{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.0,
  "eval_steps": 500,
  "global_step": 31250,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.016,
      "grad_norm": 0.06700329924693645,
      "learning_rate": 1.9968e-05,
      "loss": 0.1152,
      "step": 50
    },
    {
      "epoch": 0.032,
      "grad_norm": 0.04710424954214583,
      "learning_rate": 1.9936000000000004e-05,
      "loss": 0.0038,
      "step": 100
    },
    {
      "epoch": 0.048,
      "grad_norm": 0.04258497062208284,
      "learning_rate": 1.9904e-05,
      "loss": 0.0027,
      "step": 150
    },
    {
      "epoch": 0.064,
      "grad_norm": 0.046287169575826764,
      "learning_rate": 1.9872000000000002e-05,
      "loss": 0.0022,
      "step": 200
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.02914093414933542,
      "learning_rate": 1.9840000000000003e-05,
      "loss": 0.0018,
      "step": 250
    },
    {
      "epoch": 0.096,
      "grad_norm": 0.027612185075684643,
      "learning_rate": 1.9808e-05,
      "loss": 0.0015,
      "step": 300
    },
    {
      "epoch": 0.112,
      "grad_norm": 0.027340652289474687,
      "learning_rate": 1.9776000000000002e-05,
      "loss": 0.0013,
      "step": 350
    },
    {
      "epoch": 0.128,
      "grad_norm": 0.03207204814138842,
      "learning_rate": 1.9744e-05,
      "loss": 0.0011,
      "step": 400
    },
    {
      "epoch": 0.144,
      "grad_norm": 0.022192312524934657,
      "learning_rate": 1.9712000000000004e-05,
      "loss": 0.001,
      "step": 450
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.02528324253889177,
      "learning_rate": 1.968e-05,
      "loss": 0.0009,
      "step": 500
    },
    {
      "epoch": 0.176,
      "grad_norm": 0.02450599947369508,
      "learning_rate": 1.9648000000000002e-05,
      "loss": 0.0008,
      "step": 550
    },
    {
      "epoch": 0.192,
      "grad_norm": 0.021875653202537572,
      "learning_rate": 1.9616000000000003e-05,
      "loss": 0.0008,
      "step": 600
    },
    {
      "epoch": 0.208,
      "grad_norm": 0.024365675733290123,
      "learning_rate": 1.9584e-05,
      "loss": 0.0007,
      "step": 650
    },
    {
      "epoch": 0.224,
      "grad_norm": 0.020088699678960917,
      "learning_rate": 1.9552000000000002e-05,
      "loss": 0.0005,
      "step": 700
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.023939461961356913,
      "learning_rate": 1.9520000000000003e-05,
      "loss": 0.0005,
      "step": 750
    },
    {
      "epoch": 0.256,
      "grad_norm": 0.016654961004293048,
      "learning_rate": 1.9488000000000004e-05,
      "loss": 0.0005,
      "step": 800
    },
    {
      "epoch": 0.272,
      "grad_norm": 0.01947896740438186,
      "learning_rate": 1.9456e-05,
      "loss": 0.0005,
      "step": 850
    },
    {
      "epoch": 0.288,
      "grad_norm": 0.01632254770986144,
      "learning_rate": 1.9424e-05,
      "loss": 0.0004,
      "step": 900
    },
    {
      "epoch": 0.304,
      "grad_norm": 0.01768857594358966,
      "learning_rate": 1.9392000000000003e-05,
      "loss": 0.0004,
      "step": 950
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.016646831355600935,
      "learning_rate": 1.936e-05,
      "loss": 0.0003,
      "step": 1000
    },
    {
      "epoch": 0.336,
      "grad_norm": 0.013828212178231426,
      "learning_rate": 1.9328000000000002e-05,
      "loss": 0.0003,
      "step": 1050
    },
    {
      "epoch": 0.352,
      "grad_norm": 0.014125730332492632,
      "learning_rate": 1.9296000000000003e-05,
      "loss": 0.0003,
      "step": 1100
    },
    {
      "epoch": 0.368,
      "grad_norm": 0.014130896060127419,
      "learning_rate": 1.9264e-05,
      "loss": 0.0003,
      "step": 1150
    },
    {
      "epoch": 0.384,
      "grad_norm": 0.01326246196506426,
      "learning_rate": 1.9232e-05,
      "loss": 0.0003,
      "step": 1200
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.012809291124835588,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 0.0002,
      "step": 1250
    },
    {
      "epoch": 0.416,
      "grad_norm": 0.009475817533782105,
      "learning_rate": 1.9168000000000004e-05,
      "loss": 0.0002,
      "step": 1300
    },
    {
      "epoch": 0.432,
      "grad_norm": 0.01342329167018312,
      "learning_rate": 1.9136e-05,
      "loss": 0.0002,
      "step": 1350
    },
    {
      "epoch": 0.448,
      "grad_norm": 0.010803873420981821,
      "learning_rate": 1.9104000000000002e-05,
      "loss": 0.0002,
      "step": 1400
    },
    {
      "epoch": 0.464,
      "grad_norm": 0.008662937868043983,
      "learning_rate": 1.9072000000000003e-05,
      "loss": 0.0002,
      "step": 1450
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.009370210631900725,
      "learning_rate": 1.904e-05,
      "loss": 0.0002,
      "step": 1500
    },
    {
      "epoch": 0.496,
      "grad_norm": 0.014623062482376369,
      "learning_rate": 1.9008e-05,
      "loss": 0.0002,
      "step": 1550
    },
    {
      "epoch": 0.512,
      "grad_norm": 0.011257550251173976,
      "learning_rate": 1.8976000000000003e-05,
      "loss": 0.0001,
      "step": 1600
    },
    {
      "epoch": 0.528,
      "grad_norm": 0.007251658800690561,
      "learning_rate": 1.8944000000000004e-05,
      "loss": 0.0001,
      "step": 1650
    },
    {
      "epoch": 0.544,
      "grad_norm": 0.010009328096570643,
      "learning_rate": 1.8912e-05,
      "loss": 0.0002,
      "step": 1700
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.010324858035729835,
      "learning_rate": 1.8880000000000002e-05,
      "loss": 0.0001,
      "step": 1750
    },
    {
      "epoch": 0.576,
      "grad_norm": 0.012200136441846834,
      "learning_rate": 1.8848000000000003e-05,
      "loss": 0.0001,
      "step": 1800
    },
    {
      "epoch": 0.592,
      "grad_norm": 0.011050524306703934,
      "learning_rate": 1.8816e-05,
      "loss": 0.0001,
      "step": 1850
    },
    {
      "epoch": 0.608,
      "grad_norm": 0.009424921961091424,
      "learning_rate": 1.8784000000000002e-05,
      "loss": 0.0001,
      "step": 1900
    },
    {
      "epoch": 0.624,
      "grad_norm": 0.00893922456004691,
      "learning_rate": 1.8752000000000003e-05,
      "loss": 0.0001,
      "step": 1950
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.010239948440932777,
      "learning_rate": 1.8720000000000004e-05,
      "loss": 0.0001,
      "step": 2000
    },
    {
      "epoch": 0.656,
      "grad_norm": 0.00976946104257329,
      "learning_rate": 1.8688e-05,
      "loss": 0.0001,
      "step": 2050
    },
    {
      "epoch": 0.672,
      "grad_norm": 0.009651690086449066,
      "learning_rate": 1.8656000000000002e-05,
      "loss": 0.0001,
      "step": 2100
    },
    {
      "epoch": 0.688,
      "grad_norm": 0.05693047764323227,
      "learning_rate": 1.8624000000000003e-05,
      "loss": 0.0733,
      "step": 2150
    },
    {
      "epoch": 0.704,
      "grad_norm": 0.011338530961932,
      "learning_rate": 1.8592e-05,
      "loss": 0.0004,
      "step": 2200
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.005269926068818781,
      "learning_rate": 1.8560000000000002e-05,
      "loss": 0.0001,
      "step": 2250
    },
    {
      "epoch": 0.736,
      "grad_norm": 0.007667043664845632,
      "learning_rate": 1.8528000000000003e-05,
      "loss": 0.0001,
      "step": 2300
    },
    {
      "epoch": 0.752,
      "grad_norm": 0.004397910133064048,
      "learning_rate": 1.8496e-05,
      "loss": 0.0001,
      "step": 2350
    },
    {
      "epoch": 0.768,
      "grad_norm": 0.005138608389724877,
      "learning_rate": 1.8464e-05,
      "loss": 0.0,
      "step": 2400
    },
    {
      "epoch": 0.784,
      "grad_norm": 0.013032095144063073,
      "learning_rate": 1.8432000000000002e-05,
      "loss": 0.0,
      "step": 2450
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.0030479607356790257,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 0.0,
      "step": 2500
    },
    {
      "epoch": 0.816,
      "grad_norm": 0.002160576995308326,
      "learning_rate": 1.8368e-05,
      "loss": 0.0,
      "step": 2550
    },
    {
      "epoch": 0.832,
      "grad_norm": 0.002576068592921048,
      "learning_rate": 1.8336000000000002e-05,
      "loss": 0.0,
      "step": 2600
    },
    {
      "epoch": 0.848,
      "grad_norm": 0.004527008786251422,
      "learning_rate": 1.8304000000000003e-05,
      "loss": 0.0,
      "step": 2650
    },
    {
      "epoch": 0.864,
      "grad_norm": 0.002401493354271683,
      "learning_rate": 1.8272e-05,
      "loss": 0.0,
      "step": 2700
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.0016418474174344473,
      "learning_rate": 1.824e-05,
      "loss": 0.0,
      "step": 2750
    },
    {
      "epoch": 0.896,
      "grad_norm": 0.0029580974745545933,
      "learning_rate": 1.8208000000000003e-05,
      "loss": 0.0,
      "step": 2800
    },
    {
      "epoch": 0.912,
      "grad_norm": 0.0012505689661283781,
      "learning_rate": 1.8176000000000004e-05,
      "loss": 0.0,
      "step": 2850
    },
    {
      "epoch": 0.928,
      "grad_norm": 0.002206210406840656,
      "learning_rate": 1.8144e-05,
      "loss": 0.0,
      "step": 2900
    },
    {
      "epoch": 0.944,
      "grad_norm": 0.0006240208662934369,
      "learning_rate": 1.8112000000000002e-05,
      "loss": 0.0,
      "step": 2950
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.003067833646212094,
      "learning_rate": 1.8080000000000003e-05,
      "loss": 0.0,
      "step": 3000
    },
    {
      "epoch": 0.976,
      "grad_norm": 0.001555266902384,
      "learning_rate": 1.8048e-05,
      "loss": 0.0,
      "step": 3050
    },
    {
      "epoch": 0.992,
      "grad_norm": 0.005684931150991733,
      "learning_rate": 1.8016e-05,
      "loss": 0.0,
      "step": 3100
    },
    {
      "epoch": 1.008,
      "grad_norm": 0.00036054174942539585,
