
      "step": 3230
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9975289739651468e-05,
      "loss": 2.1546,
      "step": 3240
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.451171875,
      "learning_rate": 2.997469450050883e-05,
      "loss": 2.1857,
      "step": 3250
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.453125,
      "learning_rate": 2.9974092183206833e-05,
      "loss": 2.1829,
      "step": 3260
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4609375,
      "learning_rate": 2.997348278803018e-05,
      "loss": 2.1552,
      "step": 3270
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4609375,
      "learning_rate": 2.997286631526692e-05,
      "loss": 2.1917,
      "step": 3280
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.43359375,
      "learning_rate": 2.9972242765208432e-05,
      "loss": 2.1862,
      "step": 3290
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.435546875,
      "learning_rate": 2.997161213814947e-05,
      "loss": 2.178,
      "step": 3300
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9970974434388105e-05,
      "loss": 2.1697,
      "step": 3310
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.435546875,
      "learning_rate": 2.997032965422576e-05,
      "loss": 2.162,
      "step": 3320
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4453125,
      "learning_rate": 2.996967779796721e-05,
      "loss": 2.1701,
      "step": 3330
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.71875,
      "learning_rate": 2.9969018865920576e-05,
      "loss": 2.1776,
      "step": 3340
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9968352858397312e-05,
      "loss": 2.1803,
      "step": 3350
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4375,
      "learning_rate": 2.996767977571223e-05,
      "loss": 2.173,
      "step": 3360
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4453125,
      "learning_rate": 2.996699961818347e-05,
      "loss": 2.1556,
      "step": 3370
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9966312386132533e-05,
      "loss": 2.1581,
      "step": 3380
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.416015625,
      "learning_rate": 2.996561807988425e-05,
      "loss": 2.1964,
      "step": 3390
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.45703125,
      "learning_rate": 2.9964916699766805e-05,
      "loss": 2.1648,
      "step": 3400
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9964208246111723e-05,
      "loss": 2.208,
      "step": 3410
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9963492719253873e-05,
      "loss": 2.1908,
      "step": 3420
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.416015625,
      "learning_rate": 2.9962770119531465e-05,
      "loss": 2.1643,
      "step": 3430
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.51171875,
      "learning_rate": 2.9962040447286053e-05,
      "loss": 2.1964,
      "step": 3440
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.462890625,
      "learning_rate": 2.9961303702862538e-05,
      "loss": 2.2022,
      "step": 3450
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.45703125,
      "learning_rate": 2.9960559886609153e-05,
      "loss": 2.1782,
      "step": 3460
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.625,
      "learning_rate": 2.9959808998877487e-05,
      "loss": 2.1755,
      "step": 3470
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9959051040022468e-05,
      "loss": 2.1573,
      "step": 3480
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.453125,
      "learning_rate": 2.995828601040236e-05,
      "loss": 2.1782,
      "step": 3490
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9957513910378772e-05,
      "loss": 2.1883,
      "step": 3500
    },
    {
      "epoch": 0.07,
      "eval_accuracy": 0.5508530685001273,
      "eval_loss": 2.035475969314575,
      "eval_runtime": 16.4438,
      "eval_samples_per_second": 36.184,
      "eval_steps_per_second": 1.155,
      "step": 3500
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.458984375,
      "learning_rate": 2.995673474031666e-05,
      "loss": 2.1762,
      "step": 3510
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9955948500584312e-05,
      "loss": 2.1798,
      "step": 3520
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.482421875,
      "learning_rate": 2.995515519155337e-05,
      "loss": 2.1742,
      "step": 3530
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.427734375,
      "learning_rate": 2.995435481359881e-05,
      "loss": 2.1825,
      "step": 3540
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.42578125,
      "learning_rate": 2.9953547367098945e-05,
      "loss": 2.1573,
      "step": 3550
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.45703125,
      "learning_rate": 2.995273285243544e-05,
      "loss": 2.1711,
      "step": 3560
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.431640625,
      "learning_rate": 2.99519112699933e-05,
      "loss": 2.1965,
      "step": 3570
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.42578125,
      "learning_rate": 2.9951082620160857e-05,
      "loss": 2.1536,
      "step": 3580
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.447265625,
      "learning_rate": 2.9950246903329795e-05,
      "loss": 2.1725,
      "step": 3590
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9949404119895138e-05,
      "loss": 2.1544,
      "step": 3600
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9948554270255248e-05,
      "loss": 2.1752,
      "step": 3610
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.43359375,
      "learning_rate": 2.9947697354811822e-05,
      "loss": 2.1633,
      "step": 3620
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46484375,
      "learning_rate": 2.9946833373969915e-05,
      "loss": 2.1848,
      "step": 3630
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.458984375,
      "learning_rate": 2.9945962328137898e-05,
      "loss": 2.1922,
      "step": 3640
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.455078125,
      "learning_rate": 2.99450842177275e-05,
      "loss": 2.1747,
      "step": 3650
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9944199043153768e-05,
      "loss": 2.1722,
      "step": 3660
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.44140625,
      "learning_rate": 2.994330680483512e-05,
      "loss": 2.1634,
      "step": 3670
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.44921875,
      "learning_rate": 2.994240750319328e-05,
      "loss": 2.1441,
      "step": 3680
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.462890625,
      "learning_rate": 2.994150113865333e-05,
      "loss": 2.1594,
      "step": 3690
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46875,
      "learning_rate": 2.9940587711643693e-05,
      "loss": 2.1746,
      "step": 3700
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.419921875,
      "learning_rate": 2.993966722259611e-05,
      "loss": 2.1848,
      "step": 3710
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.482421875,
      "learning_rate": 2.9938739671945682e-05,
      "loss": 2.1767,
      "step": 3720
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9937805060130832e-05,
      "loss": 2.1459,
      "step": 3730
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9936863387593335e-05,
      "loss": 2.1789,
      "step": 3740
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46484375,
      "learning_rate": 2.9935914654778285e-05,
      "loss": 2.1755,
      "step": 3750
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4375,
      "learning_rate": 2.9934958862134138e-05,
      "loss": 2.1622,
      "step": 3760
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.439453125,
      "learning_rate": 2.993399601011266e-05,
      "loss": 2.1468,
      "step": 3770
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.447265625,
      "learning_rate": 2.993302609916898e-05,
      "loss": 2.1678,
      "step": 3780
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.42578125,
      "learning_rate": 2.9932049129761536e-05,
      "loss": 2.1863,
      "step": 3790
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9931065102352127e-05,
      "loss": 2.1686,
      "step": 3800
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46875,
      "learning_rate": 2.9930074017405872e-05,
      "loss": 2.1671,
      "step": 3810
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9929075875391237e-05,
      "loss": 2.1884,
      "step": 3820
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9928070676780016e-05,
      "loss": 2.1645,
      "step": 3830
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4453125,
      "learning_rate": 2.992705842204734e-05,
      "loss": 2.1535,
      "step": 3840
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5390625,
      "learning_rate": 2.9926039111671676e-05,
      "loss": 2.1627,
      "step": 3850
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.421875,
      "learning_rate": 2.992501274613483e-05,
      "loss": 2.1422,
      "step": 3860
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.41796875,
      "learning_rate": 2.9923979325921936e-05,
      "loss": 2.2044,
      "step": 3870
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.50390625,
      "learning_rate": 2.992293885152146e-05,
      "loss": 2.2004,
      "step": 3880
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.458984375,
      "learning_rate": 2.9921891323425224e-05,
      "loss": 2.1717,
      "step": 3890
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.423828125,
      "learning_rate": 2.9920836742128352e-05,
      "loss": 2.1866,
      "step": 3900
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4296875,
      "learning_rate": 2.9919775108129325e-05,
      "loss": 2.1442,
      "step": 3910
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.42578125,
      "learning_rate": 2.9918706421929945e-05,
      "loss": 2.1765,
      "step": 3920
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.427734375,
      "learning_rate": 2.991763068403536e-05,
      "loss": 2.1461,
      "step": 3930
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.439453125,
      "learning_rate": 2.9916547894954042e-05,
      "loss": 2.1789,
      "step": 3940
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9915458055197797e-05,
      "loss": 2.1713,
      "step": 3950
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4296875,
      "learning_rate": 2.9914361165281765e-05,
      "loss": 2.178,
      "step": 3960
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9913257225724417e-05,
      "loss": 2.1776,
      "step": 3970
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.447265625,
      "learning_rate": 2.991214623704756e-05,
      "loss": 2.169,
      "step": 3980
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4296875,
      "learning_rate": 2.991102819977633e-05,
      "loss": 2.1448,
      "step": 3990
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.57421875,
      "learning_rate": 2.990990311443919e-05,
      "loss": 2.1954,
      "step": 4000
    },
    {
      "epoch": 0.08,
      "eval_accuracy": 0.5519078012436647,
      "eval_loss": 2.029829263687134,
      "eval_runtime": 16.4327,
      "eval_samples_per_second": 36.208,
      "eval_steps_per_second": 1.156,
      "step": 4000
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9908770981567944e-05,
      "loss": 2.1643,
      "step": 4010
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9907631801697724e-05,
      "loss": 2.169,
      "step": 4020
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.43359375,
      "learning_rate": 2.990648557536699e-05,
      "loss": 2.1123,
      "step": 4030
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.470703125,
      "learning_rate": 2.990533230311753e-05,
      "loss": 2.1751,
      "step": 4040
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.453125,
      "learning_rate": 2.990417198549447e-05,
      "loss": 2.1754,
      "step": 4050
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9903004623046264e-05,
      "loss": 2.1633,
      "step": 4060
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.52734375,
      "learning_rate": 2.9901830216324694e-05,
      "loss": 2.203,
      "step": 4070
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.453125,
      "learning_rate": 2.990064876588487e-05,
      "loss": 2.1776,
      "step": 4080
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9899460272285237e-05,
      "loss": 2.1538,
      "step": 4090
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4375,
      "learning_rate": 2.9898264736087565e-05,
      "loss": 2.1649,
      "step": 4100
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9897062157856955e-05,
      "loss": 2.1647,
      "step": 4110
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9895852538161826e-05,
      "loss": 2.1567,
      "step": 4120
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4375,
      "learning_rate": 2.989463587757395e-05,
      "loss": 2.1186,
      "step": 4130
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9893412176668396e-05,
      "loss": 2.1593,
      "step": 4140
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.439453125,
      "learning_rate": 2.9892181436023592e-05,
      "loss": 2.1578,
      "step": 4150
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.439453125,
      "learning_rate": 2.9890943656221262e-05,
      "loss": 2.2018,
      "step": 4160
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.462890625,
      "learning_rate": 2.988969883784649e-05,
      "loss": 2.168,
      "step": 4170
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.474609375,
      "learning_rate": 2.9888446981487656e-05,
      "loss": 2.1636,
      "step": 4180
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.451171875,
      "learning_rate": 2.988718808773649e-05,
      "loss": 2.1722,
      "step": 4190
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.46875,
      "learning_rate": 2.988592215718803e-05,
      "loss": 2.1896,
      "step": 4200
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.470703125,
      "learning_rate": 2.9884649190440663e-05,
      "loss": 2.175,
      "step": 4210
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.453125,
      "learning_rate": 2.9883369188096074e-05,
      "loss": 2.1928,
      "step": 4220
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.478515625,
      "learning_rate": 2.9882082150759295e-05,
      "loss": 2.1761,
      "step": 4230
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.421875,
      "learning_rate": 2.9880788079038675e-05,
      "loss": 2.1722,
      "step": 4240
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9879486973545892e-05,
      "loss": 2.1427,
      "step": 4250
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.46484375,
      "learning_rate": 2.987817883489594e-05,
      "loss": 2.1735,
      "step": 4260
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.439453125,
      "learning_rate": 2.9876863663707147e-05,
      "loss": 2.1916,
      "step": 4270
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.423828125,
      "learning_rate": 2.9875541460601156e-05,
      "loss": 2.1778,
      "step": 4280
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9874212226202946e-05,
      "loss": 2.1595,
      "step": 4290
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.443359375,
      "learning_rate": 2.987287596114081e-05,
      "loss": 2.1578,
      "step": 4300
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.431640625,
      "learning_rate": 2.987153266604636e-05,
      "loss": 2.1737,
      "step": 4310
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.455078125,
      "learning_rate": 2.9870182341554552e-05,
      "loss": 2.1528,
      "step": 4320
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4921875,
      "learning_rate": 2.9868824988303637e-05,
      "loss": 2.1525,
      "step": 4330
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.443359375,
      "learning_rate": 2.986746060693521e-05,
      "loss": 2.1484,
      "step": 4340
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44140625,
      "learning_rate": 2.986608919809417e-05,
      "loss": 2.149,
      "step": 4350
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4375,
      "learning_rate": 2.9864710762428755e-05,
      "loss": 2.1657,
      "step": 4360
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.466796875,
      "learning_rate": 2.9863325300590517e-05,
      "loss": 2.1317,
      "step": 4370
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9861932813234324e-05,
      "loss": 2.1697,
      "step": 4380
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4140625,
      "learning_rate": 2.9860533301018376e-05,
      "loss": 2.1679,
      "step": 4390
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.458984375,
      "learning_rate": 2.9859126764604185e-05,
      "loss": 2.1864,
      "step": 4400
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5078125,
      "learning_rate": 2.9857713204656576e-05,
      "loss": 2.1578,
      "step": 4410
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.447265625,
      "learning_rate": 2.985629262184371e-05,
      "loss": 2.153,
      "step": 4420
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44140625,
      "learning_rate": 2.985486501683707e-05,
      "loss": 2.1819,
      "step": 4430
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4140625,
      "learning_rate": 2.9853430390311434e-05,
      "loss": 2.1473,
      "step": 4440
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9851988742944916e-05,
      "loss": 2.1792,
      "step": 4450
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9850540075418957e-05,
      "loss": 2.1412,
      "step": 4460
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9849084388418293e-05,
      "loss": 2.1679,
      "step": 4470
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.455078125,
      "learning_rate": 2.9847621682631e-05,
      "loss": 2.1457,
      "step": 4480
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9846151958748453e-05,
      "loss": 2.1461,
      "step": 4490
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6640625,
      "learning_rate": 2.984467521746536e-05,
      "loss": 2.1597,
      "step": 4500
    },
    {
      "epoch": 0.09,
      "eval_accuracy": 0.5525633127151154,
      "eval_loss": 2.025430202484131,
      "eval_runtime": 16.4538,
      "eval_samples_per_second": 36.162,
      "eval_steps_per_second": 1.155,
      "step": 4500
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.451171875,
      "learning_rate": 2.9843191459479735e-05,
      "loss": 2.1518,
      "step": 4510
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.439453125,
      "learning_rate": 2.9841700685492918e-05,
      "loss": 2.1793,
      "step": 4520
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.451171875,
      "learning_rate": 2.9840202896209557e-05,
      "loss": 2.1419,
      "step": 4530
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.435546875,
      "learning_rate": 2.983869809233762e-05,
      "loss": 2.1445,
      "step": 4540
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.416015625,
      "learning_rate": 2.9837186274588388e-05,
      "loss": 2.1577,
      "step": 4550
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.43359375,
      "learning_rate": 2.983566744367646e-05,
      "loss": 2.1443,
      "step": 4560
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.421875,
      "learning_rate": 2.983414160031975e-05,
      "loss": 2.1516,
      "step": 4570
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9832608745239484e-05,
      "loss": 2.154,
      "step": 4580
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.427734375,
      "learning_rate": 2.98310688791602e-05,
      "loss": 2.1485,
      "step": 4590
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9829522002809764e-05,
      "loss": 2.1782,
      "step": 4600
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.43359375,
      "learning_rate": 2.9827968116919338e-05,
      "loss": 2.1538,
      "step": 4610
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4296875,
      "learning_rate": 2.982640722222341e-05,
      "loss": 2.1917,
      "step": 4620
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.484375,
      "learning_rate": 2.9824839319459766e-05,
      "loss": 2.1765,
      "step": 4630
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.453125,
      "learning_rate": 2.982326440936952e-05,
      "loss": 2.1733,
      "step": 4640
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.416015625,
      "learning_rate": 2.982168249269709e-05,
      "loss": 2.1609,
      "step": 4650
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.43359375,
      "learning_rate": 2.9820093570190217e-05,
      "loss": 2.1473,
      "step": 4660
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.453125,
      "learning_rate": 2.981849764259993e-05,
      "loss": 2.1841,
      "step": 4670
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9816894710680595e-05,
      "loss": 2.1769,
      "step": 4680
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.427734375,
      "learning_rate": 2.981528477518987e-05,
      "loss": 2.1594,
      "step": 4690
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4296875,
      "learning_rate": 2.9813667836888737e-05,
      "loss": 2.157,
      "step": 4700
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9812043896541477e-05,
      "loss": 2.1651,
      "step": 4710
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44140625,
      "learning_rate": 2.981041295491569e-05,
      "loss": 2.1504,
      "step": 4720
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.42578125,
      "learning_rate": 2.9808775012782273e-05,
      "loss": 2.1422,
      "step": 4730
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.53515625,
      "learning_rate": 2.9807130070915448e-05,
      "loss": 2.183,
      "step": 4740
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9805478130092735e-05,
      "loss": 2.1432,
      "step": 4750
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9803819191094962e-05,
      "loss": 2.1552,
      "step": 4760
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9802153254706272e-05,
      "loss": 2.1795,
      "step": 4770
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.439453125,
      "learning_rate": 2.9800480321714107e-05,
      "loss": 2.1725,
      "step": 4780
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.427734375,
      "learning_rate": 2.979880039290922e-05,
      "loss": 2.1677,
      "step": 4790
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9797113469085676e-05,
      "loss": 2.1296,
      "step": 4800
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.46484375,
      "learning_rate": 2.9795419551040836e-05,
      "loss": 2.1279,
      "step": 4810
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9793718639575376e-05,
      "loss": 2.1503,
      "step": 4820
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.48046875,
      "learning_rate": 2.979201073549327e-05,
      "loss": 2.1773,
      "step": 4830
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.423828125,
      "learning_rate": 2.97902958396018e-05,
      "loss": 2.1607,
      "step": 4840
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.431640625,
      "learning_rate": 2.978857395271156e-05,
      "loss": 2.1747,
      "step": 4850
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.462890625,
      "learning_rate": 2.978684507563644e-05,
      "loss": 2.1379,
      "step": 4860
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9785109209193633e-05,
      "loss": 2.179,
      "step": 4870
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.42578125,
      "learning_rate": 2.9783366354203646e-05,
      "loss": 2.1831,
      "step": 4880
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4375,
      "learning_rate": 2.9781616511490278e-05,
      "loss": 2.1485,
      "step": 4890
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.466796875,
      "learning_rate": 2.977985968188063e-05,
      "loss": 2.1787,
      "step": 4900
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.435546875,
      "learning_rate": 2.977809586620512e-05,
      "loss": 2.1773,
      "step": 4910
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9776325065297455e-05,
      "loss": 2.151,
      "step": 4920
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9774547279994648e-05,
      "loss": 2.1736,
      "step": 4930
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.43359375,
      "learning_rate": 2.9772762511137018e-05,
      "loss": 2.1672,
      "step": 4940
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.443359375,
      "learning_rate": 2.977097075956817e-05,
      "loss": 2.1245,
      "step": 4950
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.435546875,
      "learning_rate": 2.976917202613503e-05,
      "loss": 2.1383,
      "step": 4960
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.458984375,
      "learning_rate": 2.9767366311687802e-05,
      "loss": 2.1776,
      "step": 4970
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9765553617080016e-05,
      "loss": 2.1864,
      "step": 4980
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.43359375,
      "learning_rate": 2.9763733943168478e-05,
      "loss": 2.123,
      "step": 4990
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4140625,
      "learning_rate": 2.97619072908133e-05,
      "loss": 2.1763,
      "step": 5000
    },
    {
      "epoch": 0.1,
      "eval_accuracy": 0.5531876093545923,
      "eval_loss": 2.0221822261810303,
      "eval_runtime": 16.4327,
      "eval_samples_per_second": 36.208,
      "eval_steps_per_second": 1.156,
      "step": 5000
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9760073660877904e-05,
      "loss": 2.1835,
      "step": 5010
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.42578125,
      "learning_rate": 2.975823305422899e-05,
      "loss": 2.1592,
      "step": 5020
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9756385471736573e-05,
      "loss": 2.1762,
      "step": 5030
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45703125,
      "learning_rate": 2.9754530914273953e-05,
      "loss": 2.1572,
      "step": 5040
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4296875,
      "learning_rate": 2.9752669382717735e-05,
      "loss": 2.1721,
      "step": 5050
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4453125,
      "learning_rate": 2.975080087794782e-05,
      "loss": 2.1808,
      "step": 5060
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.52734375,
      "learning_rate": 2.97489254008474e-05,
      "loss": 2.1674,
      "step": 5070
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.421875,
      "learning_rate": 2.974704295230297e-05,
      "loss": 2.1795,
      "step": 5080
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.439453125,
      "learning_rate": 2.974515353320431e-05,
      "loss": 2.1636,
      "step": 5090
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4296875,
      "learning_rate": 2.9743257144444504e-05,
      "loss": 2.1463,
      "step": 5100
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9741353786919925e-05,
      "loss": 2.1325,
      "step": 5110
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9739443461530248e-05,
      "loss": 2.1762,
      "step": 5120
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.46484375,
      "learning_rate": 2.9737526169178425e-05,
      "loss": 2.156,
      "step": 5130
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.484375,
      "learning_rate": 2.9735601910770723e-05,
      "loss": 2.1744,
      "step": 5140
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.453125,
      "learning_rate": 2.9733670687216676e-05,
      "loss": 2.1716,
      "step": 5150
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.455078125,
      "learning_rate": 2.973173249942914e-05,
      "loss": 2.163,
      "step": 5160
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9729787348324243e-05,
      "loss": 2.1575,
      "step": 5170
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44140625,
      "learning_rate": 2.97278352348214e-05,
      "loss": 2.137,
      "step": 5180
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.482421875,
      "learning_rate": 2.9725876159843334e-05,
      "loss": 2.1418,
      "step": 5190
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43359375,
      "learning_rate": 2.972391012431605e-05,
      "loss": 2.1677,
      "step": 5200
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9721937129168837e-05,
      "loss": 2.1643,
      "step": 5210
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9719957175334288e-05,
      "loss": 2.1604,
      "step": 5220
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9717970263748272e-05,
      "loss": 2.1921,
      "step": 5230
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.453125,
      "learning_rate": 2.9715976395349954e-05,
      "loss": 2.1282,
      "step": 5240
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.462890625,
      "learning_rate": 2.9713975571081785e-05,
      "loss": 2.1811,
      "step": 5250
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6484375,
      "learning_rate": 2.97119677918895e-05,
      "loss": 2.1442,
      "step": 5260
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4296875,
      "learning_rate": 2.9709953058722134e-05,
      "loss": 2.1748,
      "step": 5270
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.458984375,
      "learning_rate": 2.970793137253199e-05,
      "loss": 2.1495,
      "step": 5280
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44921875,
      "learning_rate": 2.970590273427468e-05,
      "loss": 2.1817,
      "step": 5290
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.47265625,
      "learning_rate": 2.9703867144909077e-05,
      "loss": 2.134,
      "step": 5300
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.435546875,
      "learning_rate": 2.970182460539736e-05,
      "loss": 2.1547,
      "step": 5310
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9699775116704987e-05,
      "loss": 2.1353,
      "step": 5320
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43359375,
      "learning_rate": 2.96977186798007e-05,
      "loss": 2.1861,
      "step": 5330
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9695655295656513e-05,
      "loss": 2.1293,
      "step": 5340
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.46875,
      "learning_rate": 2.9693584965247748e-05,
      "loss": 2.1599,
      "step": 5350
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4296875,
      "learning_rate": 2.9691507689553e-05,
      "loss": 2.1656,
      "step": 5360
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45703125,
      "learning_rate": 2.968942346955413e-05,
      "loss": 2.1774,
      "step": 5370
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.451171875,
      "learning_rate": 2.968733230623631e-05,
      "loss": 2.1422,
      "step": 5380
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4609375,
      "learning_rate": 2.968523420058797e-05,
      "loss": 2.1833,
      "step": 5390
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9683129153600843e-05,
      "loss": 2.1832,
      "step": 5400
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.423828125,
      "learning_rate": 2.968101716626992e-05,
      "loss": 2.1365,
      "step": 5410
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.451171875,
      "learning_rate": 2.9678898239593494e-05,
      "loss": 2.1635,
      "step": 5420
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.466796875,
      "learning_rate": 2.9676772374573118e-05,
      "loss": 2.1438,
      "step": 5430
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.439453125,
      "learning_rate": 2.9674639572213638e-05,
      "loss": 2.1447,
      "step": 5440
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4296875,
      "learning_rate": 2.9672499833523183e-05,
      "loss": 2.1231,
      "step": 5450
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9670353159513148e-05,
      "loss": 2.1225,
      "step": 5460
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.458984375,
      "learning_rate": 2.9668199551198205e-05,
      "loss": 2.1661,
      "step": 5470
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45703125,
      "learning_rate": 2.966603900959632e-05,
      "loss": 2.1532,
      "step": 5480
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.47265625,
      "learning_rate": 2.9663871535728725e-05,
      "loss": 2.1839,
      "step": 5490
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9661697130619924e-05,
      "loss": 2.1413,
      "step": 5500
    },
    {
      "epoch": 0.11,
      "eval_accuracy": 0.5540747677370068,
      "eval_loss": 2.0195205211639404,
      "eval_runtime": 16.4587,
      "eval_samples_per_second": 36.151,
      "eval_steps_per_second": 1.154,
      "step": 5500
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9659515795297703e-05,
      "loss": 2.1639,
      "step": 5510
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.47265625,
      "learning_rate": 2.9657327530793137e-05,
      "loss": 2.1437,
      "step": 5520
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.423828125,
      "learning_rate": 2.965513233814055e-05,
      "loss": 2.1736,
      "step": 5530
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4375,
      "learning_rate": 2.965293021837756e-05,
      "loss": 2.1426,
      "step": 5540
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.462890625,
      "learning_rate": 2.9650721172545048e-05,
      "loss": 2.1912,
      "step": 5550
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43359375,
      "learning_rate": 2.9648505201687184e-05,
      "loss": 2.1682,
      "step": 5560
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.55078125,
      "learning_rate": 2.964628230685139e-05,
      "loss": 2.1736,
      "step": 5570
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.470703125,
      "learning_rate": 2.9644052489088376e-05,
      "loss": 2.1522,
      "step": 5580
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44140625,
      "learning_rate": 2.964181574945213e-05,
      "loss": 2.1567,
      "step": 5590
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.455078125,
      "learning_rate": 2.9639572088999886e-05,
      "loss": 2.1937,
      "step": 5600
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9637321508792175e-05,
      "loss": 2.1349,
      "step": 5610
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.439453125,
      "learning_rate": 2.963506400989279e-05,
      "loss": 2.1544,
      "step": 5620
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.458984375,
      "learning_rate": 2.963279959336879e-05,
      "loss": 2.1762,
      "step": 5630
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4296875,
      "learning_rate": 2.963052826029051e-05,
      "loss": 2.1809,
      "step": 5640
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4296875,
      "learning_rate": 2.962825001173155e-05,
      "loss": 2.1406,
      "step": 5650
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9625964848768788e-05,
      "loss": 2.1486,
      "step": 5660
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9623672772482348e-05,
      "loss": 2.168,
      "step": 5670
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.455078125,
      "learning_rate": 2.9621373783955647e-05,
      "loss": 2.1897,
      "step": 5680
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.53515625,
      "learning_rate": 2.961906788427536e-05,
      "loss": 2.1471,
      "step": 5690
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9616755074531416e-05,
      "loss": 2.1711,
      "step": 5700
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4375,
      "learning_rate": 2.9614435355817035e-05,
      "loss": 2.1391,
      "step": 5710
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9612108729228678e-05,
      "loss": 2.1792,
      "step": 5720
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7421875,
      "learning_rate": 2.960977519586609e-05,
      "loss": 2.1334,
      "step": 5730
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.453125,
      "learning_rate": 2.9607434756832273e-05,
      "loss": 2.1689,
      "step": 5740
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9605087413233494e-05,
      "loss": 2.1577,
      "step": 5750
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.474609375,
      "learning_rate": 2.960273316617928e-05,
      "loss": 2.1605,
      "step": 5760
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9600372016782422e-05,
      "loss": 2.1539,
      "step": 5770
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4296875,
      "learning_rate": 2.959800396615898e-05,
      "loss": 2.1924,
      "step": 5780
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5390625,
      "learning_rate": 2.959562901542827e-05,
      "loss": 2.152,
      "step": 5790
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.427734375,
      "learning_rate": 2.959324716571287e-05,
      "loss": 2.1738,
      "step": 5800
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.451171875,
      "learning_rate": 2.959085841813862e-05,
      "loss": 2.1182,
      "step": 5810
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9588462773834626e-05,
      "loss": 2.1577,
      "step": 5820
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.435546875,
      "learning_rate": 2.958606023393324e-05,
      "loss": 2.1534,
      "step": 5830
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44921875,
      "learning_rate": 2.958365079957009e-05,
      "loss": 2.1774,
      "step": 5840
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43359375,
      "learning_rate": 2.9581234471884047e-05,
      "loss": 2.1479,
      "step": 5850
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45703125,
      "learning_rate": 2.9578811252017258e-05,
      "loss": 2.1877,
      "step": 5860
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.447265625,
      "learning_rate": 2.9576381141115105e-05,
      "loss": 2.1492,
      "step": 5870
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4453125,
      "learning_rate": 2.957394414032625e-05,
      "loss": 2.1479,
      "step": 5880
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9571500250802596e-05,
      "loss": 2.1611,
      "step": 5890
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44140625,
      "learning_rate": 2.956904947369931e-05,
      "loss": 2.1381,
      "step": 5900
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.439453125,
      "learning_rate": 2.9566591810174815e-05,
      "loss": 2.1702,
      "step": 5910
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.455078125,
      "learning_rate": 2.956412726139078e-05,
      "loss": 2.1371,
      "step": 5920
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.451171875,
      "learning_rate": 2.956165582851214e-05,
      "loss": 2.1248,
      "step": 5930
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.421875,
      "learning_rate": 2.9559177512707078e-05,
      "loss": 2.1574,
      "step": 5940
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9556692315147034e-05,
      "loss": 2.1891,
      "step": 5950
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.59765625,
      "learning_rate": 2.955420023700669e-05,
      "loss": 2.1647,
      "step": 5960
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.455078125,
      "learning_rate": 2.9551701279463992e-05,
      "loss": 2.1345,
      "step": 5970
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.484375,
      "learning_rate": 2.954919544370014e-05,
      "loss": 2.1311,
      "step": 5980
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.427734375,
      "learning_rate": 2.954668273089957e-05,
      "loss": 2.1413,
      "step": 5990
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.419921875,
      "learning_rate": 2.954416314224998e-05,
      "loss": 2.1812,
      "step": 6000
    },
    {
      "epoch": 0.12,
      "eval_accuracy": 0.554531490015361,
      "eval_loss": 2.016944408416748,
      "eval_runtime": 16.4992,
      "eval_samples_per_second": 36.062,
      "eval_steps_per_second": 1.152,
      "step": 6000
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.453125,
      "learning_rate": 2.9541636678942322e-05,
      "loss": 2.1293,
      "step": 6010
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.451171875,
      "learning_rate": 2.9539103342170785e-05,
      "loss": 2.1938,
      "step": 6020
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9536563133132817e-05,
      "loss": 2.1605,
      "step": 6030
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9534016053029107e-05,
      "loss": 2.1362,
      "step": 6040
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.453125,
      "learning_rate": 2.95314621030636e-05,
      "loss": 2.1439,
      "step": 6050
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9528901284443476e-05,
      "loss": 2.1792,
      "step": 6060
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4375,
      "learning_rate": 2.9526333598379177e-05,
      "loss": 2.1395,
      "step": 6070
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44140625,
      "learning_rate": 2.952375904608438e-05,
      "loss": 2.1469,
      "step": 6080
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.53125,
      "learning_rate": 2.952117762877601e-05,
      "loss": 2.1316,
      "step": 6090
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.43359375,
      "learning_rate": 2.9518589347674243e-05,
      "loss": 2.1183,
      "step": 6100
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.45703125,
      "learning_rate": 2.9515994204002485e-05,
      "loss": 2.1296,
      "step": 6110
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9513392198987398e-05,
      "loss": 2.1385,
      "step": 6120
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.486328125,
      "learning_rate": 2.9510783333858885e-05,
      "loss": 2.1327,
      "step": 6130
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9508167609850098e-05,
      "loss": 2.16,
      "step": 6140
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.48046875,
      "learning_rate": 2.950554502819741e-05,
      "loss": 2.1903,
      "step": 6150
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4375,
      "learning_rate": 2.950291559014046e-05,
      "loss": 2.1592,
      "step": 6160
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.43359375,
      "learning_rate": 2.950027929692211e-05,
      "loss": 2.1423,
      "step": 6170
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4375,
      "learning_rate": 2.9497636149788468e-05,
      "loss": 2.156,
      "step": 6180
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.447265625,
      "learning_rate": 2.949498614998889e-05,
      "loss": 2.1443,
      "step": 6190
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9492329298775963e-05,
      "loss": 2.1783,
      "step": 6200
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.45703125,
      "learning_rate": 2.9489665597405508e-05,
      "loss": 2.1605,
      "step": 6210
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.435546875,
      "learning_rate": 2.948699504713659e-05,
      "loss": 2.1638,
      "step": 6220
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.455078125,
      "learning_rate": 2.9484317649231514e-05,
      "loss": 2.1552,
      "step": 6230
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.462890625,
      "learning_rate": 2.9481633404955816e-05,
      "loss": 2.1483,
      "step": 6240
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9478942315578274e-05,
      "loss": 2.1713,
      "step": 6250
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.451171875,
      "learning_rate": 2.947624438237089e-05,
      "loss": 2.1149,
      "step": 6260
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9473539606608918e-05,
      "loss": 2.1578,
      "step": 6270
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4375,
      "learning_rate": 2.947082798957083e-05,
      "loss": 2.156,
      "step": 6280
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9468109532538346e-05,
      "loss": 2.1458,
      "step": 6290
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.462890625,
      "learning_rate": 2.9465384236796402e-05,
      "loss": 2.1533,
      "step": 6300
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9462652103633184e-05,
      "loss": 2.1582,
      "step": 6310
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.451171875,
      "learning_rate": 2.94599131343401e-05,
      "loss": 2.1627,
      "step": 6320
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.46484375,
      "learning_rate": 2.9457167330211793e-05,
      "loss": 2.152,
      "step": 6330
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9454414692546137e-05,
      "loss": 2.1506,
      "step": 6340
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44921875,
      "learning_rate": 2.945165522264422e-05,
      "loss": 2.1307,
      "step": 6350
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.453125,
      "learning_rate": 2.9448888921810394e-05,
      "loss": 2.1414,
      "step": 6360
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.412109375,
      "learning_rate": 2.9446115791352207e-05,
      "loss": 2.1924,
      "step": 6370
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9443335832580453e-05,
      "loss": 2.1812,
      "step": 6380
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9440549046809146e-05,
      "loss": 2.1624,
      "step": 6390
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.42578125,
      "learning_rate": 2.9437755435355526e-05,
      "loss": 2.1411,
      "step": 6400
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.451171875,
      "learning_rate": 2.943495499954007e-05,
      "loss": 2.1604,
      "step": 6410
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4375,
      "learning_rate": 2.9432147740686464e-05,
      "loss": 2.1493,
      "step": 6420
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.455078125,
      "learning_rate": 2.9429333660121636e-05,
      "loss": 2.1437,
      "step": 6430
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.453125,
      "learning_rate": 2.9426512759175725e-05,
      "loss": 2.169,
      "step": 6440
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.451171875,
      "learning_rate": 2.9423685039182104e-05,
      "loss": 2.136,
      "step": 6450
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9420850501477364e-05,
      "loss": 2.1452,
      "step": 6460
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4765625,
      "learning_rate": 2.9418009147401317e-05,
      "loss": 2.1624,
      "step": 6470
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9415160978297002e-05,
      "loss": 2.1795,
      "step": 6480
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.431640625,
      "learning_rate": 2.941230599551067e-05,
      "loss": 2.1615,
      "step": 6490
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.466796875,
      "learning_rate": 2.9409444200391807e-05,
      "loss": 2.1526,
      "step": 6500
    },
    {
      "epoch": 0.14,
      "eval_accuracy": 0.5546777068598701,
      "eval_loss": 2.014756441116333,
      "eval_runtime": 16.4559,
      "eval_samples_per_second": 36.157,
      "eval_steps_per_second": 1.155,
      "step": 6500
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42578125,
      "learning_rate": 2.940657559429311e-05,
      "loss": 2.1502,
      "step": 6510
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9403700178570498e-05,
      "loss": 2.1632,
      "step": 6520
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.435546875,
      "learning_rate": 2.94008179545831e-05,
      "loss": 2.1565,
      "step": 6530
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9397928923693273e-05,
      "loss": 2.1224,
      "step": 6540
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.423828125,
      "learning_rate": 2.9395033087266592e-05,
      "loss": 2.1264,
      "step": 6550
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42578125,
      "learning_rate": 2.9392130446671842e-05,
      "loss": 2.1165,
      "step": 6560
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44921875,
      "learning_rate": 2.938922100328103e-05,
      "loss": 2.1481,
      "step": 6570
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.578125,
      "learning_rate": 2.9386304758469382e-05,
      "loss": 2.1674,
      "step": 6580
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9383381713615323e-05,
      "loss": 2.1659,
      "step": 6590
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.490234375,
      "learning_rate": 2.9380451870100508e-05,
      "loss": 2.1412,
      "step": 6600
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.462890625,
      "learning_rate": 2.9377515229309792e-05,
      "loss": 2.1342,
      "step": 6610
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.482421875,
      "learning_rate": 2.9374571792631266e-05,
      "loss": 2.132,
      "step": 6620
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43359375,
      "learning_rate": 2.9371621561456205e-05,
      "loss": 2.1494,
      "step": 6630
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.439453125,
      "learning_rate": 2.9368664537179114e-05,
      "loss": 2.1398,
      "step": 6640
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9365700721197705e-05,
      "loss": 2.1788,
      "step": 6650
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.451171875,
      "learning_rate": 2.9362730114912892e-05,
      "loss": 2.1385,
      "step": 6660
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9359752719728813e-05,
      "loss": 2.1504,
      "step": 6670
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.451171875,
      "learning_rate": 2.9356768537052803e-05,
      "loss": 2.1316,
      "step": 6680
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.546875,
      "learning_rate": 2.935377756829541e-05,
      "loss": 2.1379,
      "step": 6690
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43359375,
      "learning_rate": 2.9350779814870394e-05,
      "loss": 2.1564,
      "step": 6700
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.47265625,
      "learning_rate": 2.9347775278194712e-05,
      "loss": 2.1336,
      "step": 6710
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.51171875,
      "learning_rate": 2.9344763959688532e-05,
      "loss": 2.1451,
      "step": 6720
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4296875,
      "learning_rate": 2.9341745860775234e-05,
      "loss": 2.1514,
      "step": 6730
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.458984375,
      "learning_rate": 2.933872098288139e-05,
      "loss": 2.1128,
      "step": 6740
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.455078125,
      "learning_rate": 2.933568932743679e-05,
      "loss": 2.1736,
      "step": 6750
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9332650895874413e-05,
      "loss": 2.1599,
      "step": 6760
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.435546875,
      "learning_rate": 2.932960568963046e-05,
      "loss": 2.1582,
      "step": 6770
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4375,
      "learning_rate": 2.932655371014431e-05,
      "loss": 2.1563,
      "step": 6780
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.447265625,
      "learning_rate": 2.932349495885856e-05,
      "loss": 2.1311,
      "step": 6790
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9320429437219014e-05,
      "loss": 2.1547,
      "step": 6800
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.439453125,
      "learning_rate": 2.931735714667466e-05,
      "loss": 2.1565,
      "step": 6810
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9314278088677686e-05,
      "loss": 2.1244,
      "step": 6820
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.447265625,
      "learning_rate": 2.931119226468349e-05,
      "loss": 2.1797,
      "step": 6830
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9308099676150666e-05,
      "loss": 2.1404,
      "step": 6840
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.474609375,
      "learning_rate": 2.9305000324540997e-05,
      "loss": 2.1614,
      "step": 6850
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.421875,
      "learning_rate": 2.9301894211319475e-05,
      "loss": 2.1416,
      "step": 6860
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.455078125,
      "learning_rate": 2.9298781337954277e-05,
      "loss": 2.1494,
      "step": 6870
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43359375,
      "learning_rate": 2.929566170591678e-05,
      "loss": 2.1702,
      "step": 6880
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.447265625,
      "learning_rate": 2.929253531668155e-05,
      "loss": 2.172,
      "step": 6890
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9289402171726362e-05,
      "loss": 2.1513,
      "step": 6900
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4375,
      "learning_rate": 2.9286262272532164e-05,
      "loss": 2.1364,
      "step": 6910
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.85546875,
      "learning_rate": 2.9283115620583116e-05,
      "loss": 2.1401,
      "step": 6920
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4296875,
      "learning_rate": 2.9279962217366557e-05,
      "loss": 2.1598,
      "step": 6930
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.443359375,
      "learning_rate": 2.927680206437302e-05,
      "loss": 2.1284,
      "step": 6940
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4296875,
      "learning_rate": 2.927363516309623e-05,
      "loss": 2.1634,
      "step": 6950
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.51171875,
      "learning_rate": 2.92704615150331e-05,
      "loss": 2.1751,
      "step": 6960
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.453125,
      "learning_rate": 2.9267281121683738e-05,
      "loss": 2.1339,
      "step": 6970
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4453125,
      "learning_rate": 2.926409398455143e-05,
      "loss": 2.1391,
      "step": 6980
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42578125,
      "learning_rate": 2.9260900105142657e-05,
      "loss": 2.1198,
      "step": 6990
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9257699484967084e-05,
      "loss": 2.155,
      "step": 7000
    },
    {
      "epoch": 0.15,
      "eval_accuracy": 0.5553512900761478,
      "eval_loss": 2.013082265853882,
      "eval_runtime": 16.4276,
      "eval_samples_per_second": 36.22,
      "eval_steps_per_second": 1.157,
      "step": 7000
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9254492125537565e-05,
      "loss": 2.127,
      "step": 7010
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.419921875,
      "learning_rate": 2.9251278028370137e-05,
      "loss": 2.1111,
      "step": 7020
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.458984375,
      "learning_rate": 2.9248057194984024e-05,
      "loss": 2.1568,
      "step": 7030
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.447265625,
      "learning_rate": 2.924482962690163e-05,
      "loss": 2.1332,
      "step": 7040
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.435546875,
      "learning_rate": 2.924159532564855e-05,
      "loss": 2.1805,
      "step": 7050
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9238354292753545e-05,
      "loss": 2.1403,
      "step": 7060
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.453125,
      "learning_rate": 2.923510652974858e-05,
      "loss": 2.1518,
      "step": 7070
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.48828125,
      "learning_rate": 2.9231852038168784e-05,
      "loss": 2.1383,
      "step": 7080
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41796875,
      "learning_rate": 2.922859081955248e-05,
      "loss": 2.1245,
      "step": 7090
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.421875,
      "learning_rate": 2.9225322875441158e-05,
      "loss": 2.1309,
      "step": 7100
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.431640625,
      "learning_rate": 2.9222048207379494e-05,
      "loss": 2.1364,
      "step": 7110
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.443359375,
      "learning_rate": 2.921876681691534e-05,
      "loss": 2.1302,
      "step": 7120
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.435546875,
      "learning_rate": 2.921547870559973e-05,
      "loss": 2.1439,
      "step": 7130
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9212183874986866e-05,
      "loss": 2.1611,
      "step": 7140
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44140625,
      "learning_rate": 2.920888232663413e-05,
      "loss": 2.184,
      "step": 7150
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.458984375,
      "learning_rate": 2.920557406210209e-05,
      "loss": 2.1542,
      "step": 7160
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.474609375,
      "learning_rate": 2.9202259082954464e-05,
      "loss": 2.1725,
      "step": 7170
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.474609375,
      "learning_rate": 2.9198937390758176e-05,
      "loss": 2.1364,
      "step": 7180
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.423828125,
      "learning_rate": 2.919560898708329e-05,
      "loss": 2.1639,
      "step": 7190
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.439453125,
      "learning_rate": 2.919227387350307e-05,
      "loss": 2.1327,
      "step": 7200
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.453125,
      "learning_rate": 2.9188932051593934e-05,
      "loss": 2.1416,
      "step": 7210
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4375,
      "learning_rate": 2.9185583522935476e-05,
      "loss": 2.1131,
      "step": 7220
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.451171875,
      "learning_rate": 2.9182228289110468e-05,
      "loss": 2.1464,
      "step": 7230
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5,
      "learning_rate": 2.9178866351704834e-05,
      "loss": 2.1135,
      "step": 7240
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.421875,
      "learning_rate": 2.917549771230768e-05,
      "loss": 2.1434,
      "step": 7250
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.451171875,
      "learning_rate": 2.9172122372511282e-05,
      "loss": 2.1705,
      "step": 7260
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.578125,
      "learning_rate": 2.9168740333911074e-05,
      "loss": 2.1587,
      "step": 7270
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.50390625,
      "learning_rate": 2.9165351598105657e-05,
      "loss": 2.1584,
      "step": 7280
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42578125,
      "learning_rate": 2.9161956166696805e-05,
      "loss": 2.1481,
      "step": 7290
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.474609375,
      "learning_rate": 2.9158554041289454e-05,
      "loss": 2.144,
      "step": 7300
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.451171875,
      "learning_rate": 2.91551452234917e-05,
      "loss": 2.1433,
      "step": 7310
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5078125,
      "learning_rate": 2.9151729714914805e-05,
      "loss": 2.1611,
      "step": 7320
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.466796875,
      "learning_rate": 2.914830751717319e-05,
      "loss": 2.1217,
      "step": 7330
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5078125,
      "learning_rate": 2.914487863188445e-05,
      "loss": 2.123,
      "step": 7340
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.48828125,
      "learning_rate": 2.914144306066932e-05,
      "loss": 2.1235,
      "step": 7350
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4765625,
      "learning_rate": 2.913800080515172e-05,
      "loss": 2.1306,
      "step": 7360
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45703125,
      "learning_rate": 2.9134551866958713e-05,
      "loss": 2.1256,
      "step": 7370
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43359375,
      "learning_rate": 2.9131096247720524e-05,
      "loss": 2.1392,
      "step": 7380
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9127633949070536e-05,
      "loss": 2.1704,
      "step": 7390
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45703125,
      "learning_rate": 2.912416497264529e-05,
      "loss": 2.1692,
      "step": 7400
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41796875,
      "learning_rate": 2.912068932008448e-05,
      "loss": 2.1582,
      "step": 7410
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.451171875,
      "learning_rate": 2.9117206993030967e-05,
      "loss": 2.1192,
      "step": 7420
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4609375,
      "learning_rate": 2.9113717993130754e-05,
      "loss": 2.1469,
      "step": 7430
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5234375,
      "learning_rate": 2.911022232203301e-05,
      "loss": 2.1331,
      "step": 7440
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.453125,
      "learning_rate": 2.9106719981390037e-05,
      "loss": 2.1411,
      "step": 7450
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.458984375,
      "learning_rate": 2.910321097285731e-05,
      "loss": 2.1278,
      "step": 7460
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.48828125,
      "learning_rate": 2.9099695298093454e-05,
      "loss": 2.1665,
      "step": 7470
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.474609375,
      "learning_rate": 2.909617295876023e-05,
      "loss": 2.1481,
      "step": 7480
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.453125,
      "learning_rate": 2.9092643956522568e-05,
      "loss": 2.1107,
      "step": 7490
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.412109375,
      "learning_rate": 2.908910829304853e-05,
      "loss": 2.1594,
      "step": 7500
    },
    {
      "epoch": 0.16,
      "eval_accuracy": 0.5557669402071679,
      "eval_loss": 2.011016845703125,
      "eval_runtime": 16.4416,
      "eval_samples_per_second": 36.189,
      "eval_steps_per_second": 1.156,
      "step": 7500
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.447265625,
      "learning_rate": 2.9085565970009337e-05,
      "loss": 2.1623,
      "step": 7510
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.44140625,
      "learning_rate": 2.9082016989079357e-05,
      "loss": 2.1686,
      "step": 7520
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.46875,
      "learning_rate": 2.9078461351936104e-05,
      "loss": 2.1762,
      "step": 7530
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4375,
      "learning_rate": 2.9074899060260237e-05,
      "loss": 2.1224,
      "step": 7540
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.490234375,
      "learning_rate": 2.9071330115735556e-05,
      "loss": 2.1145,
      "step": 7550
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.458984375,
      "learning_rate": 2.906775452004902e-05,
      "loss": 2.1215,
      "step": 7560
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.419921875,
      "learning_rate": 2.9064172274890718e-05,
      "loss": 2.1391,
      "step": 7570
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.72265625,
      "learning_rate": 2.906058338195389e-05,
      "loss": 2.1384,
      "step": 7580
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.46875,
      "learning_rate": 2.9056987842934908e-05,
      "loss": 2.1288,
      "step": 7590
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.45703125,
      "learning_rate": 2.9053385659533295e-05,
      "loss": 2.1429,
      "step": 7600
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.54296875,
      "learning_rate": 2.9049776833451718e-05,
      "loss": 2.1199,
      "step": 7610
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4921875,
      "learning_rate": 2.9046161366395966e-05,
      "loss": 2.1551,
      "step": 7620
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.466796875,
      "learning_rate": 2.9042539260074994e-05,
      "loss": 2.1427,
      "step": 7630
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.470703125,
      "learning_rate": 2.9038910516200867e-05,
      "loss": 2.153,
      "step": 7640
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9035275136488807e-05,
      "loss": 2.1384,
      "step": 7650
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.470703125,
      "learning_rate": 2.9031633122657163e-05,
      "loss": 2.1234,
      "step": 7660
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.455078125,
      "learning_rate": 2.9027984476427427e-05,
      "loss": 2.1772,
      "step": 7670
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.435546875,
      "learning_rate": 2.9024329199524217e-05,
      "loss": 2.1686,
      "step": 7680
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.48828125,
      "learning_rate": 2.9020667293675292e-05,
      "loss": 2.1469,
      "step": 7690
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.50390625,
      "learning_rate": 2.9016998760611542e-05,
      "loss": 2.1321,
      "step": 7700
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.474609375,
      "learning_rate": 2.9013323602066992e-05,
      "loss": 2.139,
      "step": 7710
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9009641819778795e-05,
      "loss": 2.1379,
      "step": 7720
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.419921875,
      "learning_rate": 2.9005953415487233e-05,
      "loss": 2.1558,
      "step": 7730
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.45703125,
      "learning_rate": 2.9002258390935724e-05,
      "loss": 2.1326,
      "step": 7740
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.447265625,
      "learning_rate": 2.8998556747870817e-05,
      "loss": 2.1088,
      "step": 7750
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.416015625,
      "learning_rate": 2.8994848488042177e-05,
      "loss": 2.1503,
      "step": 7760
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.466796875,
      "learning_rate": 2.8991133613202615e-05,
      "loss": 2.1422,
      "step": 7770
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.470703125,
      "learning_rate": 2.8987412125108046e-05,
      "loss": 2.1498,
      "step": 7780
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.431640625,
      "learning_rate": 2.898368402551753e-05,
      "loss": 2.1369,
      "step": 7790
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4375,
      "learning_rate": 2.897994931619325e-05,
      "loss": 2.1385,
      "step": 7800
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.546875,
      "learning_rate": 2.8976207998900498e-05,
      "loss": 2.1686,
      "step": 7810
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.478515625,
      "learning_rate": 2.8972460075407706e-05,
      "loss": 2.154,
      "step": 7820
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.470703125,
      "learning_rate": 2.8968705547486423e-05,
      "loss": 2.1063,
      "step": 7830
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.484375,
      "learning_rate": 2.8964944416911316e-05,
      "loss": 2.1428,
      "step": 7840
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.455078125,
      "learning_rate": 2.8961176685460184e-05,
      "loss": 2.1558,
      "step": 7850
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.46875,
      "learning_rate": 2.8957402354913927e-05,
      "loss": 2.1536,
      "step": 7860
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.486328125,
      "learning_rate": 2.8953621427056588e-05,
      "loss": 2.1429,
      "step": 7870
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.455078125,
      "learning_rate": 2.894983390367531e-05,
      "loss": 2.1456,
      "step": 7880
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5546875,
      "learning_rate": 2.8946039786560355e-05,
      "loss": 2.126,
      "step": 7890
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42578125,
      "learning_rate": 2.894223907750511e-05,
      "loss": 2.1207,
      "step": 7900
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.44140625,
      "learning_rate": 2.8938431778306082e-05,
      "loss": 2.1407,
      "step": 7910
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.419921875,
      "learning_rate": 2.893461789076287e-05,
      "loss": 2.1534,
      "step": 7920
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.50390625,
      "learning_rate": 2.893079741667821e-05,
      "loss": 2.1167,
      "step": 7930
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.43359375,
      "learning_rate": 2.8926970357857946e-05,
      "loss": 2.1525,
      "step": 7940
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.49609375,
      "learning_rate": 2.8923136716111028e-05,
      "loss": 2.1571,
      "step": 7950
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.47265625,
      "learning_rate": 2.8919296493249527e-05,
      "loss": 2.1512,
      "step": 7960
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.435546875,
      "learning_rate": 2.8915449691088607e-05,
      "loss": 2.1659,
      "step": 7970
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.474609375,
      "learning_rate": 2.891159631144657e-05,
      "loss": 2.1307,
      "step": 7980
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.44921875,
      "learning_rate": 2.8907736356144794e-05,
      "loss": 2.094,
      "step": 7990
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.61328125,
      "learning_rate": 2.8903869827007796e-05,
      "loss": 2.1681,
      "step": 8000
    },
    {
      "epoch": 0.17,
      "eval_accuracy": 0.5559098712798902,
      "eval_loss": 2.0096688270568848,
      "eval_runtime": 16.4393,
      "eval_samples_per_second": 36.194,
      "eval_steps_per_second": 1.156,
      "step": 8000
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4609375,
      "learning_rate": 2.889999672586318e-05,
      "loss": 2.1198,
      "step": 8010
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.44921875,
      "learning_rate": 2.8896117054541667e-05,
      "loss": 2.1312,
      "step": 8020
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.447265625,
      "learning_rate": 2.889223081487708e-05,
      "loss": 2.162,
      "step": 8030
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5234375,
      "learning_rate": 2.888833800870634e-05,
      "loss": 2.1709,
      "step": 8040
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.50390625,
      "learning_rate": 2.8884438637869482e-05,
      "loss": 2.1311,
      "step": 8050
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.47265625,
      "learning_rate": 2.8880532704209636e-05,
      "loss": 2.1262,
      "step": 8060
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.455078125,
      "learning_rate": 2.8876620209573046e-05,
      "loss": 2.136,
      "step": 8070
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.78125,
      "learning_rate": 2.8872701155809046e-05,
      "loss": 2.1387,
      "step": 8080
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.43359375,
      "learning_rate": 2.886877554477007e-05,
      "loss": 2.1656,
      "step": 8090
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.431640625,
      "learning_rate": 2.8864843378311653e-05,
      "loss": 2.1209,
      "step": 8100
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4296875,
      "learning_rate": 2.8860904658292446e-05,
      "loss": 2.1206,
      "step": 8110
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.431640625,
      "learning_rate": 2.885695938657416e-05,
      "loss": 2.1264,
      "step": 8120
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4609375,
      "learning_rate": 2.8853007565021644e-05,
      "loss": 2.1118,
      "step": 8130
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.45703125,
      "learning_rate": 2.8849049195502812e-05,
      "loss": 2.1673,
      "step": 8140
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41796875,
      "learning_rate": 2.8845084279888694e-05,
      "loss": 2.1242,
      "step": 8150
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.46875,
      "learning_rate": 2.8841112820053396e-05,
      "loss": 2.1227,
      "step": 8160
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.443359375,
      "learning_rate": 2.8837134817874138e-05,
      "loss": 2.1518,
      "step": 8170
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.455078125,
      "learning_rate": 2.883315027523121e-05,
      "loss": 2.1387,
      "step": 8180
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.44140625,
      "learning_rate": 2.8829159194008014e-05,
      "loss": 2.1144,
      "step": 8190
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.47265625,
      "learning_rate": 2.8825161576091024e-05,
      "loss": 2.1354,
      "step": 8200
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.46484375,
      "learning_rate": 2.882115742336982e-05,
      "loss": 2.1517,
      "step": 8210
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4453125,
      "learning_rate": 2.881714673773706e-05,
      "loss": 2.1294,
      "step": 8220
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.466796875,
      "learning_rate": 2.8813129521088492e-05,
      "loss": 2.1437,
      "step": 8230
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.447265625,
      "learning_rate": 2.880910577532296e-05,
      "loss": 2.1625,
      "step": 8240
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.46484375,
      "learning_rate": 2.8805075502342377e-05,
      "loss": 2.1185,
      "step": 8250
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.435546875,
      "learning_rate": 2.8801038704051765e-05,
      "loss": 2.1205,
      "step": 8260
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.431640625,
      "learning_rate": 2.87969953823592e-05,
      "loss": 2.1658,
      "step": 8270
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.44140625,
      "learning_rate": 2.879294553917587e-05,
      "loss": 2.1598,
      "step": 8280
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.470703125,
      "learning_rate": 2.8788889176416026e-05,
      "loss": 2.1324,
      "step": 8290
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.45703125,
      "learning_rate": 2.8784826295997012e-05,
      "loss": 2.1282,
      "step": 8300
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4765625,
      "learning_rate": 2.878075689983925e-05,
      "loss": 2.157,
      "step": 8310
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.45703125,
      "learning_rate": 2.877668098986624e-05,
      "loss": 2.174,
      "step": 8320
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.435546875,
      "learning_rate": 2.8772598568004557e-05,
      "loss": 2.1525,
      "step": 8330
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.474609375,
      "learning_rate": 2.8768509636183863e-05,
      "loss": 2.1448,
      "step": 8340
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5390625,
      "learning_rate": 2.876441419633689e-05,
      "loss": 2.1316,
      "step": 8350
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.43359375,
      "learning_rate": 2.8760312250399456e-05,
      "loss": 2.1271,
      "step": 8360
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4375,
      "learning_rate": 2.8756203800310433e-05,
      "loss": 2.1536,
      "step": 8370
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.439453125,
      "learning_rate": 2.875208884801179e-05,
      "loss": 2.1218,
      "step": 8380
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.427734375,
      "learning_rate": 2.874796739544857e-05,
      "loss": 2.1294,
      "step": 8390
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.451171875,
      "learning_rate": 2.8743839444568865e-05,
      "loss": 2.141,
      "step": 8400
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.484375,
      "learning_rate": 2.873970499732386e-05,
      "loss": 2.1562,
      "step": 8410
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.515625,
      "learning_rate": 2.87355640556678e-05,
      "loss": 2.1623,
      "step": 8420
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.451171875,
      "learning_rate": 2.8731416621558006e-05,
      "loss": 2.1554,
      "step": 8430
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4921875,
      "learning_rate": 2.8727262696954865e-05,
      "loss": 2.1186,
      "step": 8440
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.44921875,
      "learning_rate": 2.8723102283821826e-05,
      "loss": 2.1369,
      "step": 8450
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.453125,
      "learning_rate": 2.8718935384125423e-05,
      "loss": 2.1446,
      "step": 8460
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.466796875,
      "learning_rate": 2.8714761999835236e-05,
      "loss": 2.1343,
      "step": 8470
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.53125,
      "learning_rate": 2.871058213292392e-05,
      "loss": 2.1591,
      "step": 8480
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4296875,
      "learning_rate": 2.870639578536719e-05,
      "loss": 2.1227,
      "step": 8490
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.50390625,
      "learning_rate": 2.8702202959143833e-05,
      "loss": 2.1572,
      "step": 8500
    },
    {
      "epoch": 0.18,
      "eval_accuracy": 0.5561628757074677,
      "eval_loss": 2.008275270462036,
      "eval_runtime": 16.4042,
      "eval_samples_per_second": 36.271,
      "eval_steps_per_second": 1.158,
      "step": 8500
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.439453125,
      "learning_rate": 2.8698003656235686e-05,
      "loss": 2.0905,
      "step": 8510
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.466796875,
      "learning_rate": 2.8693797878627654e-05,
      "loss": 2.0992,
      "step": 8520
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.42578125,
      "learning_rate": 2.86895856283077e-05,
      "loss": 2.1138,
      "step": 8530
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.427734375,
      "learning_rate": 2.8685366907266855e-05,
      "loss": 2.1292,
      "step": 8540
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.453125,
      "learning_rate": 2.8681141717499198e-05,
      "loss": 2.148,
      "step": 8550
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.431640625,
      "learning_rate": 2.867691006100187e-05,
      "loss": 2.1512,
      "step": 8560
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4921875,
      "learning_rate": 2.867267193977506e-05,
      "loss": 2.1567,
      "step": 8570
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.490234375,
      "learning_rate": 2.8668427355822036e-05,
      "loss": 2.1394,
      "step": 8580
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.466796875,
      "learning_rate": 2.8664176311149097e-05,
      "loss": 2.1065,
      "step": 8590
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.44140625,
      "learning_rate": 2.86599188077656e-05,
      "loss": 2.1543,
      "step": 8600
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.478515625,
      "learning_rate": 2.8655654847683965e-05,
      "loss": 2.1116,
      "step": 8610
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4453125,
      "learning_rate": 2.8651384432919655e-05,
      "loss": 2.1389,
      "step": 8620
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41796875,
      "learning_rate": 2.864710756549119e-05,
      "loss": 2.1154,
      "step": 8630
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.431640625,
      "learning_rate": 2.8642824247420142e-05,
      "loss": 2.1443,
      "step": 8640
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.466796875,
      "learning_rate": 2.863853448073111e-05,
      "loss": 2.1235,
      "step": 8650
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.478515625,
      "learning_rate": 2.8634238267451776e-05,
      "loss": 2.1575,
      "step": 8660
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4453125,
      "learning_rate": 2.8629935609612842e-05,
      "loss": 2.1711,
      "step": 8670
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.455078125,
      "learning_rate": 2.8625626509248073e-05,
      "loss": 2.1539,
      "step": 8680
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.447265625,
      "learning_rate": 2.862131096839426e-05,
      "loss": 2.1201,
      "step": 8690
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.451171875,
      "learning_rate": 2.8616988989091263e-05,
      "loss": 2.1376,
      "step": 8700
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4375,
      "learning_rate": 2.8612660573381964e-05,
      "loss": 2.141,
      "step": 8710
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.423828125,
      "learning_rate": 2.8608325723312296e-05,
      "loss": 2.1141,
      "step": 8720
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.482421875,
      "learning_rate": 2.860398444093124e-05,
      "loss": 2.1285,
      "step": 8730
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.431640625,
      "learning_rate": 2.85996367282908e-05,
      "loss": 2.132,
      "step": 8740
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.44140625,
      "learning_rate": 2.8595282587446034e-05,
      "loss": 2.1416,
      "step": 8750
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.439453125,
      "learning_rate": 2.859092202045504e-05,
      "loss": 2.1107,
      "step": 8760
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.443359375,
      "learning_rate": 2.858655502937894e-05,
      "loss": 2.1308,
      "step": 8770
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.439453125,
      "learning_rate": 2.8582181616281904e-05,
      "loss": 2.1294,
      "step": 8780
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.458984375,
      "learning_rate": 2.857780178323113e-05,
      "loss": 2.1616,
      "step": 8790
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.44921875,
      "learning_rate": 2.8573415532296864e-05,
      "loss": 2.1197,
      "step": 8800
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.42578125,
      "learning_rate": 2.8569022865552368e-05,
      "loss": 2.1422,
      "step": 8810
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.439453125,
      "learning_rate": 2.856462378507394e-05,
      "loss": 2.1022,
      "step": 8820
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.478515625,
      "learning_rate": 2.8560218292940924e-05,
      "loss": 2.1064,
      "step": 8830
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.43359375,
      "learning_rate": 2.855580639123568e-05,
      "loss": 2.1418,
      "step": 8840
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.44921875,
      "learning_rate": 2.85513880820436e-05,
      "loss": 2.1461,
      "step": 8850
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.43359375,
      "learning_rate": 2.854696336745312e-05,
      "loss": 2.1522,
      "step": 8860
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.416015625,
      "learning_rate": 2.854253224955567e-05,
      "loss": 2.1539,
      "step": 8870
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.421875,
      "learning_rate": 2.853809473044574e-05,
      "loss": 2.1322,
      "step": 8880
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.447265625,
      "learning_rate": 2.853365081222083e-05,
      "loss": 2.0938,
      "step": 8890
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.453125,
      "learning_rate": 2.8529200496981472e-05,
      "loss": 2.1325,
      "step": 8900
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.51171875,
      "learning_rate": 2.852474378683121e-05,
      "loss": 2.127,
      "step": 8910
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.64453125,
      "learning_rate": 2.8520280683876625e-05,
      "loss": 2.1291,
      "step": 8920
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.494140625,
      "learning_rate": 2.8515811190227312e-05,
      "loss": 2.1468,
      "step": 8930
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.482421875,
      "learning_rate": 2.8511335307995882e-05,
      "loss": 2.1555,
      "step": 8940
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.421875,
      "learning_rate": 2.8506853039297976e-05,
      "loss": 2.0734,
      "step": 8950
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.44921875,
      "learning_rate": 2.850236438625225e-05,
      "loss": 2.1333,
      "step": 8960
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.43359375,
      "learning_rate": 2.8497869350980376e-05,
      "loss": 2.1369,
      "step": 8970
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.443359375,
      "learning_rate": 2.8493367935607037e-05,
      "loss": 2.1502,
      "step": 8980
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.484375,
      "learning_rate": 2.848886014225995e-05,
      "loss": 2.1398,
      "step": 8990
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.45703125,
      "learning_rate": 2.8484345973069824e-05,
      "loss": 2.0943,
      "step": 9000
    },
    {
      "epoch": 0.19,
      "eval_accuracy": 0.5566261695293954,
      "eval_loss": 2.0073742866516113,
      "eval_runtime": 16.4366,
      "eval_samples_per_second": 36.2,
      "eval_steps_per_second": 1.156,
      "step": 9000
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.47265625,
      "learning_rate": 2.8479825430170398e-05,
      "loss": 2.1152,
      "step": 9010
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.455078125,
      "learning_rate": 2.8475298515698418e-05,
      "loss": 2.1292,
      "step": 9020
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.43359375,
      "learning_rate": 2.8470765231793642e-05,
      "loss": 2.1321,
      "step": 9030
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.431640625,
      "learning_rate": 2.8466225580598836e-05,
      "loss": 2.1557,
      "step": 9040
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.474609375,
      "learning_rate": 2.8461679564259785e-05,
      "loss": 2.1168,
      "step": 9050
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.474609375,
      "learning_rate": 2.845712718492527e-05,
      "loss": 2.1334,
      "step": 9060
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.470703125,
      "learning_rate": 2.8452568444747085e-05,
      "loss": 2.1103,
      "step": 9070