      "learning_rate": 1.7984000000000003e-05,
      "loss": 0.0,
      "step": 3150
    },
    {
      "epoch": 1.024,
      "grad_norm": 0.0023891310529782278,
      "learning_rate": 1.7952e-05,
      "loss": 0.0,
      "step": 3200
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.0017593637539934622,
      "learning_rate": 1.792e-05,
      "loss": 0.0,
      "step": 3250
    },
    {
      "epoch": 1.056,
      "grad_norm": 0.00988261541456867,
      "learning_rate": 1.7888000000000002e-05,
      "loss": 0.0001,
      "step": 3300
    },
    {
      "epoch": 1.072,
      "grad_norm": 0.008606785310498308,
      "learning_rate": 1.7856000000000003e-05,
      "loss": 0.0001,
      "step": 3350
    },
    {
      "epoch": 1.088,
      "grad_norm": 0.009666636774150597,
      "learning_rate": 1.7824e-05,
      "loss": 0.0001,
      "step": 3400
    },
    {
      "epoch": 1.104,
      "grad_norm": 0.0034236747560986843,
      "learning_rate": 1.7792000000000002e-05,
      "loss": 0.0001,
      "step": 3450
    },
    {
      "epoch": 1.12,
      "grad_norm": 0.018616762957628402,
      "learning_rate": 1.7760000000000003e-05,
      "loss": 0.0094,
      "step": 3500
    },
    {
      "epoch": 1.1360000000000001,
      "grad_norm": 0.009926468045818765,
      "learning_rate": 1.7728e-05,
      "loss": 0.0002,
      "step": 3550
    },
    {
      "epoch": 1.152,
      "grad_norm": 0.007468057768454313,
      "learning_rate": 1.7696e-05,
      "loss": 0.0001,
      "step": 3600
    },
    {
      "epoch": 1.168,
      "grad_norm": 0.008415666101481943,
      "learning_rate": 1.7664000000000002e-05,
      "loss": 0.0001,
      "step": 3650
    },
    {
      "epoch": 1.184,
      "grad_norm": 0.012134121535343333,
      "learning_rate": 1.7632000000000003e-05,
      "loss": 0.0001,
      "step": 3700
    },
    {
      "epoch": 1.2,
      "grad_norm": 0.0040802073628923375,
      "learning_rate": 1.76e-05,
      "loss": 0.0,
      "step": 3750
    },
    {
      "epoch": 1.216,
      "grad_norm": 0.0012922320234944806,
      "learning_rate": 1.7568000000000002e-05,
      "loss": 0.0,
      "step": 3800
    },
    {
      "epoch": 1.232,
      "grad_norm": 0.002679388648395559,
      "learning_rate": 1.7536000000000003e-05,
      "loss": 0.0,
      "step": 3850
    },
    {
      "epoch": 1.248,
      "grad_norm": 0.0008935097752906122,
      "learning_rate": 1.7504e-05,
      "loss": 0.0,
      "step": 3900
    },
    {
      "epoch": 1.264,
      "grad_norm": 0.002432102143415828,
      "learning_rate": 1.7472e-05,
      "loss": 0.0,
      "step": 3950
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.0026062712960827376,
      "learning_rate": 1.7440000000000002e-05,
      "loss": 0.0,
      "step": 4000
    },
    {
      "epoch": 1.296,
      "grad_norm": 0.0012617909961480731,
      "learning_rate": 1.7408e-05,
      "loss": 0.0,
      "step": 4050
    },
    {
      "epoch": 1.312,
      "grad_norm": 0.005995550159991727,
      "learning_rate": 1.7376e-05,
      "loss": 0.0,
      "step": 4100
    },
    {
      "epoch": 1.328,
      "grad_norm": 0.0022878669701388655,
      "learning_rate": 1.7344000000000002e-05,
      "loss": 0.0001,
      "step": 4150
    },
    {
      "epoch": 1.3439999999999999,
      "grad_norm": 0.00472466199586374,
      "learning_rate": 1.7312000000000003e-05,
      "loss": 0.0001,
      "step": 4200
    },
    {
      "epoch": 1.3599999999999999,
      "grad_norm": 0.004622604014149787,
      "learning_rate": 1.728e-05,
      "loss": 0.0,
      "step": 4250
    },
    {
      "epoch": 1.376,
      "grad_norm": 0.006288850357467127,
      "learning_rate": 1.7248e-05,
      "loss": 0.0,
      "step": 4300
    },
    {
      "epoch": 1.392,
      "grad_norm": 0.0059456589687283045,
      "learning_rate": 1.7216000000000003e-05,
      "loss": 0.0,
      "step": 4350
    },
    {
      "epoch": 1.408,
      "grad_norm": 0.004299973129725888,
      "learning_rate": 1.7184e-05,
      "loss": 0.0,
      "step": 4400
    },
    {
      "epoch": 1.424,
      "grad_norm": 0.006384727380231639,
      "learning_rate": 1.7152e-05,
      "loss": 0.0001,
      "step": 4450
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.006961166462441037,
      "learning_rate": 1.7120000000000002e-05,
      "loss": 0.0,
      "step": 4500
    },
    {
      "epoch": 1.456,
      "grad_norm": 0.002706931118082421,
      "learning_rate": 1.7088000000000003e-05,
      "loss": 0.0,
      "step": 4550
    },
    {
      "epoch": 1.472,
      "grad_norm": 0.0044763406305933845,
      "learning_rate": 1.7056e-05,
      "loss": 0.0,
      "step": 4600
    },
    {
      "epoch": 1.488,
      "grad_norm": 0.0018676112837315154,
      "learning_rate": 1.7024e-05,
      "loss": 0.0,
      "step": 4650
    },
    {
      "epoch": 1.504,
      "grad_norm": 0.0035498650855882784,
      "learning_rate": 1.6992000000000003e-05,
      "loss": 0.0,
      "step": 4700
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.0032614453273975457,
      "learning_rate": 1.696e-05,
      "loss": 0.0,
      "step": 4750
    },
    {
      "epoch": 1.536,
      "grad_norm": 0.004704682912696666,
      "learning_rate": 1.6928e-05,
      "loss": 0.0,
      "step": 4800
    },
    {
      "epoch": 1.552,
      "grad_norm": 0.003944103313663087,
      "learning_rate": 1.6896000000000002e-05,
      "loss": 0.0,
      "step": 4850
    },
    {
      "epoch": 1.568,
      "grad_norm": 0.001028864743221988,
      "learning_rate": 1.6864e-05,
      "loss": 0.0,
      "step": 4900
    },
    {
      "epoch": 1.584,
      "grad_norm": 0.008781206361834713,
      "learning_rate": 1.6832e-05,
      "loss": 0.0001,
      "step": 4950
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.006941071455141092,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 0.0001,
      "step": 5000
    },
    {
      "epoch": 1.616,
      "grad_norm": 0.0035317612009535013,
      "learning_rate": 1.6768000000000003e-05,
      "loss": 0.0001,
      "step": 5050
    },
    {
      "epoch": 1.6320000000000001,
      "grad_norm": 0.007130755050925452,
      "learning_rate": 1.6736e-05,
      "loss": 0.0001,
      "step": 5100
    },
    {
      "epoch": 1.6480000000000001,
      "grad_norm": 0.009056134474244146,
      "learning_rate": 1.6704e-05,
      "loss": 0.0001,
      "step": 5150
    },
    {
      "epoch": 1.6640000000000001,
      "grad_norm": 0.009975608465029769,
      "learning_rate": 1.6672000000000002e-05,
      "loss": 0.0001,
      "step": 5200
    },
    {
      "epoch": 1.6800000000000002,
      "grad_norm": 0.00752603996543754,
      "learning_rate": 1.664e-05,
      "loss": 0.0001,
      "step": 5250
    },
    {
      "epoch": 1.696,
      "grad_norm": 0.005644442586376258,
      "learning_rate": 1.6608e-05,
      "loss": 0.0001,
      "step": 5300
    },
    {
      "epoch": 1.712,
      "grad_norm": 0.005278241124697153,
      "learning_rate": 1.6576000000000002e-05,
      "loss": 0.0001,
      "step": 5350
    },
    {
      "epoch": 1.728,
      "grad_norm": 0.004435698738768064,
      "learning_rate": 1.6544000000000003e-05,
      "loss": 0.0001,
      "step": 5400
    },
    {
      "epoch": 1.744,
      "grad_norm": 0.006404322123967892,
      "learning_rate": 1.6512e-05,
      "loss": 0.0,
      "step": 5450
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.005399249694087203,
      "learning_rate": 1.648e-05,
      "loss": 0.0001,
      "step": 5500
    },
    {
      "epoch": 1.776,
      "grad_norm": 0.008390970222631198,
      "learning_rate": 1.6448000000000002e-05,
      "loss": 0.0,
      "step": 5550
    },
    {
      "epoch": 1.792,
      "grad_norm": 0.0049263709067859165,
      "learning_rate": 1.6416e-05,
      "loss": 0.0001,
      "step": 5600
    },
    {
      "epoch": 1.808,
      "grad_norm": 0.0023635146513281257,
      "learning_rate": 1.6384e-05,
      "loss": 0.0,
      "step": 5650
    },
    {
      "epoch": 1.8239999999999998,
      "grad_norm": 0.005111500857791294,
      "learning_rate": 1.6352000000000002e-05,
      "loss": 0.0,
      "step": 5700
    },
    {
      "epoch": 1.8399999999999999,
      "grad_norm": 0.00844717374529172,
      "learning_rate": 1.632e-05,
      "loss": 0.0,
      "step": 5750
    },
    {
      "epoch": 1.8559999999999999,
      "grad_norm": 0.0036198931176737673,
      "learning_rate": 1.6288e-05,
      "loss": 0.0,
      "step": 5800
    },
    {
      "epoch": 1.8719999999999999,
      "grad_norm": 0.0048152236821402804,
      "learning_rate": 1.6256e-05,
      "loss": 0.0,
      "step": 5850
    },
    {
      "epoch": 1.888,
      "grad_norm": 0.0008869093396117195,
      "learning_rate": 1.6224000000000003e-05,
      "loss": 0.0,
      "step": 5900
    },
    {
      "epoch": 1.904,
      "grad_norm": 0.0025329131777146337,
      "learning_rate": 1.6192e-05,
      "loss": 0.0,
      "step": 5950
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.0037397513843206794,
      "learning_rate": 1.616e-05,
      "loss": 0.0,
      "step": 6000
    },
    {
      "epoch": 1.936,
      "grad_norm": 0.006049615714075747,
      "learning_rate": 1.6128000000000002e-05,
      "loss": 0.0,
      "step": 6050
    },
    {
      "epoch": 1.952,
      "grad_norm": 0.0017716844046744302,
      "learning_rate": 1.6096e-05,
      "loss": 0.0,
      "step": 6100
    },
    {
      "epoch": 1.968,
      "grad_norm": 0.009285851148933216,
      "learning_rate": 1.6064e-05,
      "loss": 0.0,
      "step": 6150
    },
    {
      "epoch": 1.984,
      "grad_norm": 0.0004888867999470166,
      "learning_rate": 1.6032e-05,
      "loss": 0.0,
      "step": 6200
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.006431946143017072,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.0,
      "step": 6250
    },
    {
      "epoch": 2.016,
      "grad_norm": 0.0023743769950413467,
      "learning_rate": 1.5968e-05,
      "loss": 0.0001,
      "step": 6300
    },
    {
      "epoch": 2.032,
      "grad_norm": 0.009089590999873571,
      "learning_rate": 1.5936e-05,
      "loss": 0.0001,
      "step": 6350
    },
    {
      "epoch": 2.048,
      "grad_norm": 0.006365387805287794,
      "learning_rate": 1.5904000000000002e-05,
      "loss": 0.0,
      "step": 6400
    },
    {
      "epoch": 2.064,
      "grad_norm": 0.004108736082112842,
      "learning_rate": 1.5872e-05,
      "loss": 0.0,
      "step": 6450
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.004691840217453286,
      "learning_rate": 1.584e-05,
      "loss": 0.0,
      "step": 6500
    },
    {
      "epoch": 2.096,
      "grad_norm": 0.004716746413071922,
      "learning_rate": 1.5808000000000002e-05,
      "loss": 0.0,
      "step": 6550
    },
    {
      "epoch": 2.112,
      "grad_norm": 0.008418820460181144,
      "learning_rate": 1.5776e-05,
      "loss": 0.0001,
      "step": 6600
    },
    {
      "epoch": 2.128,
      "grad_norm": 0.001730413649374287,
      "learning_rate": 1.5744e-05,
      "loss": 0.0001,
      "step": 6650
    },
    {
      "epoch": 2.144,
      "grad_norm": 0.08084393210770029,
      "learning_rate": 1.5712e-05,
      "loss": 0.0001,
      "step": 6700
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.0091071714838611,
      "learning_rate": 1.5680000000000002e-05,
      "loss": 0.0002,
      "step": 6750
    },
    {
      "epoch": 2.176,
      "grad_norm": 0.011937686894892474,
      "learning_rate": 1.5648e-05,
      "loss": 0.0007,
      "step": 6800
    },
    {
      "epoch": 2.192,
      "grad_norm": 0.007773091042443193,
      "learning_rate": 1.5616e-05,
      "loss": 0.0002,
      "step": 6850
    },
    {
      "epoch": 2.208,
      "grad_norm": 0.0059540468380226055,
      "learning_rate": 1.5584000000000002e-05,
      "loss": 0.0001,
      "step": 6900
    },
    {
      "epoch": 2.224,
      "grad_norm": 0.004014490897874684,
      "learning_rate": 1.5552e-05,
      "loss": 0.0001,
      "step": 6950
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.006875418357679233,
      "learning_rate": 1.552e-05,
      "loss": 0.0001,
      "step": 7000
    },
    {
      "epoch": 2.2560000000000002,
      "grad_norm": 0.007989346292625824,
      "learning_rate": 1.5488e-05,
      "loss": 0.0001,
      "step": 7050
    },
    {
      "epoch": 2.2720000000000002,
      "grad_norm": 0.006473879081312047,
      "learning_rate": 1.5456000000000002e-05,
      "loss": 0.0,
      "step": 7100
    },
    {
      "epoch": 2.288,
      "grad_norm": 0.001344406505995323,
      "learning_rate": 1.5424e-05,
      "loss": 0.0,
      "step": 7150
    },
    {
      "epoch": 2.304,
      "grad_norm": 0.0013279502116751898,
      "learning_rate": 1.5392e-05,
      "loss": 0.0,
      "step": 7200
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.0025834903526500636,
      "learning_rate": 1.5360000000000002e-05,
      "loss": 0.0001,
      "step": 7250
    },
    {
      "epoch": 2.336,
      "grad_norm": 0.003613767461251546,
      "learning_rate": 1.5328e-05,
      "loss": 0.0,
      "step": 7300
    },
    {
      "epoch": 2.352,
      "grad_norm": 0.007693189752853583,
      "learning_rate": 1.5296e-05,
      "loss": 0.0,
      "step": 7350
    },
    {
      "epoch": 2.368,
      "grad_norm": 0.0039145709733582204,
      "learning_rate": 1.5264e-05,
      "loss": 0.0,
      "step": 7400
    },
    {
      "epoch": 2.384,
      "grad_norm": 0.0024948404702111915,
      "learning_rate": 1.5232000000000003e-05,
      "loss": 0.0001,
      "step": 7450
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.005963883109858903,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 0.0,
      "step": 7500
    },
    {
      "epoch": 2.416,
      "grad_norm": 0.005203832761613304,
      "learning_rate": 1.5168000000000001e-05,
      "loss": 0.0,
      "step": 7550
    },
    {
      "epoch": 2.432,
      "grad_norm": 0.003019086304287391,
      "learning_rate": 1.5136000000000002e-05,
      "loss": 0.0,
      "step": 7600
    },
    {
      "epoch": 2.448,
      "grad_norm": 0.002330648219270858,
      "learning_rate": 1.5104000000000001e-05,
      "loss": 0.0,
      "step": 7650
    },
    {
      "epoch": 2.464,
      "grad_norm": 0.005547183836451343,
      "learning_rate": 1.5072000000000002e-05,
      "loss": 0.0,
      "step": 7700
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.007732333230021918,
      "learning_rate": 1.5040000000000002e-05,
      "loss": 0.0001,
      "step": 7750
    },
    {
      "epoch": 2.496,
      "grad_norm": 0.0028876572710758606,
      "learning_rate": 1.5008000000000001e-05,
      "loss": 0.0,
      "step": 7800
    },
    {
      "epoch": 2.512,
      "grad_norm": 0.00504986025914165,
      "learning_rate": 1.4976000000000002e-05,
      "loss": 0.0,
      "step": 7850
    },
    {
      "epoch": 2.528,
      "grad_norm": 0.004979440608394857,
      "learning_rate": 1.4944000000000001e-05,
      "loss": 0.0,
      "step": 7900
    },
    {
      "epoch": 2.544,
      "grad_norm": 0.004136333653101882,
      "learning_rate": 1.4912000000000002e-05,
      "loss": 0.0,
      "step": 7950
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.00870644207668528,
      "learning_rate": 1.4880000000000002e-05,
      "loss": 0.0001,
      "step": 8000
    },
    {
      "epoch": 2.576,
      "grad_norm": 0.004798469088496501,
      "learning_rate": 1.4848e-05,
      "loss": 0.0001,
      "step": 8050
    },
    {
      "epoch": 2.592,
      "grad_norm": 0.0070315691108956335,
      "learning_rate": 1.4816000000000002e-05,
      "loss": 0.0001,
      "step": 8100
    },
    {
      "epoch": 2.608,
      "grad_norm": 0.0025286695712524415,
      "learning_rate": 1.4784000000000001e-05,
      "loss": 0.0001,
      "step": 8150
    },
    {
      "epoch": 2.624,
      "grad_norm": 0.004842817224651438,
      "learning_rate": 1.4752000000000002e-05,
      "loss": 0.0001,
      "step": 8200
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.007477324219913048,
      "learning_rate": 1.4720000000000001e-05,
      "loss": 0.0,
      "step": 8250
    },
    {
      "epoch": 2.656,
      "grad_norm": 0.007776510130998169,
      "learning_rate": 1.4688000000000002e-05,
      "loss": 0.0001,
      "step": 8300
    },
    {
      "epoch": 2.672,
      "grad_norm": 0.004357492973329834,
      "learning_rate": 1.4656000000000002e-05,
      "loss": 0.0,
      "step": 8350
    },
    {
      "epoch": 2.6879999999999997,
      "grad_norm": 0.0032312366379610435,
      "learning_rate": 1.4624000000000001e-05,
      "loss": 0.0,
      "step": 8400
    },
    {
      "epoch": 2.7039999999999997,
      "grad_norm": 0.0014641972549175538,
      "learning_rate": 1.4592000000000002e-05,
      "loss": 0.0,
      "step": 8450
    },
    {
      "epoch": 2.7199999999999998,
      "grad_norm": 0.002893495767324172,
      "learning_rate": 1.4560000000000001e-05,
      "loss": 0.0,
      "step": 8500
    },
    {
      "epoch": 2.7359999999999998,
      "grad_norm": 0.003053109756022501,
      "learning_rate": 1.4528000000000002e-05,
      "loss": 0.0,
      "step": 8550
    },
    {
      "epoch": 2.752,
      "grad_norm": 0.0045809013298590984,
      "learning_rate": 1.4496000000000001e-05,
      "loss": 0.0001,
      "step": 8600
    },
    {
      "epoch": 2.768,
      "grad_norm": 0.0028170962722070987,
      "learning_rate": 1.4464e-05,
      "loss": 0.0001,
      "step": 8650
    },
    {
      "epoch": 2.784,
      "grad_norm": 0.0016398323178968255,
      "learning_rate": 1.4432000000000002e-05,
      "loss": 0.0,
      "step": 8700
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.0015106826900520753,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 0.0,
      "step": 8750
    },
    {
      "epoch": 2.816,
      "grad_norm": 0.0016227105812749837,
      "learning_rate": 1.4368000000000002e-05,
      "loss": 0.0,
      "step": 8800
    },
    {
      "epoch": 2.832,
      "grad_norm": 0.002216048526258624,
      "learning_rate": 1.4336000000000001e-05,
      "loss": 0.0,
      "step": 8850
    },
    {
      "epoch": 2.848,
      "grad_norm": 0.0028826250412742134,
      "learning_rate": 1.4304e-05,
      "loss": 0.0,
      "step": 8900
    },
    {
      "epoch": 2.864,
      "grad_norm": 0.0031560966093997045,
      "learning_rate": 1.4272000000000002e-05,
      "loss": 0.0,
      "step": 8950
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.005840061376547936,
      "learning_rate": 1.4240000000000001e-05,
      "loss": 0.0,
      "step": 9000
    },
    {
      "epoch": 2.896,
      "grad_norm": 0.0033866397581423604,
      "learning_rate": 1.4208000000000002e-05,
      "loss": 0.0,
      "step": 9050
    },
    {
      "epoch": 2.912,
      "grad_norm": 0.002624744023065921,
      "learning_rate": 1.4176000000000001e-05,
      "loss": 0.0,
      "step": 9100
    },
    {
      "epoch": 2.928,
      "grad_norm": 0.005004706105838203,
      "learning_rate": 1.4144000000000002e-05,
      "loss": 0.0,
      "step": 9150
    },
    {
      "epoch": 2.944,
      "grad_norm": 0.006743938270662594,
      "learning_rate": 1.4112000000000001e-05,
      "loss": 0.0001,
      "step": 9200
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.009483278180315352,
      "learning_rate": 1.408e-05,
      "loss": 0.0001,
      "step": 9250
    },
    {
      "epoch": 2.976,
      "grad_norm": 0.0007610189598327726,
      "learning_rate": 1.4048000000000002e-05,
      "loss": 0.0001,
      "step": 9300
    },
    {
      "epoch": 2.992,
      "grad_norm": 0.004900031620646641,
      "learning_rate": 1.4016000000000001e-05,
      "loss": 0.0001,
      "step": 9350
    },
    {
      "epoch": 3.008,
      "grad_norm": 0.0040788821051548334,
      "learning_rate": 1.3984000000000002e-05,
      "loss": 0.0,
      "step": 9400
    },
    {
      "epoch": 3.024,