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.458984375,
      "learning_rate": 2.844800334588003e-05,
      "loss": 2.17,
      "step": 9080
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.431640625,
      "learning_rate": 2.844343189048192e-05,
      "loss": 2.1498,
      "step": 9090
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.44921875,
      "learning_rate": 2.843885408071356e-05,
      "loss": 2.173,
      "step": 9100
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.439453125,
      "learning_rate": 2.8434269918738764e-05,
      "loss": 2.1362,
      "step": 9110
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.51953125,
      "learning_rate": 2.8429679406724353e-05,
      "loss": 2.1659,
      "step": 9120
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.45703125,
      "learning_rate": 2.8425082546840145e-05,
      "loss": 2.141,
      "step": 9130
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.44140625,
      "learning_rate": 2.8420479341258955e-05,
      "loss": 2.1376,
      "step": 9140
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.466796875,
      "learning_rate": 2.8415869792156612e-05,
      "loss": 2.1429,
      "step": 9150
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.435546875,
      "learning_rate": 2.8411253901711924e-05,
      "loss": 2.1491,
      "step": 9160
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4375,
      "learning_rate": 2.8406631672106706e-05,
      "loss": 2.1016,
      "step": 9170
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.44140625,
      "learning_rate": 2.8402003105525775e-05,
      "loss": 2.1172,
      "step": 9180
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.42578125,
      "learning_rate": 2.8397368204156935e-05,
      "loss": 2.1402,
      "step": 9190
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.427734375,
      "learning_rate": 2.8392726970190983e-05,
      "loss": 2.1305,
      "step": 9200
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.490234375,
      "learning_rate": 2.8388079405821718e-05,
      "loss": 2.1218,
      "step": 9210
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.439453125,
      "learning_rate": 2.838342551324592e-05,
      "loss": 2.1405,
      "step": 9220
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.427734375,
      "learning_rate": 2.8378765294663373e-05,
      "loss": 2.1289,
      "step": 9230
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.46875,
      "learning_rate": 2.8374098752276843e-05,
      "loss": 2.1442,
      "step": 9240
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.455078125,
      "learning_rate": 2.836942588829208e-05,
      "loss": 2.1256,
      "step": 9250
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.490234375,
      "learning_rate": 2.836474670491784e-05,
      "loss": 2.1373,
      "step": 9260
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.46875,
      "learning_rate": 2.8360061204365845e-05,
      "loss": 2.1321,
      "step": 9270
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.44921875,
      "learning_rate": 2.8355369388850815e-05,
      "loss": 2.1388,
      "step": 9280
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.431640625,
      "learning_rate": 2.8350671260590455e-05,
      "loss": 2.1369,
      "step": 9290
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.431640625,
      "learning_rate": 2.834596682180545e-05,
      "loss": 2.1429,
      "step": 9300
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.423828125,
      "learning_rate": 2.8341256074719465e-05,
      "loss": 2.1402,
      "step": 9310
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.87890625,
      "learning_rate": 2.833653902155916e-05,
      "loss": 2.1267,
      "step": 9320
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4296875,
      "learning_rate": 2.833181566455416e-05,
      "loss": 2.1391,
      "step": 9330
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.42578125,
      "learning_rate": 2.8327086005937078e-05,
      "loss": 2.1407,
      "step": 9340
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.43359375,
      "learning_rate": 2.8322350047943503e-05,
      "loss": 2.1258,
      "step": 9350
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.45703125,
      "learning_rate": 2.831760779281201e-05,
      "loss": 2.1461,
      "step": 9360
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.431640625,
      "learning_rate": 2.831285924278413e-05,
      "loss": 2.1264,
      "step": 9370
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.451171875,
      "learning_rate": 2.8308104400104397e-05,
      "loss": 2.1298,
      "step": 9380
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.453125,
      "learning_rate": 2.830334326702029e-05,
      "loss": 2.1558,
      "step": 9390
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.451171875,
      "learning_rate": 2.8298575845782285e-05,
      "loss": 2.1233,
      "step": 9400
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.431640625,
      "learning_rate": 2.829380213864382e-05,
      "loss": 2.1063,
      "step": 9410
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.46484375,
      "learning_rate": 2.8289022147861307e-05,
      "loss": 2.1103,
      "step": 9420
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.515625,
      "learning_rate": 2.8284235875694125e-05,
      "loss": 2.1359,
      "step": 9430
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.435546875,
      "learning_rate": 2.8279443324404623e-05,
      "loss": 2.1452,
      "step": 9440
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.44140625,
      "learning_rate": 2.827464449625812e-05,
      "loss": 2.1253,
      "step": 9450
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.42578125,
      "learning_rate": 2.82698393935229e-05,
      "loss": 2.1038,
      "step": 9460
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.435546875,
      "learning_rate": 2.8265028018470215e-05,
      "loss": 2.1573,
      "step": 9470
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.423828125,
      "learning_rate": 2.826021037337428e-05,
      "loss": 2.1392,
      "step": 9480
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4375,
      "learning_rate": 2.8255386460512273e-05,
      "loss": 2.0994,
      "step": 9490
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.423828125,
      "learning_rate": 2.8250556282164334e-05,
      "loss": 2.1421,
      "step": 9500
    },
    {
      "epoch": 0.2,
      "eval_accuracy": 0.556583454496168,
      "eval_loss": 2.0063412189483643,
      "eval_runtime": 16.4451,
      "eval_samples_per_second": 36.181,
      "eval_steps_per_second": 1.155,
      "step": 9500
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.431640625,
      "learning_rate": 2.824571984061357e-05,
      "loss": 2.1158,
      "step": 9510
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.419921875,
      "learning_rate": 2.8240877138146045e-05,
      "loss": 2.1322,
      "step": 9520
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.408203125,
      "learning_rate": 2.8236028177050776e-05,
      "loss": 2.1359,
      "step": 9530
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.462890625,
      "learning_rate": 2.823117295961975e-05,
      "loss": 2.1402,
      "step": 9540
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.427734375,
      "learning_rate": 2.8226311488147897e-05,
      "loss": 2.1369,
      "step": 9550
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.44140625,
      "learning_rate": 2.8221443764933125e-05,
      "loss": 2.1434,
      "step": 9560
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.443359375,
      "learning_rate": 2.8216569792276274e-05,
      "loss": 2.1244,
      "step": 9570
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.423828125,
      "learning_rate": 2.8211689572481156e-05,
      "loss": 2.1241,
      "step": 9580
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.439453125,
      "learning_rate": 2.820680310785452e-05,
      "loss": 2.1161,
      "step": 9590
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.43359375,
      "learning_rate": 2.820191040070607e-05,
      "loss": 2.1528,
      "step": 9600
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.453125,
      "learning_rate": 2.8197011453348474e-05,
      "loss": 2.1271,
      "step": 9610
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.443359375,
      "learning_rate": 2.8192106268097336e-05,
      "loss": 2.1448,
      "step": 9620
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.470703125,
      "learning_rate": 2.8187194847271218e-05,
      "loss": 2.1266,
      "step": 9630
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.462890625,
      "learning_rate": 2.8182277193191617e-05,
      "loss": 2.1396,
      "step": 9640
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.53125,
      "learning_rate": 2.817735330818299e-05,
      "loss": 2.1245,
      "step": 9650
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.453125,
      "learning_rate": 2.8172423194572736e-05,
      "loss": 2.0962,
      "step": 9660
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.44140625,
      "learning_rate": 2.8167486854691178e-05,
      "loss": 2.1343,
      "step": 9670
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.451171875,
      "learning_rate": 2.8162544290871616e-05,
      "loss": 2.1278,
      "step": 9680
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4375,
      "learning_rate": 2.815759550545027e-05,
      "loss": 2.1441,
      "step": 9690
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.435546875,
      "learning_rate": 2.8152640500766307e-05,
      "loss": 2.1313,
      "step": 9700
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.42578125,
      "learning_rate": 2.8147679279161826e-05,
      "loss": 2.1243,
      "step": 9710
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.52734375,
      "learning_rate": 2.8142711842981878e-05,
      "loss": 2.1317,
      "step": 9720
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.42578125,
      "learning_rate": 2.8137738194574443e-05,
      "loss": 2.1158,
      "step": 9730
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.44921875,
      "learning_rate": 2.8132758336290435e-05,
      "loss": 2.1255,
      "step": 9740
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.42578125,
      "learning_rate": 2.8127772270483714e-05,
      "loss": 2.1453,
      "step": 9750
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4296875,
      "learning_rate": 2.8122779999511062e-05,
      "loss": 2.1354,
      "step": 9760
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.447265625,
      "learning_rate": 2.8117781525732197e-05,
      "loss": 2.1403,
      "step": 9770
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4296875,
      "learning_rate": 2.8112776851509774e-05,
      "loss": 2.1206,
      "step": 9780
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.451171875,
      "learning_rate": 2.810776597920938e-05,
      "loss": 2.1437,
      "step": 9790
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41796875,
      "learning_rate": 2.8102748911199524e-05,
      "loss": 2.1329,
      "step": 9800
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.416015625,
      "learning_rate": 2.8097725649851645e-05,
      "loss": 2.1559,
      "step": 9810
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.462890625,
      "learning_rate": 2.8092696197540117e-05,
      "loss": 2.1592,
      "step": 9820
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.451171875,
      "learning_rate": 2.8087660556642222e-05,
      "loss": 2.1137,
      "step": 9830
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41796875,
      "learning_rate": 2.80826187295382e-05,
      "loss": 2.1521,
      "step": 9840
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.498046875,
      "learning_rate": 2.8077570718611184e-05,
      "loss": 2.0977,
      "step": 9850
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.478515625,
      "learning_rate": 2.8072516526247238e-05,
      "loss": 2.1278,
      "step": 9860
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4375,
      "learning_rate": 2.8067456154835354e-05,
      "loss": 2.1321,
      "step": 9870
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4375,
      "learning_rate": 2.8062389606767445e-05,
      "loss": 2.0997,
      "step": 9880
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.431640625,
      "learning_rate": 2.805731688443834e-05,
      "loss": 2.1169,
      "step": 9890
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.48046875,
      "learning_rate": 2.8052237990245778e-05,
      "loss": 2.1393,
      "step": 9900
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.435546875,
      "learning_rate": 2.8047152926590432e-05,
      "loss": 2.1279,
      "step": 9910
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4296875,
      "learning_rate": 2.8042061695875883e-05,
      "loss": 2.1392,
      "step": 9920
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.46875,
      "learning_rate": 2.8036964300508625e-05,
      "loss": 2.1144,
      "step": 9930
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.443359375,
      "learning_rate": 2.803186074289807e-05,
      "loss": 2.1193,
      "step": 9940
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4296875,
      "learning_rate": 2.802675102545653e-05,
      "loss": 2.1541,
      "step": 9950
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.43359375,
      "learning_rate": 2.8021635150599255e-05,
      "loss": 2.1294,
      "step": 9960
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.462890625,
      "learning_rate": 2.801651312074438e-05,
      "loss": 2.1278,
      "step": 9970
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.46875,
      "learning_rate": 2.801138493831296e-05,
      "loss": 2.1371,
      "step": 9980
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.443359375,
      "learning_rate": 2.8006250605728954e-05,
      "loss": 2.1129,
      "step": 9990
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.453125,
      "learning_rate": 2.8001110125419246e-05,
      "loss": 2.1196,
      "step": 10000
    },
    {
      "epoch": 0.21,
      "eval_accuracy": 0.5567691006021177,
      "eval_loss": 2.0049471855163574,
      "eval_runtime": 16.4359,
      "eval_samples_per_second": 36.201,
      "eval_steps_per_second": 1.156,
      "step": 10000
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.42578125,
      "learning_rate": 2.7995963499813596e-05,
      "loss": 2.1338,
      "step": 10010
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.43359375,
      "learning_rate": 2.7990810731344686e-05,
      "loss": 2.1141,
      "step": 10020
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.44921875,
      "learning_rate": 2.7985651822448102e-05,
      "loss": 2.1257,
      "step": 10030
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.44140625,
      "learning_rate": 2.7980486775562332e-05,
      "loss": 2.1507,
      "step": 10040
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.435546875,
      "learning_rate": 2.797531559312876e-05,
      "loss": 2.1074,
      "step": 10050
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.455078125,
      "learning_rate": 2.797013827759167e-05,
      "loss": 2.1374,
      "step": 10060
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.431640625,
      "learning_rate": 2.7964954831398252e-05,
      "loss": 2.1046,
      "step": 10070
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.447265625,
      "learning_rate": 2.7959765256998596e-05,
      "loss": 2.1433,
      "step": 10080
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4453125,
      "learning_rate": 2.795456955684567e-05,
      "loss": 2.1355,
      "step": 10090
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.470703125,
      "learning_rate": 2.794936773339536e-05,
      "loss": 2.112,
      "step": 10100
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4375,
      "learning_rate": 2.794415978910643e-05,
      "loss": 2.132,
      "step": 10110
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40625,
      "learning_rate": 2.793894572644055e-05,
      "loss": 2.1753,
      "step": 10120
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.43359375,
      "learning_rate": 2.7933725547862267e-05,
      "loss": 2.1266,
      "step": 10130
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4609375,
      "learning_rate": 2.7928499255839033e-05,
      "loss": 2.1375,
      "step": 10140
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.44921875,
      "learning_rate": 2.7923266852841182e-05,
      "loss": 2.1419,
      "step": 10150
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.49609375,
      "learning_rate": 2.7918028341341944e-05,
      "loss": 2.1547,
      "step": 10160
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.423828125,
      "learning_rate": 2.791278372381742e-05,
      "loss": 2.1191,
      "step": 10170
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.54296875,
      "learning_rate": 2.7907533002746616e-05,
      "loss": 2.1251,
      "step": 10180
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4453125,
      "learning_rate": 2.790227618061142e-05,
      "loss": 2.1291,
      "step": 10190
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.43359375,
      "learning_rate": 2.7897013259896586e-05,
      "loss": 2.1493,
      "step": 10200
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.421875,
      "learning_rate": 2.7891744243089775e-05,
      "loss": 2.1669,
      "step": 10210
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4296875,
      "learning_rate": 2.7886469132681507e-05,
      "loss": 2.1352,
      "step": 10220
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.486328125,
      "learning_rate": 2.7881187931165207e-05,
      "loss": 2.1588,
      "step": 10230
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.435546875,
      "learning_rate": 2.7875900641037155e-05,
      "loss": 2.1518,
      "step": 10240
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.419921875,
      "learning_rate": 2.787060726479652e-05,
      "loss": 2.1527,
      "step": 10250
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.443359375,
      "learning_rate": 2.7865307804945352e-05,
      "loss": 2.1195,
      "step": 10260
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.435546875,
      "learning_rate": 2.7860002263988565e-05,
      "loss": 2.1271,
      "step": 10270
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.462890625,
      "learning_rate": 2.785469064443396e-05,
      "loss": 2.1163,
      "step": 10280
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.46875,
      "learning_rate": 2.7849372948792204e-05,
      "loss": 2.1123,
      "step": 10290
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.43359375,
      "learning_rate": 2.7844049179576837e-05,
      "loss": 2.1576,
      "step": 10300
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.59765625,
      "learning_rate": 2.7838719339304265e-05,
      "loss": 2.0895,
      "step": 10310
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7734375,
      "learning_rate": 2.7833383430493776e-05,
      "loss": 2.1207,
      "step": 10320
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.435546875,
      "learning_rate": 2.782804145566751e-05,
      "loss": 2.1284,
      "step": 10330
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.51171875,
      "learning_rate": 2.78226934173505e-05,
      "loss": 2.1896,
      "step": 10340
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4453125,
      "learning_rate": 2.781733931807061e-05,
      "loss": 2.1414,
      "step": 10350
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.453125,
      "learning_rate": 2.78119791603586e-05,
      "loss": 2.1309,
      "step": 10360
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.423828125,
      "learning_rate": 2.7806612946748073e-05,
      "loss": 2.13,
      "step": 10370
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.45703125,
      "learning_rate": 2.7801240679775508e-05,
      "loss": 2.1399,
      "step": 10380
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.42578125,
      "learning_rate": 2.7795862361980237e-05,
      "loss": 2.0951,
      "step": 10390
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.431640625,
      "learning_rate": 2.7790477995904454e-05,
      "loss": 2.1012,
      "step": 10400
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.412109375,
      "learning_rate": 2.7785087584093216e-05,
      "loss": 2.1295,
      "step": 10410
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.458984375,
      "learning_rate": 2.7779691129094435e-05,
      "loss": 2.1594,
      "step": 10420
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.45703125,
      "learning_rate": 2.7774288633458865e-05,
      "loss": 2.1037,
      "step": 10430
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.416015625,
      "learning_rate": 2.776888009974015e-05,
      "loss": 2.1035,
      "step": 10440
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4453125,
      "learning_rate": 2.7763465530494753e-05,
      "loss": 2.1344,
      "step": 10450
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.453125,
      "learning_rate": 2.775804492828201e-05,
      "loss": 2.1294,
      "step": 10460
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.43359375,
      "learning_rate": 2.7752618295664103e-05,
      "loss": 2.1379,
      "step": 10470
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.443359375,
      "learning_rate": 2.7747185635206057e-05,
      "loss": 2.1426,
      "step": 10480
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.443359375,
      "learning_rate": 2.7741746949475763e-05,
      "loss": 2.1441,
      "step": 10490
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.43359375,
      "learning_rate": 2.773630224104394e-05,
      "loss": 2.1634,
      "step": 10500
    },
    {
      "epoch": 0.22,
      "eval_accuracy": 0.5568348160378521,
      "eval_loss": 2.0042004585266113,
      "eval_runtime": 16.4274,
      "eval_samples_per_second": 36.22,
      "eval_steps_per_second": 1.157,
      "step": 10500
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4609375,
      "learning_rate": 2.7730851512484168e-05,
      "loss": 2.1206,
      "step": 10510
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.451171875,
      "learning_rate": 2.7725394766372874e-05,
      "loss": 2.1319,
      "step": 10520
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4453125,
      "learning_rate": 2.7719932005289317e-05,
      "loss": 2.1057,
      "step": 10530
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4375,
      "learning_rate": 2.7714463231815614e-05,
      "loss": 2.1158,
      "step": 10540
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.435546875,
      "learning_rate": 2.77089884485367e-05,
      "loss": 2.145,
      "step": 10550
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4296875,
      "learning_rate": 2.7703507658040387e-05,
      "loss": 2.1266,
      "step": 10560
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.46875,
      "learning_rate": 2.7698020862917288e-05,
      "loss": 2.1457,
      "step": 10570
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.462890625,
      "learning_rate": 2.7692528065760886e-05,
      "loss": 2.1206,
      "step": 10580
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4375,
      "learning_rate": 2.7687029269167477e-05,
      "loss": 2.1058,
      "step": 10590
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.44921875,
      "learning_rate": 2.7681524475736212e-05,
      "loss": 2.1258,
      "step": 10600
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.494140625,
      "learning_rate": 2.767601368806906e-05,
      "loss": 2.1204,
      "step": 10610
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4375,
      "learning_rate": 2.767049690877084e-05,
      "loss": 2.1271,
      "step": 10620
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.58203125,
      "learning_rate": 2.766497414044918e-05,
      "loss": 2.1059,
      "step": 10630
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.44921875,
      "learning_rate": 2.7659445385714567e-05,
      "loss": 2.1184,
      "step": 10640
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.435546875,
      "learning_rate": 2.7653910647180298e-05,
      "loss": 2.1299,
      "step": 10650
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.416015625,
      "learning_rate": 2.7648369927462503e-05,
      "loss": 2.1336,
      "step": 10660
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.41015625,
      "learning_rate": 2.7642823229180138e-05,
      "loss": 2.113,
      "step": 10670
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4296875,
      "learning_rate": 2.763727055495499e-05,
      "loss": 2.1126,
      "step": 10680
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.44140625,
      "learning_rate": 2.763171190741167e-05,
      "loss": 2.1323,
      "step": 10690
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.43359375,
      "learning_rate": 2.7626147289177602e-05,
      "loss": 2.1229,
      "step": 10700
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4296875,
      "learning_rate": 2.7620576702883053e-05,
      "loss": 2.1137,
      "step": 10710
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.44140625,
      "learning_rate": 2.761500015116109e-05,
      "loss": 2.1521,
      "step": 10720
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4296875,
      "learning_rate": 2.7609417636647602e-05,
      "loss": 2.1017,
      "step": 10730
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4375,
      "learning_rate": 2.7603829161981317e-05,
      "loss": 2.1455,
      "step": 10740
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4453125,
      "learning_rate": 2.7598234729803753e-05,
      "loss": 2.1034,
      "step": 10750
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4453125,
      "learning_rate": 2.7592634342759265e-05,
      "loss": 2.1238,
      "step": 10760
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4375,
      "learning_rate": 2.758702800349501e-05,
      "loss": 2.1025,
      "step": 10770
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.458984375,
      "learning_rate": 2.7581415714660963e-05,
      "loss": 2.1239,
      "step": 10780
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.443359375,
      "learning_rate": 2.7575797478909914e-05,
      "loss": 2.0961,
      "step": 10790
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.47265625,
      "learning_rate": 2.7570173298897458e-05,
      "loss": 2.1189,
      "step": 10800
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.44140625,
      "learning_rate": 2.7564543177282e-05,
      "loss": 2.1192,
      "step": 10810
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.447265625,
      "learning_rate": 2.7558907116724766e-05,
      "loss": 2.0905,
      "step": 10820
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.443359375,
      "learning_rate": 2.7553265119889765e-05,
      "loss": 2.1482,
      "step": 10830
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.431640625,
      "learning_rate": 2.7547617189443835e-05,
      "loss": 2.0997,
      "step": 10840
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4453125,
      "learning_rate": 2.7541963328056616e-05,
      "loss": 2.1088,
      "step": 10850
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4296875,
      "learning_rate": 2.7536303538400532e-05,
      "loss": 2.1213,
      "step": 10860
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.46484375,
      "learning_rate": 2.753063782315083e-05,
      "loss": 2.1114,
      "step": 10870
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.447265625,
      "learning_rate": 2.752496618498555e-05,
      "loss": 2.1292,
      "step": 10880
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.447265625,
      "learning_rate": 2.7519288626585528e-05,
      "loss": 2.1307,
      "step": 10890
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.447265625,
      "learning_rate": 2.751360515063441e-05,
      "loss": 2.13,
      "step": 10900
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.42578125,
      "learning_rate": 2.7507915759818622e-05,
      "loss": 2.14,
      "step": 10910
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.48046875,
      "learning_rate": 2.75022204568274e-05,
      "loss": 2.1275,
      "step": 10920
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.447265625,
      "learning_rate": 2.749651924435277e-05,
      "loss": 2.1041,
      "step": 10930
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.421875,
      "learning_rate": 2.749081212508955e-05,
      "loss": 2.1419,
      "step": 10940
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.435546875,
      "learning_rate": 2.7485099101735348e-05,
      "loss": 2.1523,
      "step": 10950
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.421875,
      "learning_rate": 2.7479380176990575e-05,
      "loss": 2.1015,
      "step": 10960
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.44140625,
      "learning_rate": 2.747365535355841e-05,
      "loss": 2.1573,
      "step": 10970
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.427734375,
      "learning_rate": 2.746792463414484e-05,
      "loss": 2.1187,
      "step": 10980
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4296875,
      "learning_rate": 2.746218802145863e-05,
      "loss": 2.1564,
      "step": 10990
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4375,
      "learning_rate": 2.7456445518211332e-05,
      "loss": 2.1361,
      "step": 11000
    },
    {
      "epoch": 0.23,
      "eval_accuracy": 0.5572570377124457,
      "eval_loss": 2.003464937210083,
      "eval_runtime": 16.4509,
      "eval_samples_per_second": 36.168,
      "eval_steps_per_second": 1.155,
      "step": 11000
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5234375,
      "learning_rate": 2.7450697127117273e-05,
      "loss": 2.1047,
      "step": 11010
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.46484375,
      "learning_rate": 2.7444942850893588e-05,
      "loss": 2.1343,
      "step": 11020
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.423828125,
      "learning_rate": 2.743918269226017e-05,
      "loss": 2.1454,
      "step": 11030
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4296875,
      "learning_rate": 2.7433416653939696e-05,
      "loss": 2.1369,
      "step": 11040
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.435546875,
      "learning_rate": 2.7427644738657635e-05,
      "loss": 2.1412,
      "step": 11050
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4453125,
      "learning_rate": 2.7421866949142216e-05,
      "loss": 2.0932,
      "step": 11060
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4609375,
      "learning_rate": 2.7416083288124464e-05,
      "loss": 2.1111,
      "step": 11070
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.43359375,
      "learning_rate": 2.741029375833816e-05,
      "loss": 2.142,
      "step": 11080
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4375,
      "learning_rate": 2.7404498362519872e-05,
      "loss": 2.1368,
      "step": 11090
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.44140625,
      "learning_rate": 2.739869710340894e-05,
      "loss": 2.1066,
      "step": 11100
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.431640625,
      "learning_rate": 2.7392889983747473e-05,
      "loss": 2.1737,
      "step": 11110
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.443359375,
      "learning_rate": 2.738707700628034e-05,
      "loss": 2.1517,
      "step": 11120
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.443359375,
      "learning_rate": 2.73812581737552e-05,
      "loss": 2.1411,
      "step": 11130
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.43359375,
      "learning_rate": 2.737543348892246e-05,
      "loss": 2.1118,
      "step": 11140
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.427734375,
      "learning_rate": 2.7369602954535308e-05,
      "loss": 2.1332,
      "step": 11150
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.43359375,
      "learning_rate": 2.736376657334968e-05,
      "loss": 2.1295,
      "step": 11160
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.494140625,
      "learning_rate": 2.73579243481243e-05,
      "loss": 2.0897,
      "step": 11170
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.419921875,
      "learning_rate": 2.735207628162063e-05,
      "loss": 2.1324,
      "step": 11180
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.462890625,
      "learning_rate": 2.7346222376602906e-05,
      "loss": 2.1254,
      "step": 11190
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4296875,
      "learning_rate": 2.7340362635838115e-05,
      "loss": 2.1349,
      "step": 11200
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.470703125,
      "learning_rate": 2.7334497062096018e-05,
      "loss": 2.1281,
      "step": 11210
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.458984375,
      "learning_rate": 2.732862565814912e-05,
      "loss": 2.1122,
      "step": 11220
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.447265625,
      "learning_rate": 2.7322748426772678e-05,
      "loss": 2.1334,
      "step": 11230
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.46875,
      "learning_rate": 2.731686537074472e-05,
      "loss": 2.1648,
      "step": 11240
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.45703125,
      "learning_rate": 2.7310976492846013e-05,
      "loss": 2.1501,
      "step": 11250
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.455078125,
      "learning_rate": 2.7305081795860084e-05,
      "loss": 2.139,
      "step": 11260
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.474609375,
      "learning_rate": 2.7299181282573206e-05,
      "loss": 2.134,
      "step": 11270
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.427734375,
      "learning_rate": 2.72932749557744e-05,
      "loss": 2.0959,
      "step": 11280
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.421875,
      "learning_rate": 2.7287362818255438e-05,
      "loss": 2.0825,
      "step": 11290
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4296875,
      "learning_rate": 2.7281444872810843e-05,
      "loss": 2.1234,
      "step": 11300
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.478515625,
      "learning_rate": 2.727552112223787e-05,
      "loss": 2.1073,
      "step": 11310
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.427734375,
      "learning_rate": 2.7269591569336532e-05,
      "loss": 2.1179,
      "step": 11320
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.43359375,
      "learning_rate": 2.7263656216909574e-05,
      "loss": 2.1269,
      "step": 11330
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.447265625,
      "learning_rate": 2.7257715067762493e-05,
      "loss": 2.1363,
      "step": 11340
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.46484375,
      "learning_rate": 2.7251768124703512e-05,
      "loss": 2.1344,
      "step": 11350
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5,
      "learning_rate": 2.7245815390543612e-05,
      "loss": 2.0938,
      "step": 11360
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.44921875,
      "learning_rate": 2.7239856868096488e-05,
      "loss": 2.138,
      "step": 11370
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5625,
      "learning_rate": 2.7233892560178594e-05,
      "loss": 2.094,
      "step": 11380
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42578125,
      "learning_rate": 2.7227922469609094e-05,
      "loss": 2.1032,
      "step": 11390
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.447265625,
      "learning_rate": 2.722194659920991e-05,
      "loss": 2.1207,
      "step": 11400
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4375,
      "learning_rate": 2.721596495180568e-05,
      "loss": 2.1311,
      "step": 11410
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.427734375,
      "learning_rate": 2.720997753022378e-05,
      "loss": 2.1437,
      "step": 11420
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4375,
      "learning_rate": 2.720398433729431e-05,
      "loss": 2.1223,
      "step": 11430
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.419921875,
      "learning_rate": 2.71979853758501e-05,
      "loss": 2.1518,
      "step": 11440
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.427734375,
      "learning_rate": 2.7191980648726714e-05,
      "loss": 2.118,
      "step": 11450
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.41796875,
      "learning_rate": 2.7185970158762423e-05,
      "loss": 2.1094,
      "step": 11460
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.427734375,
      "learning_rate": 2.7179953908798246e-05,
      "loss": 2.1038,
      "step": 11470
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.443359375,
      "learning_rate": 2.7173931901677908e-05,
      "loss": 2.1237,
      "step": 11480
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.447265625,
      "learning_rate": 2.7167904140247855e-05,
      "loss": 2.0998,
      "step": 11490
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.462890625,
      "learning_rate": 2.7161870627357263e-05,
      "loss": 2.1614,
      "step": 11500
    },
    {
      "epoch": 0.24,
      "eval_accuracy": 0.5572373230817254,
      "eval_loss": 2.002742052078247,
      "eval_runtime": 16.4495,
      "eval_samples_per_second": 36.171,
      "eval_steps_per_second": 1.155,
      "step": 11500
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.421875,
      "learning_rate": 2.7155831365858016e-05,
      "loss": 2.123,
      "step": 11510
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4609375,
      "learning_rate": 2.714978635860472e-05,
      "loss": 2.1166,
      "step": 11520
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.435546875,
      "learning_rate": 2.7143735608454704e-05,
      "loss": 2.1447,
      "step": 11530
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.466796875,
      "learning_rate": 2.7137679118268004e-05,
      "loss": 2.1581,
      "step": 11540
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.427734375,
      "learning_rate": 2.713161689090736e-05,
      "loss": 2.1443,
      "step": 11550
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.423828125,
      "learning_rate": 2.712554892923824e-05,
      "loss": 2.1002,
      "step": 11560
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.431640625,
      "learning_rate": 2.7119475236128816e-05,
      "loss": 2.1158,
      "step": 11570
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.421875,
      "learning_rate": 2.711339581444997e-05,
      "loss": 2.1226,
      "step": 11580
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.431640625,
      "learning_rate": 2.710731066707529e-05,
      "loss": 2.1133,
      "step": 11590
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.419921875,
      "learning_rate": 2.7101219796881068e-05,
      "loss": 2.136,
      "step": 11600
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.462890625,
      "learning_rate": 2.7095123206746313e-05,
      "loss": 2.1554,
      "step": 11610
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.46484375,
      "learning_rate": 2.708902089955272e-05,
      "loss": 2.1598,
      "step": 11620
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.55078125,
      "learning_rate": 2.70829128781847e-05,
      "loss": 2.1231,
      "step": 11630
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.43359375,
      "learning_rate": 2.7076799145529353e-05,
      "loss": 2.1169,
      "step": 11640
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.48046875,
      "learning_rate": 2.7070679704476497e-05,
      "loss": 2.1103,
      "step": 11650
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.43359375,
      "learning_rate": 2.706455455791863e-05,
      "loss": 2.1337,
      "step": 11660
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.43359375,
      "learning_rate": 2.7058423708750954e-05,
      "loss": 2.1526,
      "step": 11670
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4375,
      "learning_rate": 2.7052287159871366e-05,
      "loss": 2.1409,
      "step": 11680
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.44140625,
      "learning_rate": 2.7046144914180463e-05,
      "loss": 2.126,
      "step": 11690
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.453125,
      "learning_rate": 2.7039996974581524e-05,
      "loss": 2.1366,
      "step": 11700
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.44140625,
      "learning_rate": 2.7033843343980523e-05,
      "loss": 2.1041,
      "step": 11710
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.453125,
      "learning_rate": 2.7027684025286134e-05,
      "loss": 2.1407,
      "step": 11720
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.427734375,
      "learning_rate": 2.7021519021409704e-05,
      "loss": 2.1085,
      "step": 11730
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.427734375,
      "learning_rate": 2.7015348335265276e-05,
      "loss": 2.1546,
      "step": 11740
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4609375,
      "learning_rate": 2.7009171969769576e-05,
      "loss": 2.1217,
      "step": 11750
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4453125,
      "learning_rate": 2.7002989927842013e-05,
      "loss": 2.1162,
      "step": 11760
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.435546875,
      "learning_rate": 2.699680221240469e-05,
      "loss": 2.1309,
      "step": 11770
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.443359375,
      "learning_rate": 2.699060882638238e-05,
      "loss": 2.0804,
      "step": 11780
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.451171875,
      "learning_rate": 2.6984409772702537e-05,
      "loss": 2.1226,
      "step": 11790
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.455078125,
      "learning_rate": 2.6978205054295302e-05,
      "loss": 2.1519,
      "step": 11800
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5078125,
      "learning_rate": 2.6971994674093484e-05,
      "loss": 2.1112,
      "step": 11810
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.443359375,
      "learning_rate": 2.6965778635032573e-05,
      "loss": 2.0919,
      "step": 11820
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.431640625,
      "learning_rate": 2.6959556940050736e-05,
      "loss": 2.1574,
      "step": 11830
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.431640625,
      "learning_rate": 2.695332959208881e-05,
      "loss": 2.121,
      "step": 11840
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.45703125,
      "learning_rate": 2.6947096594090306e-05,
      "loss": 2.0938,
      "step": 11850
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.47265625,
      "learning_rate": 2.6940857949001403e-05,
      "loss": 2.115,
      "step": 11860
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.447265625,
      "learning_rate": 2.693461365977095e-05,
      "loss": 2.141,
      "step": 11870
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.470703125,
      "learning_rate": 2.6928363729350465e-05,
      "loss": 2.1445,
      "step": 11880
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.421875,
      "learning_rate": 2.6922108160694133e-05,
      "loss": 2.1487,
      "step": 11890
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.419921875,
      "learning_rate": 2.69158469567588e-05,
      "loss": 2.1153,
      "step": 11900
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.439453125,
      "learning_rate": 2.6909580120503985e-05,
      "loss": 2.1072,
      "step": 11910
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.455078125,
      "learning_rate": 2.6903307654891858e-05,
      "loss": 2.0888,
      "step": 11920
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.431640625,
      "learning_rate": 2.6897029562887257e-05,
      "loss": 2.12,
      "step": 11930
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.443359375,
      "learning_rate": 2.6890745847457676e-05,
      "loss": 2.1288,
      "step": 11940
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.42578125,
      "learning_rate": 2.6884456511573266e-05,
      "loss": 2.0738,
      "step": 11950
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4140625,
      "learning_rate": 2.6878161558206845e-05,
      "loss": 2.1162,
      "step": 11960
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.458984375,
      "learning_rate": 2.6871860990333866e-05,
      "loss": 2.1009,
      "step": 11970
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.427734375,
      "learning_rate": 2.6865554810932464e-05,
      "loss": 2.147,
      "step": 11980
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.53125,
      "learning_rate": 2.68592430229834e-05,
      "loss": 2.1533,
      "step": 11990
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.42578125,
      "learning_rate": 2.6852925629470092e-05,
      "loss": 2.1205,
      "step": 12000
    },
    {
      "epoch": 0.25,
      "eval_accuracy": 0.5575905435487978,
      "eval_loss": 2.0020804405212402,
      "eval_runtime": 16.4541,
      "eval_samples_per_second": 36.161,
      "eval_steps_per_second": 1.155,
      "step": 12000
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4296875,
      "learning_rate": 2.684660263337862e-05,
      "loss": 2.1284,
      "step": 12010
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.427734375,
      "learning_rate": 2.6840274037697706e-05,
      "loss": 2.096,
      "step": 12020
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.41796875,
      "learning_rate": 2.6833939845418713e-05,
      "loss": 2.1219,
      "step": 12030
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.431640625,
      "learning_rate": 2.682760005953565e-05,
      "loss": 2.1337,
      "step": 12040
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.439453125,
      "learning_rate": 2.6821254683045184e-05,
      "loss": 2.17,
      "step": 12050
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.443359375,
      "learning_rate": 2.6814903718946605e-05,
      "loss": 2.1433,
      "step": 12060
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.451171875,
      "learning_rate": 2.6808547170241857e-05,
      "loss": 2.1263,
      "step": 12070
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.44140625,
      "learning_rate": 2.6802185039935514e-05,
      "loss": 2.0722,
      "step": 12080
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4921875,
      "learning_rate": 2.6795817331034803e-05,
      "loss": 2.1319,
      "step": 12090
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.421875,
      "learning_rate": 2.678944404654957e-05,
      "loss": 2.1281,
      "step": 12100
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.41796875,
      "learning_rate": 2.678306518949232e-05,
      "loss": 2.1321,
      "step": 12110
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4375,
      "learning_rate": 2.677668076287816e-05,
      "loss": 2.0867,
      "step": 12120
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.443359375,
      "learning_rate": 2.6770290769724856e-05,
      "loss": 2.0957,
      "step": 12130
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.427734375,
      "learning_rate": 2.676389521305279e-05,
      "loss": 2.0962,
      "step": 12140
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4453125,
      "learning_rate": 2.675749409588499e-05,
      "loss": 2.0884,
      "step": 12150
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.435546875,
      "learning_rate": 2.6751087421247096e-05,
      "loss": 2.0776,
      "step": 12160
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.42578125,
      "learning_rate": 2.674467519216738e-05,
      "loss": 2.1515,
      "step": 12170
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.453125,
      "learning_rate": 2.6738257411676746e-05,
      "loss": 2.1342,
      "step": 12180
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.439453125,
      "learning_rate": 2.673183408280871e-05,
      "loss": 2.1562,
      "step": 12190
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.439453125,
      "learning_rate": 2.6725405208599417e-05,
      "loss": 2.1546,
      "step": 12200
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.42578125,
      "learning_rate": 2.6718970792087642e-05,
      "loss": 2.1857,
      "step": 12210
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.45703125,
      "learning_rate": 2.6712530836314763e-05,
      "loss": 2.1436,
      "step": 12220
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.42578125,
      "learning_rate": 2.6706085344324786e-05,
      "loss": 2.1,
      "step": 12230
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.455078125,
      "learning_rate": 2.669963431916433e-05,
      "loss": 2.1346,
      "step": 12240
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.494140625,
      "learning_rate": 2.669317776388263e-05,
      "loss": 2.1038,
      "step": 12250
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4140625,
      "learning_rate": 2.6686715681531546e-05,
      "loss": 2.1231,
      "step": 12260
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.435546875,
      "learning_rate": 2.6680248075165535e-05,
      "loss": 2.1126,
      "step": 12270
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.419921875,
      "learning_rate": 2.6673774947841673e-05,
      "loss": 2.1341,
      "step": 12280
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.44140625,
      "learning_rate": 2.6667296302619634e-05,
      "loss": 2.0758,
      "step": 12290
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.431640625,
      "learning_rate": 2.6660812142561723e-05,
      "loss": 2.1322,
      "step": 12300
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.451171875,
      "learning_rate": 2.665432247073283e-05,
      "loss": 2.1306,
      "step": 12310
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.44921875,
      "learning_rate": 2.6647827290200467e-05,
      "loss": 2.1324,
      "step": 12320
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.435546875,
      "learning_rate": 2.6641326604034736e-05,
      "loss": 2.1077,
      "step": 12330
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0,
      "learning_rate": 2.663482041530834e-05,
      "loss": 2.1343,
      "step": 12340
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.451171875,
      "learning_rate": 2.66283087270966e-05,
      "loss": 2.1316,
      "step": 12350
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.46875,
      "learning_rate": 2.662179154247742e-05,
      "loss": 2.1421,
      "step": 12360
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.43359375,
      "learning_rate": 2.6615268864531318e-05,
      "loss": 2.1153,
      "step": 12370
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.44140625,
      "learning_rate": 2.660874069634139e-05,
      "loss": 2.1173,
      "step": 12380
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.45703125,
      "learning_rate": 2.660220704099334e-05,
      "loss": 2.1345,
      "step": 12390
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.447265625,
      "learning_rate": 2.6595667901575455e-05,
      "loss": 2.1061,
      "step": 12400
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.458984375,
      "learning_rate": 2.658912328117863e-05,
      "loss": 2.1338,
      "step": 12410
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.423828125,
      "learning_rate": 2.6582573182896336e-05,
      "loss": 2.1018,
      "step": 12420
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.44140625,
      "learning_rate": 2.657601760982464e-05,
      "loss": 2.121,
      "step": 12430
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4453125,
      "learning_rate": 2.6569456565062194e-05,
      "loss": 2.1161,
      "step": 12440
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.423828125,
      "learning_rate": 2.656289005171024e-05,
      "loss": 2.1494,
      "step": 12450
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.455078125,
      "learning_rate": 2.6556318072872607e-05,
      "loss": 2.1486,
      "step": 12460
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41796875,
      "learning_rate": 2.6549740631655695e-05,
      "loss": 2.1253,
      "step": 12470
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.42578125,
      "learning_rate": 2.6543157731168502e-05,
      "loss": 2.1218,
      "step": 12480
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.447265625,
      "learning_rate": 2.6536569374522585e-05,
      "loss": 2.1214,
      "step": 12490
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.44140625,
      "learning_rate": 2.6529975564832113e-05,
      "loss": 2.0984,
      "step": 12500
    },
    {
      "epoch": 0.26,
      "eval_accuracy": 0.5575626144886107,
      "eval_loss": 2.0010862350463867,
      "eval_runtime": 16.4543,
      "eval_samples_per_second": 36.161,
      "eval_steps_per_second": 1.155,
      "step": 12500
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4375,
      "learning_rate": 2.65233763052138e-05,
      "loss": 2.1521,
      "step": 12510
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.46875,
      "learning_rate": 2.6516771598786947e-05,
      "loss": 2.1079,
      "step": 12520
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.435546875,
      "learning_rate": 2.6510161448673443e-05,
      "loss": 2.1547,
      "step": 12530
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.462890625,
      "learning_rate": 2.650354585799773e-05,
      "loss": 2.1408,
      "step": 12540
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4296875,
      "learning_rate": 2.649692482988683e-05,
      "loss": 2.1057,
      "step": 12550
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.447265625,
      "learning_rate": 2.6490298367470344e-05,
      "loss": 2.1718,
      "step": 12560
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.453125,
      "learning_rate": 2.6483666473880425e-05,
      "loss": 2.1259,
      "step": 12570
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.435546875,
      "learning_rate": 2.6477029152251804e-05,
      "loss": 2.1342,
      "step": 12580
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.44140625,
      "learning_rate": 2.6470386405721777e-05,
      "loss": 2.1192,
      "step": 12590
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.447265625,
      "learning_rate": 2.64637382374302e-05,
      "loss": 2.0874,
      "step": 12600
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.439453125,
      "learning_rate": 2.645708465051949e-05,
      "loss": 2.1217,
      "step": 12610
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.55859375,
      "learning_rate": 2.6450425648134637e-05,
      "loss": 2.1198,
      "step": 12620
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.42578125,
      "learning_rate": 2.644376123342318e-05,
      "loss": 2.1479,
      "step": 12630
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.447265625,
      "learning_rate": 2.643709140953522e-05,
      "loss": 2.1281,
      "step": 12640
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.451171875,
      "learning_rate": 2.643041617962341e-05,
      "loss": 2.1134,
      "step": 12650
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4765625,
      "learning_rate": 2.6423735546842965e-05,
      "loss": 2.1187,
      "step": 12660
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.44140625,
      "learning_rate": 2.6417049514351655e-05,
      "loss": 2.1359,
      "step": 12670
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.42578125,
      "learning_rate": 2.6410358085309793e-05,
      "loss": 2.1364,
      "step": 12680
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.42578125,
      "learning_rate": 2.640366126288025e-05,
      "loss": 2.135,
      "step": 12690
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.443359375,
      "learning_rate": 2.6396959050228443e-05,
      "loss": 2.1092,
      "step": 12700
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.439453125,
      "learning_rate": 2.6390251450522343e-05,
      "loss": 2.1422,
      "step": 12710
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.421875,
      "learning_rate": 2.638353846693246e-05,
      "loss": 2.1304,
      "step": 12720
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4453125,
      "learning_rate": 2.637682010263185e-05,
      "loss": 2.1503,
      "step": 12730
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.46484375,
      "learning_rate": 2.6370096360796116e-05,
      "loss": 2.1212,
      "step": 12740
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.50390625,
      "learning_rate": 2.6363367244603402e-05,
      "loss": 2.1071,
      "step": 12750
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.43359375,
      "learning_rate": 2.635663275723439e-05,
      "loss": 2.1121,
      "step": 12760
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.447265625,
      "learning_rate": 2.6349892901872298e-05,
      "loss": 2.1268,
      "step": 12770
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4296875,
      "learning_rate": 2.6343147681702887e-05,
      "loss": 2.1267,
      "step": 12780
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.423828125,
      "learning_rate": 2.6336397099914454e-05,
      "loss": 2.1344,
      "step": 12790
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.451171875,
      "learning_rate": 2.6329641159697828e-05,
      "loss": 2.1353,
      "step": 12800
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.458984375,
      "learning_rate": 2.6322879864246375e-05,
      "loss": 2.1403,
      "step": 12810
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.486328125,
      "learning_rate": 2.6316113216755978e-05,
      "loss": 2.102,
      "step": 12820
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.439453125,
      "learning_rate": 2.6309341220425066e-05,
      "loss": 2.1618,
      "step": 12830
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.43359375,
      "learning_rate": 2.630256387845459e-05,
      "loss": 2.1364,
      "step": 12840
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.57421875,
      "learning_rate": 2.6295781194048033e-05,
      "loss": 2.1393,
      "step": 12850
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.439453125,
      "learning_rate": 2.628899317041139e-05,
      "loss": 2.1186,
      "step": 12860
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.421875,
      "learning_rate": 2.6282199810753192e-05,
      "loss": 2.1323,
      "step": 12870
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4296875,
      "learning_rate": 2.6275401118284494e-05,
      "loss": 2.1235,
      "step": 12880
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4296875,
      "learning_rate": 2.6268597096218853e-05,
      "loss": 2.1143,
      "step": 12890
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4453125,
      "learning_rate": 2.626178774777237e-05,
      "loss": 2.1171,
      "step": 12900
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4453125,
      "learning_rate": 2.6254973076163645e-05,
      "loss": 2.1114,
      "step": 12910
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4453125,
      "learning_rate": 2.6248153084613803e-05,
      "loss": 2.125,
      "step": 12920
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.421875,
      "learning_rate": 2.6241327776346482e-05,
      "loss": 2.1128,
      "step": 12930
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.443359375,
      "learning_rate": 2.6234497154587832e-05,
      "loss": 2.1439,
      "step": 12940
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4296875,
      "learning_rate": 2.6227661222566516e-05,
      "loss": 2.1513,
      "step": 12950
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.427734375,
      "learning_rate": 2.6220819983513706e-05,
      "loss": 2.1626,
      "step": 12960
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.427734375,
      "learning_rate": 2.6213973440663083e-05,
      "loss": 2.1124,
      "step": 12970
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4453125,
      "learning_rate": 2.6207121597250835e-05,
      "loss": 2.1114,
      "step": 12980
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4453125,
      "learning_rate": 2.620026445651565e-05,
      "loss": 2.105,
      "step": 12990
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.44140625,
      "learning_rate": 2.6193402021698732e-05,
      "loss": 2.1226,
      "step": 13000
    },
    {
      "epoch": 0.27,
      "eval_accuracy": 0.5575166136835966,
      "eval_loss": 2.0006489753723145,
      "eval_runtime": 16.4376,
      "eval_samples_per_second": 36.198,
      "eval_steps_per_second": 1.156,
      "step": 13000
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.458984375,
      "learning_rate": 2.6186534296043772e-05,
      "loss": 2.1257,
      "step": 13010
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.41796875,
      "learning_rate": 2.617966128279698e-05,
      "loss": 2.0916,
      "step": 13020
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.427734375,
      "learning_rate": 2.617278298520705e-05,
      "loss": 2.0715,
      "step": 13030
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.435546875,
      "learning_rate": 2.616589940652518e-05,
      "loss": 2.1366,
      "step": 13040
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42578125,
      "learning_rate": 2.6159010550005068e-05,
      "loss": 2.0964,
      "step": 13050
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.447265625,
      "learning_rate": 2.6152116418902892e-05,
      "loss": 2.0973,
      "step": 13060
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.46875,
      "learning_rate": 2.6145217016477344e-05,
      "loss": 2.1366,
      "step": 13070
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.46875,
      "learning_rate": 2.6138312345989594e-05,
      "loss": 2.1442,
      "step": 13080
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.44921875,
      "learning_rate": 2.6131402410703295e-05,
      "loss": 2.1418,
      "step": 13090
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4375,
      "learning_rate": 2.6124487213884616e-05,
      "loss": 2.1205,
      "step": 13100
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.478515625,
      "learning_rate": 2.6117566758802186e-05,
      "loss": 2.1157,
      "step": 13110
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.47265625,
      "learning_rate": 2.6110641048727125e-05,
      "loss": 2.1568,
      "step": 13120
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4453125,
      "learning_rate": 2.610371008693305e-05,
      "loss": 2.1396,
      "step": 13130
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.46875,
      "learning_rate": 2.6096773876696045e-05,
      "loss": 2.1056,
      "step": 13140
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.435546875,
      "learning_rate": 2.6089832421294686e-05,
      "loss": 2.116,
      "step": 13150
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.470703125,
      "learning_rate": 2.608288572401002e-05,
      "loss": 2.1278,
      "step": 13160
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.423828125,
      "learning_rate": 2.6075933788125576e-05,
      "loss": 2.1327,
      "step": 13170
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.44140625,
      "learning_rate": 2.6068976616927362e-05,
      "loss": 2.1458,
      "step": 13180
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.41796875,
      "learning_rate": 2.6062014213703848e-05,
      "loss": 2.1489,
      "step": 13190
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.447265625,
      "learning_rate": 2.6055046581745995e-05,
      "loss": 2.1528,
      "step": 13200
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.43359375,
      "learning_rate": 2.604807372434722e-05,
      "loss": 2.1142,
      "step": 13210
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.462890625,
      "learning_rate": 2.604109564480342e-05,
      "loss": 2.1181,
      "step": 13220
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4375,
      "learning_rate": 2.6034112346412957e-05,
      "loss": 2.1306,
      "step": 13230
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44921875,
      "learning_rate": 2.602712383247666e-05,
      "loss": 2.1209,
      "step": 13240
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.443359375,
      "learning_rate": 2.6020130106297814e-05,
      "loss": 2.1028,
      "step": 13250
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43359375,
      "learning_rate": 2.6013131171182185e-05,
      "loss": 2.0995,
      "step": 13260
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.45703125,
      "learning_rate": 2.6006127030437996e-05,
      "loss": 2.109,
      "step": 13270
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.45703125,
      "learning_rate": 2.5999117687375917e-05,
      "loss": 2.1417,
      "step": 13280
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4453125,
      "learning_rate": 2.5992103145309095e-05,
      "loss": 2.1016,
      "step": 13290
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.431640625,
      "learning_rate": 2.5985083407553124e-05,
      "loss": 2.0963,
      "step": 13300
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.470703125,
      "learning_rate": 2.5978058477426055e-05,
      "loss": 2.1199,
      "step": 13310
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.453125,
      "learning_rate": 2.5971028358248396e-05,
      "loss": 2.1211,
      "step": 13320
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.421875,
      "learning_rate": 2.5963993053343108e-05,
      "loss": 2.0817,
      "step": 13330
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.55078125,
      "learning_rate": 2.5956952566035602e-05,
      "loss": 2.0825,
      "step": 13340
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.451171875,
      "learning_rate": 2.5949906899653734e-05,
      "loss": 2.0881,
      "step": 13350
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4296875,
      "learning_rate": 2.594285605752782e-05,
      "loss": 2.1466,
      "step": 13360
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.439453125,
      "learning_rate": 2.5935800042990605e-05,
      "loss": 2.1196,
      "step": 13370
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.447265625,
      "learning_rate": 2.5928738859377296e-05,
      "loss": 2.1781,
      "step": 13380
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.421875,
      "learning_rate": 2.592167251002553e-05,
      "loss": 2.1188,
      "step": 13390
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44140625,
      "learning_rate": 2.59146009982754e-05,
      "loss": 2.1374,
      "step": 13400
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4453125,
      "learning_rate": 2.5907524327469424e-05,
      "loss": 2.1316,
      "step": 13410
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44921875,
      "learning_rate": 2.5900442500952566e-05,
      "loss": 2.1396,
      "step": 13420
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.421875,
      "learning_rate": 2.589335552207223e-05,
      "loss": 2.1401,
      "step": 13430
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.431640625,
      "learning_rate": 2.5886263394178246e-05,
      "loss": 2.1438,
      "step": 13440
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43359375,
      "learning_rate": 2.5879166120622887e-05,
      "loss": 2.1369,
      "step": 13450
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4296875,
      "learning_rate": 2.5872063704760856e-05,
      "loss": 2.1011,
      "step": 13460
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.46875,
      "learning_rate": 2.586495614994928e-05,
      "loss": 2.1311,
      "step": 13470
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.439453125,
      "learning_rate": 2.585784345954773e-05,
      "loss": 2.1379,
      "step": 13480
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4140625,
      "learning_rate": 2.5850725636918183e-05,
      "loss": 2.1109,
      "step": 13490
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.451171875,
      "learning_rate": 2.584360268542506e-05,
      "loss": 2.1054,
      "step": 13500
    },
    {
      "epoch": 0.28,
      "eval_accuracy": 0.5577351175074136,
      "eval_loss": 2.000054121017456,
      "eval_runtime": 16.4541,
      "eval_samples_per_second": 36.161,
      "eval_steps_per_second": 1.155,
      "step": 13500
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4609375,
      "learning_rate": 2.5836474608435202e-05,
      "loss": 2.0967,
      "step": 13510
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43359375,
      "learning_rate": 2.5829341409317866e-05,
      "loss": 2.1492,
      "step": 13520
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.421875,
      "learning_rate": 2.582220309144474e-05,
      "loss": 2.1162,
      "step": 13530
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.423828125,
      "learning_rate": 2.5815059658189924e-05,
      "loss": 2.0795,
      "step": 13540
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.490234375,
      "learning_rate": 2.5807911112929937e-05,
      "loss": 2.1174,
      "step": 13550
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.421875,
      "learning_rate": 2.580075745904371e-05,
      "loss": 2.1442,
      "step": 13560
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.447265625,
      "learning_rate": 2.5793598699912612e-05,
      "loss": 2.1377,
      "step": 13570
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.416015625,
      "learning_rate": 2.5786434838920385e-05,
      "loss": 2.1311,
      "step": 13580
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8515625,
      "learning_rate": 2.577926587945322e-05,
      "loss": 2.1132,
      "step": 13590
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.435546875,
      "learning_rate": 2.5772091824899697e-05,
      "loss": 2.1114,
      "step": 13600
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.439453125,
      "learning_rate": 2.5764912678650814e-05,
      "loss": 2.1312,
      "step": 13610
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5234375,
      "learning_rate": 2.5757728444099973e-05,
      "loss": 2.1173,
      "step": 13620
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.40625,
      "learning_rate": 2.5750539124642972e-05,
      "loss": 2.135,
      "step": 13630
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44140625,
      "learning_rate": 2.5743344723678026e-05,
      "loss": 2.1151,
      "step": 13640
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4453125,
      "learning_rate": 2.5736145244605746e-05,
      "loss": 2.125,
      "step": 13650
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4375,
      "learning_rate": 2.572894069082914e-05,
      "loss": 2.1157,
      "step": 13660
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.49609375,
      "learning_rate": 2.5721731065753626e-05,
      "loss": 2.1441,
      "step": 13670
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.431640625,
      "learning_rate": 2.5714516372787003e-05,
      "loss": 2.1452,
      "step": 13680
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.443359375,
      "learning_rate": 2.570729661533948e-05,
      "loss": 2.1182,
      "step": 13690
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43359375,
      "learning_rate": 2.5700071796823652e-05,
      "loss": 2.1234,
      "step": 13700
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4375,
      "learning_rate": 2.5692841920654504e-05,
      "loss": 2.1208,
      "step": 13710
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.439453125,
      "learning_rate": 2.5685606990249424e-05,
      "loss": 2.0994,
      "step": 13720
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.43359375,
      "learning_rate": 2.5678367009028173e-05,
      "loss": 2.1448,
      "step": 13730
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.423828125,
      "learning_rate": 2.5671121980412908e-05,
      "loss": 2.1197,
      "step": 13740
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4609375,
      "learning_rate": 2.5663871907828175e-05,
      "loss": 2.1325,
      "step": 13750
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.462890625,
      "learning_rate": 2.56566167947009e-05,
      "loss": 2.1228,
      "step": 13760
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.44921875,
      "learning_rate": 2.5649356644460385e-05,
      "loss": 2.1461,
      "step": 13770
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.455078125,
      "learning_rate": 2.5642091460538324e-05,
      "loss": 2.1437,
      "step": 13780
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.470703125,
      "learning_rate": 2.5634821246368784e-05,
      "loss": 2.1147,
      "step": 13790
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.427734375,
      "learning_rate": 2.5627546005388218e-05,
      "loss": 2.1283,
      "step": 13800
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.490234375,
      "learning_rate": 2.5620265741035438e-05,
      "loss": 2.1616,
      "step": 13810
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.458984375,
      "learning_rate": 2.5612980456751643e-05,
      "loss": 2.1316,
      "step": 13820
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.455078125,
      "learning_rate": 2.560569015598041e-05,
      "loss": 2.1151,
      "step": 13830
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.447265625,
      "learning_rate": 2.5598394842167674e-05,
      "loss": 2.0983,
      "step": 13840
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0390625,
      "learning_rate": 2.5591094518761747e-05,
      "loss": 2.1351,
      "step": 13850
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4375,
      "learning_rate": 2.5583789189213307e-05,
      "loss": 2.103,
      "step": 13860
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4296875,
      "learning_rate": 2.5576478856975394e-05,
      "loss": 2.116,
      "step": 13870
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.443359375,
      "learning_rate": 2.5569163525503426e-05,
      "loss": 2.1208,
      "step": 13880
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.435546875,
      "learning_rate": 2.556184319825517e-05,
      "loss": 2.1323,
      "step": 13890
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4296875,
      "learning_rate": 2.555451787869076e-05,
      "loss": 2.1104,
      "step": 13900
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.439453125,
      "learning_rate": 2.5547187570272687e-05,
      "loss": 2.107,
      "step": 13910
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.43359375,
      "learning_rate": 2.5539852276465806e-05,
      "loss": 2.134,
      "step": 13920
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.435546875,
      "learning_rate": 2.5532512000737325e-05,
      "loss": 2.1666,
      "step": 13930
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.44921875,
      "learning_rate": 2.5525166746556802e-05,
      "loss": 2.1158,
      "step": 13940
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.443359375,
      "learning_rate": 2.5517816517396156e-05,
      "loss": 2.1199,
      "step": 13950
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.484375,
      "learning_rate": 2.5510461316729657e-05,
      "loss": 2.1216,
      "step": 13960
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.421875,
      "learning_rate": 2.5503101148033925e-05,
      "loss": 2.0959,
      "step": 13970
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.455078125,
      "learning_rate": 2.5495736014787914e-05,
      "loss": 2.1081,
      "step": 13980
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.474609375,
      "learning_rate": 2.548836592047294e-05,
      "loss": 2.1296,
      "step": 13990
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.421875,
      "learning_rate": 2.548099086857267e-05,
      "loss": 2.1297,
      "step": 14000
    },
    {
      "epoch": 0.29,
      "eval_accuracy": 0.5578221904597617,
      "eval_loss": 1.9996856451034546,
      "eval_runtime": 16.446,
      "eval_samples_per_second": 36.179,
      "eval_steps_per_second": 1.155,
      "step": 14000
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.439453125,
      "learning_rate": 2.547361086257309e-05,
      "loss": 2.1268,
      "step": 14010
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.439453125,
      "learning_rate": 2.5466225905962556e-05,
      "loss": 2.0814,
      "step": 14020
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.515625,
      "learning_rate": 2.545883600223174e-05,
      "loss": 2.1518,
      "step": 14030
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.443359375,
      "learning_rate": 2.5451441154873666e-05,
      "loss": 2.1365,
      "step": 14040
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.435546875,
      "learning_rate": 2.5444041367383696e-05,
      "loss": 2.1052,
      "step": 14050
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.421875,
      "learning_rate": 2.5436636643259515e-05,
      "loss": 2.0966,
      "step": 14060
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.439453125,
      "learning_rate": 2.5429226986001152e-05,
      "loss": 2.1351,
      "step": 14070
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.423828125,
      "learning_rate": 2.5421812399110968e-05,
      "loss": 2.0983,
      "step": 14080
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.439453125,
      "learning_rate": 2.5414392886093645e-05,
      "loss": 2.1336,
      "step": 14090
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4296875,
      "learning_rate": 2.54069684504562e-05,
      "loss": 2.1183,
      "step": 14100
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.43359375,
      "learning_rate": 2.5399539095707977e-05,
      "loss": 2.1162,
      "step": 14110
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.451171875,
      "learning_rate": 2.539210482536065e-05,
      "loss": 2.1401,
      "step": 14120
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.51171875,
      "learning_rate": 2.5384665642928197e-05,
      "loss": 2.1168,
      "step": 14130
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.431640625,
      "learning_rate": 2.5377221551926942e-05,
      "loss": 2.1489,
      "step": 14140
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.458984375,
      "learning_rate": 2.5369772555875518e-05,
      "loss": 2.1357,
      "step": 14150
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4140625,
      "learning_rate": 2.536231865829487e-05,
      "loss": 2.0854,
      "step": 14160
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.423828125,
      "learning_rate": 2.5354859862708276e-05,
      "loss": 2.1236,
      "step": 14170
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.470703125,
      "learning_rate": 2.5347396172641313e-05,
      "loss": 2.1173,
      "step": 14180
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.439453125,
      "learning_rate": 2.5339927591621887e-05,
      "loss": 2.1211,
      "step": 14190
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.439453125,
      "learning_rate": 2.5332454123180196e-05,
      "loss": 2.0856,
      "step": 14200
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.439453125,
      "learning_rate": 2.5324975770848768e-05,
      "loss": 2.1184,
      "step": 14210
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4375,
      "learning_rate": 2.5317492538162425e-05,
      "loss": 2.107,
      "step": 14220
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.447265625,
      "learning_rate": 2.5310004428658307e-05,
      "loss": 2.1646,
      "step": 14230
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.439453125,
      "learning_rate": 2.5302511445875857e-05,
      "loss": 2.1197,
      "step": 14240
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.431640625,
      "learning_rate": 2.5295013593356805e-05,
      "loss": 2.1103,
      "step": 14250
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.431640625,
      "learning_rate": 2.5287510874645207e-05,
      "loss": 2.1076,
      "step": 14260
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.478515625,
      "learning_rate": 2.528000329328741e-05,
      "loss": 2.1217,
      "step": 14270
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.427734375,
      "learning_rate": 2.527249085283205e-05,
      "loss": 2.1151,
      "step": 14280
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.435546875,
      "learning_rate": 2.5264973556830077e-05,
      "loss": 2.1274,
      "step": 14290
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.416015625,
      "learning_rate": 2.5257451408834715e-05,
      "loss": 2.1284,
      "step": 14300
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.427734375,
      "learning_rate": 2.5249924412401503e-05,
      "loss": 2.0981,
      "step": 14310
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4921875,
      "learning_rate": 2.5242392571088257e-05,
      "loss": 2.1233,
      "step": 14320
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.427734375,
      "learning_rate": 2.523485588845509e-05,
      "loss": 2.1225,
      "step": 14330
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4375,
      "learning_rate": 2.52273143680644e-05,
      "loss": 2.1173,
      "step": 14340
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4296875,
      "learning_rate": 2.5219768013480873e-05,
      "loss": 2.1039,
      "step": 14350
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.42578125,
      "learning_rate": 2.521221682827148e-05,
      "loss": 2.1242,
      "step": 14360
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.484375,
      "learning_rate": 2.5204660816005473e-05,
      "loss": 2.1307,
      "step": 14370
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.419921875,
      "learning_rate": 2.5197099980254395e-05,
      "loss": 2.0909,
      "step": 14380
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.447265625,
      "learning_rate": 2.5189534324592054e-05,
      "loss": 2.1161,
      "step": 14390
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4140625,
      "learning_rate": 2.5181963852594548e-05,
      "loss": 2.0909,
      "step": 14400
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4375,
      "learning_rate": 2.5174388567840247e-05,
      "loss": 2.1651,
      "step": 14410
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.474609375,
      "learning_rate": 2.51668084739098e-05,
      "loss": 2.1234,
      "step": 14420
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4296875,
      "learning_rate": 2.5159223574386117e-05,
      "loss": 2.1284,
      "step": 14430
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.43359375,
      "learning_rate": 2.5151633872854393e-05,
      "loss": 2.1327,
      "step": 14440
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.42578125,
      "learning_rate": 2.5144039372902093e-05,
      "loss": 2.1369,
      "step": 14450
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.48828125,
      "learning_rate": 2.513644007811894e-05,
      "loss": 2.1351,
      "step": 14460
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.458984375,
      "learning_rate": 2.5128835992096926e-05,
      "loss": 2.1129,
      "step": 14470
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.443359375,
      "learning_rate": 2.5121227118430314e-05,
      "loss": 2.1323,
      "step": 14480
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.47265625,
      "learning_rate": 2.5113613460715625e-05,
      "loss": 2.1137,
      "step": 14490
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.431640625,
      "learning_rate": 2.510599502255164e-05,
      "loss": 2.1233,
      "step": 14500
    },
    {
      "epoch": 0.3,
      "eval_accuracy": 0.5581014810616328,
      "eval_loss": 1.9988162517547607,
      "eval_runtime": 16.4202,
      "eval_samples_per_second": 36.236,
      "eval_steps_per_second": 1.157,
      "step": 14500
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4609375,
      "learning_rate": 2.509837180753941e-05,
      "loss": 2.1153,
      "step": 14510
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.42578125,
      "learning_rate": 2.5090743819282225e-05,
      "loss": 2.124,
      "step": 14520
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.486328125,
      "learning_rate": 2.508311106138565e-05,
      "loss": 2.1196,
      "step": 14530
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.431640625,
      "learning_rate": 2.5075473537457496e-05,
      "loss": 2.1181,
      "step": 14540
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.478515625,
      "learning_rate": 2.5067831251107826e-05,
      "loss": 2.1276,
      "step": 14550
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.474609375,
      "learning_rate": 2.5060184205948953e-05,
      "loss": 2.1338,
      "step": 14560
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4296875,
      "learning_rate": 2.5052532405595444e-05,
      "loss": 2.0974,
      "step": 14570
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.427734375,
      "learning_rate": 2.504487585366411e-05,
      "loss": 2.1171,
      "step": 14580
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.42578125,
      "learning_rate": 2.5037214553774017e-05,
      "loss": 2.1265,
      "step": 14590
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.435546875,
      "learning_rate": 2.5029548509546452e-05,
      "loss": 2.1231,
      "step": 14600
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.423828125,
      "learning_rate": 2.5021877724604976e-05,
      "loss": 2.1272,
      "step": 14610
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.453125,
      "learning_rate": 2.5014202202575367e-05,
      "loss": 2.1321,
      "step": 14620
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.43359375,
      "learning_rate": 2.5006521947085654e-05,
      "loss": 2.1254,
      "step": 14630
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.44921875,
      "learning_rate": 2.4998836961766098e-05,
      "loss": 2.1157,
      "step": 14640
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.421875,
      "learning_rate": 2.4991147250249193e-05,
      "loss": 2.1514,
      "step": 14650