      "grad_norm": 0.006604571524243836,
      "learning_rate": 1.3952000000000001e-05,
      "loss": 0.0001,
      "step": 9450
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.004390543355203382,
      "learning_rate": 1.392e-05,
      "loss": 0.0001,
      "step": 9500
    },
    {
      "epoch": 3.056,
      "grad_norm": 0.006092729798201787,
      "learning_rate": 1.3888000000000002e-05,
      "loss": 0.0001,
      "step": 9550
    },
    {
      "epoch": 3.072,
      "grad_norm": 0.0024120775231851973,
      "learning_rate": 1.3856e-05,
      "loss": 0.0001,
      "step": 9600
    },
    {
      "epoch": 3.088,
      "grad_norm": 0.0023121633006781624,
      "learning_rate": 1.3824000000000002e-05,
      "loss": 0.0,
      "step": 9650
    },
    {
      "epoch": 3.104,
      "grad_norm": 0.004009670235882467,
      "learning_rate": 1.3792000000000001e-05,
      "loss": 0.0,
      "step": 9700
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.0029967749699539876,
      "learning_rate": 1.376e-05,
      "loss": 0.0,
      "step": 9750
    },
    {
      "epoch": 3.136,
      "grad_norm": 0.0011939540142138618,
      "learning_rate": 1.3728000000000001e-05,
      "loss": 0.0,
      "step": 9800
    },
    {
      "epoch": 3.152,
      "grad_norm": 0.0018367777306534149,
      "learning_rate": 1.3696e-05,
      "loss": 0.0,
      "step": 9850
    },
    {
      "epoch": 3.168,
      "grad_norm": 0.005363794390571419,
      "learning_rate": 1.3664000000000002e-05,
      "loss": 0.0,
      "step": 9900
    },
    {
      "epoch": 3.184,
      "grad_norm": 0.008476719613859292,
      "learning_rate": 1.3632000000000001e-05,
      "loss": 0.0,
      "step": 9950
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.004822397551004876,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 0.0001,
      "step": 10000
    },
    {
      "epoch": 3.216,
      "grad_norm": 0.005770066249937767,
      "learning_rate": 1.3568000000000001e-05,
      "loss": 0.0001,
      "step": 10050
    },
    {
      "epoch": 3.232,
      "grad_norm": 0.005910492274029805,
      "learning_rate": 1.3536e-05,
      "loss": 0.0001,
      "step": 10100
    },
    {
      "epoch": 3.248,
      "grad_norm": 0.003960172672416223,
      "learning_rate": 1.3504000000000001e-05,
      "loss": 0.0001,
      "step": 10150
    },
    {
      "epoch": 3.2640000000000002,
      "grad_norm": 0.001980537126146146,
      "learning_rate": 1.3472e-05,
      "loss": 0.0001,
      "step": 10200
    },
    {
      "epoch": 3.2800000000000002,
      "grad_norm": 0.0036278941646021327,
      "learning_rate": 1.3440000000000002e-05,
      "loss": 0.0,
      "step": 10250
    },
    {
      "epoch": 3.296,
      "grad_norm": 0.0018966529117990538,
      "learning_rate": 1.3408000000000001e-05,
      "loss": 0.0,
      "step": 10300
    },
    {
      "epoch": 3.312,
      "grad_norm": 0.005134519388485188,
      "learning_rate": 1.3376e-05,
      "loss": 0.0,
      "step": 10350
    },
    {
      "epoch": 3.328,
      "grad_norm": 0.0038905098172508363,
      "learning_rate": 1.3344000000000001e-05,
      "loss": 0.0,
      "step": 10400
    },
    {
      "epoch": 3.344,
      "grad_norm": 0.0031574628583158175,
      "learning_rate": 1.3312e-05,
      "loss": 0.0001,
      "step": 10450
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.002999972961544216,
      "learning_rate": 1.3280000000000002e-05,
      "loss": 0.0,
      "step": 10500
    },
    {
      "epoch": 3.376,
      "grad_norm": 0.0033055234024522047,
      "learning_rate": 1.3248000000000001e-05,
      "loss": 0.0,
      "step": 10550
    },
    {
      "epoch": 3.392,
      "grad_norm": 0.00401214391851968,
      "learning_rate": 1.3216000000000002e-05,
      "loss": 0.0,
      "step": 10600
    },
    {
      "epoch": 3.408,
      "grad_norm": 0.00021936232095850022,
      "learning_rate": 1.3184000000000001e-05,
      "loss": 0.0,
      "step": 10650
    },
    {
      "epoch": 3.424,
      "grad_norm": 0.000917802895080481,
      "learning_rate": 1.3152e-05,
      "loss": 0.0,
      "step": 10700
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.00214730949334728,
      "learning_rate": 1.3120000000000001e-05,
      "loss": 0.0,
      "step": 10750
    },
    {
      "epoch": 3.456,
      "grad_norm": 0.0030398923791812287,
      "learning_rate": 1.3088e-05,
      "loss": 0.0,
      "step": 10800
    },
    {
      "epoch": 3.472,
      "grad_norm": 0.0022242808251567757,
      "learning_rate": 1.3056000000000002e-05,
      "loss": 0.0,
      "step": 10850
    },
    {
      "epoch": 3.488,
      "grad_norm": 0.0008940846133161214,
      "learning_rate": 1.3024000000000001e-05,
      "loss": 0.0,
      "step": 10900
    },
    {
      "epoch": 3.504,
      "grad_norm": 0.0052346185053533355,
      "learning_rate": 1.2992e-05,
      "loss": 0.0,
      "step": 10950
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.002586167307537178,
      "learning_rate": 1.2960000000000001e-05,
      "loss": 0.0,
      "step": 11000
    },
    {
      "epoch": 3.536,
      "grad_norm": 0.014366158301715531,
      "learning_rate": 1.2928e-05,
      "loss": 0.0,
      "step": 11050
    },
    {
      "epoch": 3.552,
      "grad_norm": 0.005454338786955534,
      "learning_rate": 1.2896000000000002e-05,
      "loss": 0.0001,
      "step": 11100
    },
    {
      "epoch": 3.568,
      "grad_norm": 0.0021260355729629562,
      "learning_rate": 1.2864e-05,
      "loss": 0.0,
      "step": 11150
    },
    {
      "epoch": 3.584,
      "grad_norm": 0.0035810871167441354,
      "learning_rate": 1.2832e-05,
      "loss": 0.0,
      "step": 11200
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.002774145084319564,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 0.0,
      "step": 11250
    },
    {
      "epoch": 3.616,
      "grad_norm": 0.0017297614850600386,
      "learning_rate": 1.2768e-05,
      "loss": 0.0,
      "step": 11300
    },
    {
      "epoch": 3.632,
      "grad_norm": 0.005428288582783459,
      "learning_rate": 1.2736000000000001e-05,
      "loss": 0.0,
      "step": 11350
    },
    {
      "epoch": 3.648,
      "grad_norm": 0.0026717447275682845,
      "learning_rate": 1.2704e-05,
      "loss": 0.0,
      "step": 11400
    },
    {
      "epoch": 3.664,
      "grad_norm": 0.00408219616036445,
      "learning_rate": 1.2672000000000002e-05,
      "loss": 0.0,
      "step": 11450
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.0021619661272979113,
      "learning_rate": 1.2640000000000001e-05,
      "loss": 0.0,
      "step": 11500
    },
    {
      "epoch": 3.6959999999999997,
      "grad_norm": 0.0035456991848178644,
      "learning_rate": 1.2608e-05,
      "loss": 0.0001,
      "step": 11550
    },
    {
      "epoch": 3.7119999999999997,
      "grad_norm": 0.004459203173437518,
      "learning_rate": 1.2576000000000001e-05,
      "loss": 0.0,
      "step": 11600
    },
    {
      "epoch": 3.7279999999999998,
      "grad_norm": 0.005408717563076554,
      "learning_rate": 1.2544e-05,
      "loss": 0.0,
      "step": 11650
    },
    {
      "epoch": 3.7439999999999998,
      "grad_norm": 0.0035494391855729153,
      "learning_rate": 1.2512000000000002e-05,
      "loss": 0.0,
      "step": 11700
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.0020159908061204174,
      "learning_rate": 1.248e-05,
      "loss": 0.0,
      "step": 11750
    },
    {
      "epoch": 3.776,
      "grad_norm": 0.003331910844435852,
      "learning_rate": 1.2448e-05,
      "loss": 0.0,
      "step": 11800
    },
    {
      "epoch": 3.792,
      "grad_norm": 0.0012062460066491493,
      "learning_rate": 1.2416000000000001e-05,
      "loss": 0.0,
      "step": 11850
    },
    {
      "epoch": 3.808,
      "grad_norm": 0.0027807416627710986,
      "learning_rate": 1.2384e-05,
      "loss": 0.0,
      "step": 11900
    },
    {
      "epoch": 3.824,
      "grad_norm": 0.001871484410386379,
      "learning_rate": 1.2352000000000001e-05,
      "loss": 0.0,
      "step": 11950
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.0012550393993999455,
      "learning_rate": 1.232e-05,
      "loss": 0.0,
      "step": 12000
    },
    {
      "epoch": 3.856,
      "grad_norm": 0.0035320592711352403,
      "learning_rate": 1.2288e-05,
      "loss": 0.0,
      "step": 12050
    },
    {
      "epoch": 3.872,
      "grad_norm": 0.0017416538958658072,
      "learning_rate": 1.2256000000000001e-05,
      "loss": 0.0,
      "step": 12100
    },
    {
      "epoch": 3.888,
      "grad_norm": 0.0008299586122563334,
      "learning_rate": 1.2224e-05,
      "loss": 0.0,
      "step": 12150
    },
    {
      "epoch": 3.904,
      "grad_norm": 0.002838479964551143,
      "learning_rate": 1.2192000000000001e-05,
      "loss": 0.0,
      "step": 12200
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.002082164082108635,
      "learning_rate": 1.216e-05,
      "loss": 0.0,
      "step": 12250
    },
    {
      "epoch": 3.936,
      "grad_norm": 0.0037577523462426405,
      "learning_rate": 1.2128000000000001e-05,
      "loss": 0.0,
      "step": 12300
    },
    {
      "epoch": 3.952,
      "grad_norm": 0.0026759700344170553,
      "learning_rate": 1.2096e-05,
      "loss": 0.0,
      "step": 12350
    },
    {
      "epoch": 3.968,
      "grad_norm": 0.003510314103134683,
      "learning_rate": 1.2064e-05,
      "loss": 0.0001,
      "step": 12400
    },
    {
      "epoch": 3.984,
      "grad_norm": 0.0053526146673045175,
      "learning_rate": 1.2032000000000001e-05,
      "loss": 0.0001,
      "step": 12450
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.0012346099755580994,
      "learning_rate": 1.2e-05,
      "loss": 0.0001,
      "step": 12500
    },
    {
      "epoch": 4.016,
      "grad_norm": 0.0018089321919392111,
      "learning_rate": 1.1968000000000001e-05,
      "loss": 0.0,
      "step": 12550
    },
    {
      "epoch": 4.032,
      "grad_norm": 0.001913026176722158,
      "learning_rate": 1.1936e-05,
      "loss": 0.0,
      "step": 12600
    },
    {
      "epoch": 4.048,
      "grad_norm": 0.003324750552626263,
      "learning_rate": 1.1904e-05,
      "loss": 0.0,
      "step": 12650
    },
    {
      "epoch": 4.064,
      "grad_norm": 0.00024141692305707922,
      "learning_rate": 1.1872000000000001e-05,
      "loss": 0.0,
      "step": 12700
    },
    {
      "epoch": 4.08,
      "grad_norm": 0.00360992104204235,
      "learning_rate": 1.184e-05,
      "loss": 0.0,
      "step": 12750
    },
    {
      "epoch": 4.096,
      "grad_norm": 0.005440945540162379,
      "learning_rate": 1.1808000000000001e-05,
      "loss": 0.0,
      "step": 12800
    },
    {
      "epoch": 4.112,
      "grad_norm": 0.0012483986833069704,
      "learning_rate": 1.1776e-05,
      "loss": 0.0,
      "step": 12850
    },
    {
      "epoch": 4.128,
      "grad_norm": 0.0012047606022239964,
      "learning_rate": 1.1744000000000001e-05,
      "loss": 0.0,
      "step": 12900
    },
    {
      "epoch": 4.144,
      "grad_norm": 0.0012935697233135471,
      "learning_rate": 1.1712e-05,
      "loss": 0.0,
      "step": 12950
    },
    {
      "epoch": 4.16,
      "grad_norm": 0.0003751557511675381,
      "learning_rate": 1.168e-05,
      "loss": 0.0,
      "step": 13000
    },
    {
      "epoch": 4.176,
      "grad_norm": 0.0007975727216523395,
      "learning_rate": 1.1648000000000001e-05,
      "loss": 0.0,
      "step": 13050
    },
    {
      "epoch": 4.192,
      "grad_norm": 0.0024788380663340725,
      "learning_rate": 1.1616e-05,
      "loss": 0.0,
      "step": 13100
    },
    {
      "epoch": 4.208,
      "grad_norm": 0.002267500273223812,
      "learning_rate": 1.1584000000000001e-05,
      "loss": 0.0,
      "step": 13150
    },
    {
      "epoch": 4.224,
      "grad_norm": 0.00046404415579022056,
      "learning_rate": 1.1552e-05,
      "loss": 0.0,
      "step": 13200
    },
    {
      "epoch": 4.24,
      "grad_norm": 0.0003794275042058772,
      "learning_rate": 1.152e-05,
      "loss": 0.0,
      "step": 13250
    },
    {
      "epoch": 4.256,
      "grad_norm": 5.7509067974309376e-05,
      "learning_rate": 1.1488e-05,
      "loss": 0.0,
      "step": 13300
    },
    {
      "epoch": 4.272,
      "grad_norm": 0.0010141779101323779,
      "learning_rate": 1.1456e-05,
      "loss": 0.0,
      "step": 13350
    },
    {
      "epoch": 4.288,
      "grad_norm": 0.000488791600842277,
      "learning_rate": 1.1424000000000001e-05,
      "loss": 0.0,
      "step": 13400
    },
    {
      "epoch": 4.304,
      "grad_norm": 0.009154124210044096,
      "learning_rate": 1.1392e-05,
      "loss": 0.0,
      "step": 13450
    },
    {
      "epoch": 4.32,
      "grad_norm": 0.0014804520943499787,
      "learning_rate": 1.136e-05,
      "loss": 0.0,
      "step": 13500
    },
    {
      "epoch": 4.336,
      "grad_norm": 0.003087388225389283,
      "learning_rate": 1.1328e-05,
      "loss": 0.0,
      "step": 13550
    },
    {
      "epoch": 4.352,
      "grad_norm": 0.0012502357899471038,
      "learning_rate": 1.1296e-05,
      "loss": 0.0,
      "step": 13600
    },
    {
      "epoch": 4.368,
      "grad_norm": 0.004994197578764371,
      "learning_rate": 1.1264000000000001e-05,
      "loss": 0.0,
      "step": 13650
    },
    {
      "epoch": 4.384,
      "grad_norm": 0.00236250892486376,
      "learning_rate": 1.1232e-05,
      "loss": 0.0,
      "step": 13700
    },
    {
      "epoch": 4.4,
      "grad_norm": 0.0009338977271527734,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 0.0,
      "step": 13750
    },
    {
      "epoch": 4.416,
      "grad_norm": 0.0017632337662450276,
      "learning_rate": 1.1168e-05,
      "loss": 0.0,
      "step": 13800
    },
    {
      "epoch": 4.432,
      "grad_norm": 0.004662567943032515,
      "learning_rate": 1.1136e-05,
      "loss": 0.0,
      "step": 13850
    },
    {
      "epoch": 4.448,
      "grad_norm": 0.0007880804713907978,
      "learning_rate": 1.1104e-05,
      "loss": 0.0,
      "step": 13900
    },
    {
      "epoch": 4.464,
      "grad_norm": 0.004039315206219138,
      "learning_rate": 1.1072e-05,
      "loss": 0.0,
      "step": 13950
    },
    {
      "epoch": 4.48,
      "grad_norm": 0.0025060993362850705,
      "learning_rate": 1.1040000000000001e-05,
      "loss": 0.0,
      "step": 14000
    },
    {
      "epoch": 4.496,
      "grad_norm": 0.0021604759353097895,
      "learning_rate": 1.1008e-05,
      "loss": 0.0,
      "step": 14050
    },
    {
      "epoch": 4.5120000000000005,
      "grad_norm": 0.002867990975950256,
      "learning_rate": 1.0976e-05,
      "loss": 0.0,
      "step": 14100
    },
    {
      "epoch": 4.5280000000000005,
      "grad_norm": 0.00401193895429708,
      "learning_rate": 1.0944e-05,
      "loss": 0.0001,
      "step": 14150
    },
    {
      "epoch": 4.5440000000000005,
      "grad_norm": 0.00573613040417129,
      "learning_rate": 1.0912e-05,
      "loss": 0.0,
      "step": 14200
    },
    {
      "epoch": 4.5600000000000005,
      "grad_norm": 0.0030535103264292297,
      "learning_rate": 1.0880000000000001e-05,
      "loss": 0.0,
      "step": 14250
    },
    {
      "epoch": 4.576,
      "grad_norm": 0.0013740615247333295,
      "learning_rate": 1.0848e-05,
      "loss": 0.0,
      "step": 14300
    },
    {
      "epoch": 4.592,
      "grad_norm": 0.0013075234756195337,
      "learning_rate": 1.0816e-05,
      "loss": 0.0,
      "step": 14350
    },
    {
      "epoch": 4.608,
      "grad_norm": 0.003144544935729042,
      "learning_rate": 1.0784e-05,
      "loss": 0.0,
      "step": 14400
    },
    {
      "epoch": 4.624,
      "grad_norm": 0.0004717265126905979,
      "learning_rate": 1.0752e-05,
      "loss": 0.0,
      "step": 14450
    },
    {
      "epoch": 4.64,
      "grad_norm": 0.0006579410910577381,
      "learning_rate": 1.072e-05,
      "loss": 0.0,
      "step": 14500
    },
    {
      "epoch": 4.656,
      "grad_norm": 0.0011156066953653578,
      "learning_rate": 1.0688e-05,
      "loss": 0.0,
      "step": 14550
    },
    {
      "epoch": 4.672,
      "grad_norm": 0.0005837718126418017,
      "learning_rate": 1.0656000000000003e-05,
      "loss": 0.0,
      "step": 14600
    },
    {
      "epoch": 4.688,
      "grad_norm": 0.0002716341341737398,
      "learning_rate": 1.0624e-05,
      "loss": 0.0,
      "step": 14650
    },
    {
      "epoch": 4.704,
      "grad_norm": 0.0006885004604574123,
      "learning_rate": 1.0592e-05,
      "loss": 0.0,
      "step": 14700
    },
    {
      "epoch": 4.72,
      "grad_norm": 0.0001909594048696283,
      "learning_rate": 1.056e-05,
      "loss": 0.0,
      "step": 14750
    },
    {
      "epoch": 4.736,
      "grad_norm": 0.0005826950037546315,
      "learning_rate": 1.0528e-05,
      "loss": 0.0,
      "step": 14800
    },
    {
      "epoch": 4.752,
      "grad_norm": 9.968768791997796e-05,
      "learning_rate": 1.0496000000000003e-05,
      "loss": 0.0,
      "step": 14850
    },
    {
      "epoch": 4.768,
      "grad_norm": 3.934391483544783e-05,
      "learning_rate": 1.0464e-05,
      "loss": 0.0,
      "step": 14900
    },
    {
      "epoch": 4.784,
      "grad_norm": 0.0006812436780773194,
      "learning_rate": 1.0432e-05,
      "loss": 0.0,
      "step": 14950
    },
    {
      "epoch": 4.8,
      "grad_norm": 0.0023170184859336246,
      "learning_rate": 1.04e-05,
      "loss": 0.0,
      "step": 15000
    },
    {
      "epoch": 4.816,
      "grad_norm": 0.0022018970646360205,
      "learning_rate": 1.0368e-05,
      "loss": 0.0,
      "step": 15050
    },
    {
      "epoch": 4.832,
      "grad_norm": 0.000783685265043126,
      "learning_rate": 1.0336000000000002e-05,
      "loss": 0.0,
      "step": 15100
    },
    {
      "epoch": 4.848,
      "grad_norm": 0.0038512547713090923,
      "learning_rate": 1.0304e-05,
      "loss": 0.0,
      "step": 15150
    },
    {
      "epoch": 4.864,
      "grad_norm": 0.002436496154884748,
      "learning_rate": 1.0272e-05,
      "loss": 0.0,
      "step": 15200
    },
    {
      "epoch": 4.88,
      "grad_norm": 0.00219326761780263,
      "learning_rate": 1.024e-05,
      "loss": 0.0,
      "step": 15250
    },
    {
      "epoch": 4.896,
      "grad_norm": 0.0011281003341395849,
      "learning_rate": 1.0208e-05,
      "loss": 0.0,
      "step": 15300
    },
    {
      "epoch": 4.912,
      "grad_norm": 0.003253789912933492,
      "learning_rate": 1.0176000000000002e-05,
      "loss": 0.0,
      "step": 15350
    },
    {
      "epoch": 4.928,
      "grad_norm": 0.003571745357413305,
      "learning_rate": 1.0144e-05,
      "loss": 0.0,
      "step": 15400
    },
    {
      "epoch": 4.944,
      "grad_norm": 0.003456593668540485,
      "learning_rate": 1.0112000000000002e-05,
      "loss": 0.0064,
      "step": 15450
    },
    {
      "epoch": 4.96,
      "grad_norm": 0.0007709261431954543,
      "learning_rate": 1.008e-05,
      "loss": 0.0,
      "step": 15500
    },
    {
      "epoch": 4.976,
      "grad_norm": 0.0016190183412224056,
      "learning_rate": 1.0048e-05,
      "loss": 0.0,
      "step": 15550
    },
    {
      "epoch": 4.992,
      "grad_norm": 0.0004234418224673852,
      "learning_rate": 1.0016000000000002e-05,
      "loss": 0.0,
      "step": 15600
    },
    {
      "epoch": 5.008,
      "grad_norm": 0.00026016080219425926,
      "learning_rate": 9.984e-06,
      "loss": 0.0,
      "step": 15650
    },
    {
      "epoch": 5.024,
      "grad_norm": 0.00011395553690679996,
      "learning_rate": 9.952e-06,
      "loss": 0.0,
      "step": 15700
    },
    {
      "epoch": 5.04,
      "grad_norm": 0.0007346519053602048,
      "learning_rate": 9.920000000000002e-06,
      "loss": 0.0,
      "step": 15750
    },
    {
      "epoch": 5.056,
      "grad_norm": 0.000826212364328148,
      "learning_rate": 9.888000000000001e-06,