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4296875,
      "learning_rate": 2.4983452816169676e-05,
      "loss": 2.1128,
      "step": 14660
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.490234375,
      "learning_rate": 2.497575366316451e-05,
      "loss": 2.1163,
      "step": 14670
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.435546875,
      "learning_rate": 2.4968049794872895e-05,
      "loss": 2.1011,
      "step": 14680
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.453125,
      "learning_rate": 2.4960341214936244e-05,
      "loss": 2.1491,
      "step": 14690
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.439453125,
      "learning_rate": 2.4952627926998214e-05,
      "loss": 2.1284,
      "step": 14700
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.439453125,
      "learning_rate": 2.4944909934704683e-05,
      "loss": 2.1329,
      "step": 14710
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.470703125,
      "learning_rate": 2.4937187241703752e-05,
      "loss": 2.089,
      "step": 14720
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.443359375,
      "learning_rate": 2.4929459851645736e-05,
      "loss": 2.1106,
      "step": 14730
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.421875,
      "learning_rate": 2.4921727768183178e-05,
      "loss": 2.1323,
      "step": 14740
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.416015625,
      "learning_rate": 2.4913990994970838e-05,
      "loss": 2.1156,
      "step": 14750
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.423828125,
      "learning_rate": 2.49062495356657e-05,
      "loss": 2.1578,
      "step": 14760
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.427734375,
      "learning_rate": 2.489850339392695e-05,
      "loss": 2.1294,
      "step": 14770
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4375,
      "learning_rate": 2.4890752573415997e-05,
      "loss": 2.1327,
      "step": 14780
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.443359375,
      "learning_rate": 2.4882997077796453e-05,
      "loss": 2.1374,
      "step": 14790
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.447265625,
      "learning_rate": 2.4875236910734145e-05,
      "loss": 2.0998,
      "step": 14800
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.431640625,
      "learning_rate": 2.4867472075897112e-05,
      "loss": 2.1303,
      "step": 14810
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.44921875,
      "learning_rate": 2.4859702576955595e-05,
      "loss": 2.1342,
      "step": 14820
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.431640625,
      "learning_rate": 2.4851928417582034e-05,
      "loss": 2.1316,
      "step": 14830
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.427734375,
      "learning_rate": 2.484414960145108e-05,
      "loss": 2.1225,
      "step": 14840
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.431640625,
      "learning_rate": 2.4836366132239585e-05,
      "loss": 2.1203,
      "step": 14850
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.439453125,
      "learning_rate": 2.48285780136266e-05,
      "loss": 2.1207,
      "step": 14860
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4296875,
      "learning_rate": 2.4820785249293367e-05,
      "loss": 2.0991,
      "step": 14870
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.43359375,
      "learning_rate": 2.481298784292333e-05,
      "loss": 2.1467,
      "step": 14880
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.447265625,
      "learning_rate": 2.4805185798202125e-05,
      "loss": 2.1217,
      "step": 14890
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4765625,
      "learning_rate": 2.4797379118817577e-05,
      "loss": 2.1357,
      "step": 14900
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.43359375,
      "learning_rate": 2.4789567808459718e-05,
      "loss": 2.1433,
      "step": 14910
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.44921875,
      "learning_rate": 2.4781751870820746e-05,
      "loss": 2.1387,
      "step": 14920
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4296875,
      "learning_rate": 2.4773931309595064e-05,
      "loss": 2.1119,
      "step": 14930
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4296875,
      "learning_rate": 2.4766106128479248e-05,
      "loss": 2.1469,
      "step": 14940
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.458984375,
      "learning_rate": 2.4758276331172065e-05,
      "loss": 2.1553,
      "step": 14950
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.486328125,
      "learning_rate": 2.475044192137446e-05,
      "loss": 2.1087,
      "step": 14960
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.423828125,
      "learning_rate": 2.474260290278957e-05,
      "loss": 2.1025,
      "step": 14970
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.455078125,
      "learning_rate": 2.473475927912269e-05,
      "loss": 2.1313,
      "step": 14980
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.439453125,
      "learning_rate": 2.4726911054081304e-05,
      "loss": 2.1499,
      "step": 14990
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.46484375,
      "learning_rate": 2.4719058231375076e-05,
      "loss": 2.1348,
      "step": 15000
    },
    {
      "epoch": 0.31,
      "eval_accuracy": 0.5581129812628863,
      "eval_loss": 1.9984461069107056,
      "eval_runtime": 16.4389,
      "eval_samples_per_second": 36.195,
      "eval_steps_per_second": 1.156,
      "step": 15000
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5234375,
      "learning_rate": 2.471120081471583e-05,
      "loss": 2.1141,
      "step": 15010
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4375,
      "learning_rate": 2.4703338807817573e-05,
      "loss": 2.1094,
      "step": 15020
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.455078125,
      "learning_rate": 2.469547221439647e-05,
      "loss": 2.0645,
      "step": 15030
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4375,
      "learning_rate": 2.468760103817087e-05,
      "loss": 2.0682,
      "step": 15040
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.458984375,
      "learning_rate": 2.467972528286127e-05,
      "loss": 2.1166,
      "step": 15050
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.412109375,
      "learning_rate": 2.467184495219035e-05,
      "loss": 2.1059,
      "step": 15060
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.421875,
      "learning_rate": 2.4663960049882933e-05,
      "loss": 2.1045,
      "step": 15070
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.435546875,
      "learning_rate": 2.4656070579666022e-05,
      "loss": 2.1114,
      "step": 15080
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.482421875,
      "learning_rate": 2.464817654526876e-05,
      "loss": 2.1392,
      "step": 15090
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5078125,
      "learning_rate": 2.4640277950422462e-05,
      "loss": 2.1055,
      "step": 15100
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.431640625,
      "learning_rate": 2.4632374798860598e-05,
      "loss": 2.1051,
      "step": 15110
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.55078125,
      "learning_rate": 2.4624467094318783e-05,
      "loss": 2.1752,
      "step": 15120
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.416015625,
      "learning_rate": 2.4616554840534797e-05,
      "loss": 2.0669,
      "step": 15130
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.431640625,
      "learning_rate": 2.460863804124855e-05,
      "loss": 2.1496,
      "step": 15140
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.435546875,
      "learning_rate": 2.4600716700202114e-05,
      "loss": 2.1215,
      "step": 15150
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.427734375,
      "learning_rate": 2.459279082113972e-05,
      "loss": 2.0839,
      "step": 15160
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.455078125,
      "learning_rate": 2.458486040780772e-05,
      "loss": 2.1379,
      "step": 15170
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.43359375,
      "learning_rate": 2.457692546395462e-05,
      "loss": 2.1164,
      "step": 15180
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.44140625,
      "learning_rate": 2.456898599333107e-05,
      "loss": 2.1049,
      "step": 15190
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.43359375,
      "learning_rate": 2.4561041999689853e-05,
      "loss": 2.117,
      "step": 15200
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4375,
      "learning_rate": 2.45530934867859e-05,
      "loss": 2.1007,
      "step": 15210
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.419921875,
      "learning_rate": 2.454514045837627e-05,
      "loss": 2.1464,
      "step": 15220
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.451171875,
      "learning_rate": 2.453718291822015e-05,
      "loss": 2.1302,
      "step": 15230
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.419921875,
      "learning_rate": 2.4529220870078884e-05,
      "loss": 2.1042,
      "step": 15240
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.451171875,
      "learning_rate": 2.452125431771592e-05,
      "loss": 2.1049,
      "step": 15250
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.490234375,
      "learning_rate": 2.451328326489684e-05,
      "loss": 2.0955,
      "step": 15260
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.46484375,
      "learning_rate": 2.4505307715389377e-05,
      "loss": 2.1172,
      "step": 15270
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.443359375,
      "learning_rate": 2.449732767296335e-05,
      "loss": 2.1699,
      "step": 15280
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.435546875,
      "learning_rate": 2.4489343141390745e-05,
      "loss": 2.1195,
      "step": 15290
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.462890625,
      "learning_rate": 2.4481354124445633e-05,
      "loss": 2.1466,
      "step": 15300
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.43359375,
      "learning_rate": 2.447336062590422e-05,
      "loss": 2.1435,
      "step": 15310
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.46484375,
      "learning_rate": 2.4465362649544845e-05,
      "loss": 2.1078,
      "step": 15320
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4375,
      "learning_rate": 2.445736019914794e-05,
      "loss": 2.1435,
      "step": 15330
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.64453125,
      "learning_rate": 2.4449353278496052e-05,
      "loss": 2.138,
      "step": 15340
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.44921875,
      "learning_rate": 2.4441341891373866e-05,
      "loss": 2.105,
      "step": 15350
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.451171875,
      "learning_rate": 2.4433326041568152e-05,
      "loss": 2.088,
      "step": 15360
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.46484375,
      "learning_rate": 2.44253057328678e-05,
      "loss": 2.1287,
      "step": 15370
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.427734375,
      "learning_rate": 2.441728096906381e-05,
      "loss": 2.1336,
      "step": 15380
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.447265625,
      "learning_rate": 2.4409251753949284e-05,
      "loss": 2.1116,
      "step": 15390
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.419921875,
      "learning_rate": 2.4401218091319427e-05,
      "loss": 2.1251,
      "step": 15400
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.431640625,
      "learning_rate": 2.4393179984971556e-05,
      "loss": 2.1223,
      "step": 15410
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.439453125,
      "learning_rate": 2.438513743870507e-05,
      "loss": 2.1083,
      "step": 15420
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.439453125,
      "learning_rate": 2.437709045632148e-05,
      "loss": 2.1026,
      "step": 15430
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.25,
      "learning_rate": 2.4369039041624402e-05,
      "loss": 2.1047,
      "step": 15440
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.41796875,
      "learning_rate": 2.4360983198419514e-05,
      "loss": 2.1191,
      "step": 15450
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4375,
      "learning_rate": 2.4352922930514633e-05,
      "loss": 2.1407,
      "step": 15460
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4609375,
      "learning_rate": 2.4344858241719625e-05,
      "loss": 2.1377,
      "step": 15470
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.458984375,
      "learning_rate": 2.433678913584648e-05,
      "loss": 2.1688,
      "step": 15480
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.416015625,
      "learning_rate": 2.432871561670925e-05,
      "loss": 2.0989,
      "step": 15490
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4609375,
      "learning_rate": 2.4320637688124084e-05,
      "loss": 2.1494,
      "step": 15500
    },
    {
      "epoch": 0.32,
      "eval_accuracy": 0.5582411263625685,
      "eval_loss": 1.9979814291000366,
      "eval_runtime": 16.4549,
      "eval_samples_per_second": 36.159,
      "eval_steps_per_second": 1.155,
      "step": 15500
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.439453125,
      "learning_rate": 2.431255535390922e-05,
      "loss": 2.0885,
      "step": 15510
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.451171875,
      "learning_rate": 2.4304468617884965e-05,
      "loss": 2.1309,
      "step": 15520
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4453125,
      "learning_rate": 2.429637748387372e-05,
      "loss": 2.1513,
      "step": 15530
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.447265625,
      "learning_rate": 2.4288281955699963e-05,
      "loss": 2.126,
      "step": 15540
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.447265625,
      "learning_rate": 2.428018203719024e-05,
      "loss": 2.1299,
      "step": 15550
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.43359375,
      "learning_rate": 2.4272077732173177e-05,
      "loss": 2.1302,
      "step": 15560
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.439453125,
      "learning_rate": 2.4263969044479477e-05,
      "loss": 2.1572,
      "step": 15570
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.447265625,
      "learning_rate": 2.425585597794191e-05,
      "loss": 2.1321,
      "step": 15580
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.435546875,
      "learning_rate": 2.4247738536395315e-05,
      "loss": 2.1122,
      "step": 15590
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.416015625,
      "learning_rate": 2.4239616723676603e-05,
      "loss": 2.0932,
      "step": 15600
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.44140625,
      "learning_rate": 2.4231490543624752e-05,
      "loss": 2.1365,
      "step": 15610
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.421875,
      "learning_rate": 2.4223360000080803e-05,
      "loss": 2.1153,
      "step": 15620
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.455078125,
      "learning_rate": 2.4215225096887853e-05,
      "loss": 2.1355,
      "step": 15630
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.451171875,
      "learning_rate": 2.4207085837891066e-05,
      "loss": 2.1666,
      "step": 15640
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4453125,
      "learning_rate": 2.4198942226937665e-05,
      "loss": 2.1014,
      "step": 15650
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.470703125,
      "learning_rate": 2.419079426787693e-05,
      "loss": 2.1422,
      "step": 15660
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4375,
      "learning_rate": 2.41826419645602e-05,
      "loss": 2.1147,
      "step": 15670
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43359375,
      "learning_rate": 2.417448532084085e-05,
      "loss": 2.1315,
      "step": 15680
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4453125,
      "learning_rate": 2.4166324340574328e-05,
      "loss": 2.1433,
      "step": 15690
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4140625,
      "learning_rate": 2.4158159027618126e-05,
      "loss": 2.1467,
      "step": 15700
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.427734375,
      "learning_rate": 2.4149989385831775e-05,
      "loss": 2.1164,
      "step": 15710
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43359375,
      "learning_rate": 2.4141815419076853e-05,
      "loss": 2.0902,
      "step": 15720
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.439453125,
      "learning_rate": 2.4133637131216994e-05,
      "loss": 2.116,
      "step": 15730
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.466796875,
      "learning_rate": 2.412545452611787e-05,
      "loss": 2.1461,
      "step": 15740
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.458984375,
      "learning_rate": 2.411726760764719e-05,
      "loss": 2.1146,
      "step": 15750
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.421875,
      "learning_rate": 2.4109076379674702e-05,
      "loss": 2.1436,
      "step": 15760
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.455078125,
      "learning_rate": 2.4100880846072192e-05,
      "loss": 2.0966,
      "step": 15770
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.419921875,
      "learning_rate": 2.409268101071348e-05,
      "loss": 2.1472,
      "step": 15780
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43359375,
      "learning_rate": 2.4084476877474424e-05,
      "loss": 2.1254,
      "step": 15790
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4375,
      "learning_rate": 2.407626845023291e-05,
      "loss": 2.1058,
      "step": 15800
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.451171875,
      "learning_rate": 2.4068055732868855e-05,
      "loss": 2.1425,
      "step": 15810
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.42578125,
      "learning_rate": 2.4059838729264197e-05,
      "loss": 2.1306,
      "step": 15820
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.54296875,
      "learning_rate": 2.4051617443302904e-05,
      "loss": 2.0765,
      "step": 15830
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.439453125,
      "learning_rate": 2.4043391878870984e-05,
      "loss": 2.1367,
      "step": 15840
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4296875,
      "learning_rate": 2.4035162039856437e-05,
      "loss": 2.1235,
      "step": 15850
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.427734375,
      "learning_rate": 2.4026927930149318e-05,
      "loss": 2.1253,
      "step": 15860
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.427734375,
      "learning_rate": 2.4018689553641667e-05,
      "loss": 2.1068,
      "step": 15870
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.439453125,
      "learning_rate": 2.4010446914227558e-05,
      "loss": 2.1235,
      "step": 15880
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4140625,
      "learning_rate": 2.4002200015803082e-05,
      "loss": 2.094,
      "step": 15890
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.447265625,
      "learning_rate": 2.3993948862266342e-05,
      "loss": 2.101,
      "step": 15900
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.455078125,
      "learning_rate": 2.3985693457517444e-05,
      "loss": 2.1197,
      "step": 15910
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.44921875,
      "learning_rate": 2.397743380545851e-05,
      "loss": 2.1242,
      "step": 15920
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.45703125,
      "learning_rate": 2.3969169909993675e-05,
      "loss": 2.1115,
      "step": 15930
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.431640625,
      "learning_rate": 2.396090177502906e-05,
      "loss": 2.1099,
      "step": 15940
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.431640625,
      "learning_rate": 2.3952629404472816e-05,
      "loss": 2.1553,
      "step": 15950
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43359375,
      "learning_rate": 2.3944352802235075e-05,
      "loss": 2.1397,
      "step": 15960
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43359375,
      "learning_rate": 2.3936071972227988e-05,
      "loss": 2.1019,
      "step": 15970
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.458984375,
      "learning_rate": 2.392778691836568e-05,
      "loss": 2.1409,
      "step": 15980
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.427734375,
      "learning_rate": 2.3919497644564302e-05,
      "loss": 2.0796,
      "step": 15990
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.462890625,
      "learning_rate": 2.391120415474197e-05,
      "loss": 2.0827,
      "step": 16000
    },
    {
      "epoch": 0.33,
      "eval_accuracy": 0.5584021291801178,
      "eval_loss": 1.9976391792297363,
      "eval_runtime": 16.4349,
      "eval_samples_per_second": 36.203,
      "eval_steps_per_second": 1.156,
      "step": 16000
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43359375,
      "learning_rate": 2.3902906452818814e-05,
      "loss": 2.1198,
      "step": 16010
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4296875,
      "learning_rate": 2.3894604542716948e-05,
      "loss": 2.1231,
      "step": 16020
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.466796875,
      "learning_rate": 2.3886298428360472e-05,
      "loss": 2.1199,
      "step": 16030
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4140625,
      "learning_rate": 2.3877988113675472e-05,
      "loss": 2.1104,
      "step": 16040
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.447265625,
      "learning_rate": 2.3869673602590036e-05,
      "loss": 2.115,
      "step": 16050
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.48046875,
      "learning_rate": 2.3861354899034213e-05,
      "loss": 2.1439,
      "step": 16060
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.435546875,
      "learning_rate": 2.385303200694005e-05,
      "loss": 2.1313,
      "step": 16070
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.435546875,
      "learning_rate": 2.3844704930241564e-05,
      "loss": 2.0941,
      "step": 16080
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.43359375,
      "learning_rate": 2.3836373672874758e-05,
      "loss": 2.147,
      "step": 16090
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.41015625,
      "learning_rate": 2.3828038238777592e-05,
      "loss": 2.1247,
      "step": 16100
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.65625,
      "learning_rate": 2.3819698631890034e-05,
      "loss": 2.1169,
      "step": 16110
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.427734375,
      "learning_rate": 2.3811354856153996e-05,
      "loss": 2.1007,
      "step": 16120
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.412109375,
      "learning_rate": 2.3803006915513364e-05,
      "loss": 2.0962,
      "step": 16130
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.427734375,
      "learning_rate": 2.3794654813914015e-05,
      "loss": 2.0955,
      "step": 16140
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.41796875,
      "learning_rate": 2.378629855530376e-05,
      "loss": 2.107,
      "step": 16150
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43359375,
      "learning_rate": 2.37779381436324e-05,
      "loss": 2.1207,
      "step": 16160
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4453125,
      "learning_rate": 2.376957358285169e-05,
      "loss": 2.1022,
      "step": 16170
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.427734375,
      "learning_rate": 2.3761204876915338e-05,
      "loss": 2.1603,
      "step": 16180
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.46484375,
      "learning_rate": 2.375283202977903e-05,
      "loss": 2.1225,
      "step": 16190
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4375,
      "learning_rate": 2.3744455045400393e-05,
      "loss": 2.1115,
      "step": 16200
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.421875,
      "learning_rate": 2.3736073927739016e-05,
      "loss": 2.1236,
      "step": 16210
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.44140625,
      "learning_rate": 2.3727688680756443e-05,
      "loss": 2.1275,
      "step": 16220
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.431640625,
      "learning_rate": 2.3719299308416167e-05,
      "loss": 2.1374,
      "step": 16230
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.44140625,
      "learning_rate": 2.3710905814683635e-05,
      "loss": 2.1422,
      "step": 16240
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.44921875,
      "learning_rate": 2.3702508203526236e-05,
      "loss": 2.122,
      "step": 16250
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.470703125,
      "learning_rate": 2.3694106478913313e-05,
      "loss": 2.1379,
      "step": 16260
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.42578125,
      "learning_rate": 2.3685700644816143e-05,
      "loss": 2.1167,
      "step": 16270
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.435546875,
      "learning_rate": 2.367729070520795e-05,
      "loss": 2.1103,
      "step": 16280
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.423828125,
      "learning_rate": 2.3668876664063908e-05,
      "loss": 2.1135,
      "step": 16290
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4296875,
      "learning_rate": 2.3660458525361113e-05,
      "loss": 2.1097,
      "step": 16300
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.451171875,
      "learning_rate": 2.3652036293078612e-05,
      "loss": 2.1381,
      "step": 16310
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.435546875,
      "learning_rate": 2.364360997119738e-05,
      "loss": 2.1099,
      "step": 16320
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.44921875,
      "learning_rate": 2.3635179563700324e-05,
      "loss": 2.1221,
      "step": 16330
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.435546875,
      "learning_rate": 2.362674507457228e-05,
      "loss": 2.0826,
      "step": 16340
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5703125,
      "learning_rate": 2.361830650780003e-05,
      "loss": 2.1204,
      "step": 16350
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4296875,
      "learning_rate": 2.3609863867372266e-05,
      "loss": 2.121,
      "step": 16360
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.462890625,
      "learning_rate": 2.3601417157279606e-05,
      "loss": 2.1185,
      "step": 16370
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.46484375,
      "learning_rate": 2.35929663815146e-05,
      "loss": 2.1422,
      "step": 16380
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.458984375,
      "learning_rate": 2.3584511544071716e-05,
      "loss": 2.1114,
      "step": 16390
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.443359375,
      "learning_rate": 2.3576052648947338e-05,
      "loss": 2.1555,
      "step": 16400
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.439453125,
      "learning_rate": 2.3567589700139776e-05,
      "loss": 2.1213,
      "step": 16410
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4765625,
      "learning_rate": 2.3559122701649252e-05,
      "loss": 2.117,
      "step": 16420
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.474609375,
      "learning_rate": 2.35506516574779e-05,
      "loss": 2.0975,
      "step": 16430
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.42578125,
      "learning_rate": 2.354217657162977e-05,
      "loss": 2.1092,
      "step": 16440
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.44921875,
      "learning_rate": 2.3533697448110817e-05,
      "loss": 2.1054,
      "step": 16450
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.443359375,
      "learning_rate": 2.3525214290928917e-05,
      "loss": 2.0694,
      "step": 16460
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.421875,
      "learning_rate": 2.3516727104093836e-05,
      "loss": 2.1213,
      "step": 16470
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43359375,
      "learning_rate": 2.350823589161725e-05,
      "loss": 2.1395,
      "step": 16480
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.447265625,
      "learning_rate": 2.349974065751275e-05,
      "loss": 2.1251,
      "step": 16490
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.447265625,
      "learning_rate": 2.349124140579581e-05,
      "loss": 2.0991,
      "step": 16500
    },
    {
      "epoch": 0.34,
      "eval_accuracy": 0.5582000542152344,
      "eval_loss": 1.997467041015625,
      "eval_runtime": 16.4503,
      "eval_samples_per_second": 36.17,
      "eval_steps_per_second": 1.155,
      "step": 16500
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4296875,
      "learning_rate": 2.3482738140483815e-05,
      "loss": 2.1377,
      "step": 16510
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.435546875,
      "learning_rate": 2.3474230865596045e-05,
      "loss": 2.1417,
      "step": 16520
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4375,
      "learning_rate": 2.346571958515367e-05,
      "loss": 2.1126,
      "step": 16530
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.458984375,
      "learning_rate": 2.345720430317976e-05,
      "loss": 2.0925,
      "step": 16540
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.466796875,
      "learning_rate": 2.3448685023699267e-05,
      "loss": 2.1063,
      "step": 16550
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.421875,
      "learning_rate": 2.344016175073905e-05,
      "loss": 2.0826,
      "step": 16560
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.439453125,
      "learning_rate": 2.3431634488327834e-05,
      "loss": 2.1062,
      "step": 16570
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.423828125,
      "learning_rate": 2.342310324049625e-05,
      "loss": 2.1128,
      "step": 16580
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.458984375,
      "learning_rate": 2.3414568011276794e-05,
      "loss": 2.1029,
      "step": 16590
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.451171875,
      "learning_rate": 2.340602880470386e-05,
      "loss": 2.1128,
      "step": 16600
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.439453125,
      "learning_rate": 2.339748562481371e-05,
      "loss": 2.1209,
      "step": 16610
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.453125,
      "learning_rate": 2.3388938475644494e-05,
      "loss": 2.14,
      "step": 16620
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.435546875,
      "learning_rate": 2.3380387361236223e-05,
      "loss": 2.1274,
      "step": 16630
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.423828125,
      "learning_rate": 2.3371832285630805e-05,
      "loss": 2.0816,
      "step": 16640
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.416015625,
      "learning_rate": 2.3363273252872003e-05,
      "loss": 2.1338,
      "step": 16650
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4375,
      "learning_rate": 2.3354710267005452e-05,
      "loss": 2.1182,
      "step": 16660
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.43359375,
      "learning_rate": 2.334614333207866e-05,
      "loss": 2.1068,
      "step": 16670
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.470703125,
      "learning_rate": 2.3337572452141005e-05,
      "loss": 2.1285,
      "step": 16680
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.44921875,
      "learning_rate": 2.3328997631243726e-05,
      "loss": 2.1612,
      "step": 16690
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.470703125,
      "learning_rate": 2.3320418873439918e-05,
      "loss": 2.1015,
      "step": 16700
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.443359375,
      "learning_rate": 2.3311836182784544e-05,
      "loss": 2.1444,
      "step": 16710
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.439453125,
      "learning_rate": 2.3303249563334425e-05,
      "loss": 2.0945,
      "step": 16720
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.494140625,
      "learning_rate": 2.329465901914824e-05,
      "loss": 2.1494,
      "step": 16730
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4609375,
      "learning_rate": 2.3286064554286525e-05,
      "loss": 2.1233,
      "step": 16740
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.498046875,
      "learning_rate": 2.327746617281166e-05,
      "loss": 2.1219,
      "step": 16750
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.451171875,
      "learning_rate": 2.3268863878787886e-05,
      "loss": 2.1231,
      "step": 16760
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.42578125,
      "learning_rate": 2.326025767628129e-05,
      "loss": 2.1019,
      "step": 16770
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.43359375,
      "learning_rate": 2.3251647569359804e-05,
      "loss": 2.1066,
      "step": 16780
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.423828125,
      "learning_rate": 2.324303356209321e-05,
      "loss": 2.1257,
      "step": 16790
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.46484375,
      "learning_rate": 2.323441565855313e-05,
      "loss": 2.1128,
      "step": 16800
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.419921875,
      "learning_rate": 2.3225793862813033e-05,
      "loss": 2.1295,
      "step": 16810
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.431640625,
      "learning_rate": 2.3217168178948217e-05,
      "loss": 2.0857,
      "step": 16820
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.435546875,
      "learning_rate": 2.3208538611035826e-05,
      "loss": 2.1211,
      "step": 16830
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.455078125,
      "learning_rate": 2.3199905163154848e-05,
      "loss": 2.1346,
      "step": 16840
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.431640625,
      "learning_rate": 2.3191267839386082e-05,
      "loss": 2.115,
      "step": 16850
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.451171875,
      "learning_rate": 2.318262664381218e-05,
      "loss": 2.1395,
      "step": 16860
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.43359375,
      "learning_rate": 2.3173981580517615e-05,
      "loss": 2.1148,
      "step": 16870
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.443359375,
      "learning_rate": 2.3165332653588692e-05,
      "loss": 2.1701,
      "step": 16880
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.44140625,
      "learning_rate": 2.3156679867113534e-05,
      "loss": 2.1025,
      "step": 16890
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4375,
      "learning_rate": 2.3148023225182102e-05,
      "loss": 2.1646,
      "step": 16900
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.443359375,
      "learning_rate": 2.3139362731886164e-05,
      "loss": 2.1109,
      "step": 16910
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.439453125,
      "learning_rate": 2.3130698391319328e-05,
      "loss": 2.0948,
      "step": 16920
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.44140625,
      "learning_rate": 2.3122030207576993e-05,
      "loss": 2.1279,
      "step": 16930
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.443359375,
      "learning_rate": 2.31133581847564e-05,
      "loss": 2.1231,
      "step": 16940
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4375,
      "learning_rate": 2.3104682326956598e-05,
      "loss": 2.1146,
      "step": 16950
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.44140625,
      "learning_rate": 2.309600263827843e-05,
      "loss": 2.1351,
      "step": 16960
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.447265625,
      "learning_rate": 2.3087319122824582e-05,
      "loss": 2.124,
      "step": 16970
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.51171875,
      "learning_rate": 2.3078631784699526e-05,
      "loss": 2.1753,
      "step": 16980
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7265625,
      "learning_rate": 2.3069940628009543e-05,
      "loss": 2.1136,
      "step": 16990
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.443359375,
      "learning_rate": 2.306124565686273e-05,
      "loss": 2.1108,
      "step": 17000
    },
    {
      "epoch": 0.35,
      "eval_accuracy": 0.5581786966986207,
      "eval_loss": 1.997156023979187,
      "eval_runtime": 16.4385,
      "eval_samples_per_second": 36.196,
      "eval_steps_per_second": 1.156,
      "step": 17000
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4453125,
      "learning_rate": 2.305254687536898e-05,
      "loss": 2.1053,
      "step": 17010
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.439453125,
      "learning_rate": 2.304384428763998e-05,
      "loss": 2.0957,
      "step": 17020
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4375,
      "learning_rate": 2.303513789778923e-05,
      "loss": 2.1267,
      "step": 17030
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.458984375,
      "learning_rate": 2.3026427709932016e-05,
      "loss": 2.0924,
      "step": 17040
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.447265625,
      "learning_rate": 2.3017713728185416e-05,
      "loss": 2.1457,
      "step": 17050
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.43359375,
      "learning_rate": 2.3008995956668327e-05,
      "loss": 2.1346,
      "step": 17060
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.453125,
      "learning_rate": 2.300027439950141e-05,
      "loss": 2.1404,
      "step": 17070
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.455078125,
      "learning_rate": 2.299154906080712e-05,
      "loss": 2.107,
      "step": 17080
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4453125,
      "learning_rate": 2.298281994470971e-05,
      "loss": 2.1133,
      "step": 17090
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4375,
      "learning_rate": 2.2974087055335204e-05,
      "loss": 2.1116,
      "step": 17100
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.447265625,
      "learning_rate": 2.296535039681143e-05,
      "loss": 2.1429,
      "step": 17110
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.421875,
      "learning_rate": 2.2956609973267983e-05,
      "loss": 2.1368,
      "step": 17120
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.439453125,
      "learning_rate": 2.294786578883623e-05,
      "loss": 2.1275,
      "step": 17130
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.45703125,
      "learning_rate": 2.293911784764934e-05,
      "loss": 2.1393,
      "step": 17140
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.42578125,
      "learning_rate": 2.2930366153842234e-05,
      "loss": 2.1127,
      "step": 17150
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.46484375,
      "learning_rate": 2.2921610711551618e-05,
      "loss": 2.1061,
      "step": 17160
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.451171875,
      "learning_rate": 2.2912851524915968e-05,
      "loss": 2.1509,
      "step": 17170
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.427734375,
      "learning_rate": 2.2904088598075528e-05,
      "loss": 2.1013,
      "step": 17180
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.43359375,
      "learning_rate": 2.2895321935172317e-05,
      "loss": 2.1362,
      "step": 17190
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.44921875,
      "learning_rate": 2.2886551540350115e-05,
      "loss": 2.1365,
      "step": 17200
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.419921875,
      "learning_rate": 2.287777741775446e-05,
      "loss": 2.1467,
      "step": 17210
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.44140625,
      "learning_rate": 2.2868999571532664e-05,
      "loss": 2.0935,
      "step": 17220
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.451171875,
      "learning_rate": 2.2860218005833788e-05,
      "loss": 2.1297,
      "step": 17230
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4375,
      "learning_rate": 2.285143272480866e-05,
      "loss": 2.144,
      "step": 17240
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.431640625,
      "learning_rate": 2.2842643732609856e-05,
      "loss": 2.147,
      "step": 17250
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4296875,
      "learning_rate": 2.2833851033391714e-05,
      "loss": 2.1434,
      "step": 17260
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.443359375,
      "learning_rate": 2.282505463131032e-05,
      "loss": 2.1556,
      "step": 17270
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.435546875,
      "learning_rate": 2.281625453052351e-05,
      "loss": 2.0865,
      "step": 17280
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.458984375,
      "learning_rate": 2.2807450735190864e-05,
      "loss": 2.1126,
      "step": 17290
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.55859375,
      "learning_rate": 2.2798643249473727e-05,
      "loss": 2.13,
      "step": 17300
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.458984375,
      "learning_rate": 2.278983207753516e-05,
      "loss": 2.0904,
      "step": 17310
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.41796875,
      "learning_rate": 2.2781017223539988e-05,
      "loss": 2.1025,
      "step": 17320
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4296875,
      "learning_rate": 2.2772198691654768e-05,
      "loss": 2.1126,
      "step": 17330
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.431640625,
      "learning_rate": 2.27633764860478e-05,
      "loss": 2.1331,
      "step": 17340
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.439453125,
      "learning_rate": 2.2754550610889117e-05,
      "loss": 2.1165,
      "step": 17350
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.451171875,
      "learning_rate": 2.274572107035048e-05,
      "loss": 2.14,
      "step": 17360
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4296875,
      "learning_rate": 2.2736887868605396e-05,
      "loss": 2.0895,
      "step": 17370
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4375,
      "learning_rate": 2.2728051009829095e-05,
      "loss": 2.1234,
      "step": 17380
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.453125,
      "learning_rate": 2.2719210498198537e-05,
      "loss": 2.1233,
      "step": 17390
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4296875,
      "learning_rate": 2.271036633789241e-05,
      "loss": 2.1033,
      "step": 17400
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6875,
      "learning_rate": 2.270151853309112e-05,
      "loss": 2.1471,
      "step": 17410
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.46484375,
      "learning_rate": 2.2692667087976803e-05,
      "loss": 2.1319,
      "step": 17420
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.439453125,
      "learning_rate": 2.2683812006733313e-05,
      "loss": 2.091,
      "step": 17430
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.41796875,
      "learning_rate": 2.2674953293546224e-05,
      "loss": 2.0865,
      "step": 17440
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.453125,
      "learning_rate": 2.2666090952602826e-05,
      "loss": 2.1314,
      "step": 17450
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.451171875,
      "learning_rate": 2.2657224988092122e-05,
      "loss": 2.1015,
      "step": 17460
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.44140625,
      "learning_rate": 2.264835540420483e-05,
      "loss": 2.1318,
      "step": 17470
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.44140625,
      "learning_rate": 2.2639482205133375e-05,
      "loss": 2.1215,
      "step": 17480
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.451171875,
      "learning_rate": 2.2630605395071894e-05,
      "loss": 2.1075,
      "step": 17490
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.421875,
      "learning_rate": 2.262172497821624e-05,
      "loss": 2.1209,
      "step": 17500
    },
    {
      "epoch": 0.36,
      "eval_accuracy": 0.5582739840804357,
      "eval_loss": 1.9968290328979492,
      "eval_runtime": 16.4403,
      "eval_samples_per_second": 36.192,
      "eval_steps_per_second": 1.156,
      "step": 17500
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.435546875,
      "learning_rate": 2.2612840958763946e-05,
      "loss": 2.1393,
      "step": 17510
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.42578125,
      "learning_rate": 2.2603953340914282e-05,
      "loss": 2.1351,
      "step": 17520
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.439453125,
      "learning_rate": 2.2595062128868185e-05,
      "loss": 2.1133,
      "step": 17530
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4296875,
      "learning_rate": 2.258616732682831e-05,
      "loss": 2.1398,
      "step": 17540
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.55859375,
      "learning_rate": 2.2577268938999008e-05,
      "loss": 2.122,
      "step": 17550
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.43359375,
      "learning_rate": 2.2568366969586315e-05,
      "loss": 2.0974,
      "step": 17560
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.478515625,
      "learning_rate": 2.255946142279798e-05,
      "loss": 2.1085,
      "step": 17570
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.431640625,
      "learning_rate": 2.2550552302843416e-05,
      "loss": 2.1024,
      "step": 17580
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.44140625,
      "learning_rate": 2.254163961393374e-05,
      "loss": 2.1257,
      "step": 17590
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.44921875,
      "learning_rate": 2.2532723360281762e-05,
      "loss": 2.1274,
      "step": 17600
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4375,
      "learning_rate": 2.252380354610197e-05,
      "loss": 2.1184,
      "step": 17610
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5703125,
      "learning_rate": 2.2514880175610522e-05,
      "loss": 2.1198,
      "step": 17620
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.45703125,
      "learning_rate": 2.2505953253025285e-05,
      "loss": 2.1069,
      "step": 17630
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.45703125,
      "learning_rate": 2.2497022782565783e-05,
      "loss": 2.1095,
      "step": 17640
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.453125,
      "learning_rate": 2.248808876845322e-05,
      "loss": 2.1438,
      "step": 17650
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.43359375,
      "learning_rate": 2.2479151214910483e-05,
      "loss": 2.1282,
      "step": 17660
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4453125,
      "learning_rate": 2.2470210126162126e-05,
      "loss": 2.1368,
      "step": 17670
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.447265625,
      "learning_rate": 2.246126550643438e-05,
      "loss": 2.1384,
      "step": 17680
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.41796875,
      "learning_rate": 2.2452317359955133e-05,
      "loss": 2.1358,
      "step": 17690
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.42578125,
      "learning_rate": 2.244336569095395e-05,
      "loss": 2.1071,
      "step": 17700
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.42578125,
      "learning_rate": 2.2434410503662064e-05,
      "loss": 2.1211,
      "step": 17710
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.490234375,
      "learning_rate": 2.242545180231236e-05,
      "loss": 2.1031,
      "step": 17720
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.427734375,
      "learning_rate": 2.2416489591139382e-05,
      "loss": 2.1204,
      "step": 17730
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.453125,
      "learning_rate": 2.2407523874379357e-05,
      "loss": 2.1133,
      "step": 17740
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.423828125,
      "learning_rate": 2.239855465627014e-05,
      "loss": 2.0785,
      "step": 17750
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4296875,
      "learning_rate": 2.2389581941051264e-05,
      "loss": 2.1482,
      "step": 17760
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4296875,
      "learning_rate": 2.23806057329639e-05,
      "loss": 2.1116,
      "step": 17770
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.435546875,
      "learning_rate": 2.2371626036250868e-05,
      "loss": 2.1437,
      "step": 17780
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.439453125,
      "learning_rate": 2.236264285515665e-05,
      "loss": 2.1247,
      "step": 17790
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4296875,
      "learning_rate": 2.235365619392737e-05,
      "loss": 2.0982,
      "step": 17800
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.431640625,
      "learning_rate": 2.2344666056810788e-05,
      "loss": 2.1155,
      "step": 17810
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4375,
      "learning_rate": 2.2335672448056323e-05,
      "loss": 2.1112,
      "step": 17820
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.42578125,
      "learning_rate": 2.232667537191502e-05,
      "loss": 2.11,
      "step": 17830
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.474609375,
      "learning_rate": 2.2317674832639574e-05,
      "loss": 2.085,
      "step": 17840
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.435546875,
      "learning_rate": 2.230867083448431e-05,
      "loss": 2.1234,
      "step": 17850
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.41796875,
      "learning_rate": 2.2299663381705187e-05,
      "loss": 2.1444,
      "step": 17860
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4453125,
      "learning_rate": 2.2290652478559807e-05,
      "loss": 2.0778,
      "step": 17870
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4453125,
      "learning_rate": 2.2281638129307393e-05,
      "loss": 2.1029,
      "step": 17880
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.44140625,
      "learning_rate": 2.2272620338208796e-05,
      "loss": 2.1198,
      "step": 17890
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.44140625,
      "learning_rate": 2.2263599109526505e-05,
      "loss": 2.126,
      "step": 17900
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.447265625,
      "learning_rate": 2.2254574447524624e-05,
      "loss": 2.0921,
      "step": 17910
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4453125,
      "learning_rate": 2.224554635646888e-05,
      "loss": 2.1375,
      "step": 17920
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4296875,
      "learning_rate": 2.2236514840626628e-05,
      "loss": 2.1375,
      "step": 17930
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4296875,
      "learning_rate": 2.2227479904266827e-05,
      "loss": 2.1237,
      "step": 17940
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.46484375,
      "learning_rate": 2.2218441551660073e-05,
      "loss": 2.1077,
      "step": 17950
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.439453125,
      "learning_rate": 2.2209399787078565e-05,
      "loss": 2.1325,
      "step": 17960
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4375,
      "learning_rate": 2.220035461479611e-05,
      "loss": 2.137,
      "step": 17970
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.451171875,
      "learning_rate": 2.219130603908814e-05,
      "loss": 2.1194,
      "step": 17980
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.423828125,
      "learning_rate": 2.2182254064231688e-05,
      "loss": 2.0936,
      "step": 17990
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.423828125,
      "learning_rate": 2.2173198694505384e-05,
      "loss": 2.1012,
      "step": 18000
    },
    {
      "epoch": 0.37,
      "eval_accuracy": 0.5583774858917174,
      "eval_loss": 1.9962693452835083,
      "eval_runtime": 16.4418,
      "eval_samples_per_second": 36.188,
      "eval_steps_per_second": 1.156,
      "step": 18000
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.419921875,
      "learning_rate": 2.2164139934189483e-05,
      "loss": 2.1205,
      "step": 18010
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.427734375,
      "learning_rate": 2.2155077787565828e-05,
      "loss": 2.143,
      "step": 18020
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.447265625,
      "learning_rate": 2.2146012258917867e-05,
      "loss": 2.1182,
      "step": 18030
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.435546875,
      "learning_rate": 2.2136943352530643e-05,
      "loss": 2.1035,
      "step": 18040
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4296875,
      "learning_rate": 2.2127871072690798e-05,
      "loss": 2.1078,
      "step": 18050
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.44140625,
      "learning_rate": 2.2118795423686577e-05,
      "loss": 2.1071,
      "step": 18060
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4296875,
      "learning_rate": 2.2109716409807804e-05,
      "loss": 2.1282,
      "step": 18070
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.466796875,
      "learning_rate": 2.2100634035345898e-05,
      "loss": 2.077,
      "step": 18080
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4765625,
      "learning_rate": 2.2091548304593873e-05,
      "loss": 2.1142,
      "step": 18090
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.45703125,
      "learning_rate": 2.208245922184632e-05,
      "loss": 2.0945,
      "step": 18100
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.43359375,
      "learning_rate": 2.2073366791399415e-05,
      "loss": 2.1391,
      "step": 18110
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4296875,
      "learning_rate": 2.206427101755093e-05,
      "loss": 2.1022,
      "step": 18120
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.47265625,
      "learning_rate": 2.2055171904600202e-05,
      "loss": 2.1278,
      "step": 18130
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.498046875,
      "learning_rate": 2.2046069456848146e-05,
      "loss": 2.1298,
      "step": 18140
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4453125,
      "learning_rate": 2.2036963678597266e-05,
      "loss": 2.1101,
      "step": 18150
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.44140625,
      "learning_rate": 2.2027854574151634e-05,
      "loss": 2.1756,
      "step": 18160
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4453125,
      "learning_rate": 2.201874214781689e-05,
      "loss": 2.1157,
      "step": 18170
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.50390625,
      "learning_rate": 2.200962640390025e-05,
      "loss": 2.1264,
      "step": 18180
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.43359375,
      "learning_rate": 2.2000507346710487e-05,
      "loss": 2.1134,
      "step": 18190
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5078125,
      "learning_rate": 2.1991384980557958e-05,
      "loss": 2.1098,
      "step": 18200
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4296875,
      "learning_rate": 2.1982259309754574e-05,
      "loss": 2.1148,
      "step": 18210
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.431640625,
      "learning_rate": 2.1973130338613804e-05,
      "loss": 2.1257,
      "step": 18220
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.443359375,
      "learning_rate": 2.1963998071450685e-05,
      "loss": 2.1218,
      "step": 18230
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.421875,
      "learning_rate": 2.1954862512581807e-05,
      "loss": 2.1428,
      "step": 18240
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.439453125,
      "learning_rate": 2.1945723666325324e-05,
      "loss": 2.1209,
      "step": 18250
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.49609375,
      "learning_rate": 2.193658153700093e-05,
      "loss": 2.1002,
      "step": 18260
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.443359375,
      "learning_rate": 2.1927436128929875e-05,
      "loss": 2.1369,
      "step": 18270
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4453125,
      "learning_rate": 2.1918287446434974e-05,
      "loss": 2.1262,
      "step": 18280
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.423828125,
      "learning_rate": 2.1909135493840573e-05,
      "loss": 2.0885,
      "step": 18290
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4453125,
      "learning_rate": 2.1899980275472562e-05,
      "loss": 2.1326,
      "step": 18300
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.46484375,
      "learning_rate": 2.189082179565839e-05,
      "loss": 2.1098,
      "step": 18310
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.455078125,
      "learning_rate": 2.188166005872704e-05,
      "loss": 2.1133,
      "step": 18320
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4375,
      "learning_rate": 2.187249506900902e-05,
      "loss": 2.1311,
      "step": 18330
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.470703125,
      "learning_rate": 2.1863326830836396e-05,
      "loss": 2.1083,
      "step": 18340
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.431640625,
      "learning_rate": 2.1854155348542764e-05,
      "loss": 2.0964,
      "step": 18350
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.451171875,
      "learning_rate": 2.1844980626463254e-05,
      "loss": 2.1133,
      "step": 18360
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.48046875,
      "learning_rate": 2.1835802668934527e-05,
      "loss": 2.13,
      "step": 18370
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4296875,
      "learning_rate": 2.182662148029476e-05,
      "loss": 2.1313,
      "step": 18380
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.44921875,
      "learning_rate": 2.1817437064883673e-05,
      "loss": 2.1062,
      "step": 18390
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.427734375,
      "learning_rate": 2.1808249427042516e-05,
      "loss": 2.0795,
      "step": 18400
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4296875,
      "learning_rate": 2.1799058571114046e-05,
      "loss": 2.103,
      "step": 18410
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.421875,
      "learning_rate": 2.1789864501442547e-05,
      "loss": 2.1372,
      "step": 18420
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.435546875,
      "learning_rate": 2.1780667222373826e-05,
      "loss": 2.1031,
      "step": 18430
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.453125,
      "learning_rate": 2.1771466738255198e-05,
      "loss": 2.1161,
      "step": 18440
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5,
      "learning_rate": 2.176226305343551e-05,
      "loss": 2.128,
      "step": 18450
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.443359375,
      "learning_rate": 2.17530561722651e-05,
      "loss": 2.1048,
      "step": 18460
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.435546875,
      "learning_rate": 2.1743846099095837e-05,
      "loss": 2.092,
      "step": 18470
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.423828125,
      "learning_rate": 2.1734632838281082e-05,
      "loss": 2.1186,
      "step": 18480
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.42578125,
      "learning_rate": 2.1725416394175714e-05,
      "loss": 2.1216,
      "step": 18490
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.435546875,
      "learning_rate": 2.1716196771136115e-05,
      "loss": 2.1155,
      "step": 18500
    },
    {
      "epoch": 0.38,
      "eval_accuracy": 0.5584908450183592,
      "eval_loss": 1.9959198236465454,
      "eval_runtime": 16.4576,
      "eval_samples_per_second": 36.154,
      "eval_steps_per_second": 1.154,
      "step": 18500
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.431640625,
      "learning_rate": 2.1706973973520167e-05,
      "loss": 2.1317,
      "step": 18510
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.423828125,
      "learning_rate": 2.169774800568725e-05,
      "loss": 2.1393,
      "step": 18520
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.46484375,
      "learning_rate": 2.1688518871998247e-05,
      "loss": 2.1273,
      "step": 18530
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.435546875,
      "learning_rate": 2.1679286576815538e-05,
      "loss": 2.1473,
      "step": 18540
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.431640625,
      "learning_rate": 2.167005112450299e-05,
      "loss": 2.1019,
      "step": 18550
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.439453125,
      "learning_rate": 2.166081251942598e-05,
      "loss": 2.1197,
      "step": 18560
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.44921875,
      "learning_rate": 2.1651570765951348e-05,
      "loss": 2.1222,
      "step": 18570
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42578125,
      "learning_rate": 2.1642325868447444e-05,
      "loss": 2.1083,
      "step": 18580
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.447265625,
      "learning_rate": 2.16330778312841e-05,
      "loss": 2.1206,
      "step": 18590
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.435546875,
      "learning_rate": 2.162382665883262e-05,
      "loss": 2.1095,
      "step": 18600
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42578125,
      "learning_rate": 2.1614572355465814e-05,
      "loss": 2.1141,
      "step": 18610
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.43359375,
      "learning_rate": 2.160531492555794e-05,
      "loss": 2.1116,
      "step": 18620
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.45703125,
      "learning_rate": 2.1596054373484764e-05,
      "loss": 2.0925,
      "step": 18630
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.44140625,
      "learning_rate": 2.158679070362351e-05,
      "loss": 2.1143,
      "step": 18640
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.58203125,
      "learning_rate": 2.1577523920352882e-05,
      "loss": 2.0887,
      "step": 18650
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4375,
      "learning_rate": 2.1568254028053053e-05,
      "loss": 2.1091,
      "step": 18660
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.43359375,
      "learning_rate": 2.1558981031105665e-05,
      "loss": 2.1082,
      "step": 18670
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.423828125,
      "learning_rate": 2.1549704933893834e-05,
      "loss": 2.1472,
      "step": 18680
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4453125,
      "learning_rate": 2.1540425740802134e-05,
      "loss": 2.1089,
      "step": 18690
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.427734375,
      "learning_rate": 2.15311434562166e-05,
      "loss": 2.1108,
      "step": 18700
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42578125,
      "learning_rate": 2.1521858084524737e-05,
      "loss": 2.127,
      "step": 18710
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.435546875,
      "learning_rate": 2.1512569630115507e-05,
      "loss": 2.1515,
      "step": 18720
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.43359375,
      "learning_rate": 2.1503278097379323e-05,
      "loss": 2.1187,
      "step": 18730
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.427734375,
      "learning_rate": 2.1493983490708065e-05,
      "loss": 2.1243,
      "step": 18740
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4375,
      "learning_rate": 2.1484685814495054e-05,
      "loss": 2.0826,
      "step": 18750
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.431640625,
      "learning_rate": 2.147538507313506e-05,
      "loss": 2.1193,
      "step": 18760
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.421875,
      "learning_rate": 2.146608127102432e-05,
      "loss": 2.1057,
      "step": 18770
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.44921875,
      "learning_rate": 2.1456774412560502e-05,
      "loss": 2.1127,
      "step": 18780
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42578125,
      "learning_rate": 2.1447464502142723e-05,
      "loss": 2.0959,
      "step": 18790
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.44140625,
      "learning_rate": 2.143815154417154e-05,
      "loss": 2.1005,
      "step": 18800
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.44140625,
      "learning_rate": 2.142883554304895e-05,
      "loss": 2.125,
      "step": 18810
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.419921875,
      "learning_rate": 2.1419516503178392e-05,
      "loss": 2.1308,
      "step": 18820
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.443359375,
      "learning_rate": 2.1410194428964746e-05,
      "loss": 2.079,
      "step": 18830
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42578125,
      "learning_rate": 2.1400869324814315e-05,
      "loss": 2.1182,
      "step": 18840
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.52734375,
      "learning_rate": 2.1391541195134845e-05,
      "loss": 2.1309,
      "step": 18850
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.43359375,
      "learning_rate": 2.1382210044335504e-05,
      "loss": 2.1397,
      "step": 18860
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.439453125,
      "learning_rate": 2.1372875876826892e-05,
      "loss": 2.1195,
      "step": 18870
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.45703125,
      "learning_rate": 2.136353869702103e-05,
      "loss": 2.1318,
      "step": 18880
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.44921875,
      "learning_rate": 2.1354198509331378e-05,
      "loss": 2.126,
      "step": 18890
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.431640625,
      "learning_rate": 2.1344855318172788e-05,
      "loss": 2.1306,
      "step": 18900
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.431640625,
      "learning_rate": 2.1335509127961566e-05,
      "loss": 2.1333,
      "step": 18910
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42578125,
      "learning_rate": 2.1326159943115415e-05,
      "loss": 2.1297,
      "step": 18920
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.451171875,
      "learning_rate": 2.1316807768053455e-05,
      "loss": 2.1414,
      "step": 18930
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.447265625,
      "learning_rate": 2.1307452607196226e-05,
      "loss": 2.1421,
      "step": 18940
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.455078125,
      "learning_rate": 2.1298094464965677e-05,
      "loss": 2.1333,
      "step": 18950
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.466796875,
      "learning_rate": 2.1288733345785163e-05,
      "loss": 2.1258,
      "step": 18960
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.447265625,
      "learning_rate": 2.127936925407945e-05,
      "loss": 2.1289,
      "step": 18970
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.41796875,
      "learning_rate": 2.1270002194274698e-05,
      "loss": 2.1416,
      "step": 18980
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.423828125,
      "learning_rate": 2.1260632170798494e-05,
      "loss": 2.0975,
      "step": 18990
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.421875,
      "learning_rate": 2.12512591880798e-05,
      "loss": 2.1493,
      "step": 19000
    },
    {
      "epoch": 0.4,
      "eval_accuracy": 0.558476059045319,
      "eval_loss": 1.9956163167953491,
      "eval_runtime": 16.4379,
      "eval_samples_per_second": 36.197,
      "eval_steps_per_second": 1.156,
      "step": 19000
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.423828125,
      "learning_rate": 2.1241883250548996e-05,
      "loss": 2.0851,
      "step": 19010
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.44140625,
      "learning_rate": 2.1232504362637845e-05,
      "loss": 2.1371,
      "step": 19020
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.44140625,
      "learning_rate": 2.1223122528779515e-05,
      "loss": 2.1261,
      "step": 19030
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.431640625,
      "learning_rate": 2.1213737753408554e-05,
      "loss": 2.1002,
      "step": 19040
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4453125,
      "learning_rate": 2.1204350040960914e-05,
      "loss": 2.1104,
      "step": 19050
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.43359375,
      "learning_rate": 2.1194959395873927e-05,
      "loss": 2.0982,
      "step": 19060
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.486328125,
      "learning_rate": 2.1185565822586313e-05,
      "loss": 2.1183,
      "step": 19070
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.41796875,
      "learning_rate": 2.117616932553818e-05,
      "loss": 2.086,
      "step": 19080
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.443359375,
      "learning_rate": 2.116676990917101e-05,
      "loss": 2.1199,
      "step": 19090
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.443359375,
      "learning_rate": 2.115736757792767e-05,
      "loss": 2.1368,
      "step": 19100
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4453125,
      "learning_rate": 2.114796233625241e-05,
      "loss": 2.0892,
      "step": 19110
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.466796875,
      "learning_rate": 2.113855418859084e-05,
      "loss": 2.1349,
      "step": 19120
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.455078125,
      "learning_rate": 2.1129143139389966e-05,
      "loss": 2.1185,
      "step": 19130
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.478515625,
      "learning_rate": 2.1119729193098135e-05,
      "loss": 2.1421,
      "step": 19140
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.453125,
      "learning_rate": 2.1110312354165102e-05,
      "loss": 2.12,
      "step": 19150
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.43359375,
      "learning_rate": 2.1100892627041958e-05,
      "loss": 2.1586,
      "step": 19160
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4296875,
      "learning_rate": 2.109147001618117e-05,
      "loss": 2.132,
      "step": 19170
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.46875,
      "learning_rate": 2.1082044526036564e-05,
      "loss": 2.1453,
      "step": 19180
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.43359375,
      "learning_rate": 2.1072616161063334e-05,
      "loss": 2.146,
      "step": 19190
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.46875,
      "learning_rate": 2.1063184925718037e-05,
      "loss": 2.1015,
      "step": 19200
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.421875,
      "learning_rate": 2.1053750824458572e-05,
      "loss": 2.1223,
      "step": 19210
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.421875,
      "learning_rate": 2.10443138617442e-05,
      "loss": 2.1169,
      "step": 19220
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4375,
      "learning_rate": 2.1034874042035538e-05,
      "loss": 2.127,
      "step": 19230
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42578125,
      "learning_rate": 2.1025431369794546e-05,
      "loss": 2.104,
      "step": 19240
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.462890625,
      "learning_rate": 2.1015985849484534e-05,
      "loss": 2.1077,
      "step": 19250
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42578125,
      "learning_rate": 2.100653748557017e-05,
      "loss": 2.1217,
      "step": 19260
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4296875,
      "learning_rate": 2.099708628251745e-05,
      "loss": 2.1567,
      "step": 19270
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.44140625,
      "learning_rate": 2.098763224479372e-05,
      "loss": 2.1086,
      "step": 19280
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.474609375,
      "learning_rate": 2.0978175376867662e-05,
      "loss": 2.0886,
      "step": 19290
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.44921875,
      "learning_rate": 2.096871568320931e-05,
      "loss": 2.1171,
      "step": 19300
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.43359375,
      "learning_rate": 2.0959253168290006e-05,
      "loss": 2.1295,
      "step": 19310
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4375,
      "learning_rate": 2.094978783658245e-05,
      "loss": 2.1168,
      "step": 19320
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.45703125,
      "learning_rate": 2.094031969256066e-05,
      "loss": 2.1643,
      "step": 19330
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.439453125,
      "learning_rate": 2.0930848740699996e-05,
      "loss": 2.123,
      "step": 19340
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.435546875,
      "learning_rate": 2.092137498547713e-05,
      "loss": 2.1311,
      "step": 19350
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.427734375,
      "learning_rate": 2.0911898431370068e-05,
      "loss": 2.1213,
      "step": 19360
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4609375,
      "learning_rate": 2.0902419082858142e-05,
      "loss": 2.137,
      "step": 19370
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.44140625,
      "learning_rate": 2.0892936944421997e-05,
      "loss": 2.1128,
      "step": 19380
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.439453125,
      "learning_rate": 2.0883452020543598e-05,
      "loss": 2.1695,
      "step": 19390
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.439453125,
      "learning_rate": 2.0873964315706234e-05,
      "loss": 2.1137,
      "step": 19400
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4453125,
      "learning_rate": 2.08644738343945e-05,
      "loss": 2.1019,
      "step": 19410
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.423828125,
      "learning_rate": 2.0854980581094304e-05,
      "loss": 2.1201,
      "step": 19420
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.458984375,
      "learning_rate": 2.084548456029287e-05,
      "loss": 2.1292,
      "step": 19430
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.435546875,
      "learning_rate": 2.0835985776478722e-05,
      "loss": 2.1157,
      "step": 19440
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.44140625,
      "learning_rate": 2.0826484234141703e-05,
      "loss": 2.1258,
      "step": 19450
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.55078125,
      "learning_rate": 2.081697993777295e-05,
      "loss": 2.1206,
      "step": 19460
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.44921875,
      "learning_rate": 2.0807472891864894e-05,
      "loss": 2.0988,
      "step": 19470
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.435546875,
      "learning_rate": 2.0797963100911286e-05,
      "loss": 2.1064,
      "step": 19480
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.42578125,
      "learning_rate": 2.078845056940716e-05,
      "loss": 2.0966,
      "step": 19490
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.462890625,
      "learning_rate": 2.0778935301848852e-05,
      "loss": 2.1219,
      "step": 19500
    },
    {
      "epoch": 0.41,
      "eval_accuracy": 0.558676491124309,
      "eval_loss": 1.9952665567398071,
      "eval_runtime": 16.4341,
      "eval_samples_per_second": 36.205,
      "eval_steps_per_second": 1.156,
      "step": 19500
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.435546875,
      "learning_rate": 2.076941730273398e-05,
      "loss": 2.0871,
      "step": 19510
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.453125,
      "learning_rate": 2.075989657656147e-05,
      "loss": 2.1252,
      "step": 19520
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.47265625,
      "learning_rate": 2.0750373127831525e-05,
      "loss": 2.1165,
      "step": 19530
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.423828125,
      "learning_rate": 2.0740846961045637e-05,
      "loss": 2.1515,
      "step": 19540
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.482421875,
      "learning_rate": 2.073131808070659e-05,
      "loss": 2.1221,
      "step": 19550
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.42578125,
      "learning_rate": 2.0721786491318436e-05,
      "loss": 2.0964,
      "step": 19560
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.44140625,
      "learning_rate": 2.0712252197386517e-05,
      "loss": 2.1163,
      "step": 19570
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.447265625,
      "learning_rate": 2.0702715203417465e-05,
      "loss": 2.1202,
      "step": 19580
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4609375,
      "learning_rate": 2.0693175513919164e-05,
      "loss": 2.1448,
      "step": 19590
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.435546875,
      "learning_rate": 2.0683633133400784e-05,
      "loss": 2.147,
      "step": 19600
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.43359375,
      "learning_rate": 2.0674088066372773e-05,
      "loss": 2.1263,
      "step": 19610
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.431640625,
      "learning_rate": 2.066454031734684e-05,
      "loss": 2.115,
      "step": 19620
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.462890625,
      "learning_rate": 2.065498989083596e-05,
      "loss": 2.1071,
      "step": 19630
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.451171875,
      "learning_rate": 2.0645436791354387e-05,
      "loss": 2.1369,
      "step": 19640
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4375,
      "learning_rate": 2.0635881023417627e-05,
      "loss": 2.0942,
      "step": 19650
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.47265625,
      "learning_rate": 2.062632259154244e-05,
      "loss": 2.1202,
      "step": 19660
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.455078125,
      "learning_rate": 2.0616761500246866e-05,
      "loss": 2.1322,
      "step": 19670
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.80859375,
      "learning_rate": 2.060719775405019e-05,
      "loss": 2.0766,
      "step": 19680
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.43359375,
      "learning_rate": 2.059763135747295e-05,
      "loss": 2.1346,
      "step": 19690
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.455078125,
      "learning_rate": 2.0588062315036943e-05,
      "loss": 2.1397,
      "step": 19700
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4609375,
      "learning_rate": 2.0578490631265204e-05,
      "loss": 2.1286,
      "step": 19710
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4375,
      "learning_rate": 2.056891631068204e-05,
      "loss": 2.1362,
      "step": 19720
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.451171875,
      "learning_rate": 2.0559339357812976e-05,
      "loss": 2.1279,
      "step": 19730
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4296875,
      "learning_rate": 2.0549759777184803e-05,
      "loss": 2.0917,
      "step": 19740
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.486328125,
      "learning_rate": 2.0540177573325547e-05,
      "loss": 2.1212,
      "step": 19750
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.42578125,
      "learning_rate": 2.053059275076447e-05,
      "loss": 2.1251,
      "step": 19760
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.427734375,
      "learning_rate": 2.052100531403208e-05,
      "loss": 2.1012,
      "step": 19770
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.421875,
      "learning_rate": 2.0511415267660113e-05,
      "loss": 2.1164,
      "step": 19780
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.431640625,
      "learning_rate": 2.0501822616181542e-05,
      "loss": 2.1144,
      "step": 19790
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.439453125,
      "learning_rate": 2.0492227364130567e-05,
      "loss": 2.1057,
      "step": 19800
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.42578125,
      "learning_rate": 2.048262951604263e-05,
      "loss": 2.1455,
      "step": 19810
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.435546875,
      "learning_rate": 2.0473029076454378e-05,
      "loss": 2.1093,
      "step": 19820
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4375,
      "learning_rate": 2.0463426049903707e-05,
      "loss": 2.116,
      "step": 19830
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.44140625,
      "learning_rate": 2.045382044092972e-05,
      "loss": 2.1448,
      "step": 19840
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.453125,
      "learning_rate": 2.0444212254072746e-05,
      "loss": 2.1413,
      "step": 19850
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4375,
      "learning_rate": 2.0434601493874336e-05,
      "loss": 2.0833,
      "step": 19860
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.435546875,
      "learning_rate": 2.0424988164877246e-05,
      "loss": 2.1135,
      "step": 19870
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4375,
      "learning_rate": 2.0415372271625456e-05,
      "loss": 2.11,
      "step": 19880
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4609375,
      "learning_rate": 2.0405753818664162e-05,
      "loss": 2.0997,
      "step": 19890
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.458984375,
      "learning_rate": 2.0396132810539753e-05,
      "loss": 2.1178,
      "step": 19900
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.42578125,
      "learning_rate": 2.038650925179984e-05,
      "loss": 2.109,
      "step": 19910
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.439453125,
      "learning_rate": 2.0376883146993242e-05,
      "loss": 2.1309,
      "step": 19920
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.72265625,
      "learning_rate": 2.036725450066997e-05,
      "loss": 2.1595,
      "step": 19930
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.451171875,
      "learning_rate": 2.035762331738124e-05,
      "loss": 2.1005,
      "step": 19940
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.44140625,
      "learning_rate": 2.0347989601679465e-05,
      "loss": 2.1317,
      "step": 19950
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4375,
      "learning_rate": 2.0338353358118276e-05,
      "loss": 2.1407,
      "step": 19960
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.435546875,
      "learning_rate": 2.0328714591252466e-05,
      "loss": 2.1295,
      "step": 19970
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.419921875,
      "learning_rate": 2.0319073305638035e-05,
      "loss": 2.1088,
      "step": 19980
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.439453125,
      "learning_rate": 2.0309429505832184e-05,
      "loss": 2.1281,
      "step": 19990
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.515625,
      "learning_rate": 2.0299783196393288e-05,
      "loss": 2.1584,
      "step": 20000
    },
    {
      "epoch": 0.42,
      "eval_accuracy": 0.558804636223991,
      "eval_loss": 1.9951595067977905,
      "eval_runtime": 16.4485,
      "eval_samples_per_second": 36.174,
      "eval_steps_per_second": 1.155,
      "step": 20000
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.41796875,
      "learning_rate": 2.029013438188092e-05,
      "loss": 2.1485,
      "step": 20010
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.427734375,
      "learning_rate": 2.0280483066855823e-05,
      "loss": 2.1418,
      "step": 20020
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.470703125,
      "learning_rate": 2.027082925587993e-05,
      "loss": 2.1516,
      "step": 20030
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.421875,
      "learning_rate": 2.0261172953516354e-05,
      "loss": 2.1501,
      "step": 20040
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4296875,
      "learning_rate": 2.0251514164329395e-05,
      "loss": 2.1348,
      "step": 20050
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.427734375,
      "learning_rate": 2.02418528928845e-05,
      "loss": 2.1222,
      "step": 20060
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.44140625,
      "learning_rate": 2.023218914374832e-05,
      "loss": 2.1327,
      "step": 20070
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.46484375,
      "learning_rate": 2.022252292148867e-05,
      "loss": 2.1228,
      "step": 20080
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.59765625,
      "learning_rate": 2.021285423067451e-05,
      "loss": 2.1693,
      "step": 20090
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.421875,
      "learning_rate": 2.0203183075876e-05,
      "loss": 2.0866,
      "step": 20100
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4453125,
      "learning_rate": 2.019350946166445e-05,
      "loss": 2.1188,
      "step": 20110
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.42578125,
      "learning_rate": 2.0183833392612322e-05,
      "loss": 2.149,
      "step": 20120
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.451171875,
      "learning_rate": 2.0174154873293267e-05,
      "loss": 2.1203,
      "step": 20130
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4375,
      "learning_rate": 2.0164473908282058e-05,
      "loss": 2.1023,
      "step": 20140
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.44140625,
      "learning_rate": 2.0154790502154655e-05,
      "loss": 2.115,
      "step": 20150
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.427734375,
      "learning_rate": 2.0145104659488152e-05,
      "loss": 2.1455,
      "step": 20160
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.435546875,
      "learning_rate": 2.0135416384860805e-05,
      "loss": 2.0928,
      "step": 20170
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.416015625,
      "learning_rate": 2.0125725682852017e-05,
      "loss": 2.1212,
      "step": 20180
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.458984375,
      "learning_rate": 2.0116032558042337e-05,
      "loss": 2.1673,
      "step": 20190
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.462890625,
      "learning_rate": 2.0106337015013458e-05,
      "loss": 2.1458,
      "step": 20200
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.46484375,
      "learning_rate": 2.0096639058348217e-05,
      "loss": 2.1394,
      "step": 20210
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.466796875,
      "learning_rate": 2.0086938692630597e-05,
      "loss": 2.1167,
      "step": 20220
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.443359375,