      "loss": 0.0,
      "step": 15800
    },
    {
      "epoch": 5.072,
      "grad_norm": 3.820115501457758e-05,
      "learning_rate": 9.856000000000002e-06,
      "loss": 0.0,
      "step": 15850
    },
    {
      "epoch": 5.088,
      "grad_norm": 0.000369496148175044,
      "learning_rate": 9.824000000000001e-06,
      "loss": 0.0,
      "step": 15900
    },
    {
      "epoch": 5.104,
      "grad_norm": 3.493119944946329e-05,
      "learning_rate": 9.792e-06,
      "loss": 0.0,
      "step": 15950
    },
    {
      "epoch": 5.12,
      "grad_norm": 0.0011101639738022185,
      "learning_rate": 9.760000000000001e-06,
      "loss": 0.0,
      "step": 16000
    },
    {
      "epoch": 5.136,
      "grad_norm": 0.00027278241311842,
      "learning_rate": 9.728e-06,
      "loss": 0.0,
      "step": 16050
    },
    {
      "epoch": 5.152,
      "grad_norm": 7.425121803341847e-05,
      "learning_rate": 9.696000000000002e-06,
      "loss": 0.0,
      "step": 16100
    },
    {
      "epoch": 5.168,
      "grad_norm": 2.913913515572778e-05,
      "learning_rate": 9.664000000000001e-06,
      "loss": 0.0,
      "step": 16150
    },
    {
      "epoch": 5.184,
      "grad_norm": 0.00022998811212515617,
      "learning_rate": 9.632e-06,
      "loss": 0.0,
      "step": 16200
    },
    {
      "epoch": 5.2,
      "grad_norm": 0.0001390860734604394,
      "learning_rate": 9.600000000000001e-06,
      "loss": 0.0,
      "step": 16250
    },
    {
      "epoch": 5.216,
      "grad_norm": 0.00046520681578371507,
      "learning_rate": 9.568e-06,
      "loss": 0.0,
      "step": 16300
    },
    {
      "epoch": 5.232,
      "grad_norm": 1.703864278625447e-05,
      "learning_rate": 9.536000000000002e-06,
      "loss": 0.0,
      "step": 16350
    },
    {
      "epoch": 5.248,
      "grad_norm": 0.0003273884641160331,
      "learning_rate": 9.504e-06,
      "loss": 0.0,
      "step": 16400
    },
    {
      "epoch": 5.264,
      "grad_norm": 1.3852768295605306e-05,
      "learning_rate": 9.472000000000002e-06,
      "loss": 0.0,
      "step": 16450
    },
    {
      "epoch": 5.28,
      "grad_norm": 0.0008515611050291017,
      "learning_rate": 9.440000000000001e-06,
      "loss": 0.0,
      "step": 16500
    },
    {
      "epoch": 5.296,
      "grad_norm": 0.0008534837876489844,
      "learning_rate": 9.408e-06,
      "loss": 0.0,
      "step": 16550
    },
    {
      "epoch": 5.312,
      "grad_norm": 0.0006724273990838339,
      "learning_rate": 9.376000000000001e-06,
      "loss": 0.0,
      "step": 16600
    },
    {
      "epoch": 5.328,
      "grad_norm": 0.0007236852854461475,
      "learning_rate": 9.344e-06,
      "loss": 0.0,
      "step": 16650
    },
    {
      "epoch": 5.344,
      "grad_norm": 0.00010455353593376917,
      "learning_rate": 9.312000000000002e-06,
      "loss": 0.0,
      "step": 16700
    },
    {
      "epoch": 5.36,
      "grad_norm": 0.0007310347123221347,
      "learning_rate": 9.280000000000001e-06,
      "loss": 0.0,
      "step": 16750
    },
    {
      "epoch": 5.376,
      "grad_norm": 0.0005001089317212744,
      "learning_rate": 9.248e-06,
      "loss": 0.0,
      "step": 16800
    },
    {
      "epoch": 5.392,
      "grad_norm": 0.001043623838616486,
      "learning_rate": 9.216000000000001e-06,
      "loss": 0.0,
      "step": 16850
    },
    {
      "epoch": 5.408,
      "grad_norm": 0.0007706737604305622,
      "learning_rate": 9.184e-06,
      "loss": 0.0,
      "step": 16900
    },
    {
      "epoch": 5.424,
      "grad_norm": 9.706295875901001e-06,
      "learning_rate": 9.152000000000001e-06,
      "loss": 0.0,
      "step": 16950
    },
    {
      "epoch": 5.44,
      "grad_norm": 0.0008492197216223447,
      "learning_rate": 9.12e-06,
      "loss": 0.0,
      "step": 17000
    },
    {
      "epoch": 5.456,
      "grad_norm": 0.0013652762132398285,
      "learning_rate": 9.088000000000002e-06,
      "loss": 0.0,
      "step": 17050
    },
    {
      "epoch": 5.4719999999999995,
      "grad_norm": 0.0008027400483996306,
      "learning_rate": 9.056000000000001e-06,
      "loss": 0.0,
      "step": 17100
    },
    {
      "epoch": 5.4879999999999995,
      "grad_norm": 1.0452267647832807e-05,
      "learning_rate": 9.024e-06,
      "loss": 0.0,
      "step": 17150
    },
    {
      "epoch": 5.504,
      "grad_norm": 6.906557178289084e-06,
      "learning_rate": 8.992000000000001e-06,
      "loss": 0.0,
      "step": 17200
    },
    {
      "epoch": 5.52,
      "grad_norm": 0.0005904441392333963,
      "learning_rate": 8.96e-06,
      "loss": 0.0,
      "step": 17250
    },
    {
      "epoch": 5.536,
      "grad_norm": 0.0010790578585745268,
      "learning_rate": 8.928000000000002e-06,
      "loss": 0.0,
      "step": 17300
    },
    {
      "epoch": 5.552,
      "grad_norm": 0.00035965364826304414,
      "learning_rate": 8.896000000000001e-06,
      "loss": 0.0,
      "step": 17350
    },
    {
      "epoch": 5.568,
      "grad_norm": 0.00045245800981555623,
      "learning_rate": 8.864e-06,
      "loss": 0.0,
      "step": 17400
    },
    {
      "epoch": 5.584,
      "grad_norm": 0.00010256823791726092,
      "learning_rate": 8.832000000000001e-06,
      "loss": 0.0,
      "step": 17450
    },
    {
      "epoch": 5.6,
      "grad_norm": 8.117090788676675e-06,
      "learning_rate": 8.8e-06,
      "loss": 0.0,
      "step": 17500
    },
    {
      "epoch": 5.616,
      "grad_norm": 0.0007363027867556159,
      "learning_rate": 8.768000000000001e-06,
      "loss": 0.0,
      "step": 17550
    },
    {
      "epoch": 5.632,
      "grad_norm": 0.0005243403229001632,
      "learning_rate": 8.736e-06,
      "loss": 0.0,
      "step": 17600
    },
    {
      "epoch": 5.648,
      "grad_norm": 0.00040569564594367943,
      "learning_rate": 8.704e-06,
      "loss": 0.0,
      "step": 17650
    },
    {
      "epoch": 5.664,
      "grad_norm": 0.00042046445593646643,
      "learning_rate": 8.672000000000001e-06,
      "loss": 0.0,
      "step": 17700
    },
    {
      "epoch": 5.68,
      "grad_norm": 0.0004739053906413846,
      "learning_rate": 8.64e-06,
      "loss": 0.0,
      "step": 17750
    },
    {
      "epoch": 5.696,
      "grad_norm": 5.641085857115355e-05,
      "learning_rate": 8.608000000000001e-06,
      "loss": 0.0,
      "step": 17800
    },
    {
      "epoch": 5.712,
      "grad_norm": 0.000717217643766636,
      "learning_rate": 8.576e-06,
      "loss": 0.0,
      "step": 17850
    },
    {
      "epoch": 5.728,
      "grad_norm": 0.00035495184416538993,
      "learning_rate": 8.544000000000002e-06,
      "loss": 0.0,
      "step": 17900
    },
    {
      "epoch": 5.744,
      "grad_norm": 0.0007510280987038669,
      "learning_rate": 8.512e-06,
      "loss": 0.0,
      "step": 17950
    },
    {
      "epoch": 5.76,
      "grad_norm": 0.00019448124680507923,
      "learning_rate": 8.48e-06,
      "loss": 0.0,
      "step": 18000
    },
    {
      "epoch": 5.776,
      "grad_norm": 0.000741237981085058,
      "learning_rate": 8.448000000000001e-06,
      "loss": 0.0,
      "step": 18050
    },
    {
      "epoch": 5.792,
      "grad_norm": 0.0012189065717582211,
      "learning_rate": 8.416e-06,
      "loss": 0.0,
      "step": 18100
    },
    {
      "epoch": 5.808,
      "grad_norm": 0.00013649905846145337,
      "learning_rate": 8.384000000000001e-06,
      "loss": 0.0,
      "step": 18150
    },
    {
      "epoch": 5.824,
      "grad_norm": 0.0012586456173907475,
      "learning_rate": 8.352e-06,
      "loss": 0.0,
      "step": 18200
    },
    {
      "epoch": 5.84,
      "grad_norm": 0.00046421865350467135,
      "learning_rate": 8.32e-06,
      "loss": 0.0,
      "step": 18250
    },
    {
      "epoch": 5.856,
      "grad_norm": 0.0008588455951536167,
      "learning_rate": 8.288000000000001e-06,
      "loss": 0.0,
      "step": 18300
    },
    {
      "epoch": 5.872,
      "grad_norm": 0.0007684313974727155,
      "learning_rate": 8.256e-06,
      "loss": 0.0,
      "step": 18350
    },
    {
      "epoch": 5.888,
      "grad_norm": 0.0007039589600819322,
      "learning_rate": 8.224000000000001e-06,
      "loss": 0.0,
      "step": 18400
    },
    {
      "epoch": 5.904,
      "grad_norm": 3.51837159957191e-05,
      "learning_rate": 8.192e-06,
      "loss": 0.0,
      "step": 18450
    },
    {
      "epoch": 5.92,
      "grad_norm": 0.0008249995746808375,
      "learning_rate": 8.16e-06,
      "loss": 0.0,
      "step": 18500
    },
    {
      "epoch": 5.936,
      "grad_norm": 0.0005751905756043754,
      "learning_rate": 8.128e-06,
      "loss": 0.0,
      "step": 18550
    },
    {
      "epoch": 5.952,
      "grad_norm": 0.0006014537857756254,
      "learning_rate": 8.096e-06,
      "loss": 0.0,
      "step": 18600
    },
    {
      "epoch": 5.968,
      "grad_norm": 0.0004182340166299973,
      "learning_rate": 8.064000000000001e-06,
      "loss": 0.0,
      "step": 18650
    },
    {
      "epoch": 5.984,
      "grad_norm": 7.328929449883947e-06,
      "learning_rate": 8.032e-06,
      "loss": 0.0,
      "step": 18700
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.0005532029221570249,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.0,
      "step": 18750
    },
    {
      "epoch": 6.016,
      "grad_norm": 0.0004575747788336891,
      "learning_rate": 7.968e-06,
      "loss": 0.0,
      "step": 18800
    },
    {
      "epoch": 6.032,
      "grad_norm": 0.001869666267334979,
      "learning_rate": 7.936e-06,
      "loss": 0.0,
      "step": 18850
    },
    {
      "epoch": 6.048,
      "grad_norm": 0.00022141063379263522,
      "learning_rate": 7.904000000000001e-06,
      "loss": 0.0,
      "step": 18900
    },
    {
      "epoch": 6.064,
      "grad_norm": 0.0002755551414857357,
      "learning_rate": 7.872e-06,
      "loss": 0.0,
      "step": 18950
    },
    {
      "epoch": 6.08,
      "grad_norm": 0.0001262094453801626,