      "learning_rate": 2.0077235922445714e-05,
      "loss": 2.1376,
      "step": 20230
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.455078125,
      "learning_rate": 2.0067530752379823e-05,
      "loss": 2.0924,
      "step": 20240
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.423828125,
      "learning_rate": 2.005782318702031e-05,
      "loss": 2.1112,
      "step": 20250
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.42578125,
      "learning_rate": 2.00481132309557e-05,
      "loss": 2.0978,
      "step": 20260
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.439453125,
      "learning_rate": 2.0038400888775637e-05,
      "loss": 2.1434,
      "step": 20270
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.453125,
      "learning_rate": 2.00286861650709e-05,
      "loss": 2.1279,
      "step": 20280
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.423828125,
      "learning_rate": 2.00189690644334e-05,
      "loss": 2.1358,
      "step": 20290
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.439453125,
      "learning_rate": 2.0009249591456163e-05,
      "loss": 2.1218,
      "step": 20300
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.443359375,
      "learning_rate": 1.999952775073334e-05,
      "loss": 2.1323,
      "step": 20310
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.439453125,
      "learning_rate": 1.998980354686019e-05,
      "loss": 2.1324,
      "step": 20320
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.44140625,
      "learning_rate": 1.99800769844331e-05,
      "loss": 2.1094,
      "step": 20330
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4453125,
      "learning_rate": 1.9970348068049575e-05,
      "loss": 2.139,
      "step": 20340
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.423828125,
      "learning_rate": 1.996061680230823e-05,
      "loss": 2.1181,
      "step": 20350
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.44921875,
      "learning_rate": 1.995088319180878e-05,
      "loss": 2.101,
      "step": 20360
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.439453125,
      "learning_rate": 1.9941147241152062e-05,
      "loss": 2.1098,
      "step": 20370
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.455078125,
      "learning_rate": 1.9931408954940017e-05,
      "loss": 2.1317,
      "step": 20380
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.45703125,
      "learning_rate": 1.9921668337775677e-05,
      "loss": 2.1192,
      "step": 20390
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.48828125,
      "learning_rate": 1.9911925394263195e-05,
      "loss": 2.0815,
      "step": 20400
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5078125,
      "learning_rate": 1.9902180129007806e-05,
      "loss": 2.0946,
      "step": 20410
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.421875,
      "learning_rate": 1.9892432546615857e-05,
      "loss": 2.1091,
      "step": 20420
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4296875,
      "learning_rate": 1.988268265169478e-05,
      "loss": 2.108,
      "step": 20430
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.42578125,
      "learning_rate": 1.9872930448853102e-05,
      "loss": 2.123,
      "step": 20440
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.455078125,
      "learning_rate": 1.9863175942700448e-05,
      "loss": 2.1033,
      "step": 20450
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44140625,
      "learning_rate": 1.985341913784752e-05,
      "loss": 2.1119,
      "step": 20460
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.427734375,
      "learning_rate": 1.9843660038906116e-05,
      "loss": 2.1265,
      "step": 20470
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.427734375,
      "learning_rate": 1.9833898650489117e-05,
      "loss": 2.1382,
      "step": 20480
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.439453125,
      "learning_rate": 1.9824134977210485e-05,
      "loss": 2.1237,
      "step": 20490
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4609375,
      "learning_rate": 1.9814369023685257e-05,
      "loss": 2.1167,
      "step": 20500
    },
    {
      "epoch": 0.43,
      "eval_accuracy": 0.5586912770973492,
      "eval_loss": 1.9950366020202637,
      "eval_runtime": 16.449,
      "eval_samples_per_second": 36.172,
      "eval_steps_per_second": 1.155,
      "step": 20500
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44140625,
      "learning_rate": 1.9804600794529553e-05,
      "loss": 2.1172,
      "step": 20510
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.443359375,
      "learning_rate": 1.9794830294360576e-05,
      "loss": 2.1295,
      "step": 20520
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.423828125,
      "learning_rate": 1.978505752779658e-05,
      "loss": 2.1252,
      "step": 20530
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.455078125,
      "learning_rate": 1.9775282499456915e-05,
      "loss": 2.117,
      "step": 20540
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5078125,
      "learning_rate": 1.976550521396199e-05,
      "loss": 2.128,
      "step": 20550
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44921875,
      "learning_rate": 1.9755725675933274e-05,
      "loss": 2.123,
      "step": 20560
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4296875,
      "learning_rate": 1.9745943889993314e-05,
      "loss": 2.1237,
      "step": 20570
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.421875,
      "learning_rate": 1.9736159860765703e-05,
      "loss": 2.1207,
      "step": 20580
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.419921875,
      "learning_rate": 1.972637359287512e-05,
      "loss": 2.1174,
      "step": 20590
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.416015625,
      "learning_rate": 1.9716585090947272e-05,
      "loss": 2.1263,
      "step": 20600
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44921875,
      "learning_rate": 1.9706794359608938e-05,
      "loss": 2.1316,
      "step": 20610
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.470703125,
      "learning_rate": 1.9697001403487964e-05,
      "loss": 2.1233,
      "step": 20620
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.453125,
      "learning_rate": 1.9687206227213207e-05,
      "loss": 2.1083,
      "step": 20630
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.443359375,
      "learning_rate": 1.967740883541462e-05,
      "loss": 2.0951,
      "step": 20640
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.423828125,
      "learning_rate": 1.966760923272318e-05,
      "loss": 2.0794,
      "step": 20650
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44921875,
      "learning_rate": 1.96578074237709e-05,
      "loss": 2.1346,
      "step": 20660
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.41796875,
      "learning_rate": 1.9648003413190858e-05,
      "loss": 2.1101,
      "step": 20670
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.423828125,
      "learning_rate": 1.963819720561716e-05,
      "loss": 2.1105,
      "step": 20680
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4609375,
      "learning_rate": 1.9628388805684946e-05,
      "loss": 2.1145,
      "step": 20690
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.435546875,
      "learning_rate": 1.9618578218030406e-05,
      "loss": 2.1478,
      "step": 20700
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.447265625,
      "learning_rate": 1.960876544729075e-05,
      "loss": 2.1112,
      "step": 20710
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.515625,
      "learning_rate": 1.959895049810423e-05,
      "loss": 2.107,
      "step": 20720
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44140625,
      "learning_rate": 1.9589133375110132e-05,
      "loss": 2.0901,
      "step": 20730
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4453125,
      "learning_rate": 1.9579314082948746e-05,
      "loss": 2.1256,
      "step": 20740
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.416015625,
      "learning_rate": 1.9569492626261414e-05,
      "loss": 2.1244,
      "step": 20750
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.45703125,
      "learning_rate": 1.9559669009690488e-05,
      "loss": 2.1172,
      "step": 20760
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.43359375,
      "learning_rate": 1.954984323787934e-05,
      "loss": 2.1233,
      "step": 20770
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44140625,
      "learning_rate": 1.9540015315472363e-05,
      "loss": 2.0865,
      "step": 20780
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.435546875,
      "learning_rate": 1.9530185247114978e-05,
      "loss": 2.1073,
      "step": 20790
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.439453125,
      "learning_rate": 1.95203530374536e-05,
      "loss": 2.0946,
      "step": 20800
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.421875,
      "learning_rate": 1.951051869113567e-05,
      "loss": 2.1076,
      "step": 20810
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4375,
      "learning_rate": 1.9500682212809627e-05,
      "loss": 2.1217,
      "step": 20820
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.43359375,
      "learning_rate": 1.9490843607124937e-05,
      "loss": 2.1223,
      "step": 20830
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.466796875,
      "learning_rate": 1.948100287873205e-05,
      "loss": 2.1265,
      "step": 20840
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.453125,
      "learning_rate": 1.9471160032282436e-05,
      "loss": 2.1059,
      "step": 20850
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.443359375,
      "learning_rate": 1.9461315072428558e-05,
      "loss": 2.1335,
      "step": 20860
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.431640625,
      "learning_rate": 1.945146800382388e-05,
      "loss": 2.1306,
      "step": 20870
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44921875,
      "learning_rate": 1.944161883112286e-05,
      "loss": 2.0951,
      "step": 20880
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.423828125,
      "learning_rate": 1.9431767558980954e-05,
      "loss": 2.1494,
      "step": 20890
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.443359375,
      "learning_rate": 1.942191419205461e-05,
      "loss": 2.1085,
      "step": 20900
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.416015625,
      "learning_rate": 1.9412058735001265e-05,
      "loss": 2.1365,
      "step": 20910
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.51953125,
      "learning_rate": 1.940220119247934e-05,
      "loss": 2.1237,
      "step": 20920
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.435546875,
      "learning_rate": 1.9392341569148254e-05,
      "loss": 2.1004,
      "step": 20930
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42578125,
      "learning_rate": 1.93824798696684e-05,
      "loss": 2.1279,
      "step": 20940
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4609375,
      "learning_rate": 1.937261609870115e-05,
      "loss": 2.1135,
      "step": 20950
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.427734375,
      "learning_rate": 1.9362750260908855e-05,
      "loss": 2.1207,
      "step": 20960
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.453125,
      "learning_rate": 1.935288236095486e-05,
      "loss": 2.112,
      "step": 20970
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.435546875,
      "learning_rate": 1.934301240350346e-05,
      "loss": 2.1004,
      "step": 20980
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.455078125,
      "learning_rate": 1.933314039321994e-05,
      "loss": 2.1236,
      "step": 20990
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4296875,
      "learning_rate": 1.9323266334770547e-05,
      "loss": 2.1507,
      "step": 21000
    },
    {
      "epoch": 0.44,
      "eval_accuracy": 0.5586485620641218,
      "eval_loss": 1.9948058128356934,
      "eval_runtime": 16.4321,
      "eval_samples_per_second": 36.21,
      "eval_steps_per_second": 1.156,
      "step": 21000
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4453125,
      "learning_rate": 1.93133902328225e-05,
      "loss": 2.1003,
      "step": 21010
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.421875,
      "learning_rate": 1.9303512092043982e-05,
      "loss": 2.1321,
      "step": 21020
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.458984375,
      "learning_rate": 1.9293631917104143e-05,
      "loss": 2.1093,
      "step": 21030
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.44140625,
      "learning_rate": 1.928374971267309e-05,
      "loss": 2.1293,
      "step": 21040
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.439453125,
      "learning_rate": 1.927386548342189e-05,
      "loss": 2.1194,
      "step": 21050
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.58203125,
      "learning_rate": 1.9263979234022565e-05,
      "loss": 2.1374,
      "step": 21060
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.44921875,
      "learning_rate": 1.9254090969148103e-05,
      "loss": 2.0812,
      "step": 21070
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.435546875,
      "learning_rate": 1.924420069347243e-05,
      "loss": 2.0827,
      "step": 21080
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.427734375,
      "learning_rate": 1.9234308411670435e-05,
      "loss": 2.1191,
      "step": 21090
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.453125,
      "learning_rate": 1.9224414128417944e-05,
      "loss": 2.1265,
      "step": 21100
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4453125,
      "learning_rate": 1.9214517848391737e-05,
      "loss": 2.1404,
      "step": 21110
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.43359375,
      "learning_rate": 1.9204619576269532e-05,
      "loss": 2.0673,
      "step": 21120
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.45703125,
      "learning_rate": 1.919471931673e-05,
      "loss": 2.1161,
      "step": 21130
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.43359375,
      "learning_rate": 1.9184817074452735e-05,
      "loss": 2.1555,
      "step": 21140
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42578125,
      "learning_rate": 1.917491285411828e-05,
      "loss": 2.0993,
      "step": 21150
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.435546875,
      "learning_rate": 1.916500666040811e-05,
      "loss": 2.0877,
      "step": 21160
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.43359375,
      "learning_rate": 1.9155098498004632e-05,
      "loss": 2.1046,
      "step": 21170
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.416015625,
      "learning_rate": 1.9145188371591184e-05,
      "loss": 2.1036,
      "step": 21180
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.455078125,
      "learning_rate": 1.9135276285852028e-05,
      "loss": 2.1242,
      "step": 21190
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.451171875,
      "learning_rate": 1.9125362245472358e-05,
      "loss": 2.1173,
      "step": 21200
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4375,
      "learning_rate": 1.9115446255138296e-05,
      "loss": 2.1332,
      "step": 21210
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.455078125,
      "learning_rate": 1.9105528319536873e-05,
      "loss": 2.1088,
      "step": 21220
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.43359375,
      "learning_rate": 1.9095608443356048e-05,
      "loss": 2.1155,
      "step": 21230
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.45703125,
      "learning_rate": 1.9085686631284702e-05,
      "loss": 2.1532,
      "step": 21240
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.427734375,
      "learning_rate": 1.9075762888012612e-05,
      "loss": 2.1166,
      "step": 21250
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.431640625,
      "learning_rate": 1.9065837218230488e-05,
      "loss": 2.1597,
      "step": 21260
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.439453125,
      "learning_rate": 1.9055909626629946e-05,
      "loss": 2.1097,
      "step": 21270
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.439453125,
      "learning_rate": 1.9045980117903502e-05,
      "loss": 2.1096,
      "step": 21280
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.453125,
      "learning_rate": 1.9036048696744583e-05,
      "loss": 2.1596,
      "step": 21290
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.431640625,
      "learning_rate": 1.9026115367847525e-05,
      "loss": 2.1004,
      "step": 21300
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4375,
      "learning_rate": 1.9016180135907563e-05,
      "loss": 2.1107,
      "step": 21310
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.44921875,
      "learning_rate": 1.9006243005620825e-05,
      "loss": 2.125,
      "step": 21320
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4609375,
      "learning_rate": 1.899630398168434e-05,
      "loss": 2.1179,
      "step": 21330
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.447265625,
      "learning_rate": 1.8986363068796037e-05,
      "loss": 2.0901,
      "step": 21340
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.416015625,
      "learning_rate": 1.8976420271654733e-05,
      "loss": 2.0892,
      "step": 21350
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.474609375,
      "learning_rate": 1.8966475594960137e-05,
      "loss": 2.1162,
      "step": 21360
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.453125,
      "learning_rate": 1.8956529043412844e-05,
      "loss": 2.1425,
      "step": 21370
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.455078125,
      "learning_rate": 1.894658062171434e-05,
      "loss": 2.1121,
      "step": 21380
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.46484375,
      "learning_rate": 1.893663033456699e-05,
      "loss": 2.1301,
      "step": 21390
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.423828125,
      "learning_rate": 1.8926678186674047e-05,
      "loss": 2.1007,
      "step": 21400
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.435546875,
      "learning_rate": 1.8916724182739637e-05,
      "loss": 2.1118,
      "step": 21410
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.42578125,
      "learning_rate": 1.8906768327468767e-05,
      "loss": 2.1399,
      "step": 21420
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.431640625,
      "learning_rate": 1.8896810625567316e-05,
      "loss": 2.1572,
      "step": 21430
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4453125,
      "learning_rate": 1.888685108174204e-05,
      "loss": 2.1083,
      "step": 21440
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4453125,
      "learning_rate": 1.887688970070056e-05,
      "loss": 2.1104,
      "step": 21450
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4296875,
      "learning_rate": 1.8866926487151374e-05,
      "loss": 2.131,
      "step": 21460
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.455078125,
      "learning_rate": 1.8856961445803834e-05,
      "loss": 2.1342,
      "step": 21470
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.470703125,
      "learning_rate": 1.8846994581368166e-05,
      "loss": 2.118,
      "step": 21480
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.43359375,
      "learning_rate": 1.8837025898555462e-05,
      "loss": 2.1387,
      "step": 21490
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.458984375,
      "learning_rate": 1.8827055402077655e-05,
      "loss": 2.1043,
      "step": 21500
    },
    {
      "epoch": 0.45,
      "eval_accuracy": 0.5587159203857496,
      "eval_loss": 1.994558572769165,
      "eval_runtime": 16.4208,
      "eval_samples_per_second": 36.234,
      "eval_steps_per_second": 1.157,
      "step": 21500
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.44921875,
      "learning_rate": 1.8817083096647557e-05,
      "loss": 2.1209,
      "step": 21510
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.43359375,
      "learning_rate": 1.8807108986978816e-05,
      "loss": 2.1143,
      "step": 21520
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.44140625,
      "learning_rate": 1.8797133077785943e-05,
      "loss": 2.1108,
      "step": 21530
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.466796875,
      "learning_rate": 1.8787155373784302e-05,
      "loss": 2.1291,
      "step": 21540
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.462890625,
      "learning_rate": 1.8777175879690098e-05,
      "loss": 2.1061,
      "step": 21550
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4375,
      "learning_rate": 1.876719460022039e-05,
      "loss": 2.0808,
      "step": 21560
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.427734375,
      "learning_rate": 1.8757211540093073e-05,
      "loss": 2.122,
      "step": 21570
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.435546875,
      "learning_rate": 1.874722670402688e-05,
      "loss": 2.1077,
      "step": 21580
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.421875,
      "learning_rate": 1.8737240096741405e-05,
      "loss": 2.144,
      "step": 21590
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.42578125,
      "learning_rate": 1.8727251722957053e-05,
      "loss": 2.1605,
      "step": 21600
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4375,
      "learning_rate": 1.8717261587395077e-05,
      "loss": 2.1178,
      "step": 21610
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4375,
      "learning_rate": 1.870726969477757e-05,
      "loss": 2.1073,
      "step": 21620
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.421875,
      "learning_rate": 1.869727604982744e-05,
      "loss": 2.1159,
      "step": 21630
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.431640625,
      "learning_rate": 1.8687280657268428e-05,
      "loss": 2.1052,
      "step": 21640
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.451171875,
      "learning_rate": 1.8677283521825105e-05,
      "loss": 2.1352,
      "step": 21650
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.42578125,
      "learning_rate": 1.8667284648222872e-05,
      "loss": 2.1103,
      "step": 21660
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4453125,
      "learning_rate": 1.865728404118793e-05,
      "loss": 2.092,
      "step": 21670
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.451171875,
      "learning_rate": 1.864728170544733e-05,
      "loss": 2.1351,
      "step": 21680
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.470703125,
      "learning_rate": 1.863727764572891e-05,
      "loss": 2.1087,
      "step": 21690
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4453125,
      "learning_rate": 1.8627271866761335e-05,
      "loss": 2.1062,
      "step": 21700
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.431640625,
      "learning_rate": 1.861726437327409e-05,
      "loss": 2.114,
      "step": 21710
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.42578125,
      "learning_rate": 1.860725516999746e-05,
      "loss": 2.1113,
      "step": 21720
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.486328125,
      "learning_rate": 1.8597244261662547e-05,
      "loss": 2.1008,
      "step": 21730
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.76171875,
      "learning_rate": 1.8587231653001252e-05,
      "loss": 2.1187,
      "step": 21740
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4375,
      "learning_rate": 1.857721734874628e-05,
      "loss": 2.1457,
      "step": 21750
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.439453125,
      "learning_rate": 1.8567201353631143e-05,
      "loss": 2.0929,
      "step": 21760
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.431640625,
      "learning_rate": 1.8557183672390142e-05,
      "loss": 2.1049,
      "step": 21770
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.443359375,
      "learning_rate": 1.854716430975839e-05,
      "loss": 2.1545,
      "step": 21780
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.443359375,
      "learning_rate": 1.853714327047178e-05,
      "loss": 2.1205,
      "step": 21790
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.46484375,
      "learning_rate": 1.8527120559267007e-05,
      "loss": 2.1452,
      "step": 21800
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.43359375,
      "learning_rate": 1.8517096180881557e-05,
      "loss": 2.1181,
      "step": 21810
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.458984375,
      "learning_rate": 1.8507070140053688e-05,
      "loss": 2.1106,
      "step": 21820
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.447265625,
      "learning_rate": 1.8497042441522464e-05,
      "loss": 2.109,
      "step": 21830
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.453125,
      "learning_rate": 1.8487013090027727e-05,
      "loss": 2.144,
      "step": 21840
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.427734375,
      "learning_rate": 1.8476982090310094e-05,
      "loss": 2.1149,
      "step": 21850
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.423828125,
      "learning_rate": 1.8466949447110964e-05,
      "loss": 2.1142,
      "step": 21860
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4296875,
      "learning_rate": 1.8456915165172513e-05,
      "loss": 2.1116,
      "step": 21870
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4296875,
      "learning_rate": 1.8446879249237693e-05,
      "loss": 2.1393,
      "step": 21880
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4375,
      "learning_rate": 1.8436841704050237e-05,
      "loss": 2.1317,
      "step": 21890
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.439453125,
      "learning_rate": 1.8426802534354624e-05,
      "loss": 2.0958,
      "step": 21900
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.44140625,
      "learning_rate": 1.8416761744896125e-05,
      "loss": 2.1215,
      "step": 21910
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.455078125,
      "learning_rate": 1.8406719340420766e-05,
      "loss": 2.1058,
      "step": 21920
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4453125,
      "learning_rate": 1.8396675325675333e-05,
      "loss": 2.0909,
      "step": 21930
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.478515625,
      "learning_rate": 1.8386629705407382e-05,
      "loss": 2.0984,
      "step": 21940
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.44140625,
      "learning_rate": 1.8376582484365224e-05,
      "loss": 2.1219,
      "step": 21950
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.443359375,
      "learning_rate": 1.836653366729792e-05,
      "loss": 2.102,
      "step": 21960
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.427734375,
      "learning_rate": 1.8356483258955302e-05,
      "loss": 2.127,
      "step": 21970
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.41796875,
      "learning_rate": 1.8346431264087935e-05,
      "loss": 2.1234,
      "step": 21980
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.419921875,
      "learning_rate": 1.8336377687447146e-05,
      "loss": 2.1243,
      "step": 21990
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4296875,
      "learning_rate": 1.832632253378501e-05,
      "loss": 2.0864,
      "step": 22000
    },
    {
      "epoch": 0.46,
      "eval_accuracy": 0.5587339921305766,
      "eval_loss": 1.994472622871399,
      "eval_runtime": 16.4414,
      "eval_samples_per_second": 36.189,
      "eval_steps_per_second": 1.156,
      "step": 22000
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4296875,
      "learning_rate": 1.831626580785434e-05,
      "loss": 2.1219,
      "step": 22010
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.458984375,
      "learning_rate": 1.830620751440869e-05,
      "loss": 2.0869,
      "step": 22020
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.423828125,
      "learning_rate": 1.8296147658202372e-05,
      "loss": 2.1194,
      "step": 22030
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.416015625,
      "learning_rate": 1.828608624399042e-05,
      "loss": 2.1369,
      "step": 22040
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.443359375,
      "learning_rate": 1.8276023276528612e-05,
      "loss": 2.1218,
      "step": 22050
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.41796875,
      "learning_rate": 1.8265958760573458e-05,
      "loss": 2.1488,
      "step": 22060
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4609375,
      "learning_rate": 1.8255892700882193e-05,
      "loss": 2.1405,
      "step": 22070
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.451171875,
      "learning_rate": 1.8245825102212803e-05,
      "loss": 2.141,
      "step": 22080
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.431640625,
      "learning_rate": 1.823575596932398e-05,
      "loss": 2.1167,
      "step": 22090
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4453125,
      "learning_rate": 1.8225685306975135e-05,
      "loss": 2.1252,
      "step": 22100
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.44140625,
      "learning_rate": 1.821561311992644e-05,
      "loss": 2.1109,
      "step": 22110
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.419921875,
      "learning_rate": 1.8205539412938747e-05,
      "loss": 2.1224,
      "step": 22120
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4453125,
      "learning_rate": 1.8195464190773643e-05,
      "loss": 2.1245,
      "step": 22130
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.462890625,
      "learning_rate": 1.818538745819344e-05,
      "loss": 2.1176,
      "step": 22140
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.46875,
      "learning_rate": 1.8175309219961148e-05,
      "loss": 2.1291,
      "step": 22150
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.44140625,
      "learning_rate": 1.8165229480840495e-05,
      "loss": 2.1285,
      "step": 22160
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.46875,
      "learning_rate": 1.8155148245595925e-05,
      "loss": 2.1231,
      "step": 22170
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.423828125,
      "learning_rate": 1.814506551899257e-05,
      "loss": 2.1233,
      "step": 22180
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.47265625,
      "learning_rate": 1.8134981305796295e-05,
      "loss": 2.1611,
      "step": 22190
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.42578125,
      "learning_rate": 1.8124895610773645e-05,
      "loss": 2.127,
      "step": 22200
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.439453125,
      "learning_rate": 1.811480843869187e-05,
      "loss": 2.1425,
      "step": 22210
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.44921875,
      "learning_rate": 1.8104719794318924e-05,
      "loss": 2.1499,
      "step": 22220
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.470703125,
      "learning_rate": 1.8094629682423455e-05,
      "loss": 2.1263,
      "step": 22230
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.435546875,
      "learning_rate": 1.8084538107774805e-05,
      "loss": 2.1535,
      "step": 22240
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.447265625,
      "learning_rate": 1.8074445075143004e-05,
      "loss": 2.0977,
      "step": 22250
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.419921875,
      "learning_rate": 1.806435058929877e-05,
      "loss": 2.0719,
      "step": 22260
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.43359375,
      "learning_rate": 1.805425465501352e-05,
      "loss": 2.1091,
      "step": 22270
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.421875,
      "learning_rate": 1.8044157277059342e-05,
      "loss": 2.099,
      "step": 22280
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.466796875,
      "learning_rate": 1.8034058460209e-05,
      "loss": 2.081,
      "step": 22290
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.451171875,
      "learning_rate": 1.802395820923597e-05,
      "loss": 2.1037,
      "step": 22300
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.431640625,
      "learning_rate": 1.8013856528914376e-05,
      "loss": 2.1107,
      "step": 22310
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.443359375,
      "learning_rate": 1.8003753424019025e-05,
      "loss": 2.0914,
      "step": 22320
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.435546875,
      "learning_rate": 1.79936488993254e-05,
      "loss": 2.104,
      "step": 22330
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5234375,
      "learning_rate": 1.798354295960965e-05,
      "loss": 2.1512,
      "step": 22340
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4375,
      "learning_rate": 1.7973435609648605e-05,
      "loss": 2.1104,
      "step": 22350
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4375,
      "learning_rate": 1.796332685421975e-05,
      "loss": 2.1172,
      "step": 22360
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.447265625,
      "learning_rate": 1.7953216698101238e-05,
      "loss": 2.0917,
      "step": 22370
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.44140625,
      "learning_rate": 1.7943105146071887e-05,
      "loss": 2.0962,
      "step": 22380
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4609375,
      "learning_rate": 1.7932992202911165e-05,
      "loss": 2.1425,
      "step": 22390
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.42578125,
      "learning_rate": 1.792287787339921e-05,
      "loss": 2.107,
      "step": 22400
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.43359375,
      "learning_rate": 1.7912762162316807e-05,
      "loss": 2.0787,
      "step": 22410
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.435546875,
      "learning_rate": 1.7902645074445404e-05,
      "loss": 2.1219,
      "step": 22420
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4296875,
      "learning_rate": 1.7892526614567082e-05,
      "loss": 2.1216,
      "step": 22430
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.419921875,
      "learning_rate": 1.788240678746459e-05,
      "loss": 2.1209,
      "step": 22440
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.478515625,
      "learning_rate": 1.7872285597921303e-05,
      "loss": 2.1028,
      "step": 22450
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4609375,
      "learning_rate": 1.7862163050721263e-05,
      "loss": 2.1223,
      "step": 22460
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.423828125,
      "learning_rate": 1.7852039150649138e-05,
      "loss": 2.1065,
      "step": 22470
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.451171875,
      "learning_rate": 1.784191390249024e-05,
      "loss": 2.1044,
      "step": 22480
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5,
      "learning_rate": 1.7831787311030514e-05,
      "loss": 2.1274,
      "step": 22490
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.431640625,
      "learning_rate": 1.782165938105655e-05,
      "loss": 2.1074,
      "step": 22500
    },
    {
      "epoch": 0.47,
      "eval_accuracy": 0.5587454923318301,
      "eval_loss": 1.9943138360977173,
      "eval_runtime": 16.4368,
      "eval_samples_per_second": 36.199,
      "eval_steps_per_second": 1.156,
      "step": 22500
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.447265625,
      "learning_rate": 1.7811530117355557e-05,
      "loss": 2.1378,
      "step": 22510
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.435546875,
      "learning_rate": 1.7801399524715385e-05,
      "loss": 2.1277,
      "step": 22520
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4765625,
      "learning_rate": 1.779126760792451e-05,
      "loss": 2.1212,
      "step": 22530
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.439453125,
      "learning_rate": 1.7781134371772035e-05,
      "loss": 2.1201,
      "step": 22540
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.421875,
      "learning_rate": 1.777099982104768e-05,
      "loss": 2.1087,
      "step": 22550
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.439453125,
      "learning_rate": 1.7760863960541796e-05,
      "loss": 2.1147,
      "step": 22560
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.44140625,
      "learning_rate": 1.7750726795045345e-05,
      "loss": 2.1307,
      "step": 22570
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4375,
      "learning_rate": 1.7740588329349905e-05,
      "loss": 2.1069,
      "step": 22580
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4296875,
      "learning_rate": 1.7730448568247672e-05,
      "loss": 2.1233,
      "step": 22590
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.427734375,
      "learning_rate": 1.772030751653146e-05,
      "loss": 2.1263,
      "step": 22600
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4375,
      "learning_rate": 1.7710165178994694e-05,
      "loss": 2.1318,
      "step": 22610
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4453125,
      "learning_rate": 1.770002156043138e-05,
      "loss": 2.0761,
      "step": 22620
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.44140625,
      "learning_rate": 1.7689876665636172e-05,
      "loss": 2.0984,
      "step": 22630
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4453125,
      "learning_rate": 1.7679730499404288e-05,
      "loss": 2.1114,
      "step": 22640
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.443359375,
      "learning_rate": 1.7669583066531576e-05,
      "loss": 2.1403,
      "step": 22650
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.42578125,
      "learning_rate": 1.765943437181447e-05,
      "loss": 2.1127,
      "step": 22660
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.435546875,
      "learning_rate": 1.7649284420049995e-05,
      "loss": 2.1271,
      "step": 22670
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.455078125,
      "learning_rate": 1.7639133216035776e-05,
      "loss": 2.1399,
      "step": 22680
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.439453125,
      "learning_rate": 1.762898076457004e-05,
      "loss": 2.0878,
      "step": 22690
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.41796875,
      "learning_rate": 1.7618827070451578e-05,
      "loss": 2.1376,
      "step": 22700
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.44140625,
      "learning_rate": 1.76086721384798e-05,
      "loss": 2.127,
      "step": 22710
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.416015625,
      "learning_rate": 1.7598515973454677e-05,
      "loss": 2.1008,
      "step": 22720
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.423828125,
      "learning_rate": 1.7588358580176777e-05,
      "loss": 2.1554,
      "step": 22730
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.431640625,
      "learning_rate": 1.757819996344724e-05,
      "loss": 2.1145,
      "step": 22740
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.44140625,
      "learning_rate": 1.756804012806778e-05,
      "loss": 2.1001,
      "step": 22750
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.42578125,
      "learning_rate": 1.755787907884071e-05,
      "loss": 2.115,
      "step": 22760
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.44921875,
      "learning_rate": 1.7547716820568892e-05,
      "loss": 2.1515,
      "step": 22770
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.443359375,
      "learning_rate": 1.7537553358055766e-05,
      "loss": 2.0965,
      "step": 22780
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.421875,
      "learning_rate": 1.7527388696105345e-05,
      "loss": 2.0928,
      "step": 22790
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.453125,
      "learning_rate": 1.751722283952221e-05,
      "loss": 2.1073,
      "step": 22800
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.45703125,
      "learning_rate": 1.7507055793111505e-05,
      "loss": 2.1112,
      "step": 22810
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.421875,
      "learning_rate": 1.7496887561678934e-05,
      "loss": 2.1483,
      "step": 22820
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4296875,
      "learning_rate": 1.748671815003076e-05,
      "loss": 2.1108,
      "step": 22830
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.443359375,
      "learning_rate": 1.7476547562973814e-05,
      "loss": 2.1068,
      "step": 22840
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.59765625,
      "learning_rate": 1.7466375805315468e-05,
      "loss": 2.1319,
      "step": 22850
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.423828125,
      "learning_rate": 1.7456202881863656e-05,
      "loss": 2.1146,
      "step": 22860
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.42578125,
      "learning_rate": 1.7446028797426864e-05,
      "loss": 2.114,
      "step": 22870
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.455078125,
      "learning_rate": 1.7435853556814128e-05,
      "loss": 2.1013,
      "step": 22880
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.45703125,
      "learning_rate": 1.7425677164835012e-05,
      "loss": 2.0942,
      "step": 22890
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.419921875,
      "learning_rate": 1.741549962629966e-05,
      "loss": 2.1451,
      "step": 22900
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.42578125,
      "learning_rate": 1.7405320946018724e-05,
      "loss": 2.1002,
      "step": 22910
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.43359375,
      "learning_rate": 1.7395141128803405e-05,
      "loss": 2.104,
      "step": 22920
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4296875,
      "learning_rate": 1.738496017946546e-05,
      "loss": 2.092,
      "step": 22930
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.42578125,
      "learning_rate": 1.737477810281715e-05,
      "loss": 2.1153,
      "step": 22940
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.443359375,
      "learning_rate": 1.73645949036713e-05,
      "loss": 2.1225,
      "step": 22950
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.427734375,
      "learning_rate": 1.735441058684124e-05,
      "loss": 2.0868,
      "step": 22960
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8046875,
      "learning_rate": 1.7344225157140838e-05,
      "loss": 2.1242,
      "step": 22970
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.45703125,
      "learning_rate": 1.73340386193845e-05,
      "loss": 2.1192,
      "step": 22980
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.427734375,
      "learning_rate": 1.7323850978387136e-05,
      "loss": 2.1318,
      "step": 22990
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.435546875,
      "learning_rate": 1.7313662238964186e-05,
      "loss": 2.0858,
      "step": 23000
    },
    {
      "epoch": 0.48,
      "eval_accuracy": 0.5589574246120735,
      "eval_loss": 1.9941881895065308,
      "eval_runtime": 16.438,
      "eval_samples_per_second": 36.197,
      "eval_steps_per_second": 1.156,
      "step": 23000
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.419921875,
      "learning_rate": 1.7303472405931614e-05,
      "loss": 2.1364,
      "step": 23010
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.419921875,
      "learning_rate": 1.7293281484105893e-05,
      "loss": 2.1271,
      "step": 23020
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.43359375,
      "learning_rate": 1.7283089478304015e-05,
      "loss": 2.108,
      "step": 23030
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.45703125,
      "learning_rate": 1.7272896393343487e-05,
      "loss": 2.1292,
      "step": 23040
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4609375,
      "learning_rate": 1.7262702234042317e-05,
      "loss": 2.0969,
      "step": 23050
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.44140625,
      "learning_rate": 1.725250700521903e-05,
      "loss": 2.1248,
      "step": 23060
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.443359375,
      "learning_rate": 1.7242310711692647e-05,
      "loss": 2.0792,
      "step": 23070
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4453125,
      "learning_rate": 1.7232113358282705e-05,
      "loss": 2.1065,
      "step": 23080
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.43359375,
      "learning_rate": 1.7221914949809232e-05,
      "loss": 2.1182,
      "step": 23090
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.427734375,
      "learning_rate": 1.721171549109276e-05,
      "loss": 2.1064,
      "step": 23100
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.453125,
      "learning_rate": 1.720151498695431e-05,
      "loss": 2.0981,
      "step": 23110
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.427734375,
      "learning_rate": 1.7191313442215405e-05,
      "loss": 2.1122,
      "step": 23120
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.431640625,
      "learning_rate": 1.7181110861698057e-05,
      "loss": 2.1262,
      "step": 23130
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.423828125,
      "learning_rate": 1.7170907250224772e-05,
      "loss": 2.1333,
      "step": 23140
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.421875,
      "learning_rate": 1.7160702612618536e-05,
      "loss": 2.1046,
      "step": 23150
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.421875,
      "learning_rate": 1.715049695370282e-05,
      "loss": 2.1154,
      "step": 23160
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.42578125,
      "learning_rate": 1.7140290278301584e-05,
      "loss": 2.1566,
      "step": 23170
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.431640625,
      "learning_rate": 1.7130082591239265e-05,
      "loss": 2.1008,
      "step": 23180
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4453125,
      "learning_rate": 1.711987389734078e-05,
      "loss": 2.1029,
      "step": 23190
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.431640625,
      "learning_rate": 1.7109664201431517e-05,
      "loss": 2.1031,
      "step": 23200
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.431640625,
      "learning_rate": 1.709945350833734e-05,
      "loss": 2.1214,
      "step": 23210
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.443359375,
      "learning_rate": 1.708924182288459e-05,
      "loss": 2.1069,
      "step": 23220
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.455078125,
      "learning_rate": 1.7079029149900074e-05,
      "loss": 2.0887,
      "step": 23230
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4375,
      "learning_rate": 1.7068815494211056e-05,
      "loss": 2.1217,
      "step": 23240
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4375,
      "learning_rate": 1.7058600860645278e-05,
      "loss": 2.111,
      "step": 23250
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.42578125,
      "learning_rate": 1.7048385254030944e-05,
      "loss": 2.1083,
      "step": 23260
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.51171875,
      "learning_rate": 1.70381686791967e-05,
      "loss": 2.1459,
      "step": 23270
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4921875,
      "learning_rate": 1.7027951140971676e-05,
      "loss": 2.1245,
      "step": 23280
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.431640625,
      "learning_rate": 1.701773264418544e-05,
      "loss": 2.0804,
      "step": 23290
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.45703125,
      "learning_rate": 1.7007513193668003e-05,
      "loss": 2.1357,
      "step": 23300
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.435546875,
      "learning_rate": 1.699729279424986e-05,
      "loss": 2.1199,
      "step": 23310
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.43359375,
      "learning_rate": 1.6987071450761928e-05,
      "loss": 2.1095,
      "step": 23320
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.466796875,
      "learning_rate": 1.6976849168035582e-05,
      "loss": 2.1155,
      "step": 23330
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.435546875,
      "learning_rate": 1.6966625950902625e-05,
      "loss": 2.1025,
      "step": 23340
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4453125,
      "learning_rate": 1.6956401804195322e-05,
      "loss": 2.105,
      "step": 23350
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.47265625,
      "learning_rate": 1.6946176732746368e-05,
      "loss": 2.1119,
      "step": 23360
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.435546875,
      "learning_rate": 1.693595074138889e-05,
      "loss": 2.1131,
      "step": 23370
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.421875,
      "learning_rate": 1.6925723834956462e-05,
      "loss": 2.1305,
      "step": 23380
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.453125,
      "learning_rate": 1.6915496018283084e-05,
      "loss": 2.1165,
      "step": 23390
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.443359375,
      "learning_rate": 1.6905267296203182e-05,
      "loss": 2.1396,
      "step": 23400
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.470703125,
      "learning_rate": 1.6895037673551616e-05,
      "loss": 2.1197,
      "step": 23410
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5703125,
      "learning_rate": 1.6884807155163673e-05,
      "loss": 2.1365,
      "step": 23420
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4296875,
      "learning_rate": 1.6874575745875056e-05,
      "loss": 2.1316,
      "step": 23430
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.421875,
      "learning_rate": 1.6864343450521898e-05,
      "loss": 2.1134,
      "step": 23440
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.431640625,
      "learning_rate": 1.6854110273940737e-05,
      "loss": 2.0898,
      "step": 23450
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.435546875,
      "learning_rate": 1.6843876220968545e-05,
      "loss": 2.1484,
      "step": 23460
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.443359375,
      "learning_rate": 1.68336412964427e-05,
      "loss": 2.0989,
      "step": 23470
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.443359375,
      "learning_rate": 1.682340550520099e-05,
      "loss": 2.1229,
      "step": 23480
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4609375,
      "learning_rate": 1.6813168852081612e-05,
      "loss": 2.1308,
      "step": 23490
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.423828125,
      "learning_rate": 1.680293134192318e-05,
      "loss": 2.1178,
      "step": 23500
    },
    {
      "epoch": 0.49,
      "eval_accuracy": 0.5588112077675645,
      "eval_loss": 1.9940760135650635,
      "eval_runtime": 16.4168,
      "eval_samples_per_second": 36.243,
      "eval_steps_per_second": 1.157,
      "step": 23500
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.41796875,
      "learning_rate": 1.679269297956469e-05,
      "loss": 2.0734,
      "step": 23510
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.44921875,
      "learning_rate": 1.6782453769845578e-05,
      "loss": 2.1513,
      "step": 23520
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.42578125,
      "learning_rate": 1.677221371760565e-05,
      "loss": 2.1038,
      "step": 23530
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.43359375,
      "learning_rate": 1.676197282768512e-05,
      "loss": 2.101,
      "step": 23540
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.416015625,
      "learning_rate": 1.67517311049246e-05,
      "loss": 2.0985,
      "step": 23550
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.435546875,
      "learning_rate": 1.6741488554165085e-05,
      "loss": 2.1438,
      "step": 23560
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4453125,
      "learning_rate": 1.6731245180247977e-05,
      "loss": 2.133,
      "step": 23570
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.443359375,
      "learning_rate": 1.672100098801506e-05,
      "loss": 2.1621,
      "step": 23580
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.447265625,
      "learning_rate": 1.6710755982308503e-05,
      "loss": 2.1258,
      "step": 23590
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4375,
      "learning_rate": 1.670051016797086e-05,
      "loss": 2.1172,
      "step": 23600
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.478515625,
      "learning_rate": 1.669026354984507e-05,
      "loss": 2.1488,
      "step": 23610
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.51171875,
      "learning_rate": 1.6680016132774444e-05,
      "loss": 2.1051,
      "step": 23620
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.43359375,
      "learning_rate": 1.6669767921602686e-05,
      "loss": 2.1017,
      "step": 23630
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.427734375,
      "learning_rate": 1.6659518921173864e-05,
      "loss": 2.114,
      "step": 23640
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.447265625,
      "learning_rate": 1.6649269136332418e-05,
      "loss": 2.1235,
      "step": 23650
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.44921875,
      "learning_rate": 1.6639018571923165e-05,
      "loss": 2.1254,
      "step": 23660
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.439453125,
      "learning_rate": 1.662876723279129e-05,
      "loss": 2.1171,
      "step": 23670
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.421875,
      "learning_rate": 1.6618515123782337e-05,
      "loss": 2.1103,
      "step": 23680
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.46484375,
      "learning_rate": 1.6608262249742214e-05,
      "loss": 2.1078,
      "step": 23690
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.51953125,
      "learning_rate": 1.6598008615517204e-05,
      "loss": 2.1093,
      "step": 23700
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.43359375,
      "learning_rate": 1.6587754225953938e-05,
      "loss": 2.1152,
      "step": 23710
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4375,
      "learning_rate": 1.657749908589941e-05,
      "loss": 2.1565,
      "step": 23720
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4375,
      "learning_rate": 1.656724320020095e-05,
      "loss": 2.1195,
      "step": 23730
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.439453125,
      "learning_rate": 1.655698657370628e-05,
      "loss": 2.1436,
      "step": 23740
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.451171875,
      "learning_rate": 1.654672921126343e-05,
      "loss": 2.0986,
      "step": 23750
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.455078125,
      "learning_rate": 1.65364711177208e-05,
      "loss": 2.1014,
      "step": 23760
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.482421875,
      "learning_rate": 1.6526212297927133e-05,
      "loss": 2.1475,
      "step": 23770
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4140625,
      "learning_rate": 1.651595275673151e-05,
      "loss": 2.1326,
      "step": 23780
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.42578125,
      "learning_rate": 1.6505692498983367e-05,
      "loss": 2.1128,
      "step": 23790
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.453125,
      "learning_rate": 1.649543152953246e-05,
      "loss": 2.1239,
      "step": 23800
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.419921875,
      "learning_rate": 1.6485169853228893e-05,
      "loss": 2.1151,
      "step": 23810
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.421875,
      "learning_rate": 1.6474907474923093e-05,
      "loss": 2.0972,
      "step": 23820
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.43359375,
      "learning_rate": 1.6464644399465832e-05,
      "loss": 2.1181,
      "step": 23830
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.45703125,
      "learning_rate": 1.645438063170821e-05,
      "loss": 2.1397,
      "step": 23840
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.43359375,
      "learning_rate": 1.6444116176501643e-05,
      "loss": 2.0981,
      "step": 23850
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.427734375,
      "learning_rate": 1.643385103869789e-05,
      "loss": 2.1302,
      "step": 23860
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.427734375,
      "learning_rate": 1.642358522314901e-05,
      "loss": 2.1315,
      "step": 23870
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.431640625,
      "learning_rate": 1.64133187347074e-05,
      "loss": 2.1385,
      "step": 23880
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.44921875,
      "learning_rate": 1.6403051578225766e-05,
      "loss": 2.0976,
      "step": 23890
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4375,
      "learning_rate": 1.639278375855714e-05,
      "loss": 2.1141,
      "step": 23900
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.439453125,
      "learning_rate": 1.6382515280554858e-05,
      "loss": 2.1236,
      "step": 23910
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4296875,
      "learning_rate": 1.6372246149072572e-05,
      "loss": 2.086,
      "step": 23920
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.427734375,
      "learning_rate": 1.6361976368964233e-05,
      "loss": 2.1219,
      "step": 23930
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.451171875,
      "learning_rate": 1.635170594508411e-05,
      "loss": 2.1005,
      "step": 23940
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.43359375,
      "learning_rate": 1.634143488228678e-05,
      "loss": 2.1232,
      "step": 23950
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.419921875,
      "learning_rate": 1.6331163185427106e-05,
      "loss": 2.1218,
      "step": 23960
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4453125,
      "learning_rate": 1.6320890859360256e-05,
      "loss": 2.1005,
      "step": 23970
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.42578125,
      "learning_rate": 1.6310617908941707e-05,
      "loss": 2.1096,
      "step": 23980
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.455078125,
      "learning_rate": 1.630034433902722e-05,
      "loss": 2.1064,
      "step": 23990
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.46484375,
      "learning_rate": 1.6290070154472853e-05,
      "loss": 2.1148,
      "step": 24000
    },
    {
      "epoch": 0.5,
      "eval_accuracy": 0.558804636223991,
      "eval_loss": 1.993963360786438,
      "eval_runtime": 16.45,
      "eval_samples_per_second": 36.17,
      "eval_steps_per_second": 1.155,
      "step": 24000
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4453125,
      "learning_rate": 1.627979536013495e-05,
      "loss": 2.0879,
      "step": 24010
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4453125,
      "learning_rate": 1.626951996087015e-05,
      "loss": 2.1112,
      "step": 24020
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.423828125,
      "learning_rate": 1.6259243961535372e-05,
      "loss": 2.1167,
      "step": 24030
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5234375,
      "learning_rate": 1.6248967366987822e-05,
      "loss": 2.1219,
      "step": 24040
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4140625,
      "learning_rate": 1.623869018208499e-05,
      "loss": 2.0934,
      "step": 24050
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.447265625,
      "learning_rate": 1.622841241168463e-05,
      "loss": 2.096,
      "step": 24060
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.43359375,
      "learning_rate": 1.6218134060644794e-05,
      "loss": 2.1385,
      "step": 24070
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.447265625,
      "learning_rate": 1.6207855133823797e-05,
      "loss": 2.0811,
      "step": 24080
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.431640625,
      "learning_rate": 1.6197575636080226e-05,
      "loss": 2.1029,
      "step": 24090
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.419921875,
      "learning_rate": 1.618729557227294e-05,
      "loss": 2.1218,
      "step": 24100
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.439453125,
      "learning_rate": 1.6177014947261068e-05,
      "loss": 2.0949,
      "step": 24110
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.431640625,
      "learning_rate": 1.6166733765904e-05,
      "loss": 2.1215,
      "step": 24120
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.421875,
      "learning_rate": 1.6156452033061386e-05,
      "loss": 2.1416,
      "step": 24130
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.462890625,
      "learning_rate": 1.6146169753593145e-05,
      "loss": 2.1216,
      "step": 24140
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.44140625,
      "learning_rate": 1.6135886932359452e-05,
      "loss": 2.1438,
      "step": 24150
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4375,
      "learning_rate": 1.612560357422073e-05,
      "loss": 2.1182,
      "step": 24160
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.42578125,
      "learning_rate": 1.611531968403767e-05,
      "loss": 2.1294,
      "step": 24170
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.427734375,
      "learning_rate": 1.61050352666712e-05,
      "loss": 2.1341,
      "step": 24180
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.474609375,
      "learning_rate": 1.6094750326982514e-05,
      "loss": 2.0953,
      "step": 24190
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4609375,
      "learning_rate": 1.6084464869833034e-05,
      "loss": 2.1101,
      "step": 24200
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.41796875,
      "learning_rate": 1.607417890008444e-05,
      "loss": 2.1149,
      "step": 24210
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.427734375,
      "learning_rate": 1.6063892422598644e-05,
      "loss": 2.1112,
      "step": 24220
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4296875,
      "learning_rate": 1.6053605442237808e-05,
      "loss": 2.1459,
      "step": 24230
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.60546875,
      "learning_rate": 1.6043317963864326e-05,
      "loss": 2.1287,
      "step": 24240
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.447265625,
      "learning_rate": 1.603302999234083e-05,
      "loss": 2.1058,
      "step": 24250
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.443359375,
      "learning_rate": 1.6022741532530184e-05,
      "loss": 2.1185,
      "step": 24260
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.451171875,
      "learning_rate": 1.601245258929548e-05,
      "loss": 2.041,
      "step": 24270
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4375,
      "learning_rate": 1.6002163167500047e-05,
      "loss": 2.1076,
      "step": 24280
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.47265625,
      "learning_rate": 1.599187327200743e-05,
      "loss": 2.119,
      "step": 24290
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.466796875,
      "learning_rate": 1.59815829076814e-05,
      "loss": 2.1115,
      "step": 24300
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4375,
      "learning_rate": 1.5971292079385963e-05,
      "loss": 2.1172,
      "step": 24310
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.427734375,
      "learning_rate": 1.596100079198532e-05,
      "loss": 2.0953,
      "step": 24320
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.470703125,
      "learning_rate": 1.5950709050343912e-05,
      "loss": 2.1304,
      "step": 24330
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4453125,
      "learning_rate": 1.594041685932638e-05,
      "loss": 2.1117,
      "step": 24340
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.427734375,
      "learning_rate": 1.593012422379758e-05,
      "loss": 2.1197,
      "step": 24350
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.435546875,
      "learning_rate": 1.5919831148622584e-05,
      "loss": 2.1115,
      "step": 24360
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4453125,
      "learning_rate": 1.590953763866667e-05,
      "loss": 2.1122,
      "step": 24370
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4375,
      "learning_rate": 1.5899243698795317e-05,
      "loss": 2.1032,
      "step": 24380
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.462890625,
      "learning_rate": 1.588894933387421e-05,
      "loss": 2.1344,
      "step": 24390
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.447265625,
      "learning_rate": 1.587865454876924e-05,
      "loss": 2.1323,
      "step": 24400
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.45703125,
      "learning_rate": 1.586835934834648e-05,
      "loss": 2.1538,
      "step": 24410
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.453125,
      "learning_rate": 1.5858063737472222e-05,
      "loss": 2.1175,
      "step": 24420
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.44921875,
      "learning_rate": 1.5847767721012938e-05,
      "loss": 2.1313,
      "step": 24430
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.466796875,
      "learning_rate": 1.58374713038353e-05,
      "loss": 2.0902,
      "step": 24440
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.43359375,
      "learning_rate": 1.5827174490806145e-05,
      "loss": 2.1322,
      "step": 24450
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.443359375,
      "learning_rate": 1.5816877286792533e-05,
      "loss": 2.1229,
      "step": 24460
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.427734375,
      "learning_rate": 1.580657969666169e-05,
      "loss": 2.0863,
      "step": 24470
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.447265625,
      "learning_rate": 1.5796281725281026e-05,
      "loss": 2.1434,
      "step": 24480
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.443359375,
      "learning_rate": 1.5785983377518127e-05,
      "loss": 2.1387,
      "step": 24490
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.431640625,
      "learning_rate": 1.5775684658240764e-05,
      "loss": 2.1165,
      "step": 24500
    },
    {
      "epoch": 0.51,
      "eval_accuracy": 0.558840779713645,
      "eval_loss": 1.9938558340072632,
      "eval_runtime": 16.4192,
      "eval_samples_per_second": 36.238,
      "eval_steps_per_second": 1.157,
      "step": 24500
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.462890625,
      "learning_rate": 1.576538557231688e-05,
      "loss": 2.1023,
      "step": 24510
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.453125,
      "learning_rate": 1.5755086124614595e-05,
      "loss": 2.1266,
      "step": 24520
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.419921875,
      "learning_rate": 1.5744786320002197e-05,
      "loss": 2.1351,
      "step": 24530
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.419921875,
      "learning_rate": 1.5734486163348137e-05,
      "loss": 2.1247,
      "step": 24540
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.443359375,
      "learning_rate": 1.5724185659521056e-05,
      "loss": 2.1155,
      "step": 24550
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4609375,
      "learning_rate": 1.5713884813389722e-05,
      "loss": 2.1138,
      "step": 24560
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.421875,
      "learning_rate": 1.570358362982309e-05,
      "loss": 2.1336,
      "step": 24570
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.44140625,
      "learning_rate": 1.5693282113690275e-05,
      "loss": 2.1368,
      "step": 24580
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.439453125,
      "learning_rate": 1.568298026986054e-05,
      "loss": 2.1212,
      "step": 24590
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.451171875,
      "learning_rate": 1.5672678103203297e-05,
      "loss": 2.1171,
      "step": 24600
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.46875,
      "learning_rate": 1.5662375618588137e-05,
      "loss": 2.121,
      "step": 24610
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.482421875,
      "learning_rate": 1.565207282088477e-05,
      "loss": 2.1147,
      "step": 24620
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.435546875,
      "learning_rate": 1.5641769714963074e-05,
      "loss": 2.1134,
      "step": 24630
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4296875,
      "learning_rate": 1.5631466305693064e-05,
      "loss": 2.1189,
      "step": 24640
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.458984375,
      "learning_rate": 1.56211625979449e-05,
      "loss": 2.1255,
      "step": 24650
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.443359375,
      "learning_rate": 1.5610858596588888e-05,
      "loss": 2.0737,
      "step": 24660
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.439453125,
      "learning_rate": 1.5600554306495472e-05,
      "loss": 2.0988,
      "step": 24670
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.423828125,
      "learning_rate": 1.559024973253522e-05,
      "loss": 2.1238,
      "step": 24680
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.451171875,
      "learning_rate": 1.557994487957885e-05,
      "loss": 2.1258,
      "step": 24690
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.439453125,
      "learning_rate": 1.55696397524972e-05,
      "loss": 2.1146,
      "step": 24700
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.431640625,
      "learning_rate": 1.5559334356161242e-05,
      "loss": 2.1377,
      "step": 24710
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.431640625,
      "learning_rate": 1.554902869544209e-05,
      "loss": 2.1513,
      "step": 24720
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.455078125,
      "learning_rate": 1.5538722775210955e-05,
      "loss": 2.0984,
      "step": 24730
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.45703125,
      "learning_rate": 1.5528416600339188e-05,
      "loss": 2.1274,
      "step": 24740
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.439453125,
      "learning_rate": 1.551811017569826e-05,
      "loss": 2.1001,
      "step": 24750
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.443359375,
      "learning_rate": 1.5507803506159753e-05,
      "loss": 2.1188,
      "step": 24760
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.431640625,
      "learning_rate": 1.5497496596595375e-05,
      "loss": 2.1419,
      "step": 24770
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.69921875,
      "learning_rate": 1.548718945187694e-05,
      "loss": 2.0979,
      "step": 24780
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.443359375,
      "learning_rate": 1.5476882076876366e-05,
      "loss": 2.1066,
      "step": 24790
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.447265625,
      "learning_rate": 1.54665744764657e-05,
      "loss": 2.1154,
      "step": 24800
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.431640625,
      "learning_rate": 1.545626665551708e-05,
      "loss": 2.1404,
      "step": 24810
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.431640625,
      "learning_rate": 1.5445958618902752e-05,
      "loss": 2.1155,
      "step": 24820
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.44921875,
      "learning_rate": 1.5435650371495066e-05,
      "loss": 2.1252,
      "step": 24830
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.439453125,
      "learning_rate": 1.542534191816646e-05,
      "loss": 2.1651,
      "step": 24840
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4375,
      "learning_rate": 1.5415033263789493e-05,
      "loss": 2.1006,
      "step": 24850
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.462890625,
      "learning_rate": 1.5404724413236798e-05,
      "loss": 2.1414,
      "step": 24860
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.44140625,
      "learning_rate": 1.5394415371381105e-05,
      "loss": 2.1228,
      "step": 24870
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4296875,
      "learning_rate": 1.538410614309525e-05,
      "loss": 2.1602,
      "step": 24880
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.45703125,
      "learning_rate": 1.537379673325213e-05,
      "loss": 2.1302,
      "step": 24890
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.44140625,
      "learning_rate": 1.5363487146724742e-05,
      "loss": 2.1002,
      "step": 24900
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.44140625,
      "learning_rate": 1.5353177388386183e-05,
      "loss": 2.116,
      "step": 24910
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.43359375,
      "learning_rate": 1.5342867463109597e-05,
      "loss": 2.0986,
      "step": 24920
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.41796875,
      "learning_rate": 1.533255737576824e-05,
      "loss": 2.1267,
      "step": 24930
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4375,
      "learning_rate": 1.532224713123542e-05,
      "loss": 2.0892,
      "step": 24940
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.423828125,
      "learning_rate": 1.531193673438453e-05,
      "loss": 2.1235,
      "step": 24950
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.439453125,
      "learning_rate": 1.5301626190089037e-05,
      "loss": 2.0825,
      "step": 24960
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4296875,
      "learning_rate": 1.5291315503222472e-05,
      "loss": 2.1363,
      "step": 24970
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.439453125,
      "learning_rate": 1.5281004678658436e-05,
      "loss": 2.1062,
      "step": 24980
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.44921875,
      "learning_rate": 1.5270693721270595e-05,
      "loss": 2.1315,
      "step": 24990
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.439453125,
      "learning_rate": 1.526038263593268e-05,
      "loss": 2.1012,
      "step": 25000
    },
    {
      "epoch": 0.52,
      "eval_accuracy": 0.5589722105851138,
      "eval_loss": 1.9938210248947144,
      "eval_runtime": 16.4402,
      "eval_samples_per_second": 36.192,
      "eval_steps_per_second": 1.156,
      "step": 25000
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.55078125,
      "learning_rate": 1.5250071427518476e-05,
      "loss": 2.1071,
      "step": 25010
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.43359375,
      "learning_rate": 1.5239760100901834e-05,
      "loss": 2.1583,
      "step": 25020
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.447265625,
      "learning_rate": 1.5229448660956653e-05,
      "loss": 2.1198,
      "step": 25030
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.451171875,
      "learning_rate": 1.5219137112556898e-05,
      "loss": 2.1065,
      "step": 25040
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.451171875,
      "learning_rate": 1.520882546057657e-05,
      "loss": 2.1428,
      "step": 25050
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.451171875,
      "learning_rate": 1.5198513709889732e-05,
      "loss": 2.138,
      "step": 25060
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.44921875,
      "learning_rate": 1.518820186537048e-05,
      "loss": 2.1452,
      "step": 25070
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.44140625,
      "learning_rate": 1.517788993189297e-05,
      "loss": 2.1486,
      "step": 25080
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.43359375,
      "learning_rate": 1.5167577914331383e-05,
      "loss": 2.1261,
      "step": 25090
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.421875,
      "learning_rate": 1.5157265817559957e-05,
      "loss": 2.0831,
      "step": 25100
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4375,
      "learning_rate": 1.5146953646452954e-05,
      "loss": 2.1612,
      "step": 25110
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.439453125,
      "learning_rate": 1.513664140588468e-05,
      "loss": 2.0915,
      "step": 25120
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.453125,
      "learning_rate": 1.5126329100729472e-05,
      "loss": 2.0916,
      "step": 25130
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.47265625,
      "learning_rate": 1.511601673586169e-05,
      "loss": 2.1231,
      "step": 25140
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.431640625,
      "learning_rate": 1.5105704316155733e-05,
      "loss": 2.1131,
      "step": 25150
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.453125,
      "learning_rate": 1.5095391846486018e-05,
      "loss": 2.1316,
      "step": 25160
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.419921875,
      "learning_rate": 1.5085079331726993e-05,
      "loss": 2.1119,
      "step": 25170
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.453125,
      "learning_rate": 1.5074766776753118e-05,
      "loss": 2.1289,
      "step": 25180
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.42578125,
      "learning_rate": 1.5064454186438876e-05,
      "loss": 2.105,
      "step": 25190
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4375,
      "learning_rate": 1.5054141565658773e-05,
      "loss": 2.1189,
      "step": 25200
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.44921875,
      "learning_rate": 1.5043828919287322e-05,
      "loss": 2.1055,
      "step": 25210
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.453125,
      "learning_rate": 1.5033516252199047e-05,
      "loss": 2.1345,
      "step": 25220
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.48828125,
      "learning_rate": 1.5023203569268486e-05,
      "loss": 2.1026,
      "step": 25230
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.41015625,
      "learning_rate": 1.501289087537019e-05,
      "loss": 2.1427,
      "step": 25240
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.462890625,
      "learning_rate": 1.5002578175378698e-05,
      "loss": 2.149,
      "step": 25250
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.44921875,
      "learning_rate": 1.4992265474168566e-05,
      "loss": 2.1098,
      "step": 25260
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4296875,
      "learning_rate": 1.4981952776614357e-05,
      "loss": 2.0874,
      "step": 25270
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.423828125,
      "learning_rate": 1.4971640087590608e-05,
      "loss": 2.1357,
      "step": 25280
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.44921875,
      "learning_rate": 1.4961327411971871e-05,
      "loss": 2.1393,
      "step": 25290
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.43359375,
      "learning_rate": 1.4951014754632694e-05,
      "loss": 2.0795,
      "step": 25300
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.44921875,
      "learning_rate": 1.4940702120447598e-05,
      "loss": 2.1107,
      "step": 25310
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.43359375,
      "learning_rate": 1.4930389514291114e-05,
      "loss": 2.1119,
      "step": 25320
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.43359375,
      "learning_rate": 1.4920076941037747e-05,
      "loss": 2.1201,
      "step": 25330
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.447265625,
      "learning_rate": 1.4909764405561986e-05,
      "loss": 2.1366,
      "step": 25340
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.431640625,
      "learning_rate": 1.4899451912738307e-05,
      "loss": 2.0952,
      "step": 25350
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.453125,
      "learning_rate": 1.4889139467441172e-05,
      "loss": 2.1262,
      "step": 25360
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.435546875,
      "learning_rate": 1.4878827074545e-05,
      "loss": 2.1358,
      "step": 25370
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.435546875,
      "learning_rate": 1.4868514738924204e-05,
      "loss": 2.1332,
      "step": 25380
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4296875,
      "learning_rate": 1.4858202465453175e-05,
      "loss": 2.1228,
      "step": 25390
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.447265625,
      "learning_rate": 1.4847890259006244e-05,
      "loss": 2.1142,
      "step": 25400
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.41796875,
      "learning_rate": 1.4837578124457742e-05,
      "loss": 2.137,
      "step": 25410
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.453125,
      "learning_rate": 1.482726606668196e-05,
      "loss": 2.1394,
      "step": 25420
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.431640625,
      "learning_rate": 1.4816954090553133e-05,
      "loss": 2.0968,
      "step": 25430
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.435546875,
      "learning_rate": 1.480664220094548e-05,
      "loss": 2.1152,
      "step": 25440
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4296875,
      "learning_rate": 1.4796330402733166e-05,
      "loss": 2.1262,
      "step": 25450
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4375,
      "learning_rate": 1.4786018700790325e-05,
      "loss": 2.0961,
      "step": 25460
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0625,
      "learning_rate": 1.4775707099991027e-05,
      "loss": 2.1108,
      "step": 25470
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4296875,
      "learning_rate": 1.4765395605209309e-05,
      "loss": 2.1273,
      "step": 25480
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4453125,
      "learning_rate": 1.4755084221319164e-05,
      "loss": 2.1276,
      "step": 25490
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.42578125,
      "learning_rate": 1.4744772953194504e-05,
      "loss": 2.1573,
      "step": 25500
    },
    {
      "epoch": 0.53,
      "eval_accuracy": 0.5589853536722607,
      "eval_loss": 1.9936455488204956,
      "eval_runtime": 16.4618,
      "eval_samples_per_second": 36.144,
      "eval_steps_per_second": 1.154,
      "step": 25500
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.439453125,
      "learning_rate": 1.4734461805709214e-05,
      "loss": 2.1088,
      "step": 25510
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.443359375,
      "learning_rate": 1.472415078373712e-05,
      "loss": 2.1284,
      "step": 25520
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.42578125,
      "learning_rate": 1.4713839892151968e-05,
      "loss": 2.115,
      "step": 25530
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4296875,
      "learning_rate": 1.4703529135827459e-05,
      "loss": 2.1055,
      "step": 25540
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.462890625,
      "learning_rate": 1.4693218519637238e-05,
      "loss": 2.1428,
      "step": 25550
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4296875,
      "learning_rate": 1.4682908048454855e-05,
      "loss": 2.1101,
      "step": 25560
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.46875,
      "learning_rate": 1.4672597727153824e-05,
      "loss": 2.1126,
      "step": 25570
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.50390625,
      "learning_rate": 1.4662287560607565e-05,
      "loss": 2.1322,
      "step": 25580
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.42578125,
      "learning_rate": 1.4651977553689436e-05,
      "loss": 2.1119,
      "step": 25590
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.443359375,
      "learning_rate": 1.4641667711272711e-05,
      "loss": 2.1098,
      "step": 25600
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.482421875,
      "learning_rate": 1.4631358038230602e-05,
      "loss": 2.1481,
      "step": 25610
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.431640625,
      "learning_rate": 1.4621048539436221e-05,
      "loss": 2.1433,
      "step": 25620
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.427734375,
      "learning_rate": 1.4610739219762609e-05,
      "loss": 2.1147,
      "step": 25630
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.435546875,
      "learning_rate": 1.4600430084082722e-05,
      "loss": 2.1207,
      "step": 25640
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.455078125,
      "learning_rate": 1.4590121137269436e-05,
      "loss": 2.0961,
      "step": 25650
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5703125,
      "learning_rate": 1.4579812384195515e-05,
      "loss": 2.1353,
      "step": 25660
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.470703125,
      "learning_rate": 1.4569503829733658e-05,
      "loss": 2.1375,
      "step": 25670
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4375,
      "learning_rate": 1.4559195478756451e-05,
      "loss": 2.1226,
      "step": 25680
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.447265625,
      "learning_rate": 1.4548887336136394e-05,
      "loss": 2.1184,
      "step": 25690
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.44140625,
      "learning_rate": 1.4538579406745878e-05,
      "loss": 2.0869,
      "step": 25700
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4765625,
      "learning_rate": 1.4528271695457215e-05,
      "loss": 2.1118,
      "step": 25710
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.41796875,
      "learning_rate": 1.4517964207142584e-05,
      "loss": 2.0815,
      "step": 25720
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4609375,
      "learning_rate": 1.450765694667408e-05,
      "loss": 2.1251,
      "step": 25730
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.419921875,
      "learning_rate": 1.4497349918923693e-05,
      "loss": 2.1044,
      "step": 25740
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.416015625,
      "learning_rate": 1.4487043128763278e-05,
      "loss": 2.1204,
      "step": 25750
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.421875,
      "learning_rate": 1.4476736581064603e-05,
      "loss": 2.1083,
      "step": 25760
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.447265625,
      "learning_rate": 1.446643028069932e-05,
      "loss": 2.1267,
      "step": 25770
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.44921875,
      "learning_rate": 1.4456124232538941e-05,
      "loss": 2.0993,
      "step": 25780
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.44921875,
      "learning_rate": 1.4445818441454882e-05,
      "loss": 2.1483,
      "step": 25790
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.453125,
      "learning_rate": 1.4435512912318436e-05,
      "loss": 2.1093,