      "learning_rate": 7.840000000000001e-06,
      "loss": 0.0,
      "step": 19000
    },
    {
      "epoch": 6.096,
      "grad_norm": 0.00021903586619369134,
      "learning_rate": 7.808e-06,
      "loss": 0.0,
      "step": 19050
    },
    {
      "epoch": 6.112,
      "grad_norm": 0.0006519057730708936,
      "learning_rate": 7.776e-06,
      "loss": 0.0,
      "step": 19100
    },
    {
      "epoch": 6.128,
      "grad_norm": 0.000695502894667579,
      "learning_rate": 7.744e-06,
      "loss": 0.0,
      "step": 19150
    },
    {
      "epoch": 6.144,
      "grad_norm": 0.0006224173048035789,
      "learning_rate": 7.712e-06,
      "loss": 0.0,
      "step": 19200
    },
    {
      "epoch": 6.16,
      "grad_norm": 3.4614518032103047e-06,
      "learning_rate": 7.680000000000001e-06,
      "loss": 0.0,
      "step": 19250
    },
    {
      "epoch": 6.176,
      "grad_norm": 0.00033523878449984705,
      "learning_rate": 7.648e-06,
      "loss": 0.0,
      "step": 19300
    },
    {
      "epoch": 6.192,
      "grad_norm": 0.0008220658302305507,
      "learning_rate": 7.616000000000001e-06,
      "loss": 0.0,
      "step": 19350
    },
    {
      "epoch": 6.208,
      "grad_norm": 0.0003363971581878067,
      "learning_rate": 7.5840000000000006e-06,
      "loss": 0.0,
      "step": 19400
    },
    {
      "epoch": 6.224,
      "grad_norm": 0.0006817803137925653,
      "learning_rate": 7.552000000000001e-06,
      "loss": 0.0,
      "step": 19450
    },
    {
      "epoch": 6.24,
      "grad_norm": 0.0011120061826881474,
      "learning_rate": 7.520000000000001e-06,
      "loss": 0.0,
      "step": 19500
    },
    {
      "epoch": 6.256,
      "grad_norm": 1.966465122456936e-06,
      "learning_rate": 7.488000000000001e-06,
      "loss": 0.0,
      "step": 19550
    },
    {
      "epoch": 6.272,
      "grad_norm": 0.00026856259334759303,
      "learning_rate": 7.456000000000001e-06,
      "loss": 0.0,
      "step": 19600
    },
    {
      "epoch": 6.288,
      "grad_norm": 0.00044370451229992384,
      "learning_rate": 7.424e-06,
      "loss": 0.0,
      "step": 19650
    },
    {
      "epoch": 6.304,
      "grad_norm": 0.00028155440140211114,
      "learning_rate": 7.3920000000000005e-06,
      "loss": 0.0,
      "step": 19700
    },
    {
      "epoch": 6.32,
      "grad_norm": 0.0005292253112573046,
      "learning_rate": 7.360000000000001e-06,
      "loss": 0.0,
      "step": 19750
    },
    {
      "epoch": 6.336,
      "grad_norm": 0.0007954187703791572,
      "learning_rate": 7.328000000000001e-06,
      "loss": 0.0,
      "step": 19800
    },
    {
      "epoch": 6.352,
      "grad_norm": 0.00027606186196444586,
      "learning_rate": 7.296000000000001e-06,
      "loss": 0.0,
      "step": 19850
    },
    {
      "epoch": 6.368,
      "grad_norm": 6.270867123019646e-06,
      "learning_rate": 7.264000000000001e-06,
      "loss": 0.0,
      "step": 19900
    },
    {
      "epoch": 6.384,
      "grad_norm": 0.00022346505397508303,
      "learning_rate": 7.232e-06,
      "loss": 0.0,
      "step": 19950
    },
    {
      "epoch": 6.4,
      "grad_norm": 0.0005133763871789368,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 0.0,
      "step": 20000
    },
    {
      "epoch": 6.416,
      "grad_norm": 0.0012385305880268965,
      "learning_rate": 7.168000000000001e-06,
      "loss": 0.0,
      "step": 20050
    },
    {
      "epoch": 6.432,
      "grad_norm": 1.8737486868066673e-06,
      "learning_rate": 7.136000000000001e-06,
      "loss": 0.0,
      "step": 20100
    },
    {
      "epoch": 6.448,
      "grad_norm": 2.0181046177978524e-06,
      "learning_rate": 7.104000000000001e-06,
      "loss": 0.0,
      "step": 20150
    },
    {
      "epoch": 6.464,
      "grad_norm": 0.000845228083001162,
      "learning_rate": 7.072000000000001e-06,
      "loss": 0.0,
      "step": 20200
    },
    {
      "epoch": 6.48,
      "grad_norm": 0.00010442413166408255,
      "learning_rate": 7.04e-06,
      "loss": 0.0,
      "step": 20250
    },
    {
      "epoch": 6.496,
      "grad_norm": 0.0008259383918680861,
      "learning_rate": 7.0080000000000005e-06,
      "loss": 0.0,
      "step": 20300
    },
    {
      "epoch": 6.5120000000000005,
      "grad_norm": 1.9801867703023435e-06,
      "learning_rate": 6.976000000000001e-06,
      "loss": 0.0,
      "step": 20350
    },
    {
      "epoch": 6.5280000000000005,
      "grad_norm": 0.0003420242897073274,
      "learning_rate": 6.944000000000001e-06,
      "loss": 0.0,
      "step": 20400
    },
    {
      "epoch": 6.5440000000000005,
      "grad_norm": 0.000927320381200099,
      "learning_rate": 6.912000000000001e-06,
      "loss": 0.0,
      "step": 20450
    },
    {
      "epoch": 6.5600000000000005,
      "grad_norm": 0.0005350790156068339,
      "learning_rate": 6.88e-06,
      "loss": 0.0,
      "step": 20500
    },
    {
      "epoch": 6.576,
      "grad_norm": 0.0003440149318502077,
      "learning_rate": 6.848e-06,
      "loss": 0.0,
      "step": 20550
    },
    {
      "epoch": 6.592,
      "grad_norm": 0.0007668699288568762,
      "learning_rate": 6.8160000000000005e-06,
      "loss": 0.0,
      "step": 20600
    },
    {
      "epoch": 6.608,
      "grad_norm": 0.00046621701082447185,
      "learning_rate": 6.784000000000001e-06,
      "loss": 0.0,
      "step": 20650
    },
    {
      "epoch": 6.624,
      "grad_norm": 5.3773575242483376e-05,
      "learning_rate": 6.752000000000001e-06,
      "loss": 0.0,
      "step": 20700
    },
    {
      "epoch": 6.64,
      "grad_norm": 8.350580982910908e-05,
      "learning_rate": 6.720000000000001e-06,
      "loss": 0.0,
      "step": 20750
    },
    {
      "epoch": 6.656,
      "grad_norm": 0.0002395916406284074,
      "learning_rate": 6.688e-06,
      "loss": 0.0,
      "step": 20800
    },
    {
      "epoch": 6.672,
      "grad_norm": 0.0006504634991768935,
      "learning_rate": 6.656e-06,
      "loss": 0.0,
      "step": 20850
    },
    {
      "epoch": 6.688,
      "grad_norm": 0.0006017391748290424,
      "learning_rate": 6.6240000000000004e-06,
      "loss": 0.0,
      "step": 20900
    },
    {
      "epoch": 6.704,
      "grad_norm": 0.0002806311098905263,
      "learning_rate": 6.592000000000001e-06,
      "loss": 0.0,
      "step": 20950
    },
    {
      "epoch": 6.72,
      "grad_norm": 0.0008733269204726745,
      "learning_rate": 6.560000000000001e-06,
      "loss": 0.0,
      "step": 21000
    },
    {
      "epoch": 6.736,
      "grad_norm": 0.0005751125071293194,
      "learning_rate": 6.528000000000001e-06,
      "loss": 0.0,
      "step": 21050
    },
    {
      "epoch": 6.752,
      "grad_norm": 0.006290860724228949,
      "learning_rate": 6.496e-06,
      "loss": 0.0,
      "step": 21100
    },
    {
      "epoch": 6.768,
      "grad_norm": 0.0040169282359627054,
      "learning_rate": 6.464e-06,
      "loss": 0.0,
      "step": 21150
    },
    {
      "epoch": 6.784,
      "grad_norm": 0.001080348541164438,
      "learning_rate": 6.432e-06,
      "loss": 0.0,
      "step": 21200
    },
    {
      "epoch": 6.8,
      "grad_norm": 0.003811186265590068,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 0.0,
      "step": 21250
    },
    {
      "epoch": 6.816,
      "grad_norm": 0.00016534484442220018,
      "learning_rate": 6.368000000000001e-06,
      "loss": 0.0,
      "step": 21300
    },
    {
      "epoch": 6.832,
      "grad_norm": 0.0001613165255933953,
      "learning_rate": 6.336000000000001e-06,
      "loss": 0.0,
      "step": 21350
    },
    {
      "epoch": 6.848,
      "grad_norm": 0.000679046061777582,
      "learning_rate": 6.304e-06,
      "loss": 0.0,
      "step": 21400
    },
    {
      "epoch": 6.864,
      "grad_norm": 0.0010514425379430034,
      "learning_rate": 6.272e-06,
      "loss": 0.0,
      "step": 21450
    },
    {
      "epoch": 6.88,
      "grad_norm": 0.0035533924147301845,
      "learning_rate": 6.24e-06,
      "loss": 0.0,
      "step": 21500
    },
    {
      "epoch": 6.896,
      "grad_norm": 0.0010389500345308828,
      "learning_rate": 6.2080000000000005e-06,
      "loss": 0.0,
      "step": 21550
    },
    {
      "epoch": 6.912,
      "grad_norm": 0.0006256094370156619,
      "learning_rate": 6.176000000000001e-06,
      "loss": 0.0,
      "step": 21600
    },
    {
      "epoch": 6.928,
      "grad_norm": 0.00014983574866582935,
      "learning_rate": 6.144e-06,
      "loss": 0.0,
      "step": 21650
    },
    {
      "epoch": 6.944,
      "grad_norm": 7.81575212969094e-06,
      "learning_rate": 6.112e-06,
      "loss": 0.0,
      "step": 21700
    },
    {
      "epoch": 6.96,
      "grad_norm": 0.00010497171085101809,
      "learning_rate": 6.08e-06,
      "loss": 0.0,
      "step": 21750
    },
    {
      "epoch": 6.976,
      "grad_norm": 0.00026859080304041767,
      "learning_rate": 6.048e-06,
      "loss": 0.0,
      "step": 21800
    },
    {
      "epoch": 6.992,
      "grad_norm": 0.0007656211816805965,
      "learning_rate": 6.0160000000000005e-06,
      "loss": 0.0,
      "step": 21850
    },
    {
      "epoch": 7.008,
      "grad_norm": 0.00016866370577004142,
      "learning_rate": 5.984000000000001e-06,
      "loss": 0.0,
      "step": 21900
    },
    {
      "epoch": 7.024,
      "grad_norm": 0.000841725008729615,
      "learning_rate": 5.952e-06,
      "loss": 0.0,
      "step": 21950
    },
    {
      "epoch": 7.04,
      "grad_norm": 0.001403596803582104,
      "learning_rate": 5.92e-06,
      "loss": 0.0,
      "step": 22000
    },
    {
      "epoch": 7.056,
      "grad_norm": 0.0007519805421921276,
      "learning_rate": 5.888e-06,
      "loss": 0.0,
      "step": 22050
    },
    {
      "epoch": 7.072,
      "grad_norm": 3.008559634025441e-05,
      "learning_rate": 5.856e-06,
      "loss": 0.0,
      "step": 22100
    },
    {
      "epoch": 7.088,
      "grad_norm": 0.0006102491301472853,