      "step": 25800
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.44140625,
      "learning_rate": 1.4425207650000752e-05,
      "loss": 2.1248,
      "step": 25810
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.435546875,
      "learning_rate": 1.441490265937288e-05,
      "loss": 2.1181,
      "step": 25820
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.447265625,
      "learning_rate": 1.440459794530572e-05,
      "loss": 2.0949,
      "step": 25830
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.431640625,
      "learning_rate": 1.4394293512670055e-05,
      "loss": 2.1193,
      "step": 25840
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.458984375,
      "learning_rate": 1.4383989366336523e-05,
      "loss": 2.1579,
      "step": 25850
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4296875,
      "learning_rate": 1.4373685511175639e-05,
      "loss": 2.1036,
      "step": 25860
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.478515625,
      "learning_rate": 1.4363381952057779e-05,
      "loss": 2.1403,
      "step": 25870
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4296875,
      "learning_rate": 1.4353078693853162e-05,
      "loss": 2.129,
      "step": 25880
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.427734375,
      "learning_rate": 1.4342775741431885e-05,
      "loss": 2.0931,
      "step": 25890
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.443359375,
      "learning_rate": 1.43324730996639e-05,
      "loss": 2.1306,
      "step": 25900
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.447265625,
      "learning_rate": 1.432217077341899e-05,
      "loss": 2.1432,
      "step": 25910
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.42578125,
      "learning_rate": 1.4311868767566813e-05,
      "loss": 2.1181,
      "step": 25920
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.66015625,
      "learning_rate": 1.4301567086976874e-05,
      "loss": 2.1151,
      "step": 25930
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.435546875,
      "learning_rate": 1.4291265736518498e-05,
      "loss": 2.1105,
      "step": 25940
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4375,
      "learning_rate": 1.428096472106089e-05,
      "loss": 2.1074,
      "step": 25950
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4375,
      "learning_rate": 1.427066404547307e-05,
      "loss": 2.1439,
      "step": 25960
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.447265625,
      "learning_rate": 1.4260363714623912e-05,
      "loss": 2.1512,
      "step": 25970
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.44140625,
      "learning_rate": 1.4250063733382115e-05,
      "loss": 2.1296,
      "step": 25980
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.458984375,
      "learning_rate": 1.423976410661623e-05,
      "loss": 2.1344,
      "step": 25990
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.435546875,
      "learning_rate": 1.422946483919462e-05,
      "loss": 2.1674,
      "step": 26000
    },
    {
      "epoch": 0.54,
      "eval_accuracy": 0.5589360670954598,
      "eval_loss": 1.9935729503631592,
      "eval_runtime": 16.4394,
      "eval_samples_per_second": 36.194,
      "eval_steps_per_second": 1.156,
      "step": 26000
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.443359375,
      "learning_rate": 1.4219165935985491e-05,
      "loss": 2.1371,
      "step": 26010
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.466796875,
      "learning_rate": 1.4208867401856876e-05,
      "loss": 2.15,
      "step": 26020
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.47265625,
      "learning_rate": 1.4198569241676638e-05,
      "loss": 2.1126,
      "step": 26030
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4453125,
      "learning_rate": 1.4188271460312442e-05,
      "loss": 2.1363,
      "step": 26040
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.455078125,
      "learning_rate": 1.41779740626318e-05,
      "loss": 2.1472,
      "step": 26050
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4609375,
      "learning_rate": 1.416767705350203e-05,
      "loss": 2.1341,
      "step": 26060
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.46875,
      "learning_rate": 1.4157380437790266e-05,
      "loss": 2.1106,
      "step": 26070
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.51953125,
      "learning_rate": 1.4147084220363453e-05,
      "loss": 2.1236,
      "step": 26080
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.43359375,
      "learning_rate": 1.4136788406088365e-05,
      "loss": 2.1035,
      "step": 26090
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.47265625,
      "learning_rate": 1.412649299983156e-05,
      "loss": 2.1264,
      "step": 26100
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.466796875,
      "learning_rate": 1.4116198006459417e-05,
      "loss": 2.1115,
      "step": 26110
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.41796875,
      "learning_rate": 1.4105903430838135e-05,
      "loss": 2.0964,
      "step": 26120
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.42578125,
      "learning_rate": 1.409560927783368e-05,
      "loss": 2.0703,
      "step": 26130
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4375,
      "learning_rate": 1.4085315552311846e-05,
      "loss": 2.0989,
      "step": 26140
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.44140625,
      "learning_rate": 1.4075022259138223e-05,
      "loss": 2.1273,
      "step": 26150
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4375,
      "learning_rate": 1.4064729403178178e-05,
      "loss": 2.1313,
      "step": 26160
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4609375,
      "learning_rate": 1.4054436989296888e-05,
      "loss": 2.1341,
      "step": 26170
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.478515625,
      "learning_rate": 1.4044145022359325e-05,
      "loss": 2.1349,
      "step": 26180
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.458984375,
      "learning_rate": 1.403385350723023e-05,
      "loss": 2.1089,
      "step": 26190
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.423828125,
      "learning_rate": 1.402356244877415e-05,
      "loss": 2.0889,
      "step": 26200
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.427734375,
      "learning_rate": 1.40132718518554e-05,
      "loss": 2.1778,
      "step": 26210
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.43359375,
      "learning_rate": 1.4002981721338097e-05,
      "loss": 2.1216,
      "step": 26220
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.44140625,
      "learning_rate": 1.3992692062086111e-05,
      "loss": 2.0964,
      "step": 26230
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4453125,
      "learning_rate": 1.3982402878963114e-05,
      "loss": 2.1148,
      "step": 26240
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.42578125,
      "learning_rate": 1.3972114176832542e-05,
      "loss": 2.1037,
      "step": 26250
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.447265625,
      "learning_rate": 1.3961825960557598e-05,
      "loss": 2.1185,
      "step": 26260
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4296875,
      "learning_rate": 1.3951538235001262e-05,
      "loss": 2.1135,
      "step": 26270
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.43359375,
      "learning_rate": 1.3941251005026294e-05,
      "loss": 2.1159,
      "step": 26280
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.46484375,
      "learning_rate": 1.393096427549519e-05,
      "loss": 2.1127,
      "step": 26290
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4140625,
      "learning_rate": 1.3920678051270239e-05,
      "loss": 2.1071,
      "step": 26300
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4296875,
      "learning_rate": 1.3910392337213479e-05,
      "loss": 2.1105,
      "step": 26310
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.42578125,
      "learning_rate": 1.3900107138186697e-05,
      "loss": 2.0986,
      "step": 26320
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.427734375,
      "learning_rate": 1.3889822459051456e-05,
      "loss": 2.0751,
      "step": 26330
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.439453125,
      "learning_rate": 1.3879538304669063e-05,
      "loss": 2.0935,
      "step": 26340
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4453125,
      "learning_rate": 1.3869254679900572e-05,
      "loss": 2.1099,
      "step": 26350
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4765625,
      "learning_rate": 1.3858971589606793e-05,
      "loss": 2.1489,
      "step": 26360
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.453125,
      "learning_rate": 1.3848689038648292e-05,
      "loss": 2.1276,
      "step": 26370
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.470703125,
      "learning_rate": 1.3838407031885358e-05,
      "loss": 2.0907,
      "step": 26380
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.42578125,
      "learning_rate": 1.382812557417804e-05,
      "loss": 2.1162,
      "step": 26390
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.439453125,
      "learning_rate": 1.3817844670386126e-05,
      "loss": 2.1265,
      "step": 26400
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.44921875,
      "learning_rate": 1.3807564325369143e-05,
      "loss": 2.1161,
      "step": 26410
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.43359375,
      "learning_rate": 1.379728454398634e-05,
      "loss": 2.133,
      "step": 26420
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.43359375,
      "learning_rate": 1.3787005331096716e-05,
      "loss": 2.0751,
      "step": 26430
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.423828125,
      "learning_rate": 1.3776726691558995e-05,
      "loss": 2.1407,
      "step": 26440
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.431640625,
      "learning_rate": 1.3766448630231628e-05,
      "loss": 2.1342,
      "step": 26450
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.41796875,
      "learning_rate": 1.375617115197279e-05,
      "loss": 2.0845,
      "step": 26460
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.453125,
      "learning_rate": 1.3745894261640397e-05,
      "loss": 2.0888,
      "step": 26470
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.443359375,
      "learning_rate": 1.3735617964092063e-05,
      "loss": 2.1236,
      "step": 26480
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.43359375,
      "learning_rate": 1.372534226418514e-05,
      "loss": 2.0925,
      "step": 26490
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4296875,
      "learning_rate": 1.3715067166776694e-05,
      "loss": 2.1184,
      "step": 26500
    },
    {
      "epoch": 0.55,
      "eval_accuracy": 0.559028068705488,
      "eval_loss": 1.9934953451156616,
      "eval_runtime": 15.4554,
      "eval_samples_per_second": 38.498,
      "eval_steps_per_second": 1.229,
      "step": 26500
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.44140625,
      "learning_rate": 1.3704792676723496e-05,
      "loss": 2.1057,
      "step": 26510
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.439453125,
      "learning_rate": 1.3694518798882041e-05,
      "loss": 2.1311,
      "step": 26520
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.439453125,
      "learning_rate": 1.3684245538108539e-05,
      "loss": 2.1151,
      "step": 26530
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5,
      "learning_rate": 1.3673972899258886e-05,
      "loss": 2.13,
      "step": 26540
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.427734375,
      "learning_rate": 1.3663700887188708e-05,
      "loss": 2.1299,
      "step": 26550
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.435546875,
      "learning_rate": 1.3653429506753329e-05,
      "loss": 2.1389,
      "step": 26560
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.44921875,
      "learning_rate": 1.3643158762807759e-05,
      "loss": 2.1371,
      "step": 26570
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.421875,
      "learning_rate": 1.3632888660206729e-05,
      "loss": 2.1261,
      "step": 26580
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.63671875,
      "learning_rate": 1.3622619203804651e-05,
      "loss": 2.1256,
      "step": 26590
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.44140625,
      "learning_rate": 1.361235039845564e-05,
      "loss": 2.1222,
      "step": 26600
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.439453125,
      "learning_rate": 1.3602082249013498e-05,
      "loss": 2.1045,
      "step": 26610
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4375,
      "learning_rate": 1.359181476033172e-05,
      "loss": 2.1442,
      "step": 26620
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.443359375,
      "learning_rate": 1.3581547937263497e-05,
      "loss": 2.1355,
      "step": 26630
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4609375,
      "learning_rate": 1.3571281784661683e-05,
      "loss": 2.1196,
      "step": 26640
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4375,
      "learning_rate": 1.356101630737883e-05,
      "loss": 2.1325,
      "step": 26650
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.451171875,
      "learning_rate": 1.3550751510267186e-05,
      "loss": 2.1362,
      "step": 26660
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4296875,
      "learning_rate": 1.3540487398178641e-05,
      "loss": 2.1411,
      "step": 26670
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.439453125,
      "learning_rate": 1.3530223975964793e-05,
      "loss": 2.1057,
      "step": 26680
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.427734375,
      "learning_rate": 1.35199612484769e-05,
      "loss": 2.1329,
      "step": 26690
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.44921875,
      "learning_rate": 1.350969922056589e-05,
      "loss": 2.1321,
      "step": 26700
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.435546875,
      "learning_rate": 1.3499437897082363e-05,
      "loss": 2.1413,
      "step": 26710
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.44140625,
      "learning_rate": 1.3489177282876599e-05,
      "loss": 2.0994,
      "step": 26720
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.423828125,
      "learning_rate": 1.3478917382798514e-05,
      "loss": 2.0977,
      "step": 26730
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4453125,
      "learning_rate": 1.3468658201697715e-05,
      "loss": 2.1497,
      "step": 26740
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.462890625,
      "learning_rate": 1.3458399744423456e-05,
      "loss": 2.094,
      "step": 26750
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.431640625,
      "learning_rate": 1.3448142015824645e-05,
      "loss": 2.1285,
      "step": 26760
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4296875,
      "learning_rate": 1.3437885020749852e-05,
      "loss": 2.1076,
      "step": 26770
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.43359375,
      "learning_rate": 1.3427628764047308e-05,
      "loss": 2.1319,
      "step": 26780
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.416015625,
      "learning_rate": 1.3417373250564874e-05,
      "loss": 2.1056,
      "step": 26790
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.453125,
      "learning_rate": 1.3407118485150077e-05,
      "loss": 2.1156,
      "step": 26800
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.423828125,
      "learning_rate": 1.3396864472650087e-05,
      "loss": 2.1381,
      "step": 26810
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.41015625,
      "learning_rate": 1.3386611217911718e-05,
      "loss": 2.111,
      "step": 26820
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.443359375,
      "learning_rate": 1.3376358725781419e-05,
      "loss": 2.0966,
      "step": 26830
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.423828125,
      "learning_rate": 1.3366107001105283e-05,
      "loss": 2.1492,
      "step": 26840
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.443359375,
      "learning_rate": 1.3355856048729052e-05,
      "loss": 2.1115,
      "step": 26850
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.46875,
      "learning_rate": 1.3345605873498075e-05,
      "loss": 2.1053,
      "step": 26860
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4609375,
      "learning_rate": 1.333535648025736e-05,
      "loss": 2.0999,
      "step": 26870
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.42578125,
      "learning_rate": 1.3325107873851541e-05,
      "loss": 2.0953,
      "step": 26880
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.42578125,
      "learning_rate": 1.3314860059124866e-05,
      "loss": 2.1491,
      "step": 26890
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.46484375,
      "learning_rate": 1.3304613040921217e-05,
      "loss": 2.0849,
      "step": 26900
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.451171875,
      "learning_rate": 1.3294366824084114e-05,
      "loss": 2.1227,
      "step": 26910
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.439453125,
      "learning_rate": 1.3284121413456668e-05,
      "loss": 2.1092,
      "step": 26920
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.458984375,
      "learning_rate": 1.327387681388163e-05,
      "loss": 2.1102,
      "step": 26930
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4296875,
      "learning_rate": 1.3263633030201376e-05,
      "loss": 2.1417,
      "step": 26940
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.68359375,
      "learning_rate": 1.3253390067257865e-05,
      "loss": 2.1315,
      "step": 26950
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.44921875,
      "learning_rate": 1.3243147929892697e-05,
      "loss": 2.1116,
      "step": 26960
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.42578125,
      "learning_rate": 1.323290662294707e-05,
      "loss": 2.0774,
      "step": 26970
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.443359375,
      "learning_rate": 1.322266615126179e-05,
      "loss": 2.1073,
      "step": 26980
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4453125,
      "learning_rate": 1.3212426519677265e-05,
      "loss": 2.1148,
      "step": 26990
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.462890625,
      "learning_rate": 1.3202187733033512e-05,
      "loss": 2.1424,
      "step": 27000
    },
    {
      "epoch": 0.56,
      "eval_accuracy": 0.5590198542760213,
      "eval_loss": 1.9934771060943604,
      "eval_runtime": 16.4588,
      "eval_samples_per_second": 36.151,
      "eval_steps_per_second": 1.154,
      "step": 27000
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4453125,
      "learning_rate": 1.3191949796170156e-05,
      "loss": 2.0874,
      "step": 27010
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.451171875,
      "learning_rate": 1.3181712713926398e-05,
      "loss": 2.1289,
      "step": 27020
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.431640625,
      "learning_rate": 1.317147649114105e-05,
      "loss": 2.1437,
      "step": 27030
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.443359375,
      "learning_rate": 1.3161241132652526e-05,
      "loss": 2.142,
      "step": 27040
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4375,
      "learning_rate": 1.3151006643298807e-05,
      "loss": 2.1415,
      "step": 27050
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.47265625,
      "learning_rate": 1.3140773027917488e-05,
      "loss": 2.0835,
      "step": 27060
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.439453125,
      "learning_rate": 1.3130540291345734e-05,
      "loss": 2.1002,
      "step": 27070
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.443359375,
      "learning_rate": 1.3120308438420304e-05,
      "loss": 2.1329,
      "step": 27080
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4609375,
      "learning_rate": 1.311007747397753e-05,
      "loss": 2.123,
      "step": 27090
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.44921875,
      "learning_rate": 1.3099847402853345e-05,
      "loss": 2.1045,
      "step": 27100
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4453125,
      "learning_rate": 1.3089618229883224e-05,
      "loss": 2.1065,
      "step": 27110
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.44140625,
      "learning_rate": 1.3079389959902251e-05,
      "loss": 2.1408,
      "step": 27120
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.421875,
      "learning_rate": 1.3069162597745074e-05,
      "loss": 2.0833,
      "step": 27130
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.419921875,
      "learning_rate": 1.3058936148245899e-05,
      "loss": 2.1175,
      "step": 27140
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.48046875,
      "learning_rate": 1.3048710616238512e-05,
      "loss": 2.1104,
      "step": 27150
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4453125,
      "learning_rate": 1.3038486006556273e-05,
      "loss": 2.115,
      "step": 27160
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.435546875,
      "learning_rate": 1.3028262324032083e-05,
      "loss": 2.1183,
      "step": 27170
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.431640625,
      "learning_rate": 1.3018039573498422e-05,
      "loss": 2.0974,
      "step": 27180
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.419921875,
      "learning_rate": 1.3007817759787333e-05,
      "loss": 2.1345,
      "step": 27190
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.47265625,
      "learning_rate": 1.2997596887730402e-05,
      "loss": 2.0943,
      "step": 27200
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.443359375,
      "learning_rate": 1.2987376962158778e-05,
      "loss": 2.1167,
      "step": 27210
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4296875,
      "learning_rate": 1.297715798790316e-05,
      "loss": 2.1245,
      "step": 27220
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.470703125,
      "learning_rate": 1.2966939969793804e-05,
      "loss": 2.1178,
      "step": 27230
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.455078125,
      "learning_rate": 1.29567229126605e-05,
      "loss": 2.1221,
      "step": 27240
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.462890625,
      "learning_rate": 1.2946506821332599e-05,
      "loss": 2.1227,
      "step": 27250
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.451171875,
      "learning_rate": 1.2936291700638993e-05,
      "loss": 2.1038,
      "step": 27260
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4375,
      "learning_rate": 1.2926077555408098e-05,
      "loss": 2.1094,
      "step": 27270
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.44921875,
      "learning_rate": 1.291586439046789e-05,
      "loss": 2.126,
      "step": 27280
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.427734375,
      "learning_rate": 1.2905652210645882e-05,
      "loss": 2.1133,
      "step": 27290
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.421875,
      "learning_rate": 1.2895441020769096e-05,
      "loss": 2.1307,
      "step": 27300
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.443359375,
      "learning_rate": 1.2885230825664112e-05,
      "loss": 2.1296,
      "step": 27310
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.431640625,
      "learning_rate": 1.2875021630157038e-05,
      "loss": 2.1107,
      "step": 27320
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.421875,
      "learning_rate": 1.2864813439073489e-05,
      "loss": 2.105,
      "step": 27330
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.423828125,
      "learning_rate": 1.2854606257238628e-05,
      "loss": 2.1143,
      "step": 27340
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.419921875,
      "learning_rate": 1.2844400089477129e-05,
      "loss": 2.1339,
      "step": 27350
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.61328125,
      "learning_rate": 1.2834194940613188e-05,
      "loss": 2.1097,
      "step": 27360
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.439453125,
      "learning_rate": 1.282399081547052e-05,
      "loss": 2.081,
      "step": 27370
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.44140625,
      "learning_rate": 1.2813787718872358e-05,
      "loss": 2.1473,
      "step": 27380
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.435546875,
      "learning_rate": 1.2803585655641456e-05,
      "loss": 2.1149,
      "step": 27390
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.42578125,
      "learning_rate": 1.2793384630600056e-05,
      "loss": 2.1518,
      "step": 27400
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.435546875,
      "learning_rate": 1.2783184648569932e-05,
      "loss": 2.0944,
      "step": 27410
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4609375,
      "learning_rate": 1.2772985714372365e-05,
      "loss": 2.0976,
      "step": 27420
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.453125,
      "learning_rate": 1.276278783282812e-05,
      "loss": 2.1313,
      "step": 27430
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4296875,
      "learning_rate": 1.2752591008757486e-05,
      "loss": 2.1634,
      "step": 27440
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.455078125,
      "learning_rate": 1.2742395246980243e-05,
      "loss": 2.1583,
      "step": 27450
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.435546875,
      "learning_rate": 1.2732200552315668e-05,
      "loss": 2.0877,
      "step": 27460
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.435546875,
      "learning_rate": 1.2722006929582533e-05,
      "loss": 2.1509,
      "step": 27470
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.435546875,
      "learning_rate": 1.2711814383599113e-05,
      "loss": 2.134,
      "step": 27480
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.419921875,
      "learning_rate": 1.2701622919183155e-05,
      "loss": 2.1218,
      "step": 27490
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.431640625,
      "learning_rate": 1.2691432541151916e-05,
      "loss": 2.1437,
      "step": 27500
    },
    {
      "epoch": 0.57,
      "eval_accuracy": 0.5590083540747678,
      "eval_loss": 1.9934571981430054,
      "eval_runtime": 16.4953,
      "eval_samples_per_second": 36.071,
      "eval_steps_per_second": 1.152,
      "step": 27500
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.44921875,
      "learning_rate": 1.2681243254322133e-05,
      "loss": 2.124,
      "step": 27510
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.431640625,
      "learning_rate": 1.2671055063510012e-05,
      "loss": 2.115,
      "step": 27520
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.474609375,
      "learning_rate": 1.266086797353126e-05,
      "loss": 2.1179,
      "step": 27530
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4375,
      "learning_rate": 1.2650681989201062e-05,
      "loss": 2.1293,
      "step": 27540
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.462890625,
      "learning_rate": 1.2640497115334065e-05,
      "loss": 2.1372,
      "step": 27550
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.44140625,
      "learning_rate": 1.2630313356744404e-05,
      "loss": 2.1186,
      "step": 27560
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.458984375,
      "learning_rate": 1.262013071824569e-05,
      "loss": 2.1205,
      "step": 27570
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.43359375,
      "learning_rate": 1.2609949204650995e-05,
      "loss": 2.1183,
      "step": 27580
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.421875,
      "learning_rate": 1.2599768820772864e-05,
      "loss": 2.1243,
      "step": 27590
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.439453125,
      "learning_rate": 1.25895895714233e-05,
      "loss": 2.1108,
      "step": 27600
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.486328125,
      "learning_rate": 1.2579411461413792e-05,
      "loss": 2.1356,
      "step": 27610
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.412109375,
      "learning_rate": 1.2569234495555257e-05,
      "loss": 2.1289,
      "step": 27620
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.451171875,
      "learning_rate": 1.2559058678658098e-05,
      "loss": 2.1485,
      "step": 27630
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4296875,
      "learning_rate": 1.2548884015532174e-05,
      "loss": 2.1122,
      "step": 27640
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.439453125,
      "learning_rate": 1.2538710510986776e-05,
      "loss": 2.0887,
      "step": 27650
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4375,
      "learning_rate": 1.252853816983067e-05,
      "loss": 2.1251,
      "step": 27660
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.44140625,
      "learning_rate": 1.251836699687207e-05,
      "loss": 2.1284,
      "step": 27670
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.421875,
      "learning_rate": 1.2508196996918618e-05,
      "loss": 2.1318,
      "step": 27680
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4375,
      "learning_rate": 1.2498028174777428e-05,
      "loss": 2.1294,
      "step": 27690
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5625,
      "learning_rate": 1.2487860535255042e-05,
      "loss": 2.1145,
      "step": 27700
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.451171875,
      "learning_rate": 1.2477694083157445e-05,
      "loss": 2.0956,
      "step": 27710
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.443359375,
      "learning_rate": 1.2467528823290057e-05,
      "loss": 2.1108,
      "step": 27720
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.42578125,
      "learning_rate": 1.2457364760457752e-05,
      "loss": 2.122,
      "step": 27730
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.451171875,
      "learning_rate": 1.2447201899464811e-05,
      "loss": 2.0908,
      "step": 27740
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4375,
      "learning_rate": 1.2437040245114966e-05,
      "loss": 2.0817,
      "step": 27750
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.42578125,
      "learning_rate": 1.2426879802211378e-05,
      "loss": 2.1055,
      "step": 27760
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.416015625,
      "learning_rate": 1.2416720575556635e-05,
      "loss": 2.0944,
      "step": 27770
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.43359375,
      "learning_rate": 1.2406562569952736e-05,
      "loss": 2.1228,
      "step": 27780
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4375,
      "learning_rate": 1.2396405790201115e-05,
      "loss": 2.1159,
      "step": 27790
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.447265625,
      "learning_rate": 1.2386250241102635e-05,
      "loss": 2.1282,
      "step": 27800
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.439453125,
      "learning_rate": 1.2376095927457553e-05,
      "loss": 2.1084,
      "step": 27810
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.41796875,
      "learning_rate": 1.236594285406557e-05,
      "loss": 2.0959,
      "step": 27820
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.42578125,
      "learning_rate": 1.2355791025725776e-05,
      "loss": 2.1166,
      "step": 27830
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.458984375,
      "learning_rate": 1.234564044723669e-05,
      "loss": 2.1539,
      "step": 27840
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.451171875,
      "learning_rate": 1.2335491123396228e-05,
      "loss": 2.1456,
      "step": 27850
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.439453125,
      "learning_rate": 1.232534305900173e-05,
      "loss": 2.1151,
      "step": 27860
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.447265625,
      "learning_rate": 1.2315196258849914e-05,
      "loss": 2.1129,
      "step": 27870
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.43359375,
      "learning_rate": 1.2305050727736924e-05,
      "loss": 2.1252,
      "step": 27880
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.451171875,
      "learning_rate": 1.2294906470458302e-05,
      "loss": 2.0908,
      "step": 27890
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.44921875,
      "learning_rate": 1.2284763491808968e-05,
      "loss": 2.1296,
      "step": 27900
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.44140625,
      "learning_rate": 1.227462179658326e-05,
      "loss": 2.0943,
      "step": 27910
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.451171875,
      "learning_rate": 1.2264481389574903e-05,
      "loss": 2.1286,
      "step": 27920
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.470703125,
      "learning_rate": 1.2254342275577005e-05,
      "loss": 2.151,
      "step": 27930
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.435546875,
      "learning_rate": 1.224420445938207e-05,
      "loss": 2.1285,
      "step": 27940
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4140625,
      "learning_rate": 1.2234067945781991e-05,
      "loss": 2.1218,
      "step": 27950
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.078125,
      "learning_rate": 1.2223932739568038e-05,
      "loss": 2.0688,
      "step": 27960
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.45703125,
      "learning_rate": 1.2213798845530871e-05,
      "loss": 2.1179,
      "step": 27970
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.423828125,
      "learning_rate": 1.2203666268460516e-05,
      "loss": 2.1146,
      "step": 27980
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.455078125,
      "learning_rate": 1.2193535013146402e-05,
      "loss": 2.1362,
      "step": 27990
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4453125,
      "learning_rate": 1.2183405084377297e-05,
      "loss": 2.1244,
      "step": 28000
    },
    {
      "epoch": 0.58,
      "eval_accuracy": 0.5591069272283693,
      "eval_loss": 1.9933356046676636,
      "eval_runtime": 16.4523,
      "eval_samples_per_second": 36.165,
      "eval_steps_per_second": 1.155,
      "step": 28000
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.478515625,
      "learning_rate": 1.2173276486941376e-05,
      "loss": 2.1222,
      "step": 28010
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.462890625,
      "learning_rate": 1.2163149225626171e-05,
      "loss": 2.1199,
      "step": 28020
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.431640625,
      "learning_rate": 1.2153023305218575e-05,
      "loss": 2.1549,
      "step": 28030
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.447265625,
      "learning_rate": 1.2142898730504856e-05,
      "loss": 2.1344,
      "step": 28040
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.451171875,
      "learning_rate": 1.2132775506270652e-05,
      "loss": 2.1078,
      "step": 28050
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4453125,
      "learning_rate": 1.2122653637300945e-05,
      "loss": 2.1093,
      "step": 28060
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.41796875,
      "learning_rate": 1.2112533128380093e-05,
      "loss": 2.1161,
      "step": 28070
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.435546875,
      "learning_rate": 1.2102413984291804e-05,
      "loss": 2.1284,
      "step": 28080
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.41796875,
      "learning_rate": 1.2092296209819138e-05,
      "loss": 2.0985,
      "step": 28090
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.43359375,
      "learning_rate": 1.208217980974451e-05,
      "loss": 2.133,
      "step": 28100
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.44140625,
      "learning_rate": 1.2072064788849697e-05,
      "loss": 2.1152,
      "step": 28110
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4375,
      "learning_rate": 1.20619511519158e-05,
      "loss": 2.125,
      "step": 28120
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.435546875,
      "learning_rate": 1.2051838903723286e-05,
      "loss": 2.1047,
      "step": 28130
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4375,
      "learning_rate": 1.204172804905196e-05,
      "loss": 2.1467,
      "step": 28140
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5859375,
      "learning_rate": 1.2031618592680969e-05,
      "loss": 2.0989,
      "step": 28150
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.435546875,
      "learning_rate": 1.2021510539388788e-05,
      "loss": 2.1089,
      "step": 28160
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.451171875,
      "learning_rate": 1.2011403893953246e-05,
      "loss": 2.1107,
      "step": 28170
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4453125,
      "learning_rate": 1.20012986611515e-05,
      "loss": 2.125,
      "step": 28180
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.435546875,
      "learning_rate": 1.1991194845760031e-05,
      "loss": 2.0944,
      "step": 28190
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.44140625,
      "learning_rate": 1.1981092452554665e-05,
      "loss": 2.1525,
      "step": 28200
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.45703125,
      "learning_rate": 1.1970991486310543e-05,
      "loss": 2.1026,
      "step": 28210
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4296875,
      "learning_rate": 1.1960891951802137e-05,
      "loss": 2.1265,
      "step": 28220
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.419921875,
      "learning_rate": 1.1950793853803238e-05,
      "loss": 2.1504,
      "step": 28230
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.447265625,
      "learning_rate": 1.194069719708697e-05,
      "loss": 2.125,
      "step": 28240
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.431640625,
      "learning_rate": 1.1930601986425758e-05,
      "loss": 2.1218,
      "step": 28250
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4765625,
      "learning_rate": 1.1920508226591358e-05,
      "loss": 2.1364,
      "step": 28260
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.435546875,
      "learning_rate": 1.1910415922354838e-05,
      "loss": 2.0988,
      "step": 28270
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.41796875,
      "learning_rate": 1.1900325078486564e-05,
      "loss": 2.1102,
      "step": 28280
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.455078125,
      "learning_rate": 1.1890235699756233e-05,
      "loss": 2.1155,
      "step": 28290
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.421875,
      "learning_rate": 1.1880147790932839e-05,
      "loss": 2.1141,
      "step": 28300
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4453125,
      "learning_rate": 1.1870061356784674e-05,
      "loss": 2.122,
      "step": 28310
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.470703125,
      "learning_rate": 1.1859976402079346e-05,
      "loss": 2.1254,
      "step": 28320
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.435546875,
      "learning_rate": 1.1849892931583755e-05,
      "loss": 2.1076,
      "step": 28330
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.419921875,
      "learning_rate": 1.183981095006411e-05,
      "loss": 2.1195,
      "step": 28340
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.435546875,
      "learning_rate": 1.1829730462285895e-05,
      "loss": 2.1006,
      "step": 28350
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.43359375,
      "learning_rate": 1.181965147301391e-05,
      "loss": 2.1458,
      "step": 28360
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.431640625,
      "learning_rate": 1.1809573987012246e-05,
      "loss": 2.0965,
      "step": 28370
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4296875,
      "learning_rate": 1.1799498009044258e-05,
      "loss": 2.1124,
      "step": 28380
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.439453125,
      "learning_rate": 1.1789423543872616e-05,
      "loss": 2.1053,
      "step": 28390
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.431640625,
      "learning_rate": 1.1779350596259271e-05,
      "loss": 2.1208,
      "step": 28400
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.427734375,
      "learning_rate": 1.1769279170965439e-05,
      "loss": 2.0985,
      "step": 28410
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.42578125,
      "learning_rate": 1.175920927275163e-05,
      "loss": 2.1096,
      "step": 28420
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.439453125,
      "learning_rate": 1.1749140906377641e-05,
      "loss": 2.1216,
      "step": 28430
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.451171875,
      "learning_rate": 1.1739074076602519e-05,
      "loss": 2.1419,
      "step": 28440
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.43359375,
      "learning_rate": 1.172900878818461e-05,
      "loss": 2.1376,
      "step": 28450
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.44921875,
      "learning_rate": 1.171894504588152e-05,
      "loss": 2.1519,
      "step": 28460
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.43359375,
      "learning_rate": 1.1708882854450123e-05,
      "loss": 2.0907,
      "step": 28470
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.427734375,
      "learning_rate": 1.1698822218646558e-05,
      "loss": 2.095,
      "step": 28480
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4375,
      "learning_rate": 1.1688763143226247e-05,
      "loss": 2.1459,
      "step": 28490
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4453125,
      "learning_rate": 1.1678705632943844e-05,
      "loss": 2.0767,
      "step": 28500
    },
    {
      "epoch": 0.59,
      "eval_accuracy": 0.5589377099813533,
      "eval_loss": 1.9933265447616577,
      "eval_runtime": 16.4552,
      "eval_samples_per_second": 36.159,
      "eval_steps_per_second": 1.155,
      "step": 28500
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.447265625,
      "learning_rate": 1.1668649692553287e-05,
      "loss": 2.1222,
      "step": 28510
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.423828125,
      "learning_rate": 1.1658595326807766e-05,
      "loss": 2.1317,
      "step": 28520
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.455078125,
      "learning_rate": 1.1648542540459731e-05,
      "loss": 2.1365,
      "step": 28530
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.431640625,
      "learning_rate": 1.1638491338260869e-05,
      "loss": 2.1198,
      "step": 28540
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.458984375,
      "learning_rate": 1.162844172496213e-05,
      "loss": 2.155,
      "step": 28550
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4921875,
      "learning_rate": 1.1618393705313723e-05,
      "loss": 2.1436,
      "step": 28560
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.439453125,
      "learning_rate": 1.160834728406508e-05,
      "loss": 2.1108,
      "step": 28570
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.421875,
      "learning_rate": 1.1598302465964897e-05,
      "loss": 2.123,
      "step": 28580
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.427734375,
      "learning_rate": 1.1588259255761105e-05,
      "loss": 2.1177,
      "step": 28590
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4375,
      "learning_rate": 1.1578217658200875e-05,
      "loss": 2.112,
      "step": 28600
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.470703125,
      "learning_rate": 1.1568177678030612e-05,
      "loss": 2.1178,
      "step": 28610
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4296875,
      "learning_rate": 1.1558139319995971e-05,
      "loss": 2.1127,
      "step": 28620
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.439453125,
      "learning_rate": 1.1548102588841817e-05,
      "loss": 2.1346,
      "step": 28630
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.439453125,
      "learning_rate": 1.1538067489312264e-05,
      "loss": 2.1185,
      "step": 28640
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.431640625,
      "learning_rate": 1.1528034026150658e-05,
      "loss": 2.1163,
      "step": 28650
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.443359375,
      "learning_rate": 1.151800220409955e-05,
      "loss": 2.075,
      "step": 28660
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4296875,
      "learning_rate": 1.1507972027900736e-05,
      "loss": 2.1443,
      "step": 28670
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.447265625,
      "learning_rate": 1.149794350229523e-05,
      "loss": 2.1195,
      "step": 28680
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.458984375,
      "learning_rate": 1.1487916632023252e-05,
      "loss": 2.1143,
      "step": 28690
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.412109375,
      "learning_rate": 1.147789142182426e-05,
      "loss": 2.0925,
      "step": 28700
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.447265625,
      "learning_rate": 1.1467867876436909e-05,
      "loss": 2.1458,
      "step": 28710
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.451171875,
      "learning_rate": 1.1457846000599086e-05,
      "loss": 2.1353,
      "step": 28720
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.439453125,
      "learning_rate": 1.1447825799047868e-05,
      "loss": 2.0754,
      "step": 28730
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.416015625,
      "learning_rate": 1.143780727651955e-05,
      "loss": 2.1034,
      "step": 28740
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.439453125,
      "learning_rate": 1.1427790437749648e-05,
      "loss": 2.1467,
      "step": 28750
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4296875,
      "learning_rate": 1.1417775287472853e-05,
      "loss": 2.1286,
      "step": 28760
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.443359375,
      "learning_rate": 1.140776183042308e-05,
      "loss": 2.1096,
      "step": 28770
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.46875,
      "learning_rate": 1.139775007133344e-05,
      "loss": 2.1491,
      "step": 28780
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4453125,
      "learning_rate": 1.1387740014936229e-05,
      "loss": 2.1065,
      "step": 28790
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.44140625,
      "learning_rate": 1.1377731665962954e-05,
      "loss": 2.1321,
      "step": 28800
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.69140625,
      "learning_rate": 1.1367725029144315e-05,
      "loss": 2.101,
      "step": 28810
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4375,
      "learning_rate": 1.135772010921018e-05,
      "loss": 2.103,
      "step": 28820
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.427734375,
      "learning_rate": 1.1347716910889635e-05,
      "loss": 2.115,
      "step": 28830
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.443359375,
      "learning_rate": 1.1337715438910934e-05,
      "loss": 2.1064,
      "step": 28840
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.435546875,
      "learning_rate": 1.1327715698001518e-05,
      "loss": 2.0888,
      "step": 28850
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.447265625,
      "learning_rate": 1.1317717692888014e-05,
      "loss": 2.0912,
      "step": 28860
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.427734375,
      "learning_rate": 1.1307721428296229e-05,
      "loss": 2.133,
      "step": 28870
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.443359375,
      "learning_rate": 1.1297726908951137e-05,
      "loss": 2.098,
      "step": 28880
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4375,
      "learning_rate": 1.1287734139576898e-05,
      "loss": 2.081,
      "step": 28890
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.453125,
      "learning_rate": 1.1277743124896842e-05,
      "loss": 2.1387,
      "step": 28900
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.455078125,
      "learning_rate": 1.1267753869633476e-05,
      "loss": 2.1141,
      "step": 28910
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.44921875,
      "learning_rate": 1.1257766378508454e-05,
      "loss": 2.1073,
      "step": 28920
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4375,
      "learning_rate": 1.1247780656242618e-05,
      "loss": 2.117,
      "step": 28930
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.447265625,
      "learning_rate": 1.1237796707555972e-05,
      "loss": 2.1142,
      "step": 28940
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.50390625,
      "learning_rate": 1.1227814537167666e-05,
      "loss": 2.0913,
      "step": 28950
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.451171875,
      "learning_rate": 1.1217834149796028e-05,
      "loss": 2.1046,
      "step": 28960
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.43359375,
      "learning_rate": 1.1207855550158526e-05,
      "loss": 2.1097,
      "step": 28970
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.44921875,
      "learning_rate": 1.1197878742971799e-05,
      "loss": 2.1279,
      "step": 28980
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.42578125,
      "learning_rate": 1.1187903732951624e-05,
      "loss": 2.1096,
      "step": 28990
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.443359375,
      "learning_rate": 1.1177930524812946e-05,
      "loss": 2.1182,
      "step": 29000
    },
    {
      "epoch": 0.6,
      "eval_accuracy": 0.5591019985706893,
      "eval_loss": 1.993354082107544,
      "eval_runtime": 16.4416,
      "eval_samples_per_second": 36.189,
      "eval_steps_per_second": 1.156,
      "step": 29000
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.44140625,
      "learning_rate": 1.1167959123269838e-05,
      "loss": 2.0716,
      "step": 29010
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4375,
      "learning_rate": 1.1157989533035534e-05,
      "loss": 2.0974,
      "step": 29020
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.443359375,
      "learning_rate": 1.1148021758822412e-05,
      "loss": 2.1374,
      "step": 29030
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.470703125,
      "learning_rate": 1.1138055805341977e-05,
      "loss": 2.1611,
      "step": 29040
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.451171875,
      "learning_rate": 1.1128091677304888e-05,
      "loss": 2.1267,
      "step": 29050
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.42578125,
      "learning_rate": 1.1118129379420942e-05,
      "loss": 2.1332,
      "step": 29060
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.435546875,
      "learning_rate": 1.1108168916399054e-05,
      "loss": 2.0888,
      "step": 29070
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4375,
      "learning_rate": 1.109821029294729e-05,
      "loss": 2.1189,
      "step": 29080
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.453125,
      "learning_rate": 1.1088253513772837e-05,
      "loss": 2.127,
      "step": 29090
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4375,
      "learning_rate": 1.107829858358202e-05,
      "loss": 2.1072,
      "step": 29100
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4296875,
      "learning_rate": 1.106834550708027e-05,
      "loss": 2.1126,
      "step": 29110
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.458984375,
      "learning_rate": 1.105839428897216e-05,
      "loss": 2.1192,
      "step": 29120
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.419921875,
      "learning_rate": 1.1048444933961387e-05,
      "loss": 2.0924,
      "step": 29130
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.46484375,
      "learning_rate": 1.1038497446750746e-05,
      "loss": 2.1034,
      "step": 29140
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.46875,
      "learning_rate": 1.1028551832042169e-05,
      "loss": 2.1216,
      "step": 29150
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.43359375,
      "learning_rate": 1.1018608094536704e-05,
      "loss": 2.0961,
      "step": 29160
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4375,
      "learning_rate": 1.1008666238934492e-05,
      "loss": 2.1109,
      "step": 29170
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.43359375,
      "learning_rate": 1.09987262699348e-05,
      "loss": 2.1036,
      "step": 29180
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.44140625,
      "learning_rate": 1.0988788192236007e-05,
      "loss": 2.1045,
      "step": 29190
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.431640625,
      "learning_rate": 1.097885201053558e-05,
      "loss": 2.1096,
      "step": 29200
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.43359375,
      "learning_rate": 1.096891772953011e-05,
      "loss": 2.1355,
      "step": 29210
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4140625,
      "learning_rate": 1.0958985353915275e-05,
      "loss": 2.1055,
      "step": 29220
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.50390625,
      "learning_rate": 1.0949054888385862e-05,
      "loss": 2.1157,
      "step": 29230
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.49609375,
      "learning_rate": 1.0939126337635745e-05,
      "loss": 2.1231,
      "step": 29240
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.41796875,
      "learning_rate": 1.0929199706357906e-05,
      "loss": 2.1276,
      "step": 29250
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.427734375,
      "learning_rate": 1.0919274999244405e-05,
      "loss": 2.1238,
      "step": 29260
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4296875,
      "learning_rate": 1.0909352220986401e-05,
      "loss": 2.1075,
      "step": 29270
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.43359375,
      "learning_rate": 1.0899431376274147e-05,
      "loss": 2.1133,
      "step": 29280
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.451171875,
      "learning_rate": 1.0889512469796976e-05,
      "loss": 2.0715,
      "step": 29290
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5,
      "learning_rate": 1.0879595506243291e-05,
      "loss": 2.1133,
      "step": 29300
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.44140625,
      "learning_rate": 1.0869680490300606e-05,
      "loss": 2.1344,
      "step": 29310
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4765625,
      "learning_rate": 1.0859767426655488e-05,
      "loss": 2.1221,
      "step": 29320
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.451171875,
      "learning_rate": 1.0849856319993595e-05,
      "loss": 2.0857,
      "step": 29330
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.431640625,
      "learning_rate": 1.0839947174999653e-05,
      "loss": 2.0728,
      "step": 29340
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.421875,
      "learning_rate": 1.0830039996357473e-05,
      "loss": 2.1091,
      "step": 29350
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.439453125,
      "learning_rate": 1.0820134788749916e-05,
      "loss": 2.1013,
      "step": 29360
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.43359375,
      "learning_rate": 1.081023155685893e-05,
      "loss": 2.0933,
      "step": 29370
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.443359375,
      "learning_rate": 1.0800330305365526e-05,
      "loss": 2.1147,
      "step": 29380
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.455078125,
      "learning_rate": 1.0790431038949764e-05,
      "loss": 2.1455,
      "step": 29390
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.46875,
      "learning_rate": 1.0780533762290782e-05,
      "loss": 2.1274,
      "step": 29400
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.431640625,
      "learning_rate": 1.0770638480066782e-05,
      "loss": 2.1096,
      "step": 29410
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.48828125,
      "learning_rate": 1.0760745196954998e-05,
      "loss": 2.1219,
      "step": 29420
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4765625,
      "learning_rate": 1.075085391763174e-05,
      "loss": 2.1362,
      "step": 29430
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4453125,
      "learning_rate": 1.0740964646772377e-05,
      "loss": 2.1168,
      "step": 29440
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4453125,
      "learning_rate": 1.07310773890513e-05,
      "loss": 2.1243,
      "step": 29450
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.435546875,
      "learning_rate": 1.0721192149141976e-05,
      "loss": 2.1261,
      "step": 29460
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.447265625,
      "learning_rate": 1.0711308931716903e-05,
      "loss": 2.1282,
      "step": 29470
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.494140625,
      "learning_rate": 1.0701427741447635e-05,
      "loss": 2.1249,
      "step": 29480
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.427734375,
      "learning_rate": 1.0691548583004752e-05,
      "loss": 2.1242,
      "step": 29490
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4375,
      "learning_rate": 1.0681671461057884e-05,
      "loss": 2.1277,
      "step": 29500
    },
    {
      "epoch": 0.61,
      "eval_accuracy": 0.5591184274296229,
      "eval_loss": 1.9932727813720703,
      "eval_runtime": 16.4505,
      "eval_samples_per_second": 36.169,
      "eval_steps_per_second": 1.155,
      "step": 29500
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.46484375,
      "learning_rate": 1.0671796380275705e-05,
      "loss": 2.1436,
      "step": 29510
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4375,
      "learning_rate": 1.0661923345325905e-05,
      "loss": 2.1217,
      "step": 29520
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4375,
      "learning_rate": 1.0652052360875222e-05,
      "loss": 2.1073,
      "step": 29530
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.435546875,
      "learning_rate": 1.0642183431589429e-05,
      "loss": 2.1244,
      "step": 29540
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.44140625,
      "learning_rate": 1.0632316562133306e-05,
      "loss": 2.1187,
      "step": 29550
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.42578125,
      "learning_rate": 1.0622451757170674e-05,
      "loss": 2.1234,
      "step": 29560
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.44140625,
      "learning_rate": 1.0612589021364392e-05,
      "loss": 2.1303,
      "step": 29570
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4375,
      "learning_rate": 1.0602728359376307e-05,
      "loss": 2.0886,
      "step": 29580
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.443359375,
      "learning_rate": 1.0592869775867316e-05,
      "loss": 2.1196,
      "step": 29590
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4375,
      "learning_rate": 1.0583013275497318e-05,
      "loss": 2.1159,
      "step": 29600
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.44921875,
      "learning_rate": 1.0573158862925233e-05,
      "loss": 2.1427,
      "step": 29610
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.474609375,
      "learning_rate": 1.0563306542808988e-05,
      "loss": 2.1101,
      "step": 29620
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.43359375,
      "learning_rate": 1.0553456319805535e-05,
      "loss": 2.1227,
      "step": 29630
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.451171875,
      "learning_rate": 1.0543608198570815e-05,
      "loss": 2.1621,
      "step": 29640
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.44140625,
      "learning_rate": 1.0533762183759788e-05,
      "loss": 2.1057,
      "step": 29650
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5546875,
      "learning_rate": 1.052391828002642e-05,
      "loss": 2.1247,
      "step": 29660
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.42578125,
      "learning_rate": 1.0514076492023678e-05,
      "loss": 2.1071,
      "step": 29670
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.462890625,
      "learning_rate": 1.0504236824403517e-05,
      "loss": 2.1102,
      "step": 29680
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.455078125,
      "learning_rate": 1.0494399281816907e-05,
      "loss": 2.1103,
      "step": 29690
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.470703125,
      "learning_rate": 1.0484563868913804e-05,
      "loss": 2.1005,
      "step": 29700
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4375,
      "learning_rate": 1.0474730590343155e-05,
      "loss": 2.1277,
      "step": 29710
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.42578125,
      "learning_rate": 1.0464899450752905e-05,
      "loss": 2.1271,
      "step": 29720
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.431640625,
      "learning_rate": 1.045507045478999e-05,
      "loss": 2.125,
      "step": 29730
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.45703125,
      "learning_rate": 1.0445243607100315e-05,
      "loss": 2.1203,
      "step": 29740
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4453125,
      "learning_rate": 1.0435418912328789e-05,
      "loss": 2.1179,
      "step": 29750
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.419921875,
      "learning_rate": 1.0425596375119306e-05,
      "loss": 2.1003,
      "step": 29760
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.462890625,
      "learning_rate": 1.0415776000114716e-05,
      "loss": 2.0789,
      "step": 29770
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.416015625,
      "learning_rate": 1.0405957791956867e-05,
      "loss": 2.1184,
      "step": 29780
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.439453125,
      "learning_rate": 1.0396141755286586e-05,
      "loss": 2.1058,
      "step": 29790
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4375,
      "learning_rate": 1.0386327894743653e-05,
      "loss": 2.1393,
      "step": 29800
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4296875,
      "learning_rate": 1.0376516214966835e-05,
      "loss": 2.1409,
      "step": 29810
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4375,
      "learning_rate": 1.0366706720593874e-05,
      "loss": 2.1125,
      "step": 29820
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4296875,
      "learning_rate": 1.0356899416261457e-05,
      "loss": 2.1123,
      "step": 29830
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4375,
      "learning_rate": 1.0347094306605257e-05,
      "loss": 2.102,
      "step": 29840
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.421875,
      "learning_rate": 1.0337291396259897e-05,
      "loss": 2.1292,
      "step": 29850
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.435546875,
      "learning_rate": 1.0327490689858974e-05,
      "loss": 2.1375,
      "step": 29860
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.439453125,
      "learning_rate": 1.0317692192035016e-05,
      "loss": 2.1037,
      "step": 29870
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.447265625,
      "learning_rate": 1.030789590741954e-05,
      "loss": 2.1209,
      "step": 29880
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5390625,
      "learning_rate": 1.0298101840643003e-05,
      "loss": 2.1366,
      "step": 29890
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.494140625,
      "learning_rate": 1.0288309996334802e-05,
      "loss": 2.0962,
      "step": 29900
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.43359375,
      "learning_rate": 1.0278520379123297e-05,
      "loss": 2.135,
      "step": 29910
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.423828125,
      "learning_rate": 1.0268732993635803e-05,
      "loss": 2.1327,
      "step": 29920
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.43359375,
      "learning_rate": 1.0258947844498555e-05,
      "loss": 2.1022,
      "step": 29930
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.451171875,
      "learning_rate": 1.0249164936336752e-05,
      "loss": 2.1302,
      "step": 29940
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.431640625,
      "learning_rate": 1.0239384273774532e-05,
      "loss": 2.1288,
      "step": 29950
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.443359375,
      "learning_rate": 1.0229605861434957e-05,
      "loss": 2.1141,
      "step": 29960
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.419921875,
      "learning_rate": 1.0219829703940047e-05,
      "loss": 2.1062,
      "step": 29970
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.439453125,
      "learning_rate": 1.0210055805910734e-05,
      "loss": 2.1297,
      "step": 29980
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.42578125,
      "learning_rate": 1.0200284171966896e-05,
      "loss": 2.1158,
      "step": 29990
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.43359375,
      "learning_rate": 1.0190514806727339e-05,
      "loss": 2.1407,
      "step": 30000
    },
    {
      "epoch": 0.62,
      "eval_accuracy": 0.5591019985706893,
      "eval_loss": 1.9932328462600708,
      "eval_runtime": 16.4688,
      "eval_samples_per_second": 36.129,
      "eval_steps_per_second": 1.154,
      "step": 30000
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.470703125,
      "learning_rate": 1.0180747714809794e-05,
      "loss": 2.1311,
      "step": 30010
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.474609375,
      "learning_rate": 1.0170982900830914e-05,
      "loss": 2.0843,
      "step": 30020
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.458984375,
      "learning_rate": 1.0161220369406283e-05,
      "loss": 2.0854,
      "step": 30030
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.447265625,
      "learning_rate": 1.0151460125150402e-05,
      "loss": 2.1222,
      "step": 30040
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.435546875,
      "learning_rate": 1.01417021726767e-05,
      "loss": 2.1272,
      "step": 30050
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.44140625,
      "learning_rate": 1.01319465165975e-05,
      "loss": 2.1244,
      "step": 30060
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.43359375,
      "learning_rate": 1.0122193161524064e-05,
      "loss": 2.1287,
      "step": 30070
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.44140625,
      "learning_rate": 1.0112442112066552e-05,
      "loss": 2.1257,
      "step": 30080
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.427734375,
      "learning_rate": 1.0102693372834042e-05,
      "loss": 2.1363,
      "step": 30090
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.44921875,
      "learning_rate": 1.009294694843451e-05,
      "loss": 2.1155,
      "step": 30100
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.431640625,
      "learning_rate": 1.0083202843474852e-05,
      "loss": 2.0924,
      "step": 30110
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.42578125,
      "learning_rate": 1.0073461062560853e-05,
      "loss": 2.1383,
      "step": 30120
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.44921875,
      "learning_rate": 1.0063721610297207e-05,
      "loss": 2.1162,
      "step": 30130
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.421875,
      "learning_rate": 1.0053984491287517e-05,
      "loss": 2.1221,
      "step": 30140
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.421875,
      "learning_rate": 1.004424971013426e-05,
      "loss": 2.1075,
      "step": 30150
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.42578125,
      "learning_rate": 1.0034517271438826e-05,
      "loss": 2.1451,
      "step": 30160
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.431640625,
      "learning_rate": 1.00247871798015e-05,
      "loss": 2.1068,
      "step": 30170
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4296875,
      "learning_rate": 1.0015059439821434e-05,
      "loss": 2.1205,
      "step": 30180
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.458984375,
      "learning_rate": 1.0005334056096695e-05,
      "loss": 2.1149,
      "step": 30190
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.431640625,
      "learning_rate": 9.995611033224234e-06,
      "loss": 2.1312,
      "step": 30200
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4296875,
      "learning_rate": 9.985890375799858e-06,
      "loss": 2.1447,
      "step": 30210
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.44140625,
      "learning_rate": 9.976172088418291e-06,
      "loss": 2.1118,
      "step": 30220
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.435546875,
      "learning_rate": 9.966456175673114e-06,
      "loss": 2.1096,
      "step": 30230
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.419921875,
      "learning_rate": 9.956742642156803e-06,
      "loss": 2.1126,
      "step": 30240
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.466796875,
      "learning_rate": 9.947031492460686e-06,
      "loss": 2.1153,
      "step": 30250
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.431640625,
      "learning_rate": 9.937322731174985e-06,
      "loss": 2.1039,
      "step": 30260
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.42578125,
      "learning_rate": 9.927616362888795e-06,
      "loss": 2.1111,
      "step": 30270
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.455078125,
      "learning_rate": 9.917912392190051e-06,
      "loss": 2.129,
      "step": 30280
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.44140625,
      "learning_rate": 9.908210823665587e-06,
      "loss": 2.0921,
      "step": 30290
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.42578125,
      "learning_rate": 9.898511661901095e-06,
      "loss": 2.0968,
      "step": 30300
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.423828125,
      "learning_rate": 9.888814911481105e-06,
      "loss": 2.1397,
      "step": 30310
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.45703125,
      "learning_rate": 9.879120576989045e-06,
      "loss": 2.099,
      "step": 30320
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.462890625,
      "learning_rate": 9.86942866300717e-06,
      "loss": 2.1389,
      "step": 30330
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.421875,
      "learning_rate": 9.859739174116606e-06,
      "loss": 2.1263,
      "step": 30340
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.44140625,
      "learning_rate": 9.850052114897325e-06,
      "loss": 2.1329,
      "step": 30350
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.421875,
      "learning_rate": 9.840367489928166e-06,
      "loss": 2.0913,
      "step": 30360
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.443359375,
      "learning_rate": 9.830685303786792e-06,
      "loss": 2.1319,
      "step": 30370
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.44921875,
      "learning_rate": 9.821005561049735e-06,
      "loss": 2.1326,
      "step": 30380
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.443359375,
      "learning_rate": 9.811328266292373e-06,
      "loss": 2.096,
      "step": 30390
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.423828125,
      "learning_rate": 9.8016534240889e-06,
      "loss": 2.1223,
      "step": 30400
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.578125,
      "learning_rate": 9.791981039012381e-06,
      "loss": 2.1531,
      "step": 30410
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.439453125,
      "learning_rate": 9.782311115634706e-06,
      "loss": 2.1453,
      "step": 30420
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.46875,
      "learning_rate": 9.772643658526608e-06,
      "loss": 2.1339,
      "step": 30430
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4296875,
      "learning_rate": 9.762978672257639e-06,
      "loss": 2.1072,
      "step": 30440
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4453125,
      "learning_rate": 9.753316161396201e-06,
      "loss": 2.0828,
      "step": 30450
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.43359375,
      "learning_rate": 9.743656130509518e-06,
      "loss": 2.1321,
      "step": 30460
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.435546875,
      "learning_rate": 9.733998584163641e-06,
      "loss": 2.1196,
      "step": 30470
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.419921875,
      "learning_rate": 9.724343526923442e-06,
      "loss": 2.1043,
      "step": 30480
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.421875,
      "learning_rate": 9.714690963352637e-06,
      "loss": 2.1051,
      "step": 30490
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.427734375,
      "learning_rate": 9.705040898013733e-06,
      "loss": 2.1222,
      "step": 30500
    },
    {
      "epoch": 0.63,
      "eval_accuracy": 0.5590510691079951,
      "eval_loss": 1.9932135343551636,
      "eval_runtime": 16.4428,
      "eval_samples_per_second": 36.186,
      "eval_steps_per_second": 1.156,
      "step": 30500
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4296875,
      "learning_rate": 9.695393335468078e-06,
      "loss": 2.1234,
      "step": 30510
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.435546875,
      "learning_rate": 9.685748280275837e-06,
      "loss": 2.1049,
      "step": 30520
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4375,
      "learning_rate": 9.676105736995972e-06,
      "loss": 2.1076,
      "step": 30530
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.43359375,
      "learning_rate": 9.666465710186276e-06,
      "loss": 2.1172,
      "step": 30540
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43359375,
      "learning_rate": 9.656828204403352e-06,
      "loss": 2.1073,
      "step": 30550
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.421875,
      "learning_rate": 9.647193224202591e-06,
      "loss": 2.1068,
      "step": 30560
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.421875,
      "learning_rate": 9.637560774138213e-06,
      "loss": 2.0964,
      "step": 30570
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.447265625,
      "learning_rate": 9.627930858763237e-06,
      "loss": 2.1041,
      "step": 30580
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.455078125,
      "learning_rate": 9.618303482629472e-06,
      "loss": 2.1592,
      "step": 30590
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43359375,
      "learning_rate": 9.608678650287541e-06,
      "loss": 2.1262,
      "step": 30600
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.453125,
      "learning_rate": 9.599056366286854e-06,
      "loss": 2.1262,
      "step": 30610
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4375,
      "learning_rate": 9.58943663517563e-06,
      "loss": 2.1378,
      "step": 30620
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.435546875,
      "learning_rate": 9.579819461500861e-06,
      "loss": 2.1316,
      "step": 30630
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.435546875,
      "learning_rate": 9.570204849808345e-06,
      "loss": 2.1083,
      "step": 30640
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.484375,
      "learning_rate": 9.560592804642678e-06,
      "loss": 2.1088,
      "step": 30650
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.443359375,
      "learning_rate": 9.550983330547212e-06,
      "loss": 2.1397,
      "step": 30660
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4765625,
      "learning_rate": 9.541376432064113e-06,
      "loss": 2.1238,
      "step": 30670
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.451171875,
      "learning_rate": 9.531772113734325e-06,
      "loss": 2.1097,
      "step": 30680
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.46875,
      "learning_rate": 9.52217038009755e-06,
      "loss": 2.1322,
      "step": 30690
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.443359375,
      "learning_rate": 9.512571235692296e-06,
      "loss": 2.1074,
      "step": 30700
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.46484375,
      "learning_rate": 9.502974685055835e-06,
      "loss": 2.1342,
      "step": 30710
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.46875,
      "learning_rate": 9.493380732724211e-06,
      "loss": 2.137,
      "step": 30720
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43359375,
      "learning_rate": 9.483789383232241e-06,
      "loss": 2.1163,
      "step": 30730
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43359375,
      "learning_rate": 9.47420064111352e-06,
      "loss": 2.1196,
      "step": 30740
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.423828125,
      "learning_rate": 9.464614510900396e-06,
      "loss": 2.1264,
      "step": 30750
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4375,
      "learning_rate": 9.455030997123993e-06,
      "loss": 2.0968,
      "step": 30760
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.439453125,
      "learning_rate": 9.4454501043142e-06,
      "loss": 2.1388,
      "step": 30770
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.435546875,
      "learning_rate": 9.435871836999657e-06,
      "loss": 2.0876,
      "step": 30780
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4609375,
      "learning_rate": 9.426296199707769e-06,
      "loss": 2.1163,
      "step": 30790
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.455078125,
      "learning_rate": 9.4167231969647e-06,
      "loss": 2.1536,
      "step": 30800
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.419921875,
      "learning_rate": 9.407152833295373e-06,
      "loss": 2.0823,
      "step": 30810
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4375,
      "learning_rate": 9.397585113223441e-06,
      "loss": 2.1023,
      "step": 30820
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.439453125,
      "learning_rate": 9.388020041271338e-06,
      "loss": 2.1229,
      "step": 30830
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.431640625,
      "learning_rate": 9.378457621960224e-06,
      "loss": 2.0858,
      "step": 30840
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.478515625,
      "learning_rate": 9.368897859810015e-06,
      "loss": 2.1202,
      "step": 30850
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.462890625,
      "learning_rate": 9.359340759339365e-06,
      "loss": 2.1152,
      "step": 30860
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.439453125,
      "learning_rate": 9.349786325065682e-06,
      "loss": 2.0813,
      "step": 30870
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.41015625,
      "learning_rate": 9.340234561505094e-06,
      "loss": 2.1322,
      "step": 30880
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.416015625,
      "learning_rate": 9.330685473172485e-06,
      "loss": 2.1322,
      "step": 30890
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.435546875,
      "learning_rate": 9.321139064581473e-06,
      "loss": 2.107,
      "step": 30900
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43359375,
      "learning_rate": 9.311595340244392e-06,
      "loss": 2.1386,
      "step": 30910
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.458984375,
      "learning_rate": 9.302054304672324e-06,
      "loss": 2.1254,
      "step": 30920
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.451171875,
      "learning_rate": 9.292515962375081e-06,
      "loss": 2.1013,
      "step": 30930
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.439453125,
      "learning_rate": 9.282980317861188e-06,
      "loss": 2.0934,
      "step": 30940
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.427734375,
      "learning_rate": 9.273447375637903e-06,
      "loss": 2.1306,
      "step": 30950
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.50390625,
      "learning_rate": 9.263917140211218e-06,
      "loss": 2.1026,
      "step": 30960
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4453125,
      "learning_rate": 9.254389616085822e-06,
      "loss": 2.0771,
      "step": 30970
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.443359375,
      "learning_rate": 9.244864807765137e-06,
      "loss": 2.0995,
      "step": 30980
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.443359375,
      "learning_rate": 9.235342719751303e-06,
      "loss": 2.1427,
      "step": 30990
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4375,
      "learning_rate": 9.22582335654517e-06,
      "loss": 2.1146,
      "step": 31000
    },
    {
      "epoch": 0.64,
      "eval_accuracy": 0.5590855697117557,
      "eval_loss": 1.9931367635726929,
      "eval_runtime": 16.4389,
      "eval_samples_per_second": 36.195,
      "eval_steps_per_second": 1.156,
      "step": 31000
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.427734375,
      "learning_rate": 9.216306722646294e-06,
      "loss": 2.1145,
      "step": 31010
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.49609375,
      "learning_rate": 9.20679282255295e-06,
      "loss": 2.1444,
      "step": 31020
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.46484375,
      "learning_rate": 9.197281660762127e-06,
      "loss": 2.0652,
      "step": 31030
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.423828125,
      "learning_rate": 9.187773241769498e-06,
      "loss": 2.114,
      "step": 31040
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.41796875,
      "learning_rate": 9.178267570069455e-06,
      "loss": 2.1009,
      "step": 31050
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4453125,
      "learning_rate": 9.168764650155096e-06,
      "loss": 2.119,
      "step": 31060
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.451171875,
      "learning_rate": 9.1592644865182e-06,
      "loss": 2.138,
      "step": 31070
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4296875,
      "learning_rate": 9.149767083649265e-06,
      "loss": 2.1119,
      "step": 31080
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.447265625,
      "learning_rate": 9.140272446037467e-06,
      "loss": 2.1347,
      "step": 31090
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.416015625,
      "learning_rate": 9.130780578170681e-06,
      "loss": 2.0966,
      "step": 31100
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.447265625,
      "learning_rate": 9.12129148453547e-06,
      "loss": 2.1309,
      "step": 31110
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.43359375,
      "learning_rate": 9.111805169617101e-06,
      "loss": 2.1442,
      "step": 31120
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.43359375,
      "learning_rate": 9.102321637899498e-06,
      "loss": 2.0949,
      "step": 31130
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4296875,
      "learning_rate": 9.092840893865297e-06,
      "loss": 2.1191,
      "step": 31140
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.431640625,
      "learning_rate": 9.08336294199581e-06,
      "loss": 2.1242,
      "step": 31150
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.46875,
      "learning_rate": 9.073887786771011e-06,
      "loss": 2.115,
      "step": 31160
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.443359375,
      "learning_rate": 9.064415432669576e-06,
      "loss": 2.1381,
      "step": 31170
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.421875,
      "learning_rate": 9.054945884168844e-06,
      "loss": 2.0843,
      "step": 31180
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.44921875,
      "learning_rate": 9.04547914574484e-06,
      "loss": 2.0895,
      "step": 31190
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.423828125,
      "learning_rate": 9.03601522187224e-06,
      "loss": 2.1387,
      "step": 31200
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.42578125,
      "learning_rate": 9.026554117024405e-06,
      "loss": 2.0937,
      "step": 31210
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.427734375,
      "learning_rate": 9.017095835673363e-06,
      "loss": 2.1074,
      "step": 31220
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4375,
      "learning_rate": 9.007640382289804e-06,
      "loss": 2.1211,
      "step": 31230
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.439453125,
      "learning_rate": 8.998187761343075e-06,
      "loss": 2.1179,
      "step": 31240
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.421875,
      "learning_rate": 8.9887379773012e-06,
      "loss": 2.0998,
      "step": 31250
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.431640625,
      "learning_rate": 8.979291034630845e-06,
      "loss": 2.1055,
      "step": 31260
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4296875,
      "learning_rate": 8.969846937797346e-06,
      "loss": 2.0696,
      "step": 31270
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.43359375,
      "learning_rate": 8.960405691264691e-06,
      "loss": 2.1269,
      "step": 31280
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4375,
      "learning_rate": 8.95096729949551e-06,
      "loss": 2.1051,
      "step": 31290
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4375,
      "learning_rate": 8.941531766951097e-06,
      "loss": 2.1113,
      "step": 31300
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.439453125,
      "learning_rate": 8.932099098091395e-06,
      "loss": 2.1291,
      "step": 31310
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.439453125,
      "learning_rate": 8.922669297374976e-06,
      "loss": 2.1434,
      "step": 31320
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.43359375,
      "learning_rate": 8.913242369259076e-06,
      "loss": 2.0872,
      "step": 31330
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.41796875,
      "learning_rate": 8.903818318199569e-06,
      "loss": 2.0831,
      "step": 31340
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4296875,
      "learning_rate": 8.894397148650958e-06,
      "loss": 2.1078,
      "step": 31350
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.412109375,
      "learning_rate": 8.88497886506639e-06,
      "loss": 2.0866,
      "step": 31360
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.435546875,
      "learning_rate": 8.875563471897663e-06,
      "loss": 2.1092,
      "step": 31370
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.447265625,
      "learning_rate": 8.866150973595179e-06,
      "loss": 2.1217,
      "step": 31380
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.42578125,
      "learning_rate": 8.856741374607992e-06,
      "loss": 2.1097,
      "step": 31390
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.439453125,
      "learning_rate": 8.847334679383789e-06,
      "loss": 2.0865,
      "step": 31400
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.46875,
      "learning_rate": 8.837930892368876e-06,
      "loss": 2.1002,
      "step": 31410
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6875,
      "learning_rate": 8.828530018008175e-06,
      "loss": 2.1029,
      "step": 31420
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.435546875,
      "learning_rate": 8.819132060745248e-06,
      "loss": 2.0935,
      "step": 31430
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.46484375,
      "learning_rate": 8.809737025022279e-06,
      "loss": 2.1301,
      "step": 31440
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4296875,
      "learning_rate": 8.80034491528005e-06,
      "loss": 2.1212,
      "step": 31450
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.439453125,
      "learning_rate": 8.790955735957984e-06,
      "loss": 2.117,
      "step": 31460
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4296875,
      "learning_rate": 8.781569491494104e-06,
      "loss": 2.1244,
      "step": 31470
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.451171875,
      "learning_rate": 8.772186186325053e-06,
      "loss": 2.1206,
      "step": 31480
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.439453125,
      "learning_rate": 8.762805824886077e-06,
      "loss": 2.1079,
      "step": 31490
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.453125,
      "learning_rate": 8.753428411611046e-06,
      "loss": 2.1441,
      "step": 31500
    },
    {
      "epoch": 0.65,
      "eval_accuracy": 0.5591299276308764,
      "eval_loss": 1.9931890964508057,
      "eval_runtime": 16.4439,
      "eval_samples_per_second": 36.184,
      "eval_steps_per_second": 1.155,
      "step": 31500
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4296875,
      "learning_rate": 8.744053950932412e-06,
      "loss": 2.1286,
      "step": 31510
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.458984375,
      "learning_rate": 8.73468244728125e-06,
      "loss": 2.1114,
      "step": 31520
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4453125,
      "learning_rate": 8.725313905087245e-06,
      "loss": 2.1194,
      "step": 31530
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.490234375,
      "learning_rate": 8.715948328778651e-06,
      "loss": 2.1349,
      "step": 31540
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4296875,
      "learning_rate": 8.706585722782351e-06,
      "loss": 2.1535,
      "step": 31550
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.515625,
      "learning_rate": 8.697226091523814e-06,
      "loss": 2.108,
      "step": 31560
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.431640625,
      "learning_rate": 8.687869439427092e-06,
      "loss": 2.0905,
      "step": 31570
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.49609375,
      "learning_rate": 8.67851577091484e-06,
      "loss": 2.1003,
      "step": 31580
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.431640625,
      "learning_rate": 8.669165090408303e-06,
      "loss": 2.1145,
      "step": 31590
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4375,
      "learning_rate": 8.659817402327317e-06,
      "loss": 2.1379,
      "step": 31600
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.43359375,
      "learning_rate": 8.650472711090288e-06,
      "loss": 2.1164,
      "step": 31610
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.447265625,
      "learning_rate": 8.641131021114216e-06,
      "loss": 2.1077,
      "step": 31620
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.47265625,
      "learning_rate": 8.631792336814692e-06,
      "loss": 2.1436,
      "step": 31630
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.458984375,
      "learning_rate": 8.62245666260586e-06,
      "loss": 2.1146,
      "step": 31640
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.474609375,
      "learning_rate": 8.613124002900472e-06,
      "loss": 2.1544,
      "step": 31650
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.578125,
      "learning_rate": 8.60379436210983e-06,
      "loss": 2.1234,
      "step": 31660
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4375,
      "learning_rate": 8.594467744643827e-06,
      "loss": 2.1442,
      "step": 31670
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.431640625,
      "learning_rate": 8.585144154910909e-06,
      "loss": 2.1102,
      "step": 31680
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4375,
      "learning_rate": 8.575823597318113e-06,
      "loss": 2.1153,
      "step": 31690
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.416015625,
      "learning_rate": 8.566506076271019e-06,
      "loss": 2.1011,
      "step": 31700
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.453125,
      "learning_rate": 8.55719159617379e-06,
      "loss": 2.1486,
      "step": 31710
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.431640625,
      "learning_rate": 8.54788016142915e-06,
      "loss": 2.1039,
      "step": 31720
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.43359375,
      "learning_rate": 8.538571776438367e-06,
      "loss": 2.1376,
      "step": 31730
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42578125,
      "learning_rate": 8.529266445601286e-06,
      "loss": 2.1132,
      "step": 31740
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.50390625,
      "learning_rate": 8.519964173316307e-06,
      "loss": 2.1408,
      "step": 31750
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.427734375,
      "learning_rate": 8.510664963980367e-06,
      "loss": 2.1182,
      "step": 31760
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.423828125,
      "learning_rate": 8.501368821988972e-06,
      "loss": 2.0857,
      "step": 31770
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.443359375,
      "learning_rate": 8.492075751736175e-06,
      "loss": 2.0936,
      "step": 31780
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4453125,
      "learning_rate": 8.48278575761458e-06,
      "loss": 2.1202,
      "step": 31790
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42578125,
      "learning_rate": 8.47349884401532e-06,
      "loss": 2.1272,
      "step": 31800
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5078125,
      "learning_rate": 8.464215015328092e-06,
      "loss": 2.1261,
      "step": 31810
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.48828125,
      "learning_rate": 8.454934275941129e-06,
      "loss": 2.157,
      "step": 31820
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42578125,
      "learning_rate": 8.445656630241196e-06,
      "loss": 2.1043,
      "step": 31830
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.451171875,
      "learning_rate": 8.436382082613604e-06,
      "loss": 2.1176,
      "step": 31840
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42578125,
      "learning_rate": 8.427110637442202e-06,
      "loss": 2.1065,
      "step": 31850
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.91796875,
      "learning_rate": 8.417842299109356e-06,
      "loss": 2.095,
      "step": 31860
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.447265625,
      "learning_rate": 8.408577071995984e-06,
      "loss": 2.0945,
      "step": 31870
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4375,
      "learning_rate": 8.399314960481527e-06,
      "loss": 2.0827,
      "step": 31880
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.416015625,
      "learning_rate": 8.390055968943941e-06,
      "loss": 2.1257,
      "step": 31890
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.46875,
      "learning_rate": 8.380800101759727e-06,
      "loss": 2.1136,
      "step": 31900
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.474609375,
      "learning_rate": 8.37154736330389e-06,
      "loss": 2.1352,
      "step": 31910
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.431640625,
      "learning_rate": 8.362297757949979e-06,
      "loss": 2.1248,
      "step": 31920
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.453125,
      "learning_rate": 8.353051290070028e-06,
      "loss": 2.1162,
      "step": 31930
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.53515625,
      "learning_rate": 8.34380796403463e-06,
      "loss": 2.1407,
      "step": 31940
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.416015625,
      "learning_rate": 8.334567784212855e-06,
      "loss": 2.1092,
      "step": 31950
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.427734375,
      "learning_rate": 8.325330754972311e-06,
      "loss": 2.1082,
      "step": 31960
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.443359375,
      "learning_rate": 8.316096880679102e-06,
      "loss": 2.0892,
      "step": 31970
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.48046875,
      "learning_rate": 8.30686616569786e-06,
      "loss": 2.1229,
      "step": 31980
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.423828125,
      "learning_rate": 8.297638614391693e-06,
      "loss": 2.0932,
      "step": 31990
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.435546875,
      "learning_rate": 8.288414231122242e-06,
      "loss": 2.1224,
      "step": 32000
    },
    {
      "epoch": 0.67,
      "eval_accuracy": 0.5589754963569006,
      "eval_loss": 1.9931281805038452,
      "eval_runtime": 16.4271,
      "eval_samples_per_second": 36.221,
      "eval_steps_per_second": 1.157,
      "step": 32000
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4609375,
      "learning_rate": 8.279193020249642e-06,
      "loss": 2.1093,
      "step": 32010
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.48046875,
      "learning_rate": 8.269974986132514e-06,
      "loss": 2.1047,
      "step": 32020
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.435546875,
      "learning_rate": 8.260760133128e-06,
      "loss": 2.1104,
      "step": 32030
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.44921875,
      "learning_rate": 8.25154846559173e-06,
      "loss": 2.1044,
      "step": 32040
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.458984375,
      "learning_rate": 8.242339987877813e-06,
      "loss": 2.1089,
      "step": 32050
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.431640625,
      "learning_rate": 8.233134704338874e-06,
      "loss": 2.0808,
      "step": 32060
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.462890625,
      "learning_rate": 8.223932619326022e-06,
      "loss": 2.1162,
      "step": 32070
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4453125,
      "learning_rate": 8.214733737188835e-06,
      "loss": 2.1038,
      "step": 32080
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.44140625,
      "learning_rate": 8.205538062275402e-06,
      "loss": 2.0947,
      "step": 32090
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.447265625,
      "learning_rate": 8.196345598932289e-06,
      "loss": 2.1025,
      "step": 32100
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.447265625,
      "learning_rate": 8.18715635150453e-06,
      "loss": 2.1128,
      "step": 32110
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.431640625,
      "learning_rate": 8.177970324335658e-06,
      "loss": 2.11,
      "step": 32120
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.443359375,
      "learning_rate": 8.168787521767677e-06,
      "loss": 2.1063,
      "step": 32130
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.46875,
      "learning_rate": 8.159607948141061e-06,
      "loss": 2.1135,
      "step": 32140
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.451171875,
      "learning_rate": 8.150431607794765e-06,
      "loss": 2.1076,
      "step": 32150
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4375,
      "learning_rate": 8.141258505066207e-06,
      "loss": 2.1104,
      "step": 32160
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4453125,
      "learning_rate": 8.13208864429129e-06,
      "loss": 2.1295,
      "step": 32170
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4453125,
      "learning_rate": 8.122922029804365e-06,
      "loss": 2.1304,
      "step": 32180
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.435546875,
      "learning_rate": 8.113758665938264e-06,
      "loss": 2.1188,
      "step": 32190
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.435546875,
      "learning_rate": 8.104598557024284e-06,
      "loss": 2.0994,
      "step": 32200
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.43359375,
      "learning_rate": 8.095441707392164e-06,
      "loss": 2.1194,
      "step": 32210
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.44921875,
      "learning_rate": 8.086288121370122e-06,
      "loss": 2.109,
      "step": 32220
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4140625,
      "learning_rate": 8.077137803284833e-06,
      "loss": 2.1332,
      "step": 32230
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4453125,
      "learning_rate": 8.067990757461406e-06,
      "loss": 2.1041,
      "step": 32240
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.439453125,
      "learning_rate": 8.05884698822343e-06,
      "loss": 2.1283,
      "step": 32250
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.453125,
      "learning_rate": 8.049706499892931e-06,
      "loss": 2.1538,
      "step": 32260
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.419921875,
      "learning_rate": 8.040569296790383e-06,
      "loss": 2.1161,
      "step": 32270
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.41796875,
      "learning_rate": 8.031435383234714e-06,
      "loss": 2.1173,
      "step": 32280
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.416015625,
      "learning_rate": 8.022304763543297e-06,
      "loss": 2.1175,
      "step": 32290
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.44140625,
      "learning_rate": 8.013177442031935e-06,
      "loss": 2.168,
      "step": 32300
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.419921875,
      "learning_rate": 8.00405342301489e-06,
      "loss": 2.1067,
      "step": 32310
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.44140625,
      "learning_rate": 7.994932710804858e-06,
      "loss": 2.109,
      "step": 32320
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.435546875,
      "learning_rate": 7.985815309712959e-06,
      "loss": 2.0894,
      "step": 32330
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.41796875,
      "learning_rate": 7.976701224048763e-06,
      "loss": 2.1172,
      "step": 32340
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.443359375,
      "learning_rate": 7.967590458120267e-06,
      "loss": 2.1222,
      "step": 32350
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.42578125,
      "learning_rate": 7.95848301623391e-06,
      "loss": 2.1209,
      "step": 32360
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.42578125,
      "learning_rate": 7.94937890269453e-06,
      "loss": 2.1348,
      "step": 32370
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.41796875,
      "learning_rate": 7.940278121805422e-06,
      "loss": 2.1401,
      "step": 32380
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.435546875,
      "learning_rate": 7.931180677868302e-06,
      "loss": 2.1679,
      "step": 32390
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.453125,
      "learning_rate": 7.922086575183286e-06,
      "loss": 2.1095,
      "step": 32400
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.421875,
      "learning_rate": 7.912995818048942e-06,
      "loss": 2.1217,
      "step": 32410
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.435546875,
      "learning_rate": 7.903908410762225e-06,
      "loss": 2.0845,
      "step": 32420
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.458984375,
      "learning_rate": 7.894824357618537e-06,
      "loss": 2.158,
      "step": 32430
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.43359375,
      "learning_rate": 7.88574366291167e-06,
      "loss": 2.0827,
      "step": 32440
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.42578125,
      "learning_rate": 7.876666330933849e-06,
      "loss": 2.1272,
      "step": 32450
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.45703125,
      "learning_rate": 7.867592365975685e-06,
      "loss": 2.1443,
      "step": 32460
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.47265625,
      "learning_rate": 7.858521772326219e-06,
      "loss": 2.0929,
      "step": 32470
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.419921875,
      "learning_rate": 7.8494545542729e-06,
      "loss": 2.1117,
      "step": 32480
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.455078125,
      "learning_rate": 7.840390716101557e-06,
      "loss": 2.1476,
      "step": 32490
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.427734375,
      "learning_rate": 7.831330262096444e-06,
      "loss": 2.0878,
      "step": 32500
    },
    {
      "epoch": 0.68,
      "eval_accuracy": 0.5590691408528221,
      "eval_loss": 1.993199110031128,
      "eval_runtime": 16.4617,
      "eval_samples_per_second": 36.145,
      "eval_steps_per_second": 1.154,
      "step": 32500
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44921875,
      "learning_rate": 7.822273196540217e-06,
      "loss": 2.0906,
      "step": 32510
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.447265625,
      "learning_rate": 7.813219523713905e-06,
      "loss": 2.127,
      "step": 32520
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.435546875,
      "learning_rate": 7.804169247896964e-06,
      "loss": 2.1301,
      "step": 32530
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44921875,
      "learning_rate": 7.79512237336722e-06,
      "loss": 2.1127,
      "step": 32540
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.453125,
      "learning_rate": 7.786078904400916e-06,
      "loss": 2.1312,
      "step": 32550
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.42578125,
      "learning_rate": 7.777038845272656e-06,
      "loss": 2.1347,
      "step": 32560
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4453125,
      "learning_rate": 7.768002200255454e-06,
      "loss": 2.1091,
      "step": 32570
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.427734375,
      "learning_rate": 7.758968973620709e-06,
      "loss": 2.1449,
      "step": 32580
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4296875,
      "learning_rate": 7.749939169638187e-06,
      "loss": 2.1581,
      "step": 32590
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.419921875,
      "learning_rate": 7.740912792576052e-06,
      "loss": 2.0935,
      "step": 32600
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.439453125,
      "learning_rate": 7.731889846700856e-06,
      "loss": 2.0973,
      "step": 32610
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4765625,
      "learning_rate": 7.722870336277497e-06,
      "loss": 2.1215,
      "step": 32620
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.431640625,
      "learning_rate": 7.713854265569283e-06,
      "loss": 2.1051,
      "step": 32630
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4296875,
      "learning_rate": 7.704841638837883e-06,
      "loss": 2.1251,
      "step": 32640
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.443359375,
      "learning_rate": 7.695832460343331e-06,
      "loss": 2.1171,
      "step": 32650
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.43359375,
      "learning_rate": 7.686826734344045e-06,
      "loss": 2.1084,
      "step": 32660
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44140625,
      "learning_rate": 7.677824465096794e-06,
      "loss": 2.1094,
      "step": 32670
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4375,
      "learning_rate": 7.668825656856736e-06,
      "loss": 2.1289,
      "step": 32680
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.439453125,
      "learning_rate": 7.659830313877367e-06,
      "loss": 2.0985,
      "step": 32690
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4453125,
      "learning_rate": 7.650838440410566e-06,
      "loss": 2.0966,
      "step": 32700
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.451171875,
      "learning_rate": 7.641850040706562e-06,
      "loss": 2.0937,
      "step": 32710
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.42578125,
      "learning_rate": 7.63286511901394e-06,
      "loss": 2.091,
      "step": 32720
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.466796875,
      "learning_rate": 7.623883679579651e-06,
      "loss": 2.1084,
      "step": 32730
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.455078125,
      "learning_rate": 7.614905726649e-06,
      "loss": 2.1007,
      "step": 32740
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.46875,
      "learning_rate": 7.605931264465624e-06,
      "loss": 2.1229,
      "step": 32750
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.48828125,
      "learning_rate": 7.596960297271531e-06,
      "loss": 2.0927,
      "step": 32760
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.439453125,
      "learning_rate": 7.587992829307079e-06,
      "loss": 2.1181,
      "step": 32770
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44921875,
      "learning_rate": 7.579028864810948e-06,
      "loss": 2.0925,
      "step": 32780
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.427734375,
      "learning_rate": 7.570068408020184e-06,
      "loss": 2.1103,
      "step": 32790
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.435546875,
      "learning_rate": 7.561111463170176e-06,
      "loss": 2.0649,
      "step": 32800
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.466796875,
      "learning_rate": 7.55215803449463e-06,
      "loss": 2.1158,
      "step": 32810
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44921875,
      "learning_rate": 7.543208126225616e-06,
      "loss": 2.1479,
      "step": 32820
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44921875,
      "learning_rate": 7.534261742593529e-06,
      "loss": 2.1205,
      "step": 32830
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.439453125,
      "learning_rate": 7.5253188878270914e-06,
      "loss": 2.1224,
      "step": 32840
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44140625,
      "learning_rate": 7.516379566153369e-06,
      "loss": 2.1551,
      "step": 32850
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.41015625,
      "learning_rate": 7.507443781797759e-06,
      "loss": 2.12,
      "step": 32860
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.427734375,
      "learning_rate": 7.498511538983971e-06,
      "loss": 2.1026,
      "step": 32870
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4296875,
      "learning_rate": 7.4895828419340505e-06,
      "loss": 2.1311,
      "step": 32880
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.50390625,
      "learning_rate": 7.480657694868378e-06,
      "loss": 2.1483,
      "step": 32890
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.427734375,
      "learning_rate": 7.471736102005631e-06,
      "loss": 2.0904,
      "step": 32900
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.478515625,
      "learning_rate": 7.462818067562832e-06,
      "loss": 2.1419,
      "step": 32910
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44140625,
      "learning_rate": 7.4539035957553e-06,
      "loss": 2.1459,
      "step": 32920
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44140625,
      "learning_rate": 7.444992690796691e-06,
      "loss": 2.1515,
      "step": 32930
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4296875,
      "learning_rate": 7.4360853568989515e-06,
      "loss": 2.086,
      "step": 32940
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.443359375,
      "learning_rate": 7.427181598272361e-06,
      "loss": 2.1231,
      "step": 32950
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.451171875,
      "learning_rate": 7.418281419125507e-06,
      "loss": 2.0773,
      "step": 32960
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41015625,
      "learning_rate": 7.4093848236652676e-06,
      "loss": 2.0991,
      "step": 32970
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4296875,
      "learning_rate": 7.400491816096841e-06,
      "loss": 2.106,
      "step": 32980
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4609375,
      "learning_rate": 7.391602400623736e-06,
      "loss": 2.1078,
      "step": 32990
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.447265625,
      "learning_rate": 7.382716581447743e-06,
      "loss": 2.1172,
      "step": 33000
    },
    {
      "epoch": 0.69,
      "eval_accuracy": 0.5589935681017275,
      "eval_loss": 1.9931697845458984,
      "eval_runtime": 16.4436,
      "eval_samples_per_second": 36.184,
      "eval_steps_per_second": 1.155,
      "step": 33000
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.439453125,
      "learning_rate": 7.373834362768968e-06,
      "loss": 2.1184,
      "step": 33010
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.435546875,
      "learning_rate": 7.364955748785817e-06,
      "loss": 2.1333,
      "step": 33020
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4375,
      "learning_rate": 7.356080743694976e-06,
      "loss": 2.1396,
      "step": 33030
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.431640625,
      "learning_rate": 7.347209351691439e-06,
      "loss": 2.1666,
      "step": 33040
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.42578125,
      "learning_rate": 7.338341576968498e-06,
      "loss": 2.1363,
      "step": 33050
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.431640625,
      "learning_rate": 7.329477423717712e-06,
      "loss": 2.1047,
      "step": 33060
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4375,
      "learning_rate": 7.3206168961289465e-06,
      "loss": 2.1478,
      "step": 33070
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.43359375,
      "learning_rate": 7.3117599983903585e-06,
      "loss": 2.1274,
      "step": 33080
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.427734375,
      "learning_rate": 7.302906734688363e-06,
      "loss": 2.1379,
      "step": 33090
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.423828125,
      "learning_rate": 7.294057109207684e-06,
      "loss": 2.0885,
      "step": 33100
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.482421875,
      "learning_rate": 7.285211126131313e-06,
      "loss": 2.1811,
      "step": 33110
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.455078125,
      "learning_rate": 7.276368789640529e-06,
      "loss": 2.1439,
      "step": 33120
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.431640625,
      "learning_rate": 7.26753010391487e-06,
      "loss": 2.0994,
      "step": 33130
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.439453125,
      "learning_rate": 7.258695073132171e-06,
      "loss": 2.128,
      "step": 33140
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4609375,
      "learning_rate": 7.2498637014685144e-06,
      "loss": 2.1156,
      "step": 33150
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.455078125,
      "learning_rate": 7.241035993098279e-06,
      "loss": 2.1207,
      "step": 33160
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.43359375,
      "learning_rate": 7.232211952194086e-06,
      "loss": 2.0954,
      "step": 33170
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.443359375,
      "learning_rate": 7.22339158292685e-06,
      "loss": 2.1122,
      "step": 33180
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4375,
      "learning_rate": 7.2145748894657235e-06,
      "loss": 2.1424,
      "step": 33190
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4453125,
      "learning_rate": 7.205761875978144e-06,
      "loss": 2.1173,
      "step": 33200
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.431640625,
      "learning_rate": 7.196952546629799e-06,
      "loss": 2.1047,
      "step": 33210
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.443359375,
      "learning_rate": 7.18814690558463e-06,
      "loss": 2.1304,
      "step": 33220
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.447265625,
      "learning_rate": 7.179344957004843e-06,
      "loss": 2.1111,
      "step": 33230
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4453125,
      "learning_rate": 7.170546705050905e-06,
      "loss": 2.1318,
      "step": 33240
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.431640625,
      "learning_rate": 7.161752153881514e-06,
      "loss": 2.135,
      "step": 33250
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4375,
      "learning_rate": 7.152961307653638e-06,
      "loss": 2.1171,
      "step": 33260
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.43359375,
      "learning_rate": 7.144174170522494e-06,
      "loss": 2.1379,
      "step": 33270
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.455078125,
      "learning_rate": 7.135390746641527e-06,
      "loss": 2.122,
      "step": 33280
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4453125,
      "learning_rate": 7.126611040162446e-06,
      "loss": 2.1015,
      "step": 33290
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.447265625,
      "learning_rate": 7.117835055235195e-06,
      "loss": 2.1353,
      "step": 33300
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.412109375,
      "learning_rate": 7.109062796007967e-06,
      "loss": 2.0953,
      "step": 33310
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.578125,
      "learning_rate": 7.100294266627175e-06,
      "loss": 2.0993,
      "step": 33320
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.490234375,
      "learning_rate": 7.091529471237487e-06,
      "loss": 2.1095,
      "step": 33330
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.423828125,
      "learning_rate": 7.082768413981808e-06,
      "loss": 2.1414,
      "step": 33340
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4453125,
      "learning_rate": 7.074011099001252e-06,
      "loss": 2.1345,
      "step": 33350
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.439453125,
      "learning_rate": 7.065257530435188e-06,
      "loss": 2.1595,
      "step": 33360
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4453125,
      "learning_rate": 7.056507712421214e-06,
      "loss": 2.1433,
      "step": 33370
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.431640625,
      "learning_rate": 7.047761649095135e-06,
      "loss": 2.1021,
      "step": 33380
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.46484375,
      "learning_rate": 7.039019344591e-06,
      "loss": 2.1227,
      "step": 33390
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.435546875,
      "learning_rate": 7.030280803041081e-06,
      "loss": 2.1248,
      "step": 33400
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.50390625,
      "learning_rate": 7.021546028575854e-06,
      "loss": 2.1323,
      "step": 33410
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.447265625,
      "learning_rate": 7.012815025324036e-06,
      "loss": 2.1227,
      "step": 33420
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.474609375,
      "learning_rate": 7.0040877974125416e-06,
      "loss": 2.1408,
      "step": 33430
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.44140625,
      "learning_rate": 6.995364348966521e-06,
      "loss": 2.0959,
      "step": 33440
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.45703125,
      "learning_rate": 6.986644684109318e-06,
      "loss": 2.1235,
      "step": 33450
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4296875,
      "learning_rate": 6.9779288069625075e-06,
      "loss": 2.1302,
      "step": 33460
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.431640625,
      "learning_rate": 6.969216721645853e-06,
      "loss": 2.0938,
      "step": 33470
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.439453125,
      "learning_rate": 6.960508432277345e-06,
      "loss": 2.1156,
      "step": 33480
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4453125,
      "learning_rate": 6.951803942973166e-06,
      "loss": 2.099,
      "step": 33490
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4296875,
      "learning_rate": 6.943103257847719e-06,
      "loss": 2.1166,
      "step": 33500
    },
    {
      "epoch": 0.7,
      "eval_accuracy": 0.5592005717242909,
      "eval_loss": 1.9930750131607056,
      "eval_runtime": 16.4841,
      "eval_samples_per_second": 36.095,
      "eval_steps_per_second": 1.153,
      "step": 33500
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.44921875,
      "learning_rate": 6.934406381013584e-06,
      "loss": 2.1251,
      "step": 33510
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.41796875,
      "learning_rate": 6.925713316581562e-06,
      "loss": 2.079,
      "step": 33520
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.447265625,
      "learning_rate": 6.917024068660649e-06,
      "loss": 2.1075,
      "step": 33530
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.455078125,
      "learning_rate": 6.908338641358021e-06,
      "loss": 2.112,
      "step": 33540
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.435546875,
      "learning_rate": 6.899657038779067e-06,
      "loss": 2.0905,
      "step": 33550
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.43359375,
      "learning_rate": 6.890979265027365e-06,
      "loss": 2.1401,
      "step": 33560
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.46875,
      "learning_rate": 6.882305324204669e-06,
      "loss": 2.11,
      "step": 33570
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.453125,
      "learning_rate": 6.873635220410938e-06,
      "loss": 2.1147,
      "step": 33580
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42578125,
      "learning_rate": 6.864968957744314e-06,
      "loss": 2.1247,
      "step": 33590
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4609375,
      "learning_rate": 6.85630654030111e-06,
      "loss": 2.1491,
      "step": 33600
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.435546875,
      "learning_rate": 6.8476479721758375e-06,
      "loss": 2.1205,
      "step": 33610
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.515625,
      "learning_rate": 6.838993257461189e-06,
      "loss": 2.1041,
      "step": 33620
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4375,
      "learning_rate": 6.830342400248017e-06,
      "loss": 2.1365,
      "step": 33630
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4375,
      "learning_rate": 6.821695404625369e-06,
      "loss": 2.1081,
      "step": 33640
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.447265625,
      "learning_rate": 6.813052274680465e-06,
      "loss": 2.0924,
      "step": 33650
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.48046875,
      "learning_rate": 6.804413014498685e-06,
      "loss": 2.0904,
      "step": 33660
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.41015625,
      "learning_rate": 6.795777628163599e-06,
      "loss": 2.145,
      "step": 33670
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.431640625,
      "learning_rate": 6.787146119756926e-06,
      "loss": 2.0912,
      "step": 33680
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.431640625,
      "learning_rate": 6.778518493358571e-06,
      "loss": 2.1224,
      "step": 33690
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.427734375,
      "learning_rate": 6.7698947530465835e-06,
      "loss": 2.1197,
      "step": 33700
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.439453125,
      "learning_rate": 6.761274902897194e-06,
      "loss": 2.0811,
      "step": 33710
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.439453125,
      "learning_rate": 6.752658946984793e-06,
      "loss": 2.1076,
      "step": 33720
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42578125,
      "learning_rate": 6.744046889381914e-06,
      "loss": 2.1044,
      "step": 33730
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.43359375,
      "learning_rate": 6.735438734159263e-06,
      "loss": 2.1192,
      "step": 33740
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.443359375,
      "learning_rate": 6.726834485385706e-06,
      "loss": 2.1285,
      "step": 33750
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.443359375,
      "learning_rate": 6.71823414712824e-06,
      "loss": 2.097,
      "step": 33760
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4375,
      "learning_rate": 6.709637723452033e-06,
      "loss": 2.1062,
      "step": 33770
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4375,
      "learning_rate": 6.701045218420403e-06,
      "loss": 2.0759,
      "step": 33780
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.439453125,
      "learning_rate": 6.692456636094799e-06,
      "loss": 2.1428,
      "step": 33790
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4453125,
      "learning_rate": 6.683871980534833e-06,
      "loss": 2.1225,
      "step": 33800
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42578125,
      "learning_rate": 6.675291255798259e-06,
      "loss": 2.1032,
      "step": 33810
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.443359375,
      "learning_rate": 6.6667144659409565e-06,
      "loss": 2.1256,
      "step": 33820
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4296875,
      "learning_rate": 6.658141615016966e-06,
      "loss": 2.1025,
      "step": 33830
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.427734375,
      "learning_rate": 6.649572707078458e-06,
      "loss": 2.1004,
      "step": 33840
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4296875,
      "learning_rate": 6.641007746175733e-06,
      "loss": 2.1044,
      "step": 33850
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.416015625,
      "learning_rate": 6.632446736357233e-06,
      "loss": 2.1147,
      "step": 33860
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.45703125,
      "learning_rate": 6.623889681669531e-06,
      "loss": 2.096,
      "step": 33870
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.427734375,
      "learning_rate": 6.6153365861573384e-06,
      "loss": 2.1515,
      "step": 33880
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.443359375,
      "learning_rate": 6.606787453863472e-06,
      "loss": 2.0919,
      "step": 33890
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4375,
      "learning_rate": 6.598242288828906e-06,
      "loss": 2.1363,
      "step": 33900
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.455078125,
      "learning_rate": 6.589701095092712e-06,
      "loss": 2.0989,
      "step": 33910
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.462890625,
      "learning_rate": 6.581163876692103e-06,
      "loss": 2.1017,
      "step": 33920
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.41796875,
      "learning_rate": 6.5726306376624e-06,
      "loss": 2.0953,
      "step": 33930
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.421875,
      "learning_rate": 6.564101382037057e-06,
      "loss": 2.1325,
      "step": 33940
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.447265625,
      "learning_rate": 6.5555761138476306e-06,
      "loss": 2.0983,
      "step": 33950
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.412109375,
      "learning_rate": 6.5470548371238006e-06,
      "loss": 2.1065,
      "step": 33960
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.443359375,
      "learning_rate": 6.538537555893365e-06,
      "loss": 2.0793,
      "step": 33970
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4375,
      "learning_rate": 6.530024274182218e-06,
      "loss": 2.0884,
      "step": 33980
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.43359375,
      "learning_rate": 6.521514996014378e-06,
      "loss": 2.1178,
      "step": 33990
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.43359375,
      "learning_rate": 6.513009725411971e-06,
      "loss": 2.1054,
      "step": 34000
    },
    {
      "epoch": 0.71,
      "eval_accuracy": 0.559082283939969,
      "eval_loss": 1.9931409358978271,
      "eval_runtime": 16.4535,
      "eval_samples_per_second": 36.162,
      "eval_steps_per_second": 1.155,
      "step": 34000
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.462890625,
      "learning_rate": 6.504508466395211e-06,
      "loss": 2.1161,
      "step": 34010
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.462890625,
      "learning_rate": 6.496011222982434e-06,
      "loss": 2.1408,
      "step": 34020
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.42578125,
      "learning_rate": 6.487517999190078e-06,
      "loss": 2.1301,
      "step": 34030
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.427734375,
      "learning_rate": 6.479028799032664e-06,
      "loss": 2.1109,
      "step": 34040
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.421875,
      "learning_rate": 6.47054362652283e-06,
      "loss": 2.1177,
      "step": 34050
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4296875,
      "learning_rate": 6.462062485671299e-06,
      "loss": 2.1432,
      "step": 34060
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.455078125,
      "learning_rate": 6.453585380486896e-06,
      "loss": 2.1276,
      "step": 34070
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.44140625,
      "learning_rate": 6.445112314976527e-06,
      "loss": 2.1064,
      "step": 34080
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.423828125,
      "learning_rate": 6.4366432931452015e-06,
      "loss": 2.1064,
      "step": 34090
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.453125,
      "learning_rate": 6.428178318996016e-06,
      "loss": 2.1184,
      "step": 34100
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.427734375,
      "learning_rate": 6.419717396530137e-06,
      "loss": 2.0953,
      "step": 34110
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.41796875,
      "learning_rate": 6.411260529746836e-06,
      "loss": 2.124,
      "step": 34120
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.447265625,
      "learning_rate": 6.4028077226434644e-06,
      "loss": 2.1018,
      "step": 34130
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.419921875,
      "learning_rate": 6.3943589792154425e-06,
      "loss": 2.1278,
      "step": 34140
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.43359375,
      "learning_rate": 6.385914303456283e-06,
      "loss": 2.0978,
      "step": 34150
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.470703125,
      "learning_rate": 6.377473699357565e-06,
      "loss": 2.1102,
      "step": 34160
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.439453125,
      "learning_rate": 6.369037170908957e-06,
      "loss": 2.126,
      "step": 34170
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.439453125,
      "learning_rate": 6.360604722098182e-06,
      "loss": 2.0955,
      "step": 34180
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.431640625,
      "learning_rate": 6.352176356911056e-06,
      "loss": 2.1175,
      "step": 34190
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.44140625,
      "learning_rate": 6.343752079331443e-06,
      "loss": 2.1395,
      "step": 34200
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.423828125,
      "learning_rate": 6.335331893341294e-06,
      "loss": 2.1183,
      "step": 34210
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.41015625,
      "learning_rate": 6.326915802920619e-06,
      "loss": 2.1076,
      "step": 34220
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.431640625,
      "learning_rate": 6.318503812047483e-06,
      "loss": 2.1039,
      "step": 34230
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4453125,
      "learning_rate": 6.310095924698026e-06,
      "loss": 2.1221,
      "step": 34240
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.455078125,
      "learning_rate": 6.301692144846444e-06,
      "loss": 2.1192,
      "step": 34250
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.43359375,
      "learning_rate": 6.293292476464997e-06,
      "loss": 2.1157,
      "step": 34260
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.48828125,
      "learning_rate": 6.284896923523983e-06,
      "loss": 2.0826,
      "step": 34270
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.484375,
      "learning_rate": 6.276505489991775e-06,
      "loss": 2.1246,
      "step": 34280
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.41796875,
      "learning_rate": 6.268118179834796e-06,
      "loss": 2.0997,
      "step": 34290
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.44921875,
      "learning_rate": 6.259734997017506e-06,
      "loss": 2.1264,
      "step": 34300
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.43359375,
      "learning_rate": 6.2513559455024275e-06,
      "loss": 2.1287,
      "step": 34310
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.443359375,
      "learning_rate": 6.242981029250131e-06,
      "loss": 2.1016,
      "step": 34320
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.458984375,
      "learning_rate": 6.234610252219217e-06,
      "loss": 2.1278,
      "step": 34330
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.462890625,
      "learning_rate": 6.226243618366347e-06,
      "loss": 2.1235,
      "step": 34340
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.44921875,
      "learning_rate": 6.217881131646221e-06,
      "loss": 2.1028,
      "step": 34350
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.423828125,
      "learning_rate": 6.209522796011566e-06,
      "loss": 2.137,
      "step": 34360
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4453125,
      "learning_rate": 6.20116861541316e-06,
      "loss": 2.0866,
      "step": 34370
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4453125,
      "learning_rate": 6.192818593799818e-06,
      "loss": 2.0927,
      "step": 34380
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.451171875,
      "learning_rate": 6.184472735118374e-06,
      "loss": 2.1024,
      "step": 34390
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4453125,
      "learning_rate": 6.17613104331371e-06,
      "loss": 2.094,
      "step": 34400
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4296875,
      "learning_rate": 6.1677935223287365e-06,
      "loss": 2.1166,
      "step": 34410
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.462890625,
      "learning_rate": 6.159460176104379e-06,
      "loss": 2.1187,
      "step": 34420
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.435546875,
      "learning_rate": 6.15113100857961e-06,
      "loss": 2.1338,
      "step": 34430
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4296875,
      "learning_rate": 6.142806023691406e-06,
      "loss": 2.1129,
      "step": 34440
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.44140625,
      "learning_rate": 6.134485225374787e-06,
      "loss": 2.1129,
      "step": 34450
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4375,
      "learning_rate": 6.126168617562774e-06,
      "loss": 2.1016,
      "step": 34460
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.423828125,
      "learning_rate": 6.117856204186423e-06,
      "loss": 2.1158,
      "step": 34470
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.43359375,
      "learning_rate": 6.109547989174806e-06,
      "loss": 2.1232,
      "step": 34480
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4375,
      "learning_rate": 6.101243976454995e-06,
      "loss": 2.1365,
      "step": 34490
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.44921875,
      "learning_rate": 6.092944169952092e-06,
      "loss": 2.0972,
      "step": 34500
    },
    {
      "epoch": 0.72,
      "eval_accuracy": 0.5589804250145806,
      "eval_loss": 1.9931070804595947,
      "eval_runtime": 16.4414,
      "eval_samples_per_second": 36.189,
      "eval_steps_per_second": 1.156,
      "step": 34500
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.427734375,
      "learning_rate": 6.084648573589212e-06,
      "loss": 2.1009,
      "step": 34510
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.44140625,
      "learning_rate": 6.076357191287463e-06,
      "loss": 2.1011,
      "step": 34520
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.455078125,
      "learning_rate": 6.068070026965976e-06,
      "loss": 2.1411,
      "step": 34530
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.447265625,
      "learning_rate": 6.0597870845418895e-06,
      "loss": 2.1089,
      "step": 34540
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4453125,
      "learning_rate": 6.051508367930332e-06,
      "loss": 2.1593,
      "step": 34550
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.423828125,
      "learning_rate": 6.043233881044445e-06,
      "loss": 2.1074,
      "step": 34560
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.427734375,
      "learning_rate": 6.0349636277953785e-06,
      "loss": 2.1371,
      "step": 34570
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.453125,
      "learning_rate": 6.026697612092259e-06,
      "loss": 2.0958,
      "step": 34580
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.423828125,
      "learning_rate": 6.0184358378422296e-06,
      "loss": 2.149,
      "step": 34590
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.439453125,
      "learning_rate": 6.010178308950428e-06,
      "loss": 2.1108,
      "step": 34600
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.443359375,
      "learning_rate": 6.001925029319966e-06,
      "loss": 2.1496,
      "step": 34610
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.443359375,
      "learning_rate": 5.993676002851967e-06,
      "loss": 2.1262,
      "step": 34620
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.447265625,
      "learning_rate": 5.985431233445538e-06,
      "loss": 2.1116,
      "step": 34630
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.498046875,
      "learning_rate": 5.977190724997777e-06,
      "loss": 2.0973,
      "step": 34640
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.439453125,
      "learning_rate": 5.968954481403753e-06,
      "loss": 2.1023,
      "step": 34650
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4609375,
      "learning_rate": 5.960722506556543e-06,
      "loss": 2.1212,
      "step": 34660
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.43359375,
      "learning_rate": 5.95249480434718e-06,
      "loss": 2.1311,
      "step": 34670
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.435546875,
      "learning_rate": 5.944271378664703e-06,
      "loss": 2.1415,
      "step": 34680
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.435546875,
      "learning_rate": 5.936052233396106e-06,
      "loss": 2.0964,
      "step": 34690
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.439453125,
      "learning_rate": 5.927837372426381e-06,
      "loss": 2.1169,
      "step": 34700
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4453125,
      "learning_rate": 5.919626799638477e-06,
      "loss": 2.0898,
      "step": 34710
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.443359375,
      "learning_rate": 5.911420518913327e-06,
      "loss": 2.1329,
      "step": 34720
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4453125,
      "learning_rate": 5.903218534129837e-06,
      "loss": 2.1129,
      "step": 34730
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.423828125,
      "learning_rate": 5.89502084916487e-06,
      "loss": 2.1183,
      "step": 34740
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.419921875,
      "learning_rate": 5.886827467893268e-06,
      "loss": 2.1374,
      "step": 34750
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.44140625,
      "learning_rate": 5.87863839418784e-06,
      "loss": 2.1091,
      "step": 34760
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.427734375,
      "learning_rate": 5.870453631919345e-06,
      "loss": 2.1035,
      "step": 34770
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.431640625,
      "learning_rate": 5.862273184956515e-06,
      "loss": 2.1388,
      "step": 34780
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.494140625,
      "learning_rate": 5.854097057166051e-06,
      "loss": 2.1034,
      "step": 34790
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.474609375,
      "learning_rate": 5.845925252412588e-06,
      "loss": 2.1206,
      "step": 34800
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.423828125,
      "learning_rate": 5.837757774558735e-06,
      "loss": 2.1278,
      "step": 34810
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.51171875,
      "learning_rate": 5.829594627465056e-06,
      "loss": 2.092,
      "step": 34820
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.453125,
      "learning_rate": 5.821435814990067e-06,
      "loss": 2.1259,
      "step": 34830
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.419921875,
      "learning_rate": 5.8132813409902226e-06,
      "loss": 2.1105,
      "step": 34840
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.435546875,
      "learning_rate": 5.805131209319941e-06,
      "loss": 2.1231,
      "step": 34850
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.44921875,
      "learning_rate": 5.796985423831591e-06,
      "loss": 2.1064,
      "step": 34860
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.451171875,
      "learning_rate": 5.7888439883754646e-06,
      "loss": 2.1156,
      "step": 34870
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.455078125,
      "learning_rate": 5.780706906799822e-06,
      "loss": 2.1169,
      "step": 34880
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4375,
      "learning_rate": 5.772574182950857e-06,
      "loss": 2.1124,
      "step": 34890
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.42578125,
      "learning_rate": 5.764445820672695e-06,
      "loss": 2.1061,
      "step": 34900
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.494140625,
      "learning_rate": 5.756321823807419e-06,
      "loss": 2.1136,
      "step": 34910
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.431640625,
      "learning_rate": 5.748202196195022e-06,
      "loss": 2.1088,
      "step": 34920
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4296875,
      "learning_rate": 5.74008694167346e-06,
      "loss": 2.1101,
      "step": 34930
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4296875,
      "learning_rate": 5.731976064078597e-06,
      "loss": 2.1189,
      "step": 34940
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.453125,
      "learning_rate": 5.723869567244252e-06,
      "loss": 2.1191,
      "step": 34950
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4453125,
      "learning_rate": 5.715767455002149e-06,
      "loss": 2.0806,
      "step": 34960
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.48046875,
      "learning_rate": 5.707669731181955e-06,
      "loss": 2.1283,
      "step": 34970
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.44140625,
      "learning_rate": 5.699576399611267e-06,
      "loss": 2.1105,
      "step": 34980
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.46484375,
      "learning_rate": 5.6914874641155886e-06,
      "loss": 2.0672,
      "step": 34990
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.447265625,
      "learning_rate": 5.683402928518359e-06,
      "loss": 2.1228,
      "step": 35000
    },
    {
      "epoch": 0.73,
      "eval_accuracy": 0.5590444975644216,
      "eval_loss": 1.993085265159607,
      "eval_runtime": 16.4517,
      "eval_samples_per_second": 36.166,
      "eval_steps_per_second": 1.155,
      "step": 35000
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.439453125,
      "learning_rate": 5.675322796640932e-06,
      "loss": 2.0986,
      "step": 35010
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.439453125,
      "learning_rate": 5.667247072302589e-06,
      "loss": 2.1006,
      "step": 35020
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4453125,
      "learning_rate": 5.659175759320509e-06,
      "loss": 2.1034,
      "step": 35030
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.498046875,
      "learning_rate": 5.6511088615098036e-06,
      "loss": 2.1207,
      "step": 35040
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4453125,
      "learning_rate": 5.643046382683497e-06,
      "loss": 2.126,
      "step": 35050
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.466796875,
      "learning_rate": 5.634988326652507e-06,
      "loss": 2.1249,
      "step": 35060
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.48828125,
      "learning_rate": 5.626934697225679e-06,
      "loss": 2.0962,
      "step": 35070
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4375,
      "learning_rate": 5.6188854982097646e-06,
      "loss": 2.1108,
      "step": 35080
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4453125,
      "learning_rate": 5.6108407334094074e-06,
      "loss": 2.1243,
      "step": 35090
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.427734375,
      "learning_rate": 5.6028004066271686e-06,
      "loss": 2.1265,
      "step": 35100
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4296875,
      "learning_rate": 5.594764521663512e-06,
      "loss": 2.1318,
      "step": 35110
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.421875,
      "learning_rate": 5.586733082316788e-06,
      "loss": 2.1226,
      "step": 35120
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.45703125,
      "learning_rate": 5.57870609238326e-06,
      "loss": 2.125,
      "step": 35130
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.439453125,
      "learning_rate": 5.570683555657087e-06,
      "loss": 2.1059,
      "step": 35140
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4375,
      "learning_rate": 5.5626654759303085e-06,
      "loss": 2.1077,
      "step": 35150
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.423828125,
      "learning_rate": 5.554651856992882e-06,
      "loss": 2.1458,
      "step": 35160
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.435546875,
      "learning_rate": 5.546642702632627e-06,
      "loss": 2.1169,
      "step": 35170
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.427734375,
      "learning_rate": 5.538638016635284e-06,
      "loss": 2.1383,
      "step": 35180
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.44921875,
      "learning_rate": 5.530637802784449e-06,
      "loss": 2.1674,
      "step": 35190
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.435546875,
      "learning_rate": 5.522642064861633e-06,
      "loss": 2.1128,
      "step": 35200
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4375,
      "learning_rate": 5.514650806646218e-06,
      "loss": 2.1206,
      "step": 35210
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.43359375,
      "learning_rate": 5.506664031915462e-06,
      "loss": 2.1029,
      "step": 35220
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.44140625,
      "learning_rate": 5.498681744444518e-06,
      "loss": 2.1,
      "step": 35230
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.57421875,
      "learning_rate": 5.490703948006415e-06,
      "loss": 2.1042,
      "step": 35240
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.43359375,
      "learning_rate": 5.482730646372044e-06,
      "loss": 2.101,
      "step": 35250
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.42578125,
      "learning_rate": 5.47476184331019e-06,
      "loss": 2.0874,
      "step": 35260
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.447265625,
      "learning_rate": 5.466797542587509e-06,
      "loss": 2.1039,
      "step": 35270
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4453125,
      "learning_rate": 5.458837747968516e-06,
      "loss": 2.1162,
      "step": 35280
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.439453125,
      "learning_rate": 5.450882463215607e-06,
      "loss": 2.1125,
      "step": 35290
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.43359375,
      "learning_rate": 5.442931692089051e-06,
      "loss": 2.1358,
      "step": 35300
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.44140625,
      "learning_rate": 5.434985438346966e-06,
      "loss": 2.1118,
      "step": 35310
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.423828125,
      "learning_rate": 5.427043705745348e-06,
      "loss": 2.1036,
      "step": 35320
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.451171875,
      "learning_rate": 5.419106498038063e-06,
      "loss": 2.1052,
      "step": 35330
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4140625,
      "learning_rate": 5.4111738189768124e-06,
      "loss": 2.1178,
      "step": 35340
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.44140625,
      "learning_rate": 5.403245672311182e-06,
      "loss": 2.1325,
      "step": 35350
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.455078125,
      "learning_rate": 5.39532206178861e-06,
      "loss": 2.142,
      "step": 35360
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.458984375,
      "learning_rate": 5.387402991154377e-06,
      "loss": 2.1371,
      "step": 35370
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.435546875,
      "learning_rate": 5.379488464151632e-06,
      "loss": 2.1593,
      "step": 35380
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.53515625,
      "learning_rate": 5.3715784845213715e-06,
      "loss": 2.1418,
      "step": 35390
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.443359375,
      "learning_rate": 5.3636730560024484e-06,
      "loss": 2.132,
      "step": 35400
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.431640625,
      "learning_rate": 5.35577218233155e-06,
      "loss": 2.1306,
      "step": 35410
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.46484375,
      "learning_rate": 5.347875867243227e-06,
      "loss": 2.1581,
      "step": 35420
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.439453125,
      "learning_rate": 5.339984114469859e-06,
      "loss": 2.1423,
      "step": 35430
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.451171875,
      "learning_rate": 5.332096927741687e-06,
      "loss": 2.109,
      "step": 35440
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.58984375,
      "learning_rate": 5.324214310786777e-06,
      "loss": 2.1108,
      "step": 35450
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.427734375,
      "learning_rate": 5.316336267331051e-06,
      "loss": 2.1136,
      "step": 35460
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.462890625,
      "learning_rate": 5.308462801098251e-06,
      "loss": 2.1321,
      "step": 35470
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.451171875,
      "learning_rate": 5.30059391580997e-06,
      "loss": 2.1358,
      "step": 35480
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.423828125,
      "learning_rate": 5.292729615185637e-06,
      "loss": 2.15,
      "step": 35490
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.43359375,
      "learning_rate": 5.284869902942497e-06,
      "loss": 2.1231,
      "step": 35500
    },
    {
      "epoch": 0.74,
      "eval_accuracy": 0.5592170005832245,
      "eval_loss": 1.9931291341781616,
      "eval_runtime": 16.439,
      "eval_samples_per_second": 36.194,
      "eval_steps_per_second": 1.156,
      "step": 35500
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.421875,
      "learning_rate": 5.277014782795643e-06,
      "loss": 2.1469,
      "step": 35510
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4453125,
      "learning_rate": 5.269164258457997e-06,
      "loss": 2.1019,
      "step": 35520
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4453125,
      "learning_rate": 5.261318333640291e-06,
      "loss": 2.1239,
      "step": 35530
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.44140625,
      "learning_rate": 5.253477012051105e-06,
      "loss": 2.1449,
      "step": 35540
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.423828125,
      "learning_rate": 5.245640297396832e-06,
      "loss": 2.1256,
      "step": 35550
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.42578125,
      "learning_rate": 5.2378081933816845e-06,
      "loss": 2.0897,
      "step": 35560
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4375,
      "learning_rate": 5.229980703707701e-06,
      "loss": 2.1485,
      "step": 35570
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.44921875,
      "learning_rate": 5.222157832074741e-06,
      "loss": 2.111,
      "step": 35580
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4375,
      "learning_rate": 5.214339582180481e-06,
      "loss": 2.0669,
      "step": 35590
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.431640625,
      "learning_rate": 5.206525957720399e-06,
      "loss": 2.1298,
      "step": 35600
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4609375,
      "learning_rate": 5.198716962387806e-06,
      "loss": 2.1181,
      "step": 35610
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.458984375,
      "learning_rate": 5.190912599873818e-06,
      "loss": 2.1329,
      "step": 35620
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.43359375,
      "learning_rate": 5.183112873867353e-06,
      "loss": 2.1173,
      "step": 35630
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.42578125,
      "learning_rate": 5.175317788055149e-06,
      "loss": 2.1174,
      "step": 35640
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.435546875,
      "learning_rate": 5.167527346121747e-06,
      "loss": 2.1275,
      "step": 35650
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4375,
      "learning_rate": 5.159741551749487e-06,
      "loss": 2.077,
      "step": 35660
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4375,
      "learning_rate": 5.151960408618524e-06,
      "loss": 2.0951,
      "step": 35670
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.435546875,
      "learning_rate": 5.144183920406798e-06,
      "loss": 2.0967,
      "step": 35680
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.431640625,
      "learning_rate": 5.1364120907900675e-06,
      "loss": 2.1244,
      "step": 35690
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.453125,
      "learning_rate": 5.128644923441872e-06,
      "loss": 2.1489,
      "step": 35700
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.451171875,
      "learning_rate": 5.120882422033562e-06,
      "loss": 2.1294,
      "step": 35710
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.443359375,
      "learning_rate": 5.11312459023427e-06,
      "loss": 2.1346,
      "step": 35720
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.439453125,
      "learning_rate": 5.105371431710929e-06,
      "loss": 2.1249,
      "step": 35730
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.435546875,
      "learning_rate": 5.097622950128265e-06,
      "loss": 2.1315,
      "step": 35740
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4375,
      "learning_rate": 5.089879149148781e-06,
      "loss": 2.124,
      "step": 35750
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4296875,
      "learning_rate": 5.082140032432781e-06,
      "loss": 2.108,
      "step": 35760
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.419921875,
      "learning_rate": 5.074405603638349e-06,
      "loss": 2.1048,
      "step": 35770
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.48046875,
      "learning_rate": 5.0666758664213595e-06,
      "loss": 2.1274,
      "step": 35780
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4296875,
      "learning_rate": 5.058950824435452e-06,
      "loss": 2.1327,
      "step": 35790
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.419921875,
      "learning_rate": 5.051230481332065e-06,
      "loss": 2.112,
      "step": 35800
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.439453125,
      "learning_rate": 5.043514840760415e-06,
      "loss": 2.1442,
      "step": 35810
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4296875,
      "learning_rate": 5.035803906367478e-06,
      "loss": 2.1296,
      "step": 35820
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4453125,
      "learning_rate": 5.028097681798026e-06,
      "loss": 2.1406,
      "step": 35830
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4296875,
      "learning_rate": 5.020396170694598e-06,
      "loss": 2.0958,
      "step": 35840
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.455078125,
      "learning_rate": 5.012699376697496e-06,
      "loss": 2.1102,
      "step": 35850
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.439453125,
      "learning_rate": 5.005007303444804e-06,
      "loss": 2.1232,
      "step": 35860
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.443359375,
      "learning_rate": 4.9973199545723755e-06,
      "loss": 2.1201,
      "step": 35870
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.451171875,
      "learning_rate": 4.989637333713814e-06,
      "loss": 2.0996,
      "step": 35880
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.423828125,
      "learning_rate": 4.981959444500509e-06,
      "loss": 2.1139,
      "step": 35890
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.486328125,
      "learning_rate": 4.974286290561606e-06,
      "loss": 2.0859,
      "step": 35900
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.423828125,
      "learning_rate": 4.966617875524003e-06,
      "loss": 2.1293,
      "step": 35910
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.43359375,
      "learning_rate": 4.958954203012375e-06,
      "loss": 2.1068,
      "step": 35920
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.44921875,
      "learning_rate": 4.9512952766491344e-06,
      "loss": 2.1348,
      "step": 35930
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.451171875,
      "learning_rate": 4.943641100054476e-06,
      "loss": 2.1355,
      "step": 35940
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.40625,
      "learning_rate": 4.935991676846321e-06,
      "loss": 2.0624,
      "step": 35950
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.470703125,
      "learning_rate": 4.928347010640368e-06,
      "loss": 2.1113,
      "step": 35960
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.44140625,
      "learning_rate": 4.920707105050059e-06,
      "loss": 2.1175,
      "step": 35970
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.44921875,
      "learning_rate": 4.913071963686575e-06,
      "loss": 2.1449,
      "step": 35980
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.421875,
      "learning_rate": 4.905441590158859e-06,
      "loss": 2.1075,
      "step": 35990
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4453125,
      "learning_rate": 4.8978159880736e-06,
      "loss": 2.0974,
      "step": 36000
    },
    {
      "epoch": 0.75,
      "eval_accuracy": 0.5590329973631681,
      "eval_loss": 1.9931119680404663,
      "eval_runtime": 16.4794,
      "eval_samples_per_second": 36.106,
      "eval_steps_per_second": 1.153,
      "step": 36000
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.431640625,
      "learning_rate": 4.89019516103522e-06,
      "loss": 2.1181,
      "step": 36010
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.431640625,
      "learning_rate": 4.882579112645893e-06,
      "loss": 2.1242,
      "step": 36020
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5,
      "learning_rate": 4.874967846505539e-06,
      "loss": 2.1335,
      "step": 36030
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4375,
      "learning_rate": 4.8673613662118015e-06,
      "loss": 2.123,
      "step": 36040
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.439453125,
      "learning_rate": 4.8597596753600745e-06,
      "loss": 2.1119,
      "step": 36050
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.421875,
      "learning_rate": 4.852162777543491e-06,
      "loss": 2.1142,
      "step": 36060
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.51953125,
      "learning_rate": 4.844570676352903e-06,
      "loss": 2.1036,
      "step": 36070
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.453125,
      "learning_rate": 4.83698337537691e-06,
      "loss": 2.1171,
      "step": 36080
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.474609375,
      "learning_rate": 4.82940087820184e-06,
      "loss": 2.1225,
      "step": 36090
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.443359375,
      "learning_rate": 4.821823188411741e-06,
      "loss": 2.1206,
      "step": 36100
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.43359375,
      "learning_rate": 4.814250309588398e-06,
      "loss": 2.1136,
      "step": 36110
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.435546875,
      "learning_rate": 4.806682245311324e-06,
      "loss": 2.117,
      "step": 36120
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5078125,
      "learning_rate": 4.799118999157744e-06,
      "loss": 2.122,
      "step": 36130
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.43359375,
      "learning_rate": 4.791560574702614e-06,
      "loss": 2.1155,
      "step": 36140
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4375,
      "learning_rate": 4.784006975518619e-06,
      "loss": 2.0965,
      "step": 36150
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.447265625,
      "learning_rate": 4.776458205176144e-06,
      "loss": 2.1077,
      "step": 36160
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.431640625,
      "learning_rate": 4.768914267243309e-06,
      "loss": 2.1266,
      "step": 36170
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.494140625,
      "learning_rate": 4.761375165285936e-06,
      "loss": 2.1078,
      "step": 36180
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.458984375,
      "learning_rate": 4.753840902867575e-06,
      "loss": 2.1109,
      "step": 36190
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.439453125,
      "learning_rate": 4.746311483549472e-06,
      "loss": 2.1165,
      "step": 36200
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.451171875,
      "learning_rate": 4.7387869108906005e-06,
      "loss": 2.0883,
      "step": 36210
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.43359375,
      "learning_rate": 4.731267188447637e-06,
      "loss": 2.1408,
      "step": 36220
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.443359375,
      "learning_rate": 4.723752319774956e-06,
      "loss": 2.1222,
      "step": 36230
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.435546875,
      "learning_rate": 4.716242308424652e-06,
      "loss": 2.1161,
      "step": 36240
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.435546875,
      "learning_rate": 4.708737157946521e-06,
      "loss": 2.1181,
      "step": 36250
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.439453125,
      "learning_rate": 4.7012368718880476e-06,
      "loss": 2.1157,
      "step": 36260
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.42578125,
      "learning_rate": 4.693741453794433e-06,
      "loss": 2.1257,
      "step": 36270
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.458984375,
      "learning_rate": 4.686250907208575e-06,
      "loss": 2.1204,
      "step": 36280
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.427734375,
      "learning_rate": 4.6787652356710595e-06,
      "loss": 2.1552,
      "step": 36290
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.458984375,
      "learning_rate": 4.671284442720174e-06,
      "loss": 2.137,
      "step": 36300
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.46875,
      "learning_rate": 4.6638085318919094e-06,
      "loss": 2.055,
      "step": 36310
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4453125,
      "learning_rate": 4.656337506719929e-06,
      "loss": 2.1262,
      "step": 36320
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.455078125,
      "learning_rate": 4.6488713707356e-06,
      "loss": 2.1164,
      "step": 36330
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.44140625,
      "learning_rate": 4.6414101274679825e-06,
      "loss": 2.1551,
      "step": 36340
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.443359375,
      "learning_rate": 4.633953780443807e-06,
      "loss": 2.108,
      "step": 36350
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.427734375,
      "learning_rate": 4.626502333187506e-06,
      "loss": 2.1207,
      "step": 36360
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.482421875,
      "learning_rate": 4.619055789221191e-06,
      "loss": 2.1213,
      "step": 36370
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.435546875,
      "learning_rate": 4.611614152064658e-06,
      "loss": 2.1274,
      "step": 36380
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4296875,
      "learning_rate": 4.6041774252353715e-06,
      "loss": 2.0951,
      "step": 36390
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4375,
      "learning_rate": 4.596745612248488e-06,
      "loss": 2.161,
      "step": 36400
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.42578125,
      "learning_rate": 4.589318716616844e-06,
      "loss": 2.1103,
      "step": 36410
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.439453125,
      "learning_rate": 4.581896741850935e-06,
      "loss": 2.1117,
      "step": 36420
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.43359375,
      "learning_rate": 4.574479691458952e-06,
      "loss": 2.1257,
      "step": 36430
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4921875,
      "learning_rate": 4.567067568946736e-06,
      "loss": 2.1078,
      "step": 36440
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4453125,
      "learning_rate": 4.559660377817818e-06,
      "loss": 2.1118,
      "step": 36450
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.423828125,
      "learning_rate": 4.5522581215733825e-06,
      "loss": 2.1103,
      "step": 36460
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.42578125,
      "learning_rate": 4.544860803712298e-06,
      "loss": 2.1457,
      "step": 36470
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.44921875,
      "learning_rate": 4.537468427731079e-06,
      "loss": 2.109,
      "step": 36480
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.451171875,
      "learning_rate": 4.5300809971239205e-06,
      "loss": 2.1253,
      "step": 36490
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.447265625,
      "learning_rate": 4.522698515382679e-06,
      "loss": 2.1025,
      "step": 36500
    },
    {
      "epoch": 0.76,
      "eval_accuracy": 0.559105284342476,
      "eval_loss": 1.9931056499481201,
      "eval_runtime": 16.462,
      "eval_samples_per_second": 36.144,
      "eval_steps_per_second": 1.154,
      "step": 36500
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.42578125,
      "learning_rate": 4.5153209859968544e-06,
      "loss": 2.1208,
      "step": 36510
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.435546875,
      "learning_rate": 4.507948412453629e-06,
      "loss": 2.106,
      "step": 36520
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.455078125,
      "learning_rate": 4.500580798237831e-06,
      "loss": 2.1105,
      "step": 36530
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.423828125,
      "learning_rate": 4.493218146831938e-06,
      "loss": 2.1221,
      "step": 36540
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.478515625,
      "learning_rate": 4.485860461716093e-06,
      "loss": 2.1242,
      "step": 36550
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.435546875,
      "learning_rate": 4.47850774636809e-06,
      "loss": 2.113,
      "step": 36560
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.486328125,
      "learning_rate": 4.4711600042633746e-06,
      "loss": 2.1126,
      "step": 36570
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.453125,
      "learning_rate": 4.463817238875028e-06,
      "loss": 2.1519,
      "step": 36580
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.462890625,
      "learning_rate": 4.456479453673796e-06,
      "loss": 2.1234,
      "step": 36590
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.453125,
      "learning_rate": 4.449146652128069e-06,
      "loss": 2.1334,
      "step": 36600
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.431640625,
      "learning_rate": 4.4418188377038675e-06,
      "loss": 2.1007,
      "step": 36610
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4453125,
      "learning_rate": 4.4344960138648675e-06,
      "loss": 2.0964,
      "step": 36620
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.447265625,
      "learning_rate": 4.427178184072389e-06,
      "loss": 2.1164,
      "step": 36630
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.470703125,
      "learning_rate": 4.419865351785373e-06,
      "loss": 2.1119,
      "step": 36640
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.423828125,
      "learning_rate": 4.412557520460418e-06,
      "loss": 2.1117,
      "step": 36650
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.419921875,
      "learning_rate": 4.405254693551754e-06,
      "loss": 2.098,
      "step": 36660
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.435546875,
      "learning_rate": 4.397956874511234e-06,
      "loss": 2.1182,
      "step": 36670
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.443359375,
      "learning_rate": 4.3906640667883616e-06,
      "loss": 2.1273,
      "step": 36680
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4375,
      "learning_rate": 4.3833762738302524e-06,
      "loss": 2.089,
      "step": 36690
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4921875,
      "learning_rate": 4.3760934990816715e-06,
      "loss": 2.1264,
      "step": 36700
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.455078125,
      "learning_rate": 4.3688157459849945e-06,
      "loss": 2.1335,
      "step": 36710
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.419921875,
      "learning_rate": 4.361543017980239e-06,
      "loss": 2.0725,
      "step": 36720
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.421875,
      "learning_rate": 4.354275318505031e-06,
      "loss": 2.0855,
      "step": 36730
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.44140625,
      "learning_rate": 4.347012650994631e-06,
      "loss": 2.123,
      "step": 36740
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4453125,
      "learning_rate": 4.339755018881922e-06,
      "loss": 2.0903,
      "step": 36750
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.443359375,
      "learning_rate": 4.332502425597404e-06,
      "loss": 2.1431,
      "step": 36760
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.416015625,
      "learning_rate": 4.325254874569189e-06,
      "loss": 2.1281,
      "step": 36770
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.443359375,
      "learning_rate": 4.318012369223011e-06,
      "loss": 2.1016,
      "step": 36780
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.427734375,
      "learning_rate": 4.310774912982227e-06,
      "loss": 2.1429,
      "step": 36790
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.435546875,
      "learning_rate": 4.30354250926779e-06,
      "loss": 2.1594,
      "step": 36800
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.431640625,
      "learning_rate": 4.296315161498277e-06,
      "loss": 2.1184,
      "step": 36810
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.42578125,
      "learning_rate": 4.289092873089881e-06,
      "loss": 2.1207,
      "step": 36820
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.419921875,
      "learning_rate": 4.281875647456381e-06,
      "loss": 2.0888,
      "step": 36830
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4375,
      "learning_rate": 4.274663488009185e-06,
      "loss": 2.1329,
      "step": 36840
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.466796875,
      "learning_rate": 4.2674563981573015e-06,
      "loss": 2.1339,
      "step": 36850
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.435546875,
      "learning_rate": 4.2602543813073325e-06,
      "loss": 2.1374,
      "step": 36860
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4453125,
      "learning_rate": 4.2530574408634906e-06,
      "loss": 2.0959,
      "step": 36870
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4765625,
      "learning_rate": 4.245865580227594e-06,
      "loss": 2.1113,
      "step": 36880
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.494140625,
      "learning_rate": 4.238678802799043e-06,
      "loss": 2.1188,
      "step": 36890
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.42578125,
      "learning_rate": 4.231497111974853e-06,
      "loss": 2.1132,
      "step": 36900
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4296875,
      "learning_rate": 4.224320511149628e-06,
      "loss": 2.1199,
      "step": 36910
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.44140625,
      "learning_rate": 4.21714900371556e-06,
      "loss": 2.1004,
      "step": 36920
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.45703125,
      "learning_rate": 4.209982593062446e-06,
      "loss": 2.1752,
      "step": 36930
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4453125,
      "learning_rate": 4.202821282577661e-06,
      "loss": 2.1265,
      "step": 36940
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4375,
      "learning_rate": 4.19566507564618e-06,
      "loss": 2.1123,
      "step": 36950
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.44140625,
      "learning_rate": 4.188513975650555e-06,
      "loss": 2.1381,
      "step": 36960
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5078125,
      "learning_rate": 4.181367985970934e-06,
      "loss": 2.11,
      "step": 36970
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.423828125,
      "learning_rate": 4.174227109985052e-06,
      "loss": 2.1224,
      "step": 36980
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4296875,
      "learning_rate": 4.167091351068207e-06,
      "loss": 2.1297,
      "step": 36990
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.443359375,
      "learning_rate": 4.159960712593301e-06,
      "loss": 2.1217,
      "step": 37000
    },
    {
      "epoch": 0.77,
      "eval_accuracy": 0.5589886394440474,
      "eval_loss": 1.9931020736694336,
      "eval_runtime": 16.5352,
      "eval_samples_per_second": 35.984,
      "eval_steps_per_second": 1.149,
      "step": 37000
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4375,
      "learning_rate": 4.152835197930806e-06,
      "loss": 2.0892,
      "step": 37010
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4375,
      "learning_rate": 4.145714810448768e-06,
      "loss": 2.1267,
      "step": 37020
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.451171875,
      "learning_rate": 4.1385995535128175e-06,
      "loss": 2.0758,
      "step": 37030
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.46484375,
      "learning_rate": 4.131489430486159e-06,
      "loss": 2.1344,
      "step": 37040
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.43359375,
      "learning_rate": 4.124384444729561e-06,
      "loss": 2.1235,
      "step": 37050
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.453125,
      "learning_rate": 4.117284599601374e-06,
      "loss": 2.1071,
      "step": 37060
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.443359375,
      "learning_rate": 4.110189898457521e-06,
      "loss": 2.1207,
      "step": 37070
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.427734375,
      "learning_rate": 4.103100344651478e-06,
      "loss": 2.113,
      "step": 37080
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.44921875,
      "learning_rate": 4.0960159415342994e-06,
      "loss": 2.1157,
      "step": 37090
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.419921875,
      "learning_rate": 4.088936692454613e-06,
      "loss": 2.1251,
      "step": 37100
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4453125,
      "learning_rate": 4.081862600758589e-06,
      "loss": 2.103,
      "step": 37110
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.447265625,
      "learning_rate": 4.074793669789977e-06,
      "loss": 2.1519,
      "step": 37120
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.44921875,
      "learning_rate": 4.06772990289008e-06,
      "loss": 2.1609,
      "step": 37130
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.421875,
      "learning_rate": 4.06067130339777e-06,
      "loss": 2.1093,
      "step": 37140
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.455078125,
      "learning_rate": 4.053617874649455e-06,
      "loss": 2.1067,
      "step": 37150
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.47265625,
      "learning_rate": 4.046569619979124e-06,
      "loss": 2.1077,
      "step": 37160
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5234375,
      "learning_rate": 4.039526542718302e-06,
      "loss": 2.1271,
      "step": 37170
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4375,
      "learning_rate": 4.032488646196077e-06,
      "loss": 2.1231,
      "step": 37180
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.484375,
      "learning_rate": 4.025455933739079e-06,
      "loss": 2.1256,
      "step": 37190
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.435546875,
      "learning_rate": 4.018428408671503e-06,
      "loss": 2.0914,
      "step": 37200
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.453125,
      "learning_rate": 4.011406074315072e-06,
      "loss": 2.1392,
      "step": 37210
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.470703125,
      "learning_rate": 4.004388933989071e-06,
      "loss": 2.1404,
      "step": 37220
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.431640625,
      "learning_rate": 3.997376991010327e-06,
      "loss": 2.1152,
      "step": 37230
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.423828125,
      "learning_rate": 3.990370248693203e-06,
      "loss": 2.1153,
      "step": 37240
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.435546875,
      "learning_rate": 3.9833687103496095e-06,
      "loss": 2.1069,
      "step": 37250
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.423828125,
      "learning_rate": 3.976372379289005e-06,
      "loss": 2.1161,
      "step": 37260
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4296875,
      "learning_rate": 3.969381258818365e-06,
      "loss": 2.0945,
      "step": 37270
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5390625,
      "learning_rate": 3.962395352242224e-06,
      "loss": 2.1046,
      "step": 37280
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5,
      "learning_rate": 3.955414662862644e-06,
      "loss": 2.1054,
      "step": 37290
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.423828125,
      "learning_rate": 3.9484391939792145e-06,
      "loss": 2.1603,
      "step": 37300
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.421875,
      "learning_rate": 3.941468948889067e-06,
      "loss": 2.1027,
      "step": 37310
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.484375,
      "learning_rate": 3.9345039308868594e-06,
      "loss": 2.1295,
      "step": 37320
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.423828125,
      "learning_rate": 3.927544143264782e-06,
      "loss": 2.1199,
      "step": 37330
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.44140625,
      "learning_rate": 3.920589589312545e-06,
      "loss": 2.101,
      "step": 37340
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4375,
      "learning_rate": 3.913640272317394e-06,
      "loss": 2.1135,
      "step": 37350
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4296875,
      "learning_rate": 3.906696195564098e-06,
      "loss": 2.0865,
      "step": 37360
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.435546875,
      "learning_rate": 3.8997573623349385e-06,
      "loss": 2.0865,
      "step": 37370
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.451171875,
      "learning_rate": 3.892823775909731e-06,
      "loss": 2.0933,
      "step": 37380
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.44140625,
      "learning_rate": 3.885895439565811e-06,
      "loss": 2.1234,
      "step": 37390
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.42578125,
      "learning_rate": 3.878972356578019e-06,
      "loss": 2.132,
      "step": 37400
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.44921875,
      "learning_rate": 3.872054530218726e-06,
      "loss": 2.1138,
      "step": 37410
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.427734375,
      "learning_rate": 3.865141963757817e-06,
      "loss": 2.0882,
      "step": 37420
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.423828125,
      "learning_rate": 3.858234660462678e-06,
      "loss": 2.1238,
      "step": 37430
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4375,
      "learning_rate": 3.851332623598227e-06,
      "loss": 2.0945,
      "step": 37440
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4375,
      "learning_rate": 3.844435856426874e-06,
      "loss": 2.1293,
      "step": 37450
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.43359375,
      "learning_rate": 3.837544362208554e-06,
      "loss": 2.1185,
      "step": 37460
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4296875,
      "learning_rate": 3.830658144200691e-06,
      "loss": 2.1271,
      "step": 37470
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.453125,
      "learning_rate": 3.82377720565824e-06,
      "loss": 2.1114,
      "step": 37480
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.431640625,
      "learning_rate": 3.816901549833633e-06,
      "loss": 2.1179,
      "step": 37490
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.431640625,
      "learning_rate": 3.8100311799768256e-06,
      "loss": 2.1227,
      "step": 37500
    },
    {
      "epoch": 0.78,
      "eval_accuracy": 0.5591479993757034,
      "eval_loss": 1.992994785308838,
      "eval_runtime": 16.4756,
      "eval_samples_per_second": 36.114,
      "eval_steps_per_second": 1.153,
      "step": 37500
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.421875,
      "learning_rate": 3.803166099335267e-06,
      "loss": 2.1182,
      "step": 37510
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4296875,
      "learning_rate": 3.796306311153913e-06,
      "loss": 2.106,
      "step": 37520
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.859375,
      "learning_rate": 3.7894518186752007e-06,
      "loss": 2.1329,
      "step": 37530
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.427734375,
      "learning_rate": 3.7826026251390816e-06,
      "loss": 2.113,
      "step": 37540
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.51171875,
      "learning_rate": 3.7757587337830007e-06,
      "loss": 2.0975,
      "step": 37550
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.423828125,
      "learning_rate": 3.768920147841882e-06,
      "loss": 2.1341,
      "step": 37560
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.42578125,
      "learning_rate": 3.7620868705481586e-06,
      "loss": 2.146,
      "step": 37570
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.458984375,
      "learning_rate": 3.75525890513175e-06,
      "loss": 2.1283,
      "step": 37580
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4375,
      "learning_rate": 3.7484362548200558e-06,
      "loss": 2.1078,
      "step": 37590
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.421875,
      "learning_rate": 3.7416189228379732e-06,
      "loss": 2.1181,
      "step": 37600
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.474609375,
      "learning_rate": 3.734806912407888e-06,
      "loss": 2.13,
      "step": 37610
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.462890625,
      "learning_rate": 3.7280002267496584e-06,
      "loss": 2.1177,
      "step": 37620
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4375,
      "learning_rate": 3.7211988690806313e-06,
      "loss": 2.1211,
      "step": 37630
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.439453125,
      "learning_rate": 3.7144028426156463e-06,
      "loss": 2.1043,
      "step": 37640
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.421875,
      "learning_rate": 3.707612150567003e-06,
      "loss": 2.0999,
      "step": 37650
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.458984375,
      "learning_rate": 3.7008267961444924e-06,
      "loss": 2.1014,
      "step": 37660
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.431640625,
      "learning_rate": 3.6940467825553857e-06,
      "loss": 2.1203,
      "step": 37670
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4609375,
      "learning_rate": 3.6872721130044145e-06,
      "loss": 2.1194,
      "step": 37680
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.443359375,
      "learning_rate": 3.6805027906938038e-06,
      "loss": 2.1392,
      "step": 37690
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.46484375,
      "learning_rate": 3.6737388188232305e-06,
      "loss": 2.1037,
      "step": 37700
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.431640625,
      "learning_rate": 3.6669802005898644e-06,
      "loss": 2.1136,
      "step": 37710
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.447265625,
      "learning_rate": 3.6602269391883214e-06,
      "loss": 2.1088,
      "step": 37720
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4375,
      "learning_rate": 3.653479037810705e-06,
      "loss": 2.1291,
      "step": 37730
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.453125,
      "learning_rate": 3.646736499646578e-06,
      "loss": 2.1073,
      "step": 37740
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.439453125,
      "learning_rate": 3.639999327882962e-06,
      "loss": 2.1245,
      "step": 37750
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4375,
      "learning_rate": 3.633267525704352e-06,
      "loss": 2.1101,
      "step": 37760
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4296875,
      "learning_rate": 3.6265410962927027e-06,
      "loss": 2.1155,
      "step": 37770
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4296875,
      "learning_rate": 3.619820042827421e-06,
      "loss": 2.1103,
      "step": 37780
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.44140625,
      "learning_rate": 3.6131043684853824e-06,
      "loss": 2.0879,
      "step": 37790
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.431640625,
      "learning_rate": 3.6063940764409205e-06,
      "loss": 2.1121,
      "step": 37800
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.453125,
      "learning_rate": 3.59968916986581e-06,
      "loss": 2.0914,
      "step": 37810
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.431640625,
      "learning_rate": 3.592989651929299e-06,
      "loss": 2.1075,
      "step": 37820
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.52734375,
      "learning_rate": 3.5862955257980813e-06,
      "loss": 2.1359,
      "step": 37830
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.447265625,
      "learning_rate": 3.579606794636293e-06,
      "loss": 2.1123,
      "step": 37840
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.443359375,
      "learning_rate": 3.572923461605534e-06,
      "loss": 2.1447,
      "step": 37850
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.427734375,
      "learning_rate": 3.56624552986485e-06,
      "loss": 2.1121,
      "step": 37860
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.451171875,
      "learning_rate": 3.5595730025707214e-06,
      "loss": 2.1075,
      "step": 37870
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.44140625,
      "learning_rate": 3.552905882877087e-06,
      "loss": 2.1,
      "step": 37880
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.431640625,
      "learning_rate": 3.546244173935327e-06,
      "loss": 2.1252,
      "step": 37890
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4453125,
      "learning_rate": 3.539587878894268e-06,
      "loss": 2.137,
      "step": 37900
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4375,
      "learning_rate": 3.5329370009001606e-06,
      "loss": 2.1511,
      "step": 37910
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5078125,
      "learning_rate": 3.5262915430967175e-06,
      "loss": 2.1462,
      "step": 37920
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4453125,
      "learning_rate": 3.51965150862507e-06,
      "loss": 2.1215,
      "step": 37930
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.41796875,
      "learning_rate": 3.5130169006238016e-06,
      "loss": 2.12,
      "step": 37940
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.43359375,
      "learning_rate": 3.5063877222289146e-06,
      "loss": 2.1103,
      "step": 37950
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.48828125,
      "learning_rate": 3.499763976573866e-06,
      "loss": 2.1292,
      "step": 37960
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4296875,
      "learning_rate": 3.4931456667895206e-06,
      "loss": 2.1046,
      "step": 37970
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.427734375,
      "learning_rate": 3.486532796004193e-06,
      "loss": 2.1059,
      "step": 37980
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.447265625,
      "learning_rate": 3.4799253673436215e-06,
      "loss": 2.1328,
      "step": 37990
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.43359375,
      "learning_rate": 3.4733233839309642e-06,
      "loss": 2.1272,
      "step": 38000
    },
    {
      "epoch": 0.79,
      "eval_accuracy": 0.5592038574960776,
      "eval_loss": 1.99307382106781,
      "eval_runtime": 16.4339,
      "eval_samples_per_second": 36.206,
      "eval_steps_per_second": 1.156,
      "step": 38000
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.443359375,
      "learning_rate": 3.4667268488868166e-06,
      "loss": 2.0993,
      "step": 38010
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.431640625,
      "learning_rate": 3.460135765329196e-06,
      "loss": 2.0954,
      "step": 38020
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4453125,
      "learning_rate": 3.453550136373532e-06,
      "loss": 2.0937,
      "step": 38030
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4375,
      "learning_rate": 3.446969965132692e-06,
      "loss": 2.1,
      "step": 38040
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.421875,
      "learning_rate": 3.440395254716959e-06,
      "loss": 2.108,
      "step": 38050
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.439453125,
      "learning_rate": 3.4338260082340233e-06,
      "loss": 2.1147,
      "step": 38060
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4375,
      "learning_rate": 3.427262228789007e-06,
      "loss": 2.1175,
      "step": 38070
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.427734375,
      "learning_rate": 3.420703919484443e-06,
      "loss": 2.0821,
      "step": 38080
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.42578125,
      "learning_rate": 3.41415108342028e-06,
      "loss": 2.1068,
      "step": 38090
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4765625,
      "learning_rate": 3.407603723693872e-06,
      "loss": 2.1375,
      "step": 38100
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.462890625,
      "learning_rate": 3.4010618433999928e-06,
      "loss": 2.1335,
      "step": 38110
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4375,
      "learning_rate": 3.39452544563083e-06,
      "loss": 2.1256,
      "step": 38120
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.435546875,
      "learning_rate": 3.387994533475963e-06,
      "loss": 2.1404,
      "step": 38130
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.447265625,
      "learning_rate": 3.381469110022395e-06,
      "loss": 2.0768,
      "step": 38140
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.423828125,
      "learning_rate": 3.374949178354531e-06,
      "loss": 2.1113,
      "step": 38150
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4296875,
      "learning_rate": 3.36843474155417e-06,
      "loss": 2.1124,
      "step": 38160
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.51171875,
      "learning_rate": 3.36192580270053e-06,
      "loss": 2.1087,
      "step": 38170
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.451171875,
      "learning_rate": 3.3554223648702136e-06,
      "loss": 2.109,
      "step": 38180
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.419921875,
      "learning_rate": 3.34892443113724e-06,
      "loss": 2.1273,
      "step": 38190
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.447265625,
      "learning_rate": 3.3424320045730076e-06,
      "loss": 2.1078,
      "step": 38200
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.42578125,
      "learning_rate": 3.3359450882463347e-06,
      "loss": 2.1389,
      "step": 38210
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.427734375,
      "learning_rate": 3.329463685223411e-06,
      "loss": 2.1324,
      "step": 38220
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.419921875,
      "learning_rate": 3.322987798567836e-06,
      "loss": 2.1135,
      "step": 38230
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.486328125,
      "learning_rate": 3.3165174313406023e-06,
      "loss": 2.1338,
      "step": 38240
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.447265625,
      "learning_rate": 3.3100525866000814e-06,
      "loss": 2.1119,
      "step": 38250
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4375,
      "learning_rate": 3.3035932674020457e-06,
      "loss": 2.1136,
      "step": 38260
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.435546875,
      "learning_rate": 3.2971394767996514e-06,
      "loss": 2.1327,
      "step": 38270
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.443359375,
      "learning_rate": 3.290691217843447e-06,
      "loss": 2.1511,
      "step": 38280
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.443359375,
      "learning_rate": 3.2842484935813533e-06,
      "loss": 2.1398,
      "step": 38290
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4296875,
      "learning_rate": 3.2778113070586852e-06,
      "loss": 2.138,
      "step": 38300
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.44140625,
      "learning_rate": 3.271379661318145e-06,
      "loss": 2.1258,
      "step": 38310
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.431640625,
      "learning_rate": 3.264953559399798e-06,
      "loss": 2.0918,
      "step": 38320
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4375,
      "learning_rate": 3.258533004341106e-06,
      "loss": 2.1346,
      "step": 38330
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.427734375,
      "learning_rate": 3.252117999176906e-06,
      "loss": 2.1038,
      "step": 38340
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.443359375,
      "learning_rate": 3.2457085469394015e-06,
      "loss": 2.1346,
      "step": 38350
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.44921875,
      "learning_rate": 3.239304650658182e-06,
      "loss": 2.1366,
      "step": 38360
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4375,
      "learning_rate": 3.232906313360211e-06,
      "loss": 2.1097,
      "step": 38370
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.455078125,
      "learning_rate": 3.226513538069813e-06,
      "loss": 2.1326,
      "step": 38380
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.427734375,
      "learning_rate": 3.2201263278086956e-06,
      "loss": 2.0995,
      "step": 38390
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.466796875,
      "learning_rate": 3.2137446855959356e-06,
      "loss": 2.1398,
      "step": 38400
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.439453125,
      "learning_rate": 3.207368614447965e-06,
      "loss": 2.0944,
      "step": 38410
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4375,
      "learning_rate": 3.200998117378597e-06,
      "loss": 2.1035,
      "step": 38420
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.439453125,
      "learning_rate": 3.1946331973990066e-06,
      "loss": 2.0984,
      "step": 38430
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4375,
      "learning_rate": 3.1882738575177234e-06,
      "loss": 2.0896,
      "step": 38440
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.453125,
      "learning_rate": 3.181920100740654e-06,
      "loss": 2.0856,
      "step": 38450
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.427734375,
      "learning_rate": 3.175571930071051e-06,
      "loss": 2.1063,
      "step": 38460
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.427734375,
      "learning_rate": 3.169229348509543e-06,
      "loss": 2.1152,
      "step": 38470
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.431640625,
      "learning_rate": 3.162892359054098e-06,
      "loss": 2.0908,
      "step": 38480
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.466796875,
      "learning_rate": 3.156560964700055e-06,
      "loss": 2.1044,
      "step": 38490
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.439453125,
      "learning_rate": 3.1502351684401094e-06,
      "loss": 2.117,
      "step": 38500
    },
    {
      "epoch": 0.8,
      "eval_accuracy": 0.559105284342476,
      "eval_loss": 1.9930951595306396,
      "eval_runtime": 16.4469,
      "eval_samples_per_second": 36.177,
      "eval_steps_per_second": 1.155,
      "step": 38500
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.43359375,
      "learning_rate": 3.1439149732642964e-06,
      "loss": 2.0981,
      "step": 38510
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4296875,
      "learning_rate": 3.1376003821600153e-06,
      "loss": 2.1308,
      "step": 38520
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4375,
      "learning_rate": 3.13129139811202e-06,
      "loss": 2.1007,
      "step": 38530
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.427734375,
      "learning_rate": 3.124988024102397e-06,
      "loss": 2.1312,
      "step": 38540
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.474609375,
      "learning_rate": 3.118690263110596e-06,
      "loss": 2.1244,
      "step": 38550
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.419921875,
      "learning_rate": 3.1123981181134167e-06,
      "loss": 2.1276,
      "step": 38560
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.447265625,
      "learning_rate": 3.1061115920849853e-06,
      "loss": 2.1307,
      "step": 38570
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.45703125,
      "learning_rate": 3.099830687996789e-06,
      "loss": 2.1206,
      "step": 38580
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.419921875,
      "learning_rate": 3.0935554088176565e-06,
      "loss": 2.1353,
      "step": 38590
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.44140625,
      "learning_rate": 3.087285757513745e-06,
      "loss": 2.1118,
      "step": 38600
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.431640625,
      "learning_rate": 3.081021737048565e-06,
      "loss": 2.1263,
      "step": 38610
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.439453125,
      "learning_rate": 3.0747633503829635e-06,
      "loss": 2.1086,
      "step": 38620
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.435546875,
      "learning_rate": 3.0685106004751146e-06,
      "loss": 2.1548,
      "step": 38630
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.466796875,
      "learning_rate": 3.0622634902805367e-06,
      "loss": 2.0745,
      "step": 38640
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.427734375,
      "learning_rate": 3.0560220227520836e-06,
      "loss": 2.1024,
      "step": 38650
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.50390625,
      "learning_rate": 3.0497862008399428e-06,
      "loss": 2.0821,
      "step": 38660
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4921875,
      "learning_rate": 3.0435560274916207e-06,
      "loss": 2.1247,
      "step": 38670
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.419921875,
      "learning_rate": 3.0373315056519697e-06,
      "loss": 2.1326,
      "step": 38680
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.42578125,
      "learning_rate": 3.0311126382631574e-06,
      "loss": 2.1253,
      "step": 38690
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.44140625,
      "learning_rate": 3.0248994282646936e-06,
      "loss": 2.1393,
      "step": 38700
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.44921875,
      "learning_rate": 3.0186918785933965e-06,
      "loss": 2.1067,
      "step": 38710
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.421875,
      "learning_rate": 3.012489992183426e-06,
      "loss": 2.1137,
      "step": 38720
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.447265625,
      "learning_rate": 3.0062937719662485e-06,
      "loss": 2.1006,
      "step": 38730
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.43359375,
      "learning_rate": 3.0001032208706653e-06,
      "loss": 2.1159,
      "step": 38740
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4375,
      "learning_rate": 2.9939183418227966e-06,
      "loss": 2.1229,
      "step": 38750
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.44140625,
      "learning_rate": 2.987739137746071e-06,
      "loss": 2.124,
      "step": 38760
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.421875,
      "learning_rate": 2.9815656115612473e-06,
      "loss": 2.1329,
      "step": 38770
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4296875,
      "learning_rate": 2.9753977661863956e-06,
      "loss": 2.0899,
      "step": 38780
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4453125,
      "learning_rate": 2.9692356045368964e-06,
      "loss": 2.0893,
      "step": 38790
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.43359375,
      "learning_rate": 2.963079129525447e-06,
      "loss": 2.0987,
      "step": 38800
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.42578125,
      "learning_rate": 2.956928344062066e-06,
      "loss": 2.1152,
      "step": 38810
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.423828125,
      "learning_rate": 2.9507832510540633e-06,
      "loss": 2.0963,
      "step": 38820
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.45703125,
      "learning_rate": 2.944643853406076e-06,
      "loss": 2.1232,
      "step": 38830
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.453125,
      "learning_rate": 2.9385101540200375e-06,
      "loss": 2.085,
      "step": 38840
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.44921875,
      "learning_rate": 2.9323821557952007e-06,
      "loss": 2.1208,
      "step": 38850
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.427734375,
      "learning_rate": 2.926259861628105e-06,
      "loss": 2.0946,
      "step": 38860
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.443359375,
      "learning_rate": 2.9201432744126074e-06,
      "loss": 2.1473,
      "step": 38870
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4765625,
      "learning_rate": 2.9140323970398695e-06,
      "loss": 2.1226,
      "step": 38880
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.427734375,
      "learning_rate": 2.907927232398342e-06,
      "loss": 2.1028,
      "step": 38890
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.44140625,
      "learning_rate": 2.901827783373782e-06,
      "loss": 2.1287,
      "step": 38900
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.431640625,
      "learning_rate": 2.895734052849251e-06,
      "loss": 2.1321,
      "step": 38910
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.43359375,
      "learning_rate": 2.889646043705094e-06,
      "loss": 2.1108,
      "step": 38920
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.439453125,
      "learning_rate": 2.883563758818965e-06,
      "loss": 2.1251,
      "step": 38930
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7734375,
      "learning_rate": 2.8774872010658005e-06,
      "loss": 2.1175,
      "step": 38940
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.43359375,
      "learning_rate": 2.8714163733178412e-06,
      "loss": 2.122,
      "step": 38950
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.44921875,
      "learning_rate": 2.865351278444607e-06,
      "loss": 2.1043,
      "step": 38960
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.427734375,
      "learning_rate": 2.859291919312924e-06,
      "loss": 2.1147,
      "step": 38970
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.447265625,
      "learning_rate": 2.853238298786887e-06,
      "loss": 2.1021,
      "step": 38980
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4296875,
      "learning_rate": 2.847190419727897e-06,
      "loss": 2.1217,
      "step": 38990
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.443359375,
      "learning_rate": 2.8411482849946352e-06,
      "loss": 2.1325,
      "step": 39000
    },
    {
      "epoch": 0.81,
      "eval_accuracy": 0.5590789981681822,
      "eval_loss": 1.993077278137207,
      "eval_runtime": 16.4287,
      "eval_samples_per_second": 36.217,
      "eval_steps_per_second": 1.157,
      "step": 39000
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.423828125,
      "learning_rate": 2.835111897443057e-06,
      "loss": 2.1213,
      "step": 39010
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4296875,
      "learning_rate": 2.8290812599264178e-06,
      "loss": 2.1199,
      "step": 39020
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.439453125,
      "learning_rate": 2.8230563752952426e-06,
      "loss": 2.1274,
      "step": 39030
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.490234375,
      "learning_rate": 2.817037246397351e-06,
      "loss": 2.1116,
      "step": 39040
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.451171875,
      "learning_rate": 2.81102387607782e-06,
      "loss": 2.1164,
      "step": 39050
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.423828125,
      "learning_rate": 2.8050162671790264e-06,
      "loss": 2.0473,
      "step": 39060
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.427734375,
      "learning_rate": 2.799014422540615e-06,
      "loss": 2.0956,
      "step": 39070
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.435546875,
      "learning_rate": 2.7930183449995e-06,
      "loss": 2.1364,
      "step": 39080
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4296875,
      "learning_rate": 2.7870280373898792e-06,
      "loss": 2.1309,
      "step": 39090
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4375,
      "learning_rate": 2.7810435025432244e-06,
      "loss": 2.1126,
      "step": 39100
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.43359375,
      "learning_rate": 2.775064743288265e-06,
      "loss": 2.119,
      "step": 39110
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.439453125,
      "learning_rate": 2.7690917624510127e-06,
      "loss": 2.1179,
      "step": 39120
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.458984375,
      "learning_rate": 2.7631245628547496e-06,
      "loss": 2.1025,
      "step": 39130
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.462890625,
      "learning_rate": 2.7571631473200115e-06,
      "loss": 2.1248,
      "step": 39140
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4296875,
      "learning_rate": 2.7512075186646125e-06,
      "loss": 2.0777,
      "step": 39150
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.427734375,
      "learning_rate": 2.745257679703632e-06,
      "loss": 2.0887,
      "step": 39160
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.48828125,
      "learning_rate": 2.739313633249401e-06,
      "loss": 2.0912,
      "step": 39170
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4765625,
      "learning_rate": 2.7333753821115252e-06,
      "loss": 2.0972,
      "step": 39180
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.439453125,
      "learning_rate": 2.7274429290968626e-06,
      "loss": 2.1552,
      "step": 39190
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.451171875,
      "learning_rate": 2.721516277009537e-06,
      "loss": 2.1281,
      "step": 39200
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.431640625,
      "learning_rate": 2.715595428650922e-06,
      "loss": 2.1185,
      "step": 39210
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.423828125,
      "learning_rate": 2.7096803868196546e-06,
      "loss": 2.1236,
      "step": 39220
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.455078125,
      "learning_rate": 2.703771154311631e-06,
      "loss": 2.0983,
      "step": 39230
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.44140625,
      "learning_rate": 2.6978677339199877e-06,
      "loss": 2.1234,
      "step": 39240
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.421875,
      "learning_rate": 2.691970128435125e-06,
      "loss": 2.1048,
      "step": 39250
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.43359375,
      "learning_rate": 2.6860783406446986e-06,
      "loss": 2.1394,
      "step": 39260
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.44140625,
      "learning_rate": 2.6801923733335975e-06,
      "loss": 2.1226,
      "step": 39270
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.455078125,
      "learning_rate": 2.6743122292839726e-06,
      "loss": 2.1134,
      "step": 39280
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.43359375,
      "learning_rate": 2.668437911275226e-06,
      "loss": 2.0929,
      "step": 39290
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4375,
      "learning_rate": 2.6625694220839896e-06,
      "loss": 2.1166,
      "step": 39300
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.416015625,
      "learning_rate": 2.656706764484153e-06,
      "loss": 2.1016,
      "step": 39310
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.49609375,
      "learning_rate": 2.650849941246853e-06,
      "loss": 2.137,
      "step": 39320
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.427734375,
      "learning_rate": 2.6449989551404514e-06,
      "loss": 2.1449,
      "step": 39330
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.470703125,
      "learning_rate": 2.6391538089305663e-06,
      "loss": 2.1063,
      "step": 39340
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.431640625,
      "learning_rate": 2.633314505380052e-06,
      "loss": 2.115,
      "step": 39350
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.42578125,
      "learning_rate": 2.627481047248997e-06,
      "loss": 2.1144,
      "step": 39360
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.44921875,
      "learning_rate": 2.6216534372947274e-06,
      "loss": 2.0999,
      "step": 39370
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.46875,
      "learning_rate": 2.615831678271816e-06,
      "loss": 2.1344,
      "step": 39380
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.439453125,
      "learning_rate": 2.6100157729320497e-06,
      "loss": 2.1486,
      "step": 39390
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4453125,
      "learning_rate": 2.6042057240244653e-06,
      "loss": 2.1418,
      "step": 39400
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.439453125,
      "learning_rate": 2.598401534295327e-06,
      "loss": 2.1089,
      "step": 39410
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.451171875,
      "learning_rate": 2.5926032064881298e-06,
      "loss": 2.1405,
      "step": 39420
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4765625,
      "learning_rate": 2.58681074334359e-06,
      "loss": 2.104,
      "step": 39430
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.44921875,
      "learning_rate": 2.581024147599667e-06,
      "loss": 2.1205,
      "step": 39440
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4296875,
      "learning_rate": 2.5752434219915298e-06,
      "loss": 2.1532,
      "step": 39450
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.51171875,
      "learning_rate": 2.5694685692515895e-06,
      "loss": 2.1023,
      "step": 39460
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.439453125,
      "learning_rate": 2.563699592109464e-06,
      "loss": 2.1162,
      "step": 39470
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4296875,
      "learning_rate": 2.5579364932920106e-06,
      "loss": 2.1218,
      "step": 39480
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4375,
      "learning_rate": 2.5521792755232955e-06,
      "loss": 2.121,
      "step": 39490
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.42578125,
      "learning_rate": 2.5464279415246093e-06,
      "loss": 2.1046,
      "step": 39500
    },
    {
      "epoch": 0.82,
      "eval_accuracy": 0.55914635648981,
      "eval_loss": 1.9930146932601929,
      "eval_runtime": 16.4521,
      "eval_samples_per_second": 36.166,
      "eval_steps_per_second": 1.155,
      "step": 39500
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.443359375,
      "learning_rate": 2.54068249401447e-06,
      "loss": 2.1028,
      "step": 39510
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.423828125,
      "learning_rate": 2.5349429357085952e-06,
      "loss": 2.0948,
      "step": 39520
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.431640625,
      "learning_rate": 2.5292092693199344e-06,
      "loss": 2.0923,
      "step": 39530
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.49609375,
      "learning_rate": 2.523481497558648e-06,
      "loss": 2.1247,
      "step": 39540
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.43359375,
      "learning_rate": 2.5177596231321037e-06,
      "loss": 2.1331,
      "step": 39550
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4140625,
      "learning_rate": 2.51204364874489e-06,
      "loss": 2.1433,
      "step": 39560
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.431640625,
      "learning_rate": 2.5063335770988053e-06,
      "loss": 2.1148,
      "step": 39570
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.419921875,
      "learning_rate": 2.5006294108928528e-06,
      "loss": 2.1084,
      "step": 39580
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.486328125,
      "learning_rate": 2.494931152823247e-06,
      "loss": 2.0711,
      "step": 39590
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.453125,
      "learning_rate": 2.4892388055834137e-06,
      "loss": 2.1357,
      "step": 39600
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.439453125,
      "learning_rate": 2.483552371863982e-06,
      "loss": 2.1171,
      "step": 39610
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.431640625,
      "learning_rate": 2.47787185435278e-06,
      "loss": 2.109,
      "step": 39620
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.443359375,
      "learning_rate": 2.472197255734847e-06,
      "loss": 2.1119,
      "step": 39630
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.427734375,
      "learning_rate": 2.4665285786924292e-06,
      "loss": 2.1521,
      "step": 39640
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.423828125,
      "learning_rate": 2.4608658259049544e-06,
      "loss": 2.1119,
      "step": 39650
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.45703125,
      "learning_rate": 2.455209000049072e-06,
      "loss": 2.1421,
      "step": 39660
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.44921875,
      "learning_rate": 2.449558103798619e-06,
      "loss": 2.162,
      "step": 39670
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4609375,
      "learning_rate": 2.443913139824625e-06,
      "loss": 2.1142,
      "step": 39680
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.451171875,
      "learning_rate": 2.4382741107953317e-06,
      "loss": 2.1177,
      "step": 39690
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.431640625,
      "learning_rate": 2.432641019376157e-06,
      "loss": 2.1082,
      "step": 39700
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.48828125,
      "learning_rate": 2.4270138682297273e-06,
      "loss": 2.0948,
      "step": 39710
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.423828125,
      "learning_rate": 2.421392660015847e-06,
      "loss": 2.1198,
      "step": 39720
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.46875,
      "learning_rate": 2.415777397391526e-06,
      "loss": 2.1266,
      "step": 39730
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4609375,
      "learning_rate": 2.4101680830109523e-06,
      "loss": 2.1253,
      "step": 39740
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.435546875,
      "learning_rate": 2.404564719525507e-06,
      "loss": 2.0963,
      "step": 39750
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.44921875,
      "learning_rate": 2.398967309583765e-06,
      "loss": 2.1115,
      "step": 39760
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.451171875,
      "learning_rate": 2.393375855831473e-06,
      "loss": 2.1266,
      "step": 39770
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.43359375,
      "learning_rate": 2.3877903609115697e-06,
      "loss": 2.1145,
      "step": 39780
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.466796875,
      "learning_rate": 2.3822108274641823e-06,
      "loss": 2.1107,
      "step": 39790
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4296875,
      "learning_rate": 2.3766372581266146e-06,
      "loss": 2.1349,
      "step": 39800
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41796875,
      "learning_rate": 2.3710696555333468e-06,
      "loss": 2.1091,
      "step": 39810
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.423828125,
      "learning_rate": 2.3655080223160476e-06,
      "loss": 2.0956,
      "step": 39820
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.427734375,
      "learning_rate": 2.359952361103564e-06,
      "loss": 2.1263,
      "step": 39830
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.435546875,
      "learning_rate": 2.354402674521907e-06,
      "loss": 2.11,
      "step": 39840
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.431640625,
      "learning_rate": 2.3488589651942803e-06,
      "loss": 2.1226,
      "step": 39850
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.45703125,
      "learning_rate": 2.343321235741056e-06,
      "loss": 2.1063,
      "step": 39860
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5234375,
      "learning_rate": 2.337789488779772e-06,
      "loss": 2.1354,
      "step": 39870
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4296875,
      "learning_rate": 2.3322637269251497e-06,
      "loss": 2.1069,
      "step": 39880
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.447265625,
      "learning_rate": 2.32674395278908e-06,
      "loss": 2.1531,
      "step": 39890
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.427734375,
      "learning_rate": 2.3212301689806127e-06,
      "loss": 2.1052,
      "step": 39900
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4765625,
      "learning_rate": 2.3157223781059776e-06,
      "loss": 2.1108,
      "step": 39910
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.423828125,
      "learning_rate": 2.3102205827685747e-06,
      "loss": 2.1199,
      "step": 39920
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4453125,
      "learning_rate": 2.304724785568953e-06,
      "loss": 2.1061,
      "step": 39930
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4375,
      "learning_rate": 2.299234989104843e-06,
      "loss": 2.1323,
      "step": 39940
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.45703125,
      "learning_rate": 2.2937511959711293e-06,
      "loss": 2.1359,
      "step": 39950
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4375,
      "learning_rate": 2.2882734087598685e-06,
      "loss": 2.0921,
      "step": 39960
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.44921875,
      "learning_rate": 2.2828016300602623e-06,
      "loss": 2.1156,
      "step": 39970
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.455078125,
      "learning_rate": 2.2773358624586887e-06,
      "loss": 2.1434,
      "step": 39980
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.44921875,
      "learning_rate": 2.271876108538678e-06,
      "loss": 2.1101,
      "step": 39990
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.427734375,
      "learning_rate": 2.266422370880914e-06,
      "loss": 2.1096,
      "step": 40000
    },
    {
      "epoch": 0.83,
      "eval_accuracy": 0.5590970699130092,
      "eval_loss": 1.9930177927017212,
      "eval_runtime": 16.4304,
      "eval_samples_per_second": 36.213,
      "eval_steps_per_second": 1.156,
      "step": 40000
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4296875,
      "learning_rate": 2.260974652063243e-06,
      "loss": 2.1459,
      "step": 40010
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.453125,
      "learning_rate": 2.255532954660667e-06,
      "loss": 2.116,
      "step": 40020
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.43359375,
      "learning_rate": 2.250097281245333e-06,
      "loss": 2.0934,
      "step": 40030
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.421875,
      "learning_rate": 2.2446676343865464e-06,
      "loss": 2.1055,
      "step": 40040
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4375,
      "learning_rate": 2.239244016650772e-06,
      "loss": 2.1163,
      "step": 40050
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.435546875,
      "learning_rate": 2.2338264306016065e-06,
      "loss": 2.1134,
      "step": 40060
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.453125,
      "learning_rate": 2.228414878799811e-06,
      "loss": 2.1396,
      "step": 40070
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.455078125,
      "learning_rate": 2.2230093638032912e-06,
      "loss": 2.1483,
      "step": 40080
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.451171875,
      "learning_rate": 2.2176098881670913e-06,
      "loss": 2.0784,
      "step": 40090
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.46875,
      "learning_rate": 2.2122164544434103e-06,
      "loss": 2.1203,
      "step": 40100
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.43359375,
      "learning_rate": 2.206829065181592e-06,
      "loss": 2.1252,
      "step": 40110
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.44140625,
      "learning_rate": 2.201447722928111e-06,
      "loss": 2.1072,
      "step": 40120
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.43359375,
      "learning_rate": 2.1960724302265946e-06,
      "loss": 2.1589,
      "step": 40130
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41796875,
      "learning_rate": 2.190703189617813e-06,
      "loss": 2.1173,
      "step": 40140
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41796875,
      "learning_rate": 2.1853400036396636e-06,
      "loss": 2.1002,
      "step": 40150
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4453125,
      "learning_rate": 2.1799828748271897e-06,
      "loss": 2.1007,
      "step": 40160
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.42578125,
      "learning_rate": 2.174631805712576e-06,
      "loss": 2.1291,
      "step": 40170
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.431640625,
      "learning_rate": 2.1692867988251315e-06,
      "loss": 2.1131,
      "step": 40180
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.435546875,
      "learning_rate": 2.1639478566913107e-06,
      "loss": 2.0844,
      "step": 40190
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.435546875,
      "learning_rate": 2.158614981834691e-06,
      "loss": 2.1219,
      "step": 40200
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.46484375,
      "learning_rate": 2.1532881767759915e-06,
      "loss": 2.1261,
      "step": 40210
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.42578125,
      "learning_rate": 2.147967444033056e-06,
      "loss": 2.105,
      "step": 40220
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44140625,
      "learning_rate": 2.1426527861208605e-06,
      "loss": 2.1067,
      "step": 40230
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.443359375,
      "learning_rate": 2.137344205551514e-06,
      "loss": 2.1079,
      "step": 40240
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.462890625,
      "learning_rate": 2.1320417048342395e-06,
      "loss": 2.1024,
      "step": 40250
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.47265625,
      "learning_rate": 2.1267452864754015e-06,
      "loss": 2.1224,
      "step": 40260
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.515625,
      "learning_rate": 2.121454952978485e-06,
      "loss": 2.1138,
      "step": 40270
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.447265625,
      "learning_rate": 2.1161707068440887e-06,
      "loss": 2.1279,
      "step": 40280
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44921875,
      "learning_rate": 2.110892550569947e-06,
      "loss": 2.1078,
      "step": 40290
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44921875,
      "learning_rate": 2.1056204866509145e-06,
      "loss": 2.0859,
      "step": 40300
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.431640625,
      "learning_rate": 2.100354517578955e-06,
      "loss": 2.11,
      "step": 40310
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.455078125,
      "learning_rate": 2.095094645843162e-06,
      "loss": 2.1157,
      "step": 40320
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.43359375,
      "learning_rate": 2.0898408739297476e-06,
      "loss": 2.1055,
      "step": 40330
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.42578125,
      "learning_rate": 2.0845932043220307e-06,
      "loss": 2.0706,
      "step": 40340
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.423828125,
      "learning_rate": 2.0793516395004554e-06,
      "loss": 2.1144,
      "step": 40350
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.439453125,
      "learning_rate": 2.0741161819425748e-06,
      "loss": 2.1332,
      "step": 40360
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.427734375,
      "learning_rate": 2.0688868341230626e-06,
      "loss": 2.1171,
      "step": 40370
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.42578125,
      "learning_rate": 2.063663598513694e-06,
      "loss": 2.0994,
      "step": 40380
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.462890625,
      "learning_rate": 2.0584464775833596e-06,
      "loss": 2.0858,
      "step": 40390
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.43359375,
      "learning_rate": 2.053235473798066e-06,
      "loss": 2.0817,
      "step": 40400
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44140625,
      "learning_rate": 2.048030589620917e-06,
      "loss": 2.1484,
      "step": 40410
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.43359375,
      "learning_rate": 2.0428318275121327e-06,
      "loss": 2.1246,
      "step": 40420
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.53125,
      "learning_rate": 2.0376391899290397e-06,
      "loss": 2.1304,
      "step": 40430
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.451171875,
      "learning_rate": 2.0324526793260586e-06,
      "loss": 2.1256,
      "step": 40440
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.42578125,
      "learning_rate": 2.0272722981547294e-06,
      "loss": 2.1141,
      "step": 40450
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.427734375,
      "learning_rate": 2.0220980488636803e-06,
      "loss": 2.1487,
      "step": 40460
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.41015625,
      "learning_rate": 2.016929933898657e-06,
      "loss": 2.1142,
      "step": 40470
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.68359375,
      "learning_rate": 2.0117679557024866e-06,
      "loss": 2.1062,
      "step": 40480
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4609375,
      "learning_rate": 2.0066121167151125e-06,
      "loss": 2.1048,
      "step": 40490
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5078125,
      "learning_rate": 2.001462419373565e-06,
      "loss": 2.1149,
      "step": 40500
    },
    {
      "epoch": 0.84,
      "eval_accuracy": 0.559087212597649,
      "eval_loss": 1.9930588006973267,
      "eval_runtime": 16.4374,
      "eval_samples_per_second": 36.198,
      "eval_steps_per_second": 1.156,
      "step": 40500
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.431640625,
      "learning_rate": 1.9963188661119798e-06,
      "loss": 2.1208,
      "step": 40510
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.46484375,
      "learning_rate": 1.991181459361584e-06,
      "loss": 2.1136,
      "step": 40520
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44140625,
      "learning_rate": 1.986050201550696e-06,
      "loss": 2.1268,
      "step": 40530
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4375,
      "learning_rate": 1.9809250951047335e-06,
      "loss": 2.105,
      "step": 40540
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44140625,
      "learning_rate": 1.9758061424462058e-06,
      "loss": 2.1074,
      "step": 40550
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.423828125,
      "learning_rate": 1.9706933459947156e-06,
      "loss": 2.1022,
      "step": 40560
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4453125,
      "learning_rate": 1.9655867081669463e-06,
      "loss": 2.1218,
      "step": 40570
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.439453125,
      "learning_rate": 1.9604862313766775e-06,
      "loss": 2.0989,
      "step": 40580
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.427734375,
      "learning_rate": 1.9553919180347818e-06,
      "loss": 2.138,
      "step": 40590
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.453125,
      "learning_rate": 1.950303770549203e-06,
      "loss": 2.1152,
      "step": 40600
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.447265625,
      "learning_rate": 1.945221791324985e-06,
      "loss": 2.0858,
      "step": 40610
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.439453125,
      "learning_rate": 1.9401459827642516e-06,
      "loss": 2.1118,
      "step": 40620
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44921875,
      "learning_rate": 1.9350763472662057e-06,
      "loss": 2.1138,
      "step": 40630
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.439453125,
      "learning_rate": 1.930012887227138e-06,
      "loss": 2.1353,
      "step": 40640
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.435546875,
      "learning_rate": 1.924955605040421e-06,
      "loss": 2.1248,
      "step": 40650
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4375,
      "learning_rate": 1.919904503096497e-06,
      "loss": 2.1035,
      "step": 40660
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.427734375,
      "learning_rate": 1.914859583782897e-06,
      "loss": 2.1136,
      "step": 40670
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.458984375,
      "learning_rate": 1.9098208494842323e-06,
      "loss": 2.136,
      "step": 40680
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4375,
      "learning_rate": 1.9047883025821777e-06,
      "loss": 2.1199,
      "step": 40690
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.447265625,
      "learning_rate": 1.8997619454554955e-06,
      "loss": 2.1209,
      "step": 40700
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.47265625,
      "learning_rate": 1.8947417804800131e-06,
      "loss": 2.1306,
      "step": 40710
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.447265625,
      "learning_rate": 1.8897278100286413e-06,
      "loss": 2.1293,
      "step": 40720
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.443359375,
      "learning_rate": 1.8847200364713507e-06,
      "loss": 2.1038,
      "step": 40730
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.44140625,
      "learning_rate": 1.879718462175193e-06,
      "loss": 2.1225,
      "step": 40740
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4296875,
      "learning_rate": 1.8747230895042883e-06,
      "loss": 2.1301,
      "step": 40750
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.427734375,
      "learning_rate": 1.8697339208198183e-06,
      "loss": 2.1155,
      "step": 40760
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.427734375,
      "learning_rate": 1.8647509584800365e-06,
      "loss": 2.1167,
      "step": 40770
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41796875,
      "learning_rate": 1.8597742048402687e-06,
      "loss": 2.1075,
      "step": 40780
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4375,
      "learning_rate": 1.8548036622528952e-06,
      "loss": 2.1266,
      "step": 40790
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41015625,
      "learning_rate": 1.8498393330673663e-06,
      "loss": 2.1219,
      "step": 40800
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.404296875,
      "learning_rate": 1.8448812196302e-06,
      "loss": 2.1229,
      "step": 40810
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.439453125,
      "learning_rate": 1.8399293242849635e-06,
      "loss": 2.1481,
      "step": 40820
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.458984375,
      "learning_rate": 1.834983649372296e-06,
      "loss": 2.1021,
      "step": 40830
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.423828125,
      "learning_rate": 1.8300441972298976e-06,
      "loss": 2.1458,
      "step": 40840
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.43359375,
      "learning_rate": 1.825110970192514e-06,
      "loss": 2.1128,
      "step": 40850
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.44140625,
      "learning_rate": 1.8201839705919598e-06,
      "loss": 2.1155,
      "step": 40860
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.453125,
      "learning_rate": 1.815263200757109e-06,
      "loss": 2.151,
      "step": 40870
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.45703125,
      "learning_rate": 1.810348663013876e-06,
      "loss": 2.1232,
      "step": 40880
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.427734375,
      "learning_rate": 1.8054403596852425e-06,
      "loss": 2.0951,
      "step": 40890
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.42578125,
      "learning_rate": 1.8005382930912446e-06,
      "loss": 2.1109,
      "step": 40900
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4296875,
      "learning_rate": 1.795642465548955e-06,
      "loss": 2.1178,
      "step": 40910
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.609375,
      "learning_rate": 1.7907528793725132e-06,
      "loss": 2.1073,
      "step": 40920
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.435546875,
      "learning_rate": 1.7858695368731048e-06,
      "loss": 2.1175,
      "step": 40930
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4453125,
      "learning_rate": 1.7809924403589589e-06,
      "loss": 2.091,
      "step": 40940
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.42578125,
      "learning_rate": 1.776121592135358e-06,
      "loss": 2.0829,
      "step": 40950
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.443359375,
      "learning_rate": 1.7712569945046253e-06,
      "loss": 2.1106,
      "step": 40960
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4296875,
      "learning_rate": 1.7663986497661383e-06,
      "loss": 2.1407,
      "step": 40970
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4375,
      "learning_rate": 1.7615465602163067e-06,
      "loss": 2.1278,
      "step": 40980
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.458984375,
      "learning_rate": 1.756700728148594e-06,
      "loss": 2.1042,
      "step": 40990
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.421875,
      "learning_rate": 1.7518611558535086e-06,
      "loss": 2.122,
      "step": 41000
    },
    {
      "epoch": 0.85,
      "eval_accuracy": 0.559128284744983,
      "eval_loss": 1.9930520057678223,
      "eval_runtime": 16.4331,
      "eval_samples_per_second": 36.207,
      "eval_steps_per_second": 1.156,
      "step": 41000
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.431640625,
      "learning_rate": 1.7470278456185834e-06,
      "loss": 2.1016,
      "step": 41010
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.439453125,
      "learning_rate": 1.7422007997284063e-06,
      "loss": 2.1012,
      "step": 41020
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4453125,
      "learning_rate": 1.737380020464605e-06,
      "loss": 2.1105,
      "step": 41030
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.447265625,
      "learning_rate": 1.7325655101058308e-06,
      "loss": 2.1164,
      "step": 41040
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4296875,
      "learning_rate": 1.7277572709277861e-06,
      "loss": 2.1155,
      "step": 41050
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.439453125,
      "learning_rate": 1.7229553052032066e-06,
      "loss": 2.1464,
      "step": 41060
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.423828125,
      "learning_rate": 1.718159615201853e-06,
      "loss": 2.1352,
      "step": 41070
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.44140625,
      "learning_rate": 1.7133702031905307e-06,
      "loss": 2.1204,
      "step": 41080
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.42578125,
      "learning_rate": 1.7085870714330753e-06,
      "loss": 2.084,
      "step": 41090
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.458984375,
      "learning_rate": 1.7038102221903483e-06,
      "loss": 2.111,
      "step": 41100
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.43359375,
      "learning_rate": 1.699039657720247e-06,
      "loss": 2.1301,
      "step": 41110
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.412109375,
      "learning_rate": 1.694275380277701e-06,
      "loss": 2.0874,
      "step": 41120
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.435546875,
      "learning_rate": 1.6895173921146555e-06,
      "loss": 2.1138,
      "step": 41130
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4375,
      "learning_rate": 1.684765695480095e-06,
      "loss": 2.0863,
      "step": 41140
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.478515625,
      "learning_rate": 1.6800202926200276e-06,
      "loss": 2.1258,
      "step": 41150
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.416015625,
      "learning_rate": 1.6752811857774874e-06,
      "loss": 2.0856,
      "step": 41160
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.416015625,
      "learning_rate": 1.6705483771925244e-06,
      "loss": 2.0796,
      "step": 41170
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.453125,
      "learning_rate": 1.6658218691022242e-06,
      "loss": 2.1002,
      "step": 41180
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.46484375,
      "learning_rate": 1.6611016637406796e-06,
      "loss": 2.1254,
      "step": 41190
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.443359375,
      "learning_rate": 1.656387763339019e-06,
      "loss": 2.1359,
      "step": 41200
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.443359375,
      "learning_rate": 1.6516801701253802e-06,
      "loss": 2.1073,
      "step": 41210
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.421875,
      "learning_rate": 1.6469788863249268e-06,
      "loss": 2.0931,
      "step": 41220
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.458984375,
      "learning_rate": 1.6422839141598311e-06,
      "loss": 2.1148,
      "step": 41230
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.419921875,
      "learning_rate": 1.6375952558492912e-06,
      "loss": 2.1125,
      "step": 41240
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.419921875,
      "learning_rate": 1.632912913609521e-06,
      "loss": 2.1401,
      "step": 41250
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.44140625,
      "learning_rate": 1.6282368896537386e-06,
      "loss": 2.1214,
      "step": 41260
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4296875,
      "learning_rate": 1.6235671861921858e-06,
      "loss": 2.1231,
      "step": 41270
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.453125,
      "learning_rate": 1.6189038054321153e-06,
      "loss": 2.1373,
      "step": 41280
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.234375,
      "learning_rate": 1.6142467495777824e-06,
      "loss": 2.1255,
      "step": 41290
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.435546875,
      "learning_rate": 1.609596020830466e-06,
      "loss": 2.0925,
      "step": 41300
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.431640625,
      "learning_rate": 1.604951621388448e-06,
      "loss": 2.0955,
      "step": 41310
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.439453125,
      "learning_rate": 1.6003135534470137e-06,
      "loss": 2.1064,
      "step": 41320
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.46484375,
      "learning_rate": 1.5956818191984628e-06,
      "loss": 2.0881,
      "step": 41330
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4296875,
      "learning_rate": 1.5910564208320993e-06,
      "loss": 2.1502,
      "step": 41340
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.443359375,
      "learning_rate": 1.5864373605342342e-06,
      "loss": 2.0895,
      "step": 41350
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.435546875,
      "learning_rate": 1.5818246404881775e-06,
      "loss": 2.1122,
      "step": 41360
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4296875,
      "learning_rate": 1.5772182628742454e-06,
      "loss": 2.12,
      "step": 41370
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.419921875,
      "learning_rate": 1.5726182298697594e-06,
      "loss": 2.081,
      "step": 41380
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.42578125,
      "learning_rate": 1.5680245436490338e-06,
      "loss": 2.0825,
      "step": 41390
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.462890625,
      "learning_rate": 1.5634372063833896e-06,
      "loss": 2.0768,
      "step": 41400
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4375,
      "learning_rate": 1.5588562202411495e-06,
      "loss": 2.1102,
      "step": 41410
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4453125,
      "learning_rate": 1.5542815873876225e-06,
      "loss": 2.1071,
      "step": 41420
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.421875,
      "learning_rate": 1.5497133099851263e-06,
      "loss": 2.1297,
      "step": 41430
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.42578125,
      "learning_rate": 1.5451513901929714e-06,
      "loss": 2.1385,
      "step": 41440
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4921875,
      "learning_rate": 1.5405958301674582e-06,
      "loss": 2.1197,
      "step": 41450
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4609375,
      "learning_rate": 1.5360466320618876e-06,
      "loss": 2.104,
      "step": 41460
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.431640625,
      "learning_rate": 1.5315037980265478e-06,
      "loss": 2.1565,
      "step": 41470
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.44921875,
      "learning_rate": 1.5269673302087244e-06,
      "loss": 2.1101,
      "step": 41480
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.439453125,
      "learning_rate": 1.5224372307526874e-06,
      "loss": 2.136,
      "step": 41490
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.451171875,
      "learning_rate": 1.5179135017997053e-06,
      "loss": 2.1137,
      "step": 41500
    },
    {
      "epoch": 0.86,
      "eval_accuracy": 0.5590559977656752,
      "eval_loss": 1.9930741786956787,
      "eval_runtime": 16.5205,
      "eval_samples_per_second": 36.016,
      "eval_steps_per_second": 1.15,
      "step": 41500
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4296875,
      "learning_rate": 1.5133961454880247e-06,
      "loss": 2.1122,
      "step": 41510
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.439453125,
      "learning_rate": 1.5088851639528872e-06,
      "loss": 2.101,
      "step": 41520
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.44921875,
      "learning_rate": 1.5043805593265226e-06,
      "loss": 2.1284,
      "step": 41530
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4296875,
      "learning_rate": 1.499882333738144e-06,
      "loss": 2.1305,
      "step": 41540
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.478515625,
      "learning_rate": 1.4953904893139424e-06,
      "loss": 2.0943,
      "step": 41550
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.435546875,
      "learning_rate": 1.4909050281771047e-06,
      "loss": 2.0968,
      "step": 41560
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.423828125,
      "learning_rate": 1.486425952447798e-06,
      "loss": 2.1306,
      "step": 41570
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.443359375,
      "learning_rate": 1.4819532642431588e-06,
      "loss": 2.1324,
      "step": 41580
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.453125,
      "learning_rate": 1.4774869656773194e-06,
      "loss": 2.1278,
      "step": 41590
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.41796875,
      "learning_rate": 1.4730270588613876e-06,
      "loss": 2.1217,
      "step": 41600
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.478515625,
      "learning_rate": 1.468573545903444e-06,
      "loss": 2.138,
      "step": 41610
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.47265625,
      "learning_rate": 1.4641264289085543e-06,
      "loss": 2.1068,
      "step": 41620
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.48046875,
      "learning_rate": 1.4596857099787609e-06,
      "loss": 2.1027,
      "step": 41630
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4375,
      "learning_rate": 1.4552513912130726e-06,
      "loss": 2.1088,
      "step": 41640
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.44140625,
      "learning_rate": 1.4508234747074834e-06,
      "loss": 2.1131,
      "step": 41650
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.44140625,
      "learning_rate": 1.44640196255496e-06,
      "loss": 2.1198,
      "step": 41660
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.44921875,
      "learning_rate": 1.4419868568454347e-06,
      "loss": 2.1491,
      "step": 41670
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.427734375,
      "learning_rate": 1.4375781596658194e-06,
      "loss": 2.1198,
      "step": 41680
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.45703125,
      "learning_rate": 1.4331758730999939e-06,
      "loss": 2.0786,
      "step": 41690
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5,
      "learning_rate": 1.4287799992288037e-06,
      "loss": 2.1014,
      "step": 41700
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.44921875,
      "learning_rate": 1.4243905401300734e-06,
      "loss": 2.134,
      "step": 41710
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.42578125,
      "learning_rate": 1.4200074978785832e-06,
      "loss": 2.1039,
      "step": 41720
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.419921875,
      "learning_rate": 1.4156308745460928e-06,
      "loss": 2.0882,
      "step": 41730
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4765625,
      "learning_rate": 1.411260672201314e-06,
      "loss": 2.151,
      "step": 41740
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.427734375,
      "learning_rate": 1.406896892909934e-06,
      "loss": 2.1136,
      "step": 41750
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.43359375,
      "learning_rate": 1.4025395387346035e-06,
      "loss": 2.103,
      "step": 41760
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.455078125,
      "learning_rate": 1.3981886117349312e-06,
      "loss": 2.1098,
      "step": 41770
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.43359375,
      "learning_rate": 1.393844113967489e-06,
      "loss": 2.1477,
      "step": 41780
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.419921875,
      "learning_rate": 1.3895060474858158e-06,
      "loss": 2.1405,
      "step": 41790
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.45703125,
      "learning_rate": 1.385174414340401e-06,
      "loss": 2.1344,
      "step": 41800
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.466796875,
      "learning_rate": 1.380849216578699e-06,
      "loss": 2.1131,
      "step": 41810
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.44921875,
      "learning_rate": 1.3765304562451242e-06,
      "loss": 2.1128,
      "step": 41820
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.423828125,
      "learning_rate": 1.372218135381042e-06,
      "loss": 2.1234,
      "step": 41830
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4453125,
      "learning_rate": 1.3679122560247787e-06,
      "loss": 2.123,
      "step": 41840
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.482421875,
      "learning_rate": 1.3636128202116177e-06,
      "loss": 2.101,
      "step": 41850
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.423828125,
      "learning_rate": 1.3593198299737869e-06,
      "loss": 2.1083,
      "step": 41860
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4296875,
      "learning_rate": 1.3550332873404798e-06,
      "loss": 2.1425,
      "step": 41870
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.435546875,
      "learning_rate": 1.3507531943378354e-06,
      "loss": 2.1073,
      "step": 41880
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.427734375,
      "learning_rate": 1.3464795529889424e-06,
      "loss": 2.1124,
      "step": 41890
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.43359375,
      "learning_rate": 1.3422123653138464e-06,
      "loss": 2.0959,
      "step": 41900
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.451171875,
      "learning_rate": 1.3379516333295372e-06,
      "loss": 2.1069,
      "step": 41910
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4453125,
      "learning_rate": 1.3336973590499606e-06,
      "loss": 2.1041,
      "step": 41920
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.419921875,
      "learning_rate": 1.329449544485995e-06,
      "loss": 2.1082,
      "step": 41930
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.44140625,
      "learning_rate": 1.3252081916454823e-06,
      "loss": 2.1362,
      "step": 41940
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.439453125,
      "learning_rate": 1.3209733025331988e-06,
      "loss": 2.1294,
      "step": 41950
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.43359375,
      "learning_rate": 1.3167448791508734e-06,
      "loss": 2.1381,
      "step": 41960
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4453125,
      "learning_rate": 1.3125229234971691e-06,
      "loss": 2.1294,
      "step": 41970
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.443359375,
      "learning_rate": 1.3083074375677029e-06,
      "loss": 2.1277,
      "step": 41980
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.443359375,
      "learning_rate": 1.3040984233550242e-06,
      "loss": 2.1164,
      "step": 41990
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.43359375,
      "learning_rate": 1.2998958828486295e-06,
      "loss": 2.0983,
      "step": 42000
    },
    {
      "epoch": 0.87,
      "eval_accuracy": 0.5590182113901279,
      "eval_loss": 1.9930047988891602,
      "eval_runtime": 16.4418,
      "eval_samples_per_second": 36.188,
      "eval_steps_per_second": 1.156,
      "step": 42000
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.87890625,
      "learning_rate": 1.2956998180349555e-06,
      "loss": 2.1191,
      "step": 42010
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.431640625,
      "learning_rate": 1.291510230897372e-06,
      "loss": 2.1201,
      "step": 42020
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.453125,
      "learning_rate": 1.2873271234161932e-06,
      "loss": 2.1202,
      "step": 42030
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.453125,
      "learning_rate": 1.2831504975686693e-06,
      "loss": 2.1257,
      "step": 42040
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.42578125,
      "learning_rate": 1.2789803553289825e-06,
      "loss": 2.1017,
      "step": 42050
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.427734375,
      "learning_rate": 1.2748166986682546e-06,
      "loss": 2.086,
      "step": 42060
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.431640625,
      "learning_rate": 1.2706595295545449e-06,
      "loss": 2.1073,
      "step": 42070
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.44140625,
      "learning_rate": 1.2665088499528344e-06,
      "loss": 2.105,
      "step": 42080
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.423828125,
      "learning_rate": 1.2623646618250505e-06,
      "loss": 2.1214,
      "step": 42090
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.453125,
      "learning_rate": 1.2582269671300435e-06,
      "loss": 2.118,
      "step": 42100
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.447265625,
      "learning_rate": 1.2540957678235998e-06,
      "loss": 2.1249,
      "step": 42110
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.431640625,
      "learning_rate": 1.2499710658584296e-06,
      "loss": 2.1289,
      "step": 42120
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.435546875,
      "learning_rate": 1.2458528631841748e-06,
      "loss": 2.1342,
      "step": 42130
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.42578125,
      "learning_rate": 1.2417411617474117e-06,
      "loss": 2.1441,
      "step": 42140
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.431640625,
      "learning_rate": 1.237635963491629e-06,
      "loss": 2.0913,
      "step": 42150
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.43359375,
      "learning_rate": 1.2335372703572546e-06,
      "loss": 2.0991,
      "step": 42160
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.435546875,
      "learning_rate": 1.2294450842816403e-06,
      "loss": 2.097,
      "step": 42170
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.44921875,
      "learning_rate": 1.225359407199052e-06,
      "loss": 2.1143,
      "step": 42180
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.435546875,
      "learning_rate": 1.221280241040692e-06,
      "loss": 2.1487,
      "step": 42190
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.427734375,
      "learning_rate": 1.2172075877346728e-06,
      "loss": 2.1163,
      "step": 42200
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4296875,
      "learning_rate": 1.2131414492060412e-06,
      "loss": 2.1128,
      "step": 42210
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.431640625,
      "learning_rate": 1.2090818273767529e-06,
      "loss": 2.099,
      "step": 42220
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.453125,
      "learning_rate": 1.2050287241656931e-06,
      "loss": 2.0942,
      "step": 42230
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4375,
      "learning_rate": 1.2009821414886545e-06,
      "loss": 2.1352,
      "step": 42240
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.423828125,
      "learning_rate": 1.1969420812583603e-06,
      "loss": 2.1387,
      "step": 42250
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.45703125,
      "learning_rate": 1.1929085453844446e-06,
      "loss": 2.1062,
      "step": 42260
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.439453125,
      "learning_rate": 1.188881535773454e-06,
      "loss": 2.1165,
      "step": 42270
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.46875,
      "learning_rate": 1.1848610543288557e-06,
      "loss": 2.1221,
      "step": 42280
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.427734375,
      "learning_rate": 1.18084710295103e-06,
      "loss": 2.0941,
      "step": 42290
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.453125,
      "learning_rate": 1.1768396835372736e-06,
      "loss": 2.1231,
      "step": 42300
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.46484375,
      "learning_rate": 1.1728387979817868e-06,
      "loss": 2.1246,
      "step": 42310
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.42578125,
      "learning_rate": 1.1688444481756882e-06,
      "loss": 2.1094,
      "step": 42320
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.427734375,
      "learning_rate": 1.1648566360070095e-06,
      "loss": 2.1421,
      "step": 42330
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.44921875,
      "learning_rate": 1.1608753633606845e-06,
      "loss": 2.1195,
      "step": 42340
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.431640625,
      "learning_rate": 1.156900632118561e-06,
      "loss": 2.098,
      "step": 42350
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.458984375,
      "learning_rate": 1.152932444159398e-06,
      "loss": 2.1224,
      "step": 42360
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.431640625,
      "learning_rate": 1.1489708013588502e-06,
      "loss": 2.1212,
      "step": 42370
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.439453125,
      "learning_rate": 1.145015705589491e-06,
      "loss": 2.1188,
      "step": 42380
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.43359375,
      "learning_rate": 1.1410671587207954e-06,
      "loss": 2.1269,
      "step": 42390
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.43359375,
      "learning_rate": 1.137125162619137e-06,
      "loss": 2.0952,
      "step": 42400
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.470703125,
      "learning_rate": 1.1331897191478001e-06,
      "loss": 2.1496,
      "step": 42410
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.435546875,
      "learning_rate": 1.1292608301669721e-06,
      "loss": 2.1091,
      "step": 42420
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.439453125,
      "learning_rate": 1.1253384975337322e-06,
      "loss": 2.0972,
      "step": 42430
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.447265625,
      "learning_rate": 1.121422723102074e-06,
      "loss": 2.1133,
      "step": 42440
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.455078125,
      "learning_rate": 1.1175135087228854e-06,
      "loss": 2.1323,
      "step": 42450
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4453125,
      "learning_rate": 1.1136108562439519e-06,
      "loss": 2.0985,
      "step": 42460
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4375,
      "learning_rate": 1.1097147675099539e-06,
      "loss": 2.1211,
      "step": 42470
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.423828125,
      "learning_rate": 1.1058252443624773e-06,
      "loss": 2.097,
      "step": 42480
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.439453125,
      "learning_rate": 1.101942288640006e-06,
      "loss": 2.1088,
      "step": 42490
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.458984375,
      "learning_rate": 1.098065902177907e-06,
      "loss": 2.1109,
      "step": 42500
    },
    {
      "epoch": 0.88,
      "eval_accuracy": 0.5590888554835424,
      "eval_loss": 1.9930570125579834,
      "eval_runtime": 16.5088,
      "eval_samples_per_second": 36.041,
      "eval_steps_per_second": 1.151,
      "step": 42500
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4375,
      "learning_rate": 1.0941960868084549e-06,
      "loss": 2.1218,
      "step": 42510
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.443359375,
      "learning_rate": 1.0903328443608145e-06,
      "loss": 2.0967,
      "step": 42520
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.439453125,
      "learning_rate": 1.0864761766610388e-06,
      "loss": 2.1258,
      "step": 42530
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.439453125,
      "learning_rate": 1.082626085532079e-06,
      "loss": 2.1216,
      "step": 42540
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.44921875,
      "learning_rate": 1.0787825727937783e-06,
      "loss": 2.1111,
      "step": 42550
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4609375,
      "learning_rate": 1.0749456402628632e-06,
      "loss": 2.1542,
      "step": 42560
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.482421875,
      "learning_rate": 1.071115289752958e-06,
      "loss": 2.1169,
      "step": 42570
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.42578125,
      "learning_rate": 1.0672915230745728e-06,
      "loss": 2.1001,
      "step": 42580
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.453125,
      "learning_rate": 1.0634743420351033e-06,
      "loss": 2.0937,
      "step": 42590
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4609375,
      "learning_rate": 1.0596637484388334e-06,
      "loss": 2.1009,
      "step": 42600
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4375,
      "learning_rate": 1.0558597440869399e-06,
      "loss": 2.0736,
      "step": 42610
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.431640625,
      "learning_rate": 1.0520623307774735e-06,
      "loss": 2.0985,
      "step": 42620
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4296875,
      "learning_rate": 1.048271510305378e-06,
      "loss": 2.1038,
      "step": 42630
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.46484375,
      "learning_rate": 1.0444872844624804e-06,
      "loss": 2.1185,
      "step": 42640
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.42578125,
      "learning_rate": 1.040709655037483e-06,
      "loss": 2.1106,
      "step": 42650
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.46484375,
      "learning_rate": 1.036938623815979e-06,
      "loss": 2.1104,
      "step": 42660
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.431640625,
      "learning_rate": 1.0331741925804394e-06,
      "loss": 2.1271,
      "step": 42670
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.44921875,
      "learning_rate": 1.0294163631102193e-06,
      "loss": 2.1043,
      "step": 42680
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.435546875,
      "learning_rate": 1.025665137181543e-06,
      "loss": 2.1021,
      "step": 42690
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.48828125,
      "learning_rate": 1.021920516567525e-06,
      "loss": 2.0928,
      "step": 42700
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.44140625,
      "learning_rate": 1.0181825030381497e-06,
      "loss": 2.1148,
      "step": 42710
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.470703125,
      "learning_rate": 1.014451098360284e-06,
      "loss": 2.1105,
      "step": 42720
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.408203125,
      "learning_rate": 1.010726304297665e-06,
      "loss": 2.1245,
      "step": 42730
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.43359375,
      "learning_rate": 1.0070081226109146e-06,
      "loss": 2.0956,
      "step": 42740
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.43359375,
      "learning_rate": 1.0032965550575151e-06,
      "loss": 2.1112,
      "step": 42750
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4375,
      "learning_rate": 9.995916033918367e-07,
      "loss": 2.1226,
      "step": 42760
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.443359375,
      "learning_rate": 9.958932693651168e-07,
      "loss": 2.0988,
      "step": 42770
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4609375,
      "learning_rate": 9.922015547254587e-07,
      "loss": 2.1113,
      "step": 42780
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4296875,
      "learning_rate": 9.885164612178482e-07,
      "loss": 2.103,
      "step": 42790
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.44140625,
      "learning_rate": 9.848379905841337e-07,
      "loss": 2.0847,
      "step": 42800
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4375,
      "learning_rate": 9.811661445630343e-07,
      "loss": 2.1191,
      "step": 42810
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.453125,
      "learning_rate": 9.7750092489014e-07,
      "loss": 2.1427,
      "step": 42820
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4296875,
      "learning_rate": 9.738423332979085e-07,
      "loss": 2.1083,
      "step": 42830
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.419921875,
      "learning_rate": 9.701903715156613e-07,
      "loss": 2.0986,
      "step": 42840
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.453125,
      "learning_rate": 9.665450412695875e-07,
      "loss": 2.1249,
      "step": 42850
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4296875,
      "learning_rate": 9.629063442827458e-07,
      "loss": 2.0911,
      "step": 42860
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.41796875,
      "learning_rate": 9.592742822750568e-07,
      "loss": 2.1047,
      "step": 42870
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.421875,
      "learning_rate": 9.556488569632988e-07,
      "loss": 2.1153,
      "step": 42880
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.458984375,
      "learning_rate": 9.520300700611228e-07,
      "loss": 2.1311,
      "step": 42890
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.447265625,
      "learning_rate": 9.484179232790391e-07,
      "loss": 2.1254,
      "step": 42900
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.453125,
      "learning_rate": 9.448124183244134e-07,
      "loss": 2.1288,
      "step": 42910
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.435546875,
      "learning_rate": 9.412135569014807e-07,
      "loss": 2.1384,
      "step": 42920
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.439453125,
      "learning_rate": 9.376213407113321e-07,
      "loss": 2.083,
      "step": 42930
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.421875,
      "learning_rate": 9.340357714519143e-07,
      "loss": 2.1177,
      "step": 42940
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.451171875,
      "learning_rate": 9.304568508180384e-07,
      "loss": 2.1337,
      "step": 42950
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.419921875,
      "learning_rate": 9.268845805013642e-07,
      "loss": 2.1216,
      "step": 42960
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.453125,
      "learning_rate": 9.233189621904215e-07,
      "loss": 2.1355,
      "step": 42970
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.44921875,
      "learning_rate": 9.1975999757058e-07,
      "loss": 2.1295,
      "step": 42980
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.484375,
      "learning_rate": 9.162076883240794e-07,
      "loss": 2.0949,
      "step": 42990
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.42578125,
      "learning_rate": 9.126620361299998e-07,
      "loss": 2.172,
      "step": 43000
    },
    {
      "epoch": 0.89,
      "eval_accuracy": 0.559046140450315,
      "eval_loss": 1.9930146932601929,
      "eval_runtime": 16.4531,
      "eval_samples_per_second": 36.163,
      "eval_steps_per_second": 1.155,
      "step": 43000
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4375,
      "learning_rate": 9.091230426642854e-07,
      "loss": 2.1312,
      "step": 43010
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4453125,
      "learning_rate": 9.055907095997302e-07,
      "loss": 2.1183,
      "step": 43020
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.435546875,
      "learning_rate": 9.020650386059731e-07,
      "loss": 2.0889,
      "step": 43030
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.423828125,
      "learning_rate": 8.985460313495131e-07,
      "loss": 2.1385,
      "step": 43040
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.423828125,
      "learning_rate": 8.950336894936978e-07,
      "loss": 2.1234,
      "step": 43050
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.435546875,
      "learning_rate": 8.915280146987209e-07,
      "loss": 2.1088,
      "step": 43060
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.44921875,
      "learning_rate": 8.880290086216236e-07,
      "loss": 2.1122,
      "step": 43070
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.427734375,
      "learning_rate": 8.845366729162996e-07,
      "loss": 2.1481,
      "step": 43080
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.447265625,
      "learning_rate": 8.8105100923349e-07,
      "loss": 2.0863,
      "step": 43090
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.435546875,
      "learning_rate": 8.775720192207753e-07,
      "loss": 2.1109,
      "step": 43100
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.458984375,
      "learning_rate": 8.740997045225901e-07,
      "loss": 2.1144,
      "step": 43110
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.447265625,
      "learning_rate": 8.706340667802082e-07,
      "loss": 2.1011,
      "step": 43120
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.453125,
      "learning_rate": 8.671751076317475e-07,
      "loss": 2.0959,
      "step": 43130
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43359375,
      "learning_rate": 8.637228287121717e-07,
      "loss": 2.1237,
      "step": 43140
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.435546875,
      "learning_rate": 8.602772316532875e-07,
      "loss": 2.1407,
      "step": 43150
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.447265625,
      "learning_rate": 8.568383180837369e-07,
      "loss": 2.0922,
      "step": 43160
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.458984375,
      "learning_rate": 8.534060896290113e-07,
      "loss": 2.0881,
      "step": 43170
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.447265625,
      "learning_rate": 8.499805479114397e-07,
      "loss": 2.1546,
      "step": 43180
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43359375,
      "learning_rate": 8.465616945501836e-07,
      "loss": 2.1102,
      "step": 43190
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.474609375,
      "learning_rate": 8.431495311612536e-07,
      "loss": 2.1159,
      "step": 43200
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.451171875,
      "learning_rate": 8.397440593574895e-07,
      "loss": 2.1184,
      "step": 43210
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.44140625,
      "learning_rate": 8.363452807485755e-07,
      "loss": 2.1455,
      "step": 43220
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43359375,
      "learning_rate": 8.329531969410232e-07,
      "loss": 2.0902,
      "step": 43230
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.453125,
      "learning_rate": 8.295678095381865e-07,
      "loss": 2.1179,
      "step": 43240
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.427734375,
      "learning_rate": 8.261891201402572e-07,
      "loss": 2.1354,
      "step": 43250
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.44140625,
      "learning_rate": 8.228171303442494e-07,
      "loss": 2.1055,
      "step": 43260
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.421875,
      "learning_rate": 8.1945184174402e-07,
      "loss": 2.1178,
      "step": 43270
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4375,
      "learning_rate": 8.160932559302598e-07,
      "loss": 2.1001,
      "step": 43280
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.443359375,
      "learning_rate": 8.127413744904805e-07,
      "loss": 2.1227,
      "step": 43290
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.455078125,
      "learning_rate": 8.093961990090365e-07,
      "loss": 2.1074,
      "step": 43300
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.455078125,
      "learning_rate": 8.060577310671064e-07,
      "loss": 2.1241,
      "step": 43310
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.427734375,
      "learning_rate": 8.027259722426994e-07,
      "loss": 2.1054,
      "step": 43320
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.48828125,
      "learning_rate": 7.994009241106526e-07,
      "loss": 2.1314,
      "step": 43330
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.44921875,
      "learning_rate": 7.960825882426354e-07,
      "loss": 2.1079,
      "step": 43340
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.423828125,
      "learning_rate": 7.927709662071364e-07,
      "loss": 2.1226,
      "step": 43350
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.455078125,
      "learning_rate": 7.894660595694786e-07,
      "loss": 2.1269,
      "step": 43360
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.443359375,
      "learning_rate": 7.86167869891809e-07,
      "loss": 2.1135,
      "step": 43370
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43359375,
      "learning_rate": 7.828763987330956e-07,
      "loss": 2.1089,
      "step": 43380
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.447265625,
      "learning_rate": 7.795916476491355e-07,
      "loss": 2.1283,
      "step": 43390
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.46875,
      "learning_rate": 7.763136181925467e-07,
      "loss": 2.1094,
      "step": 43400
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4375,
      "learning_rate": 7.7304231191277e-07,
      "loss": 2.12,
      "step": 43410
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4296875,
      "learning_rate": 7.697777303560699e-07,
      "loss": 2.1002,
      "step": 43420
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.427734375,
      "learning_rate": 7.665198750655339e-07,
      "loss": 2.1274,
      "step": 43430
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.435546875,
      "learning_rate": 7.632687475810634e-07,
      "loss": 2.1221,
      "step": 43440
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4453125,
      "learning_rate": 7.600243494393894e-07,
      "loss": 2.1052,
      "step": 43450
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.435546875,
      "learning_rate": 7.567866821740499e-07,
      "loss": 2.1455,
      "step": 43460
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.427734375,
      "learning_rate": 7.535557473154142e-07,
      "loss": 2.1079,
      "step": 43470
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.42578125,
      "learning_rate": 7.503315463906607e-07,
      "loss": 2.1307,
      "step": 43480
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.42578125,
      "learning_rate": 7.47114080923787e-07,
      "loss": 2.1143,
      "step": 43490
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.44921875,
      "learning_rate": 7.439033524356131e-07,
      "loss": 2.0882,
      "step": 43500
    },
    {
      "epoch": 0.9,
      "eval_accuracy": 0.5590527119938885,
      "eval_loss": 1.9930274486541748,
      "eval_runtime": 16.4545,
      "eval_samples_per_second": 36.16,
      "eval_steps_per_second": 1.155,
      "step": 43500
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.421875,
      "learning_rate": 7.406993624437619e-07,
      "loss": 2.1017,
      "step": 43510
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4375,
      "learning_rate": 7.375021124626802e-07,
      "loss": 2.0924,
      "step": 43520
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.419921875,
      "learning_rate": 7.343116040036324e-07,
      "loss": 2.1026,
      "step": 43530
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4375,
      "learning_rate": 7.311278385746855e-07,
      "loss": 2.1089,
      "step": 43540
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.443359375,
      "learning_rate": 7.279508176807259e-07,
      "loss": 2.1175,
      "step": 43550
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.45703125,
      "learning_rate": 7.247805428234538e-07,
      "loss": 2.0975,
      "step": 43560
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.439453125,
      "learning_rate": 7.21617015501374e-07,
      "loss": 2.1424,
      "step": 43570
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.451171875,
      "learning_rate": 7.184602372098087e-07,
      "loss": 2.0906,
      "step": 43580
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.41796875,
      "learning_rate": 7.153102094408893e-07,
      "loss": 2.083,
      "step": 43590
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.451171875,
      "learning_rate": 7.121669336835479e-07,
      "loss": 2.1038,
      "step": 43600
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.451171875,
      "learning_rate": 7.090304114235347e-07,
      "loss": 2.1372,
      "step": 43610
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.447265625,
      "learning_rate": 7.059006441434051e-07,
      "loss": 2.1194,
      "step": 43620
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.45703125,
      "learning_rate": 7.027776333225211e-07,
      "loss": 2.1084,
      "step": 43630
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.44140625,
      "learning_rate": 6.9966138043705e-07,
      "loss": 2.1085,
      "step": 43640
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.435546875,
      "learning_rate": 6.965518869599652e-07,
      "loss": 2.0933,
      "step": 43650
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.419921875,
      "learning_rate": 6.934491543610494e-07,
      "loss": 2.0934,
      "step": 43660
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.46875,
      "learning_rate": 6.903531841068795e-07,
      "loss": 2.1324,
      "step": 43670
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.423828125,
      "learning_rate": 6.872639776608464e-07,
      "loss": 2.1084,
      "step": 43680
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.423828125,
      "learning_rate": 6.841815364831405e-07,
      "loss": 2.126,
      "step": 43690
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.408203125,
      "learning_rate": 6.811058620307514e-07,
      "loss": 2.0742,
      "step": 43700
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.423828125,
      "learning_rate": 6.780369557574762e-07,
      "loss": 2.145,
      "step": 43710
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.43359375,
      "learning_rate": 6.749748191139027e-07,
      "loss": 2.1475,
      "step": 43720
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4375,
      "learning_rate": 6.719194535474332e-07,
      "loss": 2.1074,
      "step": 43730
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.43359375,
      "learning_rate": 6.688708605022558e-07,
      "loss": 2.127,
      "step": 43740
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.41796875,
      "learning_rate": 6.658290414193663e-07,
      "loss": 2.1133,
      "step": 43750
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.439453125,
      "learning_rate": 6.627939977365544e-07,
      "loss": 2.1232,
      "step": 43760
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.15625,
      "learning_rate": 6.597657308884075e-07,
      "loss": 2.0822,
      "step": 43770
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.439453125,
      "learning_rate": 6.567442423063158e-07,
      "loss": 2.1159,
      "step": 43780
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.53125,
      "learning_rate": 6.537295334184534e-07,
      "loss": 2.1127,
      "step": 43790
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4453125,
      "learning_rate": 6.507216056498006e-07,
      "loss": 2.1474,
      "step": 43800
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.44921875,
      "learning_rate": 6.477204604221298e-07,
      "loss": 2.1023,
      "step": 43810
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.423828125,
      "learning_rate": 6.447260991540066e-07,
      "loss": 2.1411,
      "step": 43820
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.49609375,
      "learning_rate": 6.41738523260787e-07,
      "loss": 2.0886,
      "step": 43830
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.462890625,
      "learning_rate": 6.387577341546247e-07,
      "loss": 2.1056,
      "step": 43840
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.43359375,
      "learning_rate": 6.357837332444627e-07,
      "loss": 2.1066,
      "step": 43850
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.451171875,
      "learning_rate": 6.328165219360365e-07,
      "loss": 2.1097,
      "step": 43860
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.484375,
      "learning_rate": 6.298561016318693e-07,
      "loss": 2.103,
      "step": 43870
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.443359375,
      "learning_rate": 6.269024737312817e-07,
      "loss": 2.1166,
      "step": 43880
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.423828125,
      "learning_rate": 6.239556396303753e-07,
      "loss": 2.0711,
      "step": 43890
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.44921875,
      "learning_rate": 6.210156007220441e-07,
      "loss": 2.1139,
      "step": 43900
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.439453125,
      "learning_rate": 6.180823583959733e-07,
      "loss": 2.0997,
      "step": 43910
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.431640625,
      "learning_rate": 6.15155914038627e-07,
      "loss": 2.1208,
      "step": 43920
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.435546875,
      "learning_rate": 6.122362690332655e-07,
      "loss": 2.1486,
      "step": 43930
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4375,
      "learning_rate": 6.093234247599317e-07,
      "loss": 2.1087,
      "step": 43940
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.431640625,
      "learning_rate": 6.064173825954506e-07,
      "loss": 2.1032,
      "step": 43950
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.427734375,
      "learning_rate": 6.035181439134368e-07,
      "loss": 2.1044,
      "step": 43960
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.427734375,
      "learning_rate": 6.00625710084286e-07,
      "loss": 2.1427,
      "step": 43970
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4453125,
      "learning_rate": 5.977400824751794e-07,
      "loss": 2.0962,
      "step": 43980
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.44921875,
      "learning_rate": 5.948612624500776e-07,
      "loss": 2.1323,
      "step": 43990
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.42578125,
      "learning_rate": 5.919892513697306e-07,
      "loss": 2.0646,
      "step": 44000
    },
    {
      "epoch": 0.91,
      "eval_accuracy": 0.5590773552822889,
      "eval_loss": 1.993033528327942,
      "eval_runtime": 16.4535,
      "eval_samples_per_second": 36.163,
      "eval_steps_per_second": 1.155,
      "step": 44000
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.43359375,
      "learning_rate": 5.891240505916623e-07,
      "loss": 2.1339,
      "step": 44010
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.451171875,
      "learning_rate": 5.862656614701795e-07,
      "loss": 2.1037,
      "step": 44020
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4453125,
      "learning_rate": 5.83414085356373e-07,
      "loss": 2.1265,
      "step": 44030
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.44140625,
      "learning_rate": 5.805693235981113e-07,
      "loss": 2.127,
      "step": 44040
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4609375,
      "learning_rate": 5.777313775400389e-07,
      "loss": 2.1356,
      "step": 44050
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4296875,
      "learning_rate": 5.749002485235827e-07,
      "loss": 2.1389,
      "step": 44060
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4375,
      "learning_rate": 5.720759378869455e-07,
      "loss": 2.0945,
      "step": 44070
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4375,
      "learning_rate": 5.692584469651063e-07,
      "loss": 2.1135,
      "step": 44080
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4453125,
      "learning_rate": 5.664477770898214e-07,
      "loss": 2.1239,
      "step": 44090
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4375,
      "learning_rate": 5.636439295896279e-07,
      "loss": 2.1188,
      "step": 44100
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.455078125,
      "learning_rate": 5.608469057898291e-07,
      "loss": 2.1432,
      "step": 44110
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.447265625,
      "learning_rate": 5.580567070125076e-07,
      "loss": 2.1249,
      "step": 44120
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.44921875,
      "learning_rate": 5.552733345765232e-07,
      "loss": 2.1214,
      "step": 44130
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.431640625,
      "learning_rate": 5.52496789797502e-07,
      "loss": 2.1195,
      "step": 44140
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.43359375,
      "learning_rate": 5.497270739878474e-07,
      "loss": 2.1074,
      "step": 44150
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4140625,
      "learning_rate": 5.469641884567355e-07,
      "loss": 2.1507,
      "step": 44160
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.431640625,
      "learning_rate": 5.442081345101102e-07,
      "loss": 2.1158,
      "step": 44170
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.431640625,
      "learning_rate": 5.414589134506892e-07,
      "loss": 2.0916,
      "step": 44180
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4453125,
      "learning_rate": 5.387165265779631e-07,
      "loss": 2.0888,
      "step": 44190
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.443359375,
      "learning_rate": 5.359809751881834e-07,
      "loss": 2.1165,
      "step": 44200
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.458984375,
      "learning_rate": 5.332522605743805e-07,
      "loss": 2.137,
      "step": 44210
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.43359375,
      "learning_rate": 5.305303840263465e-07,
      "loss": 2.1321,
      "step": 44220
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.455078125,
      "learning_rate": 5.278153468306435e-07,
      "loss": 2.0847,
      "step": 44230
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.42578125,
      "learning_rate": 5.251071502706018e-07,
      "loss": 2.1404,
      "step": 44240
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.490234375,
      "learning_rate": 5.224057956263157e-07,
      "loss": 2.1265,
      "step": 44250
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.42578125,
      "learning_rate": 5.197112841746504e-07,
      "loss": 2.1401,
      "step": 44260
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.427734375,
      "learning_rate": 5.170236171892307e-07,
      "loss": 2.0868,
      "step": 44270
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.431640625,
      "learning_rate": 5.143427959404484e-07,
      "loss": 2.1228,
      "step": 44280
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.453125,
      "learning_rate": 5.116688216954635e-07,
      "loss": 2.0813,
      "step": 44290
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4140625,
      "learning_rate": 5.090016957181914e-07,
      "loss": 2.0937,
      "step": 44300
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.451171875,
      "learning_rate": 5.063414192693172e-07,
      "loss": 2.1192,
      "step": 44310
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.43359375,
      "learning_rate": 5.036879936062883e-07,
      "loss": 2.0863,
      "step": 44320
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.439453125,
      "learning_rate": 5.010414199833097e-07,
      "loss": 2.1189,
      "step": 44330
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.435546875,
      "learning_rate": 4.984016996513474e-07,
      "loss": 2.0778,
      "step": 44340
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.419921875,
      "learning_rate": 4.957688338581368e-07,
      "loss": 2.1197,
      "step": 44350
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.423828125,
      "learning_rate": 4.931428238481606e-07,
      "loss": 2.108,
      "step": 44360
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.427734375,
      "learning_rate": 4.905236708626709e-07,
      "loss": 2.055,
      "step": 44370
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.427734375,
      "learning_rate": 4.879113761396741e-07,
      "loss": 2.0908,
      "step": 44380
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.423828125,
      "learning_rate": 4.853059409139376e-07,
      "loss": 2.1413,
      "step": 44390
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.43359375,
      "learning_rate": 4.827073664169812e-07,
      "loss": 2.1274,
      "step": 44400
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4453125,
      "learning_rate": 4.801156538770857e-07,
      "loss": 2.0934,
      "step": 44410
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4375,
      "learning_rate": 4.775308045192927e-07,
      "loss": 2.1133,
      "step": 44420
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.435546875,
      "learning_rate": 4.7495281956539016e-07,
      "loss": 2.1339,
      "step": 44430
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.45703125,
      "learning_rate": 4.7238170023393147e-07,
      "loss": 2.128,
      "step": 44440
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.43359375,
      "learning_rate": 4.6981744774021284e-07,
      "loss": 2.1505,
      "step": 44450
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4296875,
      "learning_rate": 4.6726006329629645e-07,
      "loss": 2.093,
      "step": 44460
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.546875,
      "learning_rate": 4.6470954811099195e-07,
      "loss": 2.105,
      "step": 44470
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.482421875,
      "learning_rate": 4.621659033898634e-07,
      "loss": 2.0961,
      "step": 44480
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.44140625,
      "learning_rate": 4.5962913033522733e-07,
      "loss": 2.154,
      "step": 44490
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.421875,
      "learning_rate": 4.570992301461513e-07,
      "loss": 2.1223,
      "step": 44500
    },
    {
      "epoch": 0.93,
      "eval_accuracy": 0.5591348562885565,
      "eval_loss": 1.9930362701416016,
      "eval_runtime": 16.4511,
      "eval_samples_per_second": 36.168,
      "eval_steps_per_second": 1.155,
      "step": 44500
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.59765625,
      "learning_rate": 4.54576204018457e-07,
      "loss": 2.0922,
      "step": 44510
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43359375,
      "learning_rate": 4.520600531447139e-07,
      "loss": 2.1062,
      "step": 44520
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4296875,
      "learning_rate": 4.4955077871424223e-07,
      "loss": 2.1182,
      "step": 44530
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4296875,
      "learning_rate": 4.4704838191311494e-07,
      "loss": 2.1066,
      "step": 44540
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.439453125,
      "learning_rate": 4.4455286392414916e-07,
      "loss": 2.1063,
      "step": 44550
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4296875,
      "learning_rate": 4.420642259269148e-07,
      "loss": 2.108,
      "step": 44560
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.453125,
      "learning_rate": 4.395824690977257e-07,
      "loss": 2.1246,
      "step": 44570
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.42578125,
      "learning_rate": 4.371075946096503e-07,
      "loss": 2.1458,
      "step": 44580
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4453125,
      "learning_rate": 4.3463960363249286e-07,
      "loss": 2.1224,
      "step": 44590
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.447265625,
      "learning_rate": 4.3217849733281523e-07,
      "loss": 2.1193,
      "step": 44600
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.421875,
      "learning_rate": 4.297242768739185e-07,
      "loss": 2.1001,
      "step": 44610
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.435546875,
      "learning_rate": 4.272769434158497e-07,
      "loss": 2.1277,
      "step": 44620
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.42578125,
      "learning_rate": 4.2483649811540346e-07,
      "loss": 2.1026,
      "step": 44630
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.427734375,
      "learning_rate": 4.2240294212611553e-07,
      "loss": 2.1033,
      "step": 44640
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.427734375,
      "learning_rate": 4.1997627659826564e-07,
      "loss": 2.1107,
      "step": 44650
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.451171875,
      "learning_rate": 4.17556502678878e-07,
      "loss": 2.1202,
      "step": 44660
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.421875,
      "learning_rate": 4.1514362151171934e-07,
      "loss": 2.126,
      "step": 44670
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.435546875,
      "learning_rate": 4.12737634237294e-07,
      "loss": 2.1066,
      "step": 44680
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.44140625,
      "learning_rate": 4.10338541992854e-07,
      "loss": 2.1077,
      "step": 44690
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.455078125,
      "learning_rate": 4.079463459123922e-07,
      "loss": 2.1392,
      "step": 44700
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.447265625,
      "learning_rate": 4.055610471266341e-07,
      "loss": 2.1258,
      "step": 44710
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4453125,
      "learning_rate": 4.031826467630545e-07,
      "loss": 2.1216,
      "step": 44720
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.44921875,
      "learning_rate": 4.008111459458591e-07,
      "loss": 2.1069,
      "step": 44730
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.423828125,
      "learning_rate": 3.984465457959996e-07,
      "loss": 2.129,
      "step": 44740
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.44921875,
      "learning_rate": 3.9608884743116025e-07,
      "loss": 2.1299,
      "step": 44750
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4453125,
      "learning_rate": 3.9373805196576627e-07,
      "loss": 2.1517,
      "step": 44760
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.421875,
      "learning_rate": 3.9139416051098053e-07,
      "loss": 2.1009,
      "step": 44770
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.431640625,
      "learning_rate": 3.890571741746984e-07,
      "loss": 2.1082,
      "step": 44780
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.42578125,
      "learning_rate": 3.867270940615564e-07,
      "loss": 2.1467,
      "step": 44790
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43359375,
      "learning_rate": 3.844039212729267e-07,
      "loss": 2.1072,
      "step": 44800
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.431640625,
      "learning_rate": 3.8208765690690925e-07,
      "loss": 2.1238,
      "step": 44810
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.431640625,
      "learning_rate": 3.7977830205834676e-07,
      "loss": 2.143,
      "step": 44820
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.439453125,
      "learning_rate": 3.7747585781881253e-07,
      "loss": 2.1126,
      "step": 44830
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.41015625,
      "learning_rate": 3.751803252766128e-07,
      "loss": 2.1183,
      "step": 44840
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.48828125,
      "learning_rate": 3.7289170551678633e-07,
      "loss": 2.1425,
      "step": 44850
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.435546875,
      "learning_rate": 3.7060999962111095e-07,
      "loss": 2.1106,
      "step": 44860
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.490234375,
      "learning_rate": 3.6833520866808745e-07,
      "loss": 2.1488,
      "step": 44870
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43359375,
      "learning_rate": 3.6606733373295064e-07,
      "loss": 2.1408,
      "step": 44880
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5,
      "learning_rate": 3.6380637588767306e-07,
      "loss": 2.1283,
      "step": 44890
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4375,
      "learning_rate": 3.615523362009465e-07,
      "loss": 2.1038,
      "step": 44900
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40625,
      "learning_rate": 3.593052157382004e-07,
      "loss": 2.0932,
      "step": 44910
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43359375,
      "learning_rate": 3.570650155615951e-07,
      "loss": 2.1117,
      "step": 44920
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.427734375,
      "learning_rate": 3.548317367300119e-07,
      "loss": 2.1107,
      "step": 44930
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.41796875,
      "learning_rate": 3.526053802990664e-07,
      "loss": 2.0999,
      "step": 44940
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.423828125,
      "learning_rate": 3.5038594732110354e-07,
      "loss": 2.0925,
      "step": 44950
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.412109375,
      "learning_rate": 3.481734388451907e-07,
      "loss": 2.1169,
      "step": 44960
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.453125,
      "learning_rate": 3.459678559171248e-07,
      "loss": 2.1306,
      "step": 44970
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4296875,
      "learning_rate": 3.437691995794301e-07,
      "loss": 2.0983,
      "step": 44980
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.44140625,
      "learning_rate": 3.415774708713554e-07,
      "loss": 2.0743,
      "step": 44990
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.42578125,
      "learning_rate": 3.393926708288753e-07,
      "loss": 2.1342,
      "step": 45000
    },
    {
      "epoch": 0.94,
      "eval_accuracy": 0.5590576406515686,
      "eval_loss": 1.9930360317230225,
      "eval_runtime": 16.4305,
      "eval_samples_per_second": 36.213,
      "eval_steps_per_second": 1.156,
      "step": 45000
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.427734375,
      "learning_rate": 3.37214800484687e-07,
      "loss": 2.1057,
      "step": 45010
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.423828125,
      "learning_rate": 3.3504386086822046e-07,
      "loss": 2.1398,
      "step": 45020
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.431640625,
      "learning_rate": 3.3287985300561985e-07,
      "loss": 2.0922,
      "step": 45030
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.439453125,
      "learning_rate": 3.307227779197569e-07,
      "loss": 2.1144,
      "step": 45040
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.52734375,
      "learning_rate": 3.2857263663022943e-07,
      "loss": 2.1353,
      "step": 45050
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.462890625,
      "learning_rate": 3.26429430153351e-07,
      "loss": 2.1495,
      "step": 45060
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.458984375,
      "learning_rate": 3.242931595021631e-07,
      "loss": 2.0845,
      "step": 45070
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.451171875,
      "learning_rate": 3.221638256864279e-07,
      "loss": 2.1206,
      "step": 45080
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.423828125,
      "learning_rate": 3.200414297126253e-07,
      "loss": 2.1069,
      "step": 45090
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.46484375,
      "learning_rate": 3.179259725839595e-07,
      "loss": 2.1014,
      "step": 45100
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.435546875,
      "learning_rate": 3.15817455300354e-07,
      "loss": 2.138,
      "step": 45110
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.447265625,
      "learning_rate": 3.137158788584499e-07,
      "loss": 2.1401,
      "step": 45120
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.42578125,
      "learning_rate": 3.1162124425161087e-07,
      "loss": 2.1208,
      "step": 45130
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.46875,
      "learning_rate": 3.0953355246991663e-07,
      "loss": 2.1036,
      "step": 45140
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.421875,
      "learning_rate": 3.0745280450016954e-07,
      "loss": 2.1082,
      "step": 45150
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.498046875,
      "learning_rate": 3.0537900132588113e-07,
      "loss": 2.1344,
      "step": 45160
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.455078125,
      "learning_rate": 3.0331214392728903e-07,
      "loss": 2.1449,
      "step": 45170
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.447265625,
      "learning_rate": 3.0125223328134334e-07,
      "loss": 2.114,
      "step": 45180
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.44921875,
      "learning_rate": 2.991992703617136e-07,
      "loss": 2.1032,
      "step": 45190
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.419921875,
      "learning_rate": 2.9715325613878184e-07,
      "loss": 2.142,
      "step": 45200
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.427734375,
      "learning_rate": 2.9511419157964625e-07,
      "loss": 2.1107,
      "step": 45210
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.494140625,
      "learning_rate": 2.9308207764812244e-07,
      "loss": 2.1332,
      "step": 45220
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4375,
      "learning_rate": 2.910569153047404e-07,
      "loss": 2.1236,
      "step": 45230
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4375,
      "learning_rate": 2.890387055067412e-07,
      "loss": 2.106,
      "step": 45240
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.431640625,
      "learning_rate": 2.8702744920808154e-07,
      "loss": 2.1194,
      "step": 45250
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.421875,
      "learning_rate": 2.850231473594328e-07,
      "loss": 2.1311,
      "step": 45260
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.44921875,
      "learning_rate": 2.830258009081771e-07,
      "loss": 2.1452,
      "step": 45270
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.44140625,
      "learning_rate": 2.8103541079840924e-07,
      "loss": 2.1229,
      "step": 45280
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.423828125,
      "learning_rate": 2.7905197797093663e-07,
      "loss": 2.1255,
      "step": 45290
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.265625,
      "learning_rate": 2.770755033632777e-07,
      "loss": 2.1103,
      "step": 45300
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.490234375,
      "learning_rate": 2.751059879096618e-07,
      "loss": 2.1238,
      "step": 45310
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.421875,
      "learning_rate": 2.7314343254102925e-07,
      "loss": 2.098,
      "step": 45320
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.447265625,
      "learning_rate": 2.7118783818503136e-07,
      "loss": 2.1116,
      "step": 45330
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.419921875,
      "learning_rate": 2.69239205766027e-07,
      "loss": 2.129,
      "step": 45340
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.435546875,
      "learning_rate": 2.6729753620508613e-07,
      "loss": 2.135,
      "step": 45350
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.44921875,
      "learning_rate": 2.6536283041998454e-07,
      "loss": 2.1212,
      "step": 45360
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.439453125,
      "learning_rate": 2.6343508932521243e-07,
      "loss": 2.1066,
      "step": 45370
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.423828125,
      "learning_rate": 2.615143138319609e-07,
      "loss": 2.1065,
      "step": 45380
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.47265625,
      "learning_rate": 2.5960050484813536e-07,
      "loss": 2.1595,
      "step": 45390
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.435546875,
      "learning_rate": 2.576936632783422e-07,
      "loss": 2.0861,
      "step": 45400
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4296875,
      "learning_rate": 2.5579379002389716e-07,
      "loss": 2.114,
      "step": 45410
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.451171875,
      "learning_rate": 2.5390088598282344e-07,
      "loss": 2.1287,
      "step": 45420
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.427734375,
      "learning_rate": 2.520149520498488e-07,
      "loss": 2.0905,
      "step": 45430
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.412109375,
      "learning_rate": 2.5013598911640514e-07,
      "loss": 2.1268,
      "step": 45440
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.447265625,
      "learning_rate": 2.482639980706336e-07,
      "loss": 2.0868,
      "step": 45450
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4375,
      "learning_rate": 2.463989797973748e-07,
      "loss": 2.0986,
      "step": 45460
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.43359375,
      "learning_rate": 2.4454093517817687e-07,
      "loss": 2.1135,
      "step": 45470
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.447265625,
      "learning_rate": 2.4268986509128733e-07,
      "loss": 2.0712,
      "step": 45480
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.435546875,
      "learning_rate": 2.408457704116662e-07,
      "loss": 2.0803,
      "step": 45490
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.435546875,
      "learning_rate": 2.39008652010963e-07,
      "loss": 2.0991,
      "step": 45500
    },
    {
      "epoch": 0.95,
      "eval_accuracy": 0.5589902823299407,
      "eval_loss": 1.9930479526519775,
      "eval_runtime": 16.443,
      "eval_samples_per_second": 36.186,
      "eval_steps_per_second": 1.156,
      "step": 45500
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.416015625,
      "learning_rate": 2.3717851075754305e-07,
      "loss": 2.1284,
      "step": 45510
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.439453125,
      "learning_rate": 2.3535534751646436e-07,
      "loss": 2.1352,
      "step": 45520
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4453125,
      "learning_rate": 2.335391631494943e-07,
      "loss": 2.1524,
      "step": 45530
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.447265625,
      "learning_rate": 2.317299585150928e-07,
      "loss": 2.1166,
      "step": 45540
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4453125,
      "learning_rate": 2.2992773446842585e-07,
      "loss": 2.1092,
      "step": 45550
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.59765625,
      "learning_rate": 2.2813249186136197e-07,
      "loss": 2.1172,
      "step": 45560
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44921875,
      "learning_rate": 2.263442315424641e-07,
      "loss": 2.1471,
      "step": 45570
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.43359375,
      "learning_rate": 2.245629543569977e-07,
      "loss": 2.1398,
      "step": 45580
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4375,
      "learning_rate": 2.2278866114693098e-07,
      "loss": 2.1437,
      "step": 45590
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.46875,
      "learning_rate": 2.2102135275092307e-07,
      "loss": 2.1167,
      "step": 45600
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.447265625,
      "learning_rate": 2.192610300043374e-07,
      "loss": 2.1036,
      "step": 45610
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.435546875,
      "learning_rate": 2.1750769373923508e-07,
      "loss": 2.0965,
      "step": 45620
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44921875,
      "learning_rate": 2.1576134478437315e-07,
      "loss": 2.1495,
      "step": 45630
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44140625,
      "learning_rate": 2.1402198396520633e-07,
      "loss": 2.1311,
      "step": 45640
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.46484375,
      "learning_rate": 2.1228961210388698e-07,
      "loss": 2.1074,
      "step": 45650
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.435546875,
      "learning_rate": 2.105642300192634e-07,
      "loss": 2.1014,
      "step": 45660
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44140625,
      "learning_rate": 2.088458385268799e-07,
      "loss": 2.135,
      "step": 45670
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.48828125,
      "learning_rate": 2.0713443843897674e-07,
      "loss": 2.0879,
      "step": 45680
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.431640625,
      "learning_rate": 2.054300305644885e-07,
      "loss": 2.1002,
      "step": 45690
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.421875,
      "learning_rate": 2.0373261570904744e-07,
      "loss": 2.1279,
      "step": 45700
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.443359375,
      "learning_rate": 2.0204219467497842e-07,
      "loss": 2.0985,
      "step": 45710
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4375,
      "learning_rate": 2.0035876826129728e-07,
      "loss": 2.1147,
      "step": 45720
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44140625,
      "learning_rate": 1.9868233726372253e-07,
      "loss": 2.0787,
      "step": 45730
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.42578125,
      "learning_rate": 1.9701290247465697e-07,
      "loss": 2.127,
      "step": 45740
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4453125,
      "learning_rate": 1.9535046468320272e-07,
      "loss": 2.0871,
      "step": 45750
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.490234375,
      "learning_rate": 1.9369502467514788e-07,
      "loss": 2.1296,
      "step": 45760
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.47265625,
      "learning_rate": 1.9204658323298152e-07,
      "loss": 2.1679,
      "step": 45770
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.45703125,
      "learning_rate": 1.904051411358787e-07,
      "loss": 2.1201,
      "step": 45780
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.439453125,
      "learning_rate": 1.8877069915970712e-07,
      "loss": 2.1464,
      "step": 45790
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.466796875,
      "learning_rate": 1.8714325807702548e-07,
      "loss": 2.1096,
      "step": 45800
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.439453125,
      "learning_rate": 1.8552281865708675e-07,
      "loss": 2.0995,
      "step": 45810
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.439453125,
      "learning_rate": 1.8390938166582826e-07,
      "loss": 2.0881,
      "step": 45820
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4453125,
      "learning_rate": 1.823029478658833e-07,
      "loss": 2.1175,
      "step": 45830
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44140625,
      "learning_rate": 1.8070351801657114e-07,
      "loss": 2.1802,
      "step": 45840
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4375,
      "learning_rate": 1.791110928739037e-07,
      "loss": 2.1145,
      "step": 45850
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44140625,
      "learning_rate": 1.7752567319057723e-07,
      "loss": 2.0945,
      "step": 45860
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.431640625,
      "learning_rate": 1.7594725971598225e-07,
      "loss": 2.0999,
      "step": 45870
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44921875,
      "learning_rate": 1.74375853196192e-07,
      "loss": 2.1349,
      "step": 45880
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44140625,
      "learning_rate": 1.7281145437397394e-07,
      "loss": 2.1026,
      "step": 45890
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.462890625,
      "learning_rate": 1.712540639887783e-07,
      "loss": 2.1191,
      "step": 45900
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.43359375,
      "learning_rate": 1.6970368277674285e-07,
      "loss": 2.1523,
      "step": 45910
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.435546875,
      "learning_rate": 1.6816031147069476e-07,
      "loss": 2.1087,
      "step": 45920
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.43359375,
      "learning_rate": 1.6662395080014547e-07,
      "loss": 2.0808,
      "step": 45930
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4765625,
      "learning_rate": 1.650946014912974e-07,
      "loss": 2.1197,
      "step": 45940
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4296875,
      "learning_rate": 1.6357226426703064e-07,
      "loss": 2.1047,
      "step": 45950
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.451171875,
      "learning_rate": 1.620569398469196e-07,
      "loss": 2.0961,
      "step": 45960
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4609375,
      "learning_rate": 1.60548628947218e-07,
      "loss": 2.1058,
      "step": 45970
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.423828125,
      "learning_rate": 1.5904733228086553e-07,
      "loss": 2.1122,
      "step": 45980
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4375,
      "learning_rate": 1.5755305055748625e-07,
      "loss": 2.1416,
      "step": 45990
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.455078125,
      "learning_rate": 1.560657844833918e-07,
      "loss": 2.1431,
      "step": 46000
    },
    {
      "epoch": 0.96,
      "eval_accuracy": 0.5591742855499972,
      "eval_loss": 1.992998480796814,
      "eval_runtime": 16.4485,
      "eval_samples_per_second": 36.174,
      "eval_steps_per_second": 1.155,
      "step": 46000
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.427734375,
      "learning_rate": 1.5458553476157488e-07,
      "loss": 2.0971,
      "step": 46010
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.423828125,
      "learning_rate": 1.5311230209171078e-07,
      "loss": 2.1007,
      "step": 46020
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.43359375,
      "learning_rate": 1.5164608717016082e-07,
      "loss": 2.1256,
      "step": 46030
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.43359375,
      "learning_rate": 1.501868906899656e-07,
      "loss": 2.1464,
      "step": 46040
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.48828125,
      "learning_rate": 1.4873471334085175e-07,
      "loss": 2.1391,
      "step": 46050
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.439453125,
      "learning_rate": 1.4728955580922853e-07,
      "loss": 2.1074,
      "step": 46060
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.451171875,
      "learning_rate": 1.458514187781812e-07,
      "loss": 2.1233,
      "step": 46070
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.46484375,
      "learning_rate": 1.4442030292748432e-07,
      "loss": 2.1176,
      "step": 46080
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.431640625,
      "learning_rate": 1.429962089335901e-07,
      "loss": 2.1019,
      "step": 46090
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.435546875,
      "learning_rate": 1.4157913746962846e-07,
      "loss": 2.0939,
      "step": 46100
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.423828125,
      "learning_rate": 1.4016908920541695e-07,
      "loss": 2.0803,
      "step": 46110
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.455078125,
      "learning_rate": 1.387660648074457e-07,
      "loss": 2.1201,
      "step": 46120
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.431640625,
      "learning_rate": 1.3737006493889427e-07,
      "loss": 2.134,
      "step": 46130
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.423828125,
      "learning_rate": 1.3598109025961315e-07,
      "loss": 2.1016,
      "step": 46140
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.439453125,
      "learning_rate": 1.3459914142613384e-07,
      "loss": 2.1291,
      "step": 46150
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4375,
      "learning_rate": 1.3322421909167216e-07,
      "loss": 2.1362,
      "step": 46160
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.44921875,
      "learning_rate": 1.3185632390611823e-07,
      "loss": 2.1282,
      "step": 46170
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.44140625,
      "learning_rate": 1.3049545651603989e-07,
      "loss": 2.1112,
      "step": 46180
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.474609375,
      "learning_rate": 1.291416175646859e-07,
      "loss": 2.1117,
      "step": 46190
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.470703125,
      "learning_rate": 1.277948076919827e-07,
      "loss": 2.1086,
      "step": 46200
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.447265625,
      "learning_rate": 1.264550275345311e-07,
      "loss": 2.1295,
      "step": 46210
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.453125,
      "learning_rate": 1.251222777256128e-07,
      "loss": 2.1282,
      "step": 46220
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.431640625,
      "learning_rate": 1.2379655889518393e-07,
      "loss": 2.1342,
      "step": 46230
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.453125,
      "learning_rate": 1.2247787166987823e-07,
      "loss": 2.1258,
      "step": 46240
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4453125,
      "learning_rate": 1.211662166730071e-07,
      "loss": 2.1023,
      "step": 46250
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.443359375,
      "learning_rate": 1.1986159452455626e-07,
      "loss": 2.1328,
      "step": 46260
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.43359375,
      "learning_rate": 1.1856400584118577e-07,
      "loss": 2.1128,
      "step": 46270
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.44140625,
      "learning_rate": 1.1727345123623667e-07,
      "loss": 2.1226,
      "step": 46280
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.447265625,
      "learning_rate": 1.1598993131971769e-07,
      "loss": 2.1344,
      "step": 46290
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.431640625,
      "learning_rate": 1.1471344669831852e-07,
      "loss": 2.1125,
      "step": 46300
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.44921875,
      "learning_rate": 1.1344399797539983e-07,
      "loss": 2.1422,
      "step": 46310
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.42578125,
      "learning_rate": 1.1218158575100001e-07,
      "loss": 2.1421,
      "step": 46320
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.431640625,
      "learning_rate": 1.1092621062182506e-07,
      "loss": 2.124,
      "step": 46330
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4296875,
      "learning_rate": 1.0967787318126366e-07,
      "loss": 2.1198,
      "step": 46340
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.431640625,
      "learning_rate": 1.0843657401937213e-07,
      "loss": 2.1193,
      "step": 46350
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.443359375,
      "learning_rate": 1.0720231372288114e-07,
      "loss": 2.1048,
      "step": 46360
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.44921875,
      "learning_rate": 1.0597509287519236e-07,
      "loss": 2.1352,
      "step": 46370
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.451171875,
      "learning_rate": 1.047549120563851e-07,
      "loss": 2.1113,
      "step": 46380
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4453125,
      "learning_rate": 1.0354177184320468e-07,
      "loss": 2.1259,
      "step": 46390
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.42578125,
      "learning_rate": 1.0233567280907408e-07,
      "loss": 2.1186,
      "step": 46400
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.435546875,
      "learning_rate": 1.011366155240856e-07,
      "loss": 2.1275,
      "step": 46410
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.42578125,
      "learning_rate": 9.994460055500254e-08,
      "loss": 2.1298,
      "step": 46420
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.498046875,
      "learning_rate": 9.875962846526087e-08,
      "loss": 2.1294,
      "step": 46430
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.447265625,
      "learning_rate": 9.758169981496756e-08,
      "loss": 2.1182,
      "step": 46440
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4296875,
      "learning_rate": 9.641081516089722e-08,
      "loss": 2.0929,
      "step": 46450
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.443359375,
      "learning_rate": 9.524697505650048e-08,
      "loss": 2.1053,
      "step": 46460
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.44921875,
      "learning_rate": 9.409018005189229e-08,
      "loss": 2.1157,
      "step": 46470
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.427734375,
      "learning_rate": 9.29404306938636e-08,
      "loss": 2.1169,
      "step": 46480
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4375,
      "learning_rate": 9.179772752586802e-08,
      "loss": 2.1062,
      "step": 46490
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.42578125,
      "learning_rate": 9.066207108803515e-08,
      "loss": 2.0965,
      "step": 46500
    },
    {
      "epoch": 0.97,
      "eval_accuracy": 0.5589984967594076,
      "eval_loss": 1.9930667877197266,
      "eval_runtime": 16.443,
      "eval_samples_per_second": 36.186,
      "eval_steps_per_second": 1.156,
      "step": 46500
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.447265625,
      "learning_rate": 8.953346191716228e-08,
      "loss": 2.1052,
      "step": 46510
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.419921875,
      "learning_rate": 8.841190054670933e-08,
      "loss": 2.1407,
      "step": 46520
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4921875,
      "learning_rate": 8.729738750681559e-08,
      "loss": 2.0972,
      "step": 46530
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.443359375,
      "learning_rate": 8.618992332427966e-08,
      "loss": 2.1144,
      "step": 46540
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.439453125,
      "learning_rate": 8.50895085225728e-08,
      "loss": 2.0964,
      "step": 46550
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.43359375,
      "learning_rate": 8.399614362183228e-08,
      "loss": 2.1371,
      "step": 46560
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.455078125,
      "learning_rate": 8.290982913886802e-08,
      "loss": 2.1149,
      "step": 46570
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.458984375,
      "learning_rate": 8.183056558714763e-08,
      "loss": 2.0917,
      "step": 46580
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.478515625,
      "learning_rate": 8.075835347681638e-08,
      "loss": 2.0968,
      "step": 46590
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4296875,
      "learning_rate": 7.96931933146805e-08,
      "loss": 2.0971,
      "step": 46600
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.447265625,
      "learning_rate": 7.863508560421228e-08,
      "loss": 2.1231,
      "step": 46610
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.734375,
      "learning_rate": 7.758403084555499e-08,
      "loss": 2.1269,
      "step": 46620
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.427734375,
      "learning_rate": 7.654002953551453e-08,
      "loss": 2.1248,
      "step": 46630
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9453125,
      "learning_rate": 7.550308216756285e-08,
      "loss": 2.0894,
      "step": 46640
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.427734375,
      "learning_rate": 7.447318923184121e-08,
      "loss": 2.1127,
      "step": 46650
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.431640625,
      "learning_rate": 7.345035121515186e-08,
      "loss": 2.1164,
      "step": 46660
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.431640625,
      "learning_rate": 7.243456860096476e-08,
      "loss": 2.0763,
      "step": 46670
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.439453125,
      "learning_rate": 7.14258418694158e-08,
      "loss": 2.1182,
      "step": 46680
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.48828125,
      "learning_rate": 7.042417149730363e-08,
      "loss": 2.1176,
      "step": 46690
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.435546875,
      "learning_rate": 6.942955795809447e-08,
      "loss": 2.1102,
      "step": 46700
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.439453125,
      "learning_rate": 6.844200172191395e-08,
      "loss": 2.1534,
      "step": 46710
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.431640625,
      "learning_rate": 6.7461503255557e-08,
      "loss": 2.1617,
      "step": 46720
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.427734375,
      "learning_rate": 6.648806302247956e-08,
      "loss": 2.0943,
      "step": 46730
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.447265625,
      "learning_rate": 6.552168148280191e-08,
      "loss": 2.1025,
      "step": 46740
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.423828125,
      "learning_rate": 6.456235909330866e-08,
      "loss": 2.1174,
      "step": 46750
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.416015625,
      "learning_rate": 6.361009630744708e-08,
      "loss": 2.1029,
      "step": 46760
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.43359375,
      "learning_rate": 6.266489357532878e-08,
      "loss": 2.1107,
      "step": 46770
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.423828125,
      "learning_rate": 6.172675134372474e-08,
      "loss": 2.1372,
      "step": 46780
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.435546875,
      "learning_rate": 6.079567005607523e-08,
      "loss": 2.1395,
      "step": 46790
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.416015625,
      "learning_rate": 5.98716501524732e-08,
      "loss": 2.1642,
      "step": 46800
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4296875,
      "learning_rate": 5.895469206968429e-08,
      "loss": 2.1287,
      "step": 46810
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.43359375,
      "learning_rate": 5.8044796241128465e-08,
      "loss": 2.1205,
      "step": 46820
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.419921875,
      "learning_rate": 5.7141963096893344e-08,
      "loss": 2.12,
      "step": 46830
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.421875,
      "learning_rate": 5.624619306372092e-08,
      "loss": 2.1247,
      "step": 46840
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.447265625,
      "learning_rate": 5.5357486565024154e-08,
      "loss": 2.1244,
      "step": 46850
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.484375,
      "learning_rate": 5.4475844020868696e-08,
      "loss": 2.127,
      "step": 46860
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4296875,
      "learning_rate": 5.360126584798453e-08,
      "loss": 2.1243,
      "step": 46870
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4375,
      "learning_rate": 5.273375245976597e-08,
      "loss": 2.1084,
      "step": 46880
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.466796875,
      "learning_rate": 5.187330426626169e-08,
      "loss": 2.1259,
      "step": 46890
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.43359375,
      "learning_rate": 5.101992167418301e-08,
      "loss": 2.0995,
      "step": 46900
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.435546875,
      "learning_rate": 5.0173605086905606e-08,
      "loss": 2.1291,
      "step": 46910
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.431640625,
      "learning_rate": 4.9334354904459475e-08,
      "loss": 2.1173,
      "step": 46920
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.43359375,
      "learning_rate": 4.850217152353731e-08,
      "loss": 2.134,
      "step": 46930
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.458984375,
      "learning_rate": 4.767705533749112e-08,
      "loss": 2.1278,
      "step": 46940
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.439453125,
      "learning_rate": 4.685900673633392e-08,
      "loss": 2.1123,
      "step": 46950
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4140625,
      "learning_rate": 4.604802610673475e-08,
      "loss": 2.1184,
      "step": 46960
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.43359375,
      "learning_rate": 4.5244113832025314e-08,
      "loss": 2.1002,
      "step": 46970
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.416015625,
      "learning_rate": 4.4447270292191646e-08,
      "loss": 2.1362,
      "step": 46980
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.421875,
      "learning_rate": 4.365749586388579e-08,
      "loss": 2.1261,
      "step": 46990
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.42578125,
      "learning_rate": 4.28747909204108e-08,
      "loss": 2.1377,
      "step": 47000
    },
    {
      "epoch": 0.98,
      "eval_accuracy": 0.5591907144089308,
      "eval_loss": 1.99307382106781,
      "eval_runtime": 16.463,
      "eval_samples_per_second": 36.142,
      "eval_steps_per_second": 1.154,
      "step": 47000
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.44921875,
      "learning_rate": 4.209915583173241e-08,
      "loss": 2.1158,
      "step": 47010
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.431640625,
      "learning_rate": 4.1330590964474e-08,
      "loss": 2.0902,
      "step": 47020
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.435546875,
      "learning_rate": 4.056909668191666e-08,
      "loss": 2.1167,
      "step": 47030
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.43359375,
      "learning_rate": 3.981467334399913e-08,
      "loss": 2.0849,
      "step": 47040
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.43359375,
      "learning_rate": 3.906732130731949e-08,
      "loss": 2.1406,
      "step": 47050
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.44140625,
      "learning_rate": 3.8327040925130175e-08,
      "loss": 2.1419,
      "step": 47060
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4296875,
      "learning_rate": 3.759383254734461e-08,
      "loss": 2.1096,
      "step": 47070
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.45703125,
      "learning_rate": 3.686769652053224e-08,
      "loss": 2.1092,
      "step": 47080
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.44140625,
      "learning_rate": 3.6148633187916834e-08,
      "loss": 2.1151,
      "step": 47090
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.451171875,
      "learning_rate": 3.54366428893832e-08,
      "loss": 2.1464,
      "step": 47100
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.43359375,
      "learning_rate": 3.473172596147045e-08,
      "loss": 2.1102,
      "step": 47110
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.439453125,
      "learning_rate": 3.40338827373754e-08,
      "loss": 2.0963,
      "step": 47120
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.439453125,
      "learning_rate": 3.3343113546950855e-08,
      "loss": 2.0916,
      "step": 47130
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.431640625,
      "learning_rate": 3.26594187167073e-08,
      "loss": 2.1226,
      "step": 47140
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.421875,
      "learning_rate": 3.19827985698079e-08,
      "loss": 2.1285,
      "step": 47150
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.41796875,
      "learning_rate": 3.1313253426073496e-08,
      "loss": 2.1429,
      "step": 47160
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.427734375,
      "learning_rate": 3.065078360198426e-08,
      "loss": 2.111,
      "step": 47170
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.494140625,
      "learning_rate": 2.999538941066804e-08,
      "loss": 2.1493,
      "step": 47180
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4375,
      "learning_rate": 2.9347071161918703e-08,
      "loss": 2.1358,
      "step": 47190
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4296875,
      "learning_rate": 2.8705829162176122e-08,
      "loss": 2.0828,
      "step": 47200
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.59375,
      "learning_rate": 2.8071663714539508e-08,
      "loss": 2.1029,
      "step": 47210
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.439453125,
      "learning_rate": 2.744457511876408e-08,
      "loss": 2.1115,
      "step": 47220
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.423828125,
      "learning_rate": 2.6824563671259406e-08,
      "loss": 2.1062,
      "step": 47230
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.42578125,
      "learning_rate": 2.6211629665086056e-08,
      "loss": 2.1235,
      "step": 47240
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.451171875,
      "learning_rate": 2.5605773389965616e-08,
      "loss": 2.1205,
      "step": 47250
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.427734375,
      "learning_rate": 2.5006995132269007e-08,
      "loss": 2.0969,
      "step": 47260
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.423828125,
      "learning_rate": 2.4415295175024832e-08,
      "loss": 2.0746,
      "step": 47270
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.43359375,
      "learning_rate": 2.3830673797914372e-08,
      "loss": 2.1345,
      "step": 47280
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.455078125,
      "learning_rate": 2.325313127727158e-08,
      "loss": 2.1141,
      "step": 47290
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.439453125,
      "learning_rate": 2.2682667886089747e-08,
      "loss": 2.1138,
      "step": 47300
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.447265625,
      "learning_rate": 2.2119283894009856e-08,
      "loss": 2.1214,
      "step": 47310
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.466796875,
      "learning_rate": 2.1562979567330554e-08,
      "loss": 2.1126,
      "step": 47320
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.458984375,
      "learning_rate": 2.1013755169001503e-08,
      "loss": 2.0662,
      "step": 47330
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.443359375,
      "learning_rate": 2.0471610958628374e-08,
      "loss": 2.0994,
      "step": 47340
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.435546875,
      "learning_rate": 1.9936547192467845e-08,
      "loss": 2.1121,
      "step": 47350
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.44921875,
      "learning_rate": 1.9408564123432614e-08,
      "loss": 2.1105,
      "step": 47360
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.453125,
      "learning_rate": 1.888766200108638e-08,
      "loss": 2.1276,
      "step": 47370
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.423828125,
      "learning_rate": 1.8373841071645526e-08,
      "loss": 2.1132,
      "step": 47380
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.431640625,
      "learning_rate": 1.7867101577980773e-08,
      "loss": 2.1257,
      "step": 47390
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.42578125,
      "learning_rate": 1.736744375961552e-08,
      "loss": 2.1129,
      "step": 47400
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.421875,
      "learning_rate": 1.6874867852724186e-08,
      "loss": 2.1268,
      "step": 47410
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.419921875,
      "learning_rate": 1.6389374090135524e-08,
      "loss": 2.1189,
      "step": 47420
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4375,
      "learning_rate": 1.5910962701330966e-08,
      "loss": 2.0854,
      "step": 47430
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.439453125,
      "learning_rate": 1.5439633912442962e-08,
      "loss": 2.1192,
      "step": 47440
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.421875,
      "learning_rate": 1.4975387946256634e-08,
      "loss": 2.1096,
      "step": 47450
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4609375,
      "learning_rate": 1.451822502220812e-08,
      "loss": 2.1462,
      "step": 47460
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.45703125,
      "learning_rate": 1.4068145356389561e-08,
      "loss": 2.1359,
      "step": 47470
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.482421875,
      "learning_rate": 1.3625149161539118e-08,
      "loss": 2.1303,
      "step": 47480
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4375,
      "learning_rate": 1.3189236647052626e-08,
      "loss": 2.1349,
      "step": 47490
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.431640625,
      "learning_rate": 1.2760408018973602e-08,
      "loss": 2.1118,
      "step": 47500
    },
    {
      "epoch": 0.99,
      "eval_accuracy": 0.5591677140064237,
      "eval_loss": 1.9930524826049805,
      "eval_runtime": 16.4253,
      "eval_samples_per_second": 36.225,
      "eval_steps_per_second": 1.157,
      "step": 47500
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.43359375,
      "learning_rate": 1.233866347999657e-08,
      "loss": 2.122,
      "step": 47510
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.453125,
      "learning_rate": 1.1924003229473735e-08,
      "loss": 2.0844,
      "step": 47520
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.423828125,
      "learning_rate": 1.1516427463401646e-08,
      "loss": 2.1384,
      "step": 47530
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.44140625,
      "learning_rate": 1.1115936374431202e-08,
      "loss": 2.1229,
      "step": 47540
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.455078125,
      "learning_rate": 1.0722530151864307e-08,
      "loss": 2.1078,
      "step": 47550
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4375,
      "learning_rate": 1.0336208981655549e-08,
      "loss": 2.1337,
      "step": 47560
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4453125,
      "learning_rate": 9.956973046407192e-09,
      "loss": 2.1397,
      "step": 47570
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.470703125,
      "learning_rate": 9.584822525377512e-09,
      "loss": 2.1335,
      "step": 47580
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.44921875,
      "learning_rate": 9.219757594469136e-09,
      "loss": 2.1263,
      "step": 47590
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.45703125,
      "learning_rate": 8.861778426242362e-09,
      "loss": 2.1176,
      "step": 47600
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.419921875,
      "learning_rate": 8.510885189901841e-09,
      "loss": 2.0973,
      "step": 47610
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.427734375,
      "learning_rate": 8.167078051306565e-09,
      "loss": 2.086,
      "step": 47620
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.431640625,
      "learning_rate": 7.830357172966541e-09,
      "loss": 2.1037,
      "step": 47630
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.43359375,
      "learning_rate": 7.50072271404112e-09,
      "loss": 2.1,
      "step": 47640
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4765625,
      "learning_rate": 7.1781748303406665e-09,
      "loss": 2.1152,
      "step": 47650
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.419921875,
      "learning_rate": 6.862713674323229e-09,
      "loss": 2.1094,
      "step": 47660
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.44921875,
      "learning_rate": 6.554339395101194e-09,
      "loss": 2.1537,
      "step": 47670
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4609375,
      "learning_rate": 6.253052138434634e-09,
      "loss": 2.0739,
      "step": 47680
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8203125,
      "learning_rate": 5.958852046736296e-09,
      "loss": 2.1077,
      "step": 47690
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.423828125,
      "learning_rate": 5.67173925906328e-09,
      "loss": 2.1207,
      "step": 47700
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.44140625,
      "learning_rate": 5.391713911128693e-09,
      "loss": 2.1074,
      "step": 47710
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.470703125,
      "learning_rate": 5.118776135294989e-09,
      "loss": 2.1124,
      "step": 47720
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.44921875,
      "learning_rate": 4.8529260605706396e-09,
      "loss": 2.116,
      "step": 47730
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.45703125,
      "learning_rate": 4.594163812615126e-09,
      "loss": 2.1099,
      "step": 47740
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.45703125,
      "learning_rate": 4.3424895137439415e-09,
      "loss": 2.1091,
      "step": 47750
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.427734375,
      "learning_rate": 4.09790328291193e-09,
      "loss": 2.1067,
      "step": 47760
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.447265625,
      "learning_rate": 3.860405235729947e-09,
      "loss": 2.0861,
      "step": 47770
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.41796875,
      "learning_rate": 3.6299954844598582e-09,
      "loss": 2.1638,
      "step": 47780
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4296875,
      "learning_rate": 3.4066741380078815e-09,
      "loss": 2.1137,
      "step": 47790
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.419921875,
      "learning_rate": 3.1904413019329114e-09,
      "loss": 2.1118,
      "step": 47800
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.412109375,
      "learning_rate": 2.9812970784448556e-09,
      "loss": 2.0686,
      "step": 47810
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.443359375,
      "learning_rate": 2.779241566397972e-09,
      "loss": 2.1276,
      "step": 47820
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.439453125,
      "learning_rate": 2.5842748612991963e-09,
      "loss": 2.1231,
      "step": 47830
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.431640625,
      "learning_rate": 2.396397055306476e-09,
      "loss": 2.1088,
      "step": 47840
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.423828125,
      "learning_rate": 2.2156082372221108e-09,
      "loss": 2.116,
      "step": 47850
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4375,
      "learning_rate": 2.041908492502742e-09,
      "loss": 2.0902,
      "step": 47860
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.455078125,
      "learning_rate": 1.8752979032510275e-09,
      "loss": 2.1146,
      "step": 47870
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.451171875,
      "learning_rate": 1.7157765482189724e-09,
      "loss": 2.1416,
      "step": 47880
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.423828125,
      "learning_rate": 1.563344502809594e-09,
      "loss": 2.1123,
      "step": 47890
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.435546875,
      "learning_rate": 1.4180018390735906e-09,
      "loss": 2.1166,
      "step": 47900
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4453125,
      "learning_rate": 1.2797486257076774e-09,
      "loss": 2.0844,
      "step": 47910
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.427734375,
      "learning_rate": 1.1485849280645777e-09,
      "loss": 2.1241,
      "step": 47920
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.439453125,
      "learning_rate": 1.0245108081413656e-09,
      "loss": 2.1271,
      "step": 47930
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.423828125,
      "learning_rate": 9.075263245827969e-10,
      "loss": 2.0772,
      "step": 47940
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.515625,
      "learning_rate": 7.976315326846396e-10,
      "loss": 2.1259,
      "step": 47950
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.423828125,
      "learning_rate": 6.948264843936735e-10,
      "loss": 2.1273,
      "step": 47960
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.427734375,
      "learning_rate": 5.991112283026956e-10,
      "loss": 2.1402,
      "step": 47970
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.451171875,
      "learning_rate": 5.104858096505183e-10,
      "loss": 2.1194,
      "step": 47980
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.455078125,
      "learning_rate": 4.2895027033196254e-10,
      "loss": 2.1542,
      "step": 47990
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0625,
      "learning_rate": 3.5450464888620027e-10,
      "loss": 2.089,
      "step": 48000
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.5590494262221017,
      "eval_loss": 1.993034839630127,
      "eval_runtime": 16.4309,
      "eval_samples_per_second": 36.212,
      "eval_steps_per_second": 1.156,
      "step": 48000
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.44140625,
      "learning_rate": 2.8714898050008485e-10,
      "loss": 2.1068,
      "step": 48010
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.42578125,
      "learning_rate": 2.2688329701314737e-10,
      "loss": 2.0929,
      "step": 48020
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.44140625,
      "learning_rate": 1.7370762690926968e-10,
      "loss": 2.0989,
      "step": 48030
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.435546875,
      "learning_rate": 1.276219953250113e-10,
      "loss": 2.1373,
      "step": 48040
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.43359375,
      "learning_rate": 8.862642404294797e-11,
      "loss": 2.1411,
      "step": 48050
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.42578125,
      "learning_rate": 5.6720931495002346e-11,
      "loss": 2.1095,
      "step": 48060
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.42578125,
      "learning_rate": 3.1905532764109347e-11,
      "loss": 2.1358,
      "step": 48070
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.427734375,
      "learning_rate": 1.418023957588943e-11,
      "loss": 2.1348,
      "step": 48080
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.435546875,
      "learning_rate": 3.545060313636661e-12,
      "loss": 2.1329,
      "step": 48090
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.419921875,
      "learning_rate": 0.0,
      "loss": 2.1093,
      "step": 48100
    },
    {
      "epoch": 1.0,
      "step": 48100,
      "total_flos": 3.101101287470069e+20,
      "train_loss": 2.1293136361829954,
      "train_runtime": 442446.2037,
      "train_samples_per_second": 13.915,
      "train_steps_per_second": 0.109
    }
  ],
  "logging_steps": 10,
  "max_steps": 48100,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 3.101101287470069e+20,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}