      "learning_rate": 5.8240000000000005e-06,
      "loss": 0.0,
      "step": 22150
    },
    {
      "epoch": 7.104,
      "grad_norm": 0.000600562913095834,
      "learning_rate": 5.792000000000001e-06,
      "loss": 0.0,
      "step": 22200
    },
    {
      "epoch": 7.12,
      "grad_norm": 0.0004817236180086057,
      "learning_rate": 5.76e-06,
      "loss": 0.0,
      "step": 22250
    },
    {
      "epoch": 7.136,
      "grad_norm": 0.000686443266077338,
      "learning_rate": 5.728e-06,
      "loss": 0.0,
      "step": 22300
    },
    {
      "epoch": 7.152,
      "grad_norm": 0.001062635129484694,
      "learning_rate": 5.696e-06,
      "loss": 0.0,
      "step": 22350
    },
    {
      "epoch": 7.168,
      "grad_norm": 0.000404941651506133,
      "learning_rate": 5.664e-06,
      "loss": 0.0,
      "step": 22400
    },
    {
      "epoch": 7.184,
      "grad_norm": 0.000816725754364363,
      "learning_rate": 5.6320000000000005e-06,
      "loss": 0.0,
      "step": 22450
    },
    {
      "epoch": 7.2,
      "grad_norm": 0.00037021587973858533,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.0,
      "step": 22500
    },
    {
      "epoch": 7.216,
      "grad_norm": 4.088127671811198e-06,
      "learning_rate": 5.568e-06,
      "loss": 0.0,
      "step": 22550
    },
    {
      "epoch": 7.232,
      "grad_norm": 0.00013304569592629414,
      "learning_rate": 5.536e-06,
      "loss": 0.0,
      "step": 22600
    },
    {
      "epoch": 7.248,
      "grad_norm": 0.00011693788654007106,
      "learning_rate": 5.504e-06,
      "loss": 0.0,
      "step": 22650
    },
    {
      "epoch": 7.264,
      "grad_norm": 0.00039873529911523876,
      "learning_rate": 5.472e-06,
      "loss": 0.0,
      "step": 22700
    },
    {
      "epoch": 7.28,
      "grad_norm": 0.0005861985910699136,
      "learning_rate": 5.4400000000000004e-06,
      "loss": 0.0,
      "step": 22750
    },
    {
      "epoch": 7.296,
      "grad_norm": 0.0001408305739675062,
      "learning_rate": 5.408e-06,
      "loss": 0.0,
      "step": 22800
    },
    {
      "epoch": 7.312,
      "grad_norm": 0.0010182175879277604,
      "learning_rate": 5.376e-06,
      "loss": 0.0,
      "step": 22850
    },
    {
      "epoch": 7.328,
      "grad_norm": 0.00032526688809646643,
      "learning_rate": 5.344e-06,
      "loss": 0.0,
      "step": 22900
    },
    {
      "epoch": 7.344,
      "grad_norm": 0.00017128057733419155,
      "learning_rate": 5.312e-06,
      "loss": 0.0,
      "step": 22950
    },
    {
      "epoch": 7.36,
      "grad_norm": 2.4948106999283475e-06,
      "learning_rate": 5.28e-06,
      "loss": 0.0,
      "step": 23000
    },
    {
      "epoch": 7.376,
      "grad_norm": 0.000593923369887912,
      "learning_rate": 5.248000000000001e-06,
      "loss": 0.0,
      "step": 23050
    },
    {
      "epoch": 7.392,
      "grad_norm": 0.0011459595030973938,
      "learning_rate": 5.216e-06,
      "loss": 0.0,
      "step": 23100
    },
    {
      "epoch": 7.408,
      "grad_norm": 0.00026313583806989295,
      "learning_rate": 5.184e-06,
      "loss": 0.0,
      "step": 23150
    },
    {
      "epoch": 7.424,
      "grad_norm": 0.00039830820671334343,
      "learning_rate": 5.152e-06,
      "loss": 0.0,
      "step": 23200
    },
    {
      "epoch": 7.44,
      "grad_norm": 0.0006286453125070797,
      "learning_rate": 5.12e-06,
      "loss": 0.0,
      "step": 23250
    },
    {
      "epoch": 7.456,
      "grad_norm": 0.0006427562567428625,
      "learning_rate": 5.088000000000001e-06,
      "loss": 0.0,
      "step": 23300
    },
    {
      "epoch": 7.4719999999999995,
      "grad_norm": 0.00015946006697521752,
      "learning_rate": 5.056000000000001e-06,
      "loss": 0.0,
      "step": 23350
    },
    {
      "epoch": 7.4879999999999995,
      "grad_norm": 0.0004289380452787707,
      "learning_rate": 5.024e-06,
      "loss": 0.0,
      "step": 23400
    },
    {
      "epoch": 7.504,
      "grad_norm": 0.000861813993896526,
      "learning_rate": 4.992e-06,
      "loss": 0.0,
      "step": 23450
    },
    {
      "epoch": 7.52,
      "grad_norm": 0.0005979597824620298,
      "learning_rate": 4.960000000000001e-06,
      "loss": 0.0,
      "step": 23500
    },
    {
      "epoch": 7.536,
      "grad_norm": 3.6426689561550785e-06,
      "learning_rate": 4.928000000000001e-06,
      "loss": 0.0,
      "step": 23550
    },
    {
      "epoch": 7.552,
      "grad_norm": 0.0004956499352349563,
      "learning_rate": 4.896e-06,
      "loss": 0.0,
      "step": 23600
    },
    {
      "epoch": 7.568,
      "grad_norm": 2.025048998589199e-05,
      "learning_rate": 4.864e-06,
      "loss": 0.0,
      "step": 23650
    },
    {
      "epoch": 7.584,
      "grad_norm": 0.0003160837457656607,
      "learning_rate": 4.8320000000000005e-06,
      "loss": 0.0,
      "step": 23700
    },
    {
      "epoch": 7.6,
      "grad_norm": 0.0006549989193442926,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.0,
      "step": 23750
    },
    {
      "epoch": 7.616,
      "grad_norm": 0.00022446916196028902,
      "learning_rate": 4.768000000000001e-06,
      "loss": 0.0,
      "step": 23800
    },
    {
      "epoch": 7.632,
      "grad_norm": 0.00018036750821287214,
      "learning_rate": 4.736000000000001e-06,
      "loss": 0.0,
      "step": 23850
    },
    {
      "epoch": 7.648,
      "grad_norm": 0.0004573385403025739,
      "learning_rate": 4.704e-06,
      "loss": 0.0,
      "step": 23900
    },
    {
      "epoch": 7.664,
      "grad_norm": 0.00019710767122580646,
      "learning_rate": 4.672e-06,
      "loss": 0.0,
      "step": 23950
    },
    {
      "epoch": 7.68,
      "grad_norm": 1.5084117814095418e-06,
      "learning_rate": 4.6400000000000005e-06,
      "loss": 0.0,
      "step": 24000
    },
    {
      "epoch": 7.696,
      "grad_norm": 0.0006600246803147537,
      "learning_rate": 4.608000000000001e-06,
      "loss": 0.0,
      "step": 24050
    },
    {
      "epoch": 7.712,
      "grad_norm": 0.0004764085088233622,
      "learning_rate": 4.576000000000001e-06,
      "loss": 0.0,
      "step": 24100
    },
    {
      "epoch": 7.728,
      "grad_norm": 0.00044471308815751895,
      "learning_rate": 4.544000000000001e-06,
      "loss": 0.0,
      "step": 24150
    },
    {
      "epoch": 7.744,
      "grad_norm": 0.0001657220726173927,
      "learning_rate": 4.512e-06,
      "loss": 0.0,
      "step": 24200
    },
    {
      "epoch": 7.76,
      "grad_norm": 0.00021805960530738044,
      "learning_rate": 4.48e-06,
      "loss": 0.0,
      "step": 24250
    },
    {
      "epoch": 7.776,
      "grad_norm": 0.0006216156458408036,
      "learning_rate": 4.4480000000000004e-06,
      "loss": 0.0,
      "step": 24300
    },
    {
      "epoch": 7.792,
      "grad_norm": 1.279911660415784e-06,
      "learning_rate": 4.416000000000001e-06,
      "loss": 0.0,
      "step": 24350
    },
    {
      "epoch": 7.808,
      "grad_norm": 0.0001357939729174033,
      "learning_rate": 4.384000000000001e-06,
      "loss": 0.0,
      "step": 24400
    },
    {
      "epoch": 7.824,
      "grad_norm": 0.0005642799025411847,
      "learning_rate": 4.352e-06,
      "loss": 0.0,
      "step": 24450
    },
    {
      "epoch": 7.84,
      "grad_norm": 0.00018091340913126554,
      "learning_rate": 4.32e-06,
      "loss": 0.0,
      "step": 24500
    },
    {
      "epoch": 7.856,
      "grad_norm": 0.0007611652586088176,
      "learning_rate": 4.288e-06,
      "loss": 0.0,
      "step": 24550
    },
    {
      "epoch": 7.872,
      "grad_norm": 0.0008024893561623562,
      "learning_rate": 4.256e-06,
      "loss": 0.0,
      "step": 24600
    },
    {
      "epoch": 7.888,
      "grad_norm": 1.2498287700551542e-06,
      "learning_rate": 4.2240000000000006e-06,
      "loss": 0.0,
      "step": 24650
    },
    {
      "epoch": 7.904,
      "grad_norm": 1.5673886902933284e-06,
      "learning_rate": 4.192000000000001e-06,
      "loss": 0.0,
      "step": 24700
    },
    {
      "epoch": 7.92,
      "grad_norm": 0.0006167687246851426,
      "learning_rate": 4.16e-06,
      "loss": 0.0,
      "step": 24750
    },
    {
      "epoch": 7.936,
      "grad_norm": 1.2371482089162035e-06,
      "learning_rate": 4.128e-06,
      "loss": 0.0,
      "step": 24800
    },
    {
      "epoch": 7.952,
      "grad_norm": 0.00020817390844313262,
      "learning_rate": 4.096e-06,
      "loss": 0.0,
      "step": 24850
    },
    {
      "epoch": 7.968,
      "grad_norm": 0.000763205279133764,
      "learning_rate": 4.064e-06,
      "loss": 0.0,
      "step": 24900
    },
    {
      "epoch": 7.984,
      "grad_norm": 0.0010338995161218454,
      "learning_rate": 4.0320000000000005e-06,
      "loss": 0.0,
      "step": 24950
    },
    {
      "epoch": 8.0,
      "grad_norm": 0.0005393700329852327,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0,
      "step": 25000
    },
    {
      "epoch": 8.016,
      "grad_norm": 1.5599262561285773e-06,
      "learning_rate": 3.968e-06,
      "loss": 0.0,
      "step": 25050
    },
    {
      "epoch": 8.032,
      "grad_norm": 0.00012475265397112023,
      "learning_rate": 3.936e-06,
      "loss": 0.0,
      "step": 25100
    },
    {
      "epoch": 8.048,
      "grad_norm": 0.0004963011431024235,
      "learning_rate": 3.904e-06,
      "loss": 0.0,
      "step": 25150
    },
    {
      "epoch": 8.064,
      "grad_norm": 2.1454502270095894e-05,
      "learning_rate": 3.872e-06,
      "loss": 0.0,
      "step": 25200
    },
    {
      "epoch": 8.08,
      "grad_norm": 7.98369110830542e-05,
      "learning_rate": 3.8400000000000005e-06,
      "loss": 0.0,
      "step": 25250
    },
    {
      "epoch": 8.096,
      "grad_norm": 0.0001569271601524967,
      "learning_rate": 3.8080000000000006e-06,
      "loss": 0.0,
      "step": 25300
    },
    {
      "epoch": 8.112,
      "grad_norm": 3.448395868399914e-06,
      "learning_rate": 3.7760000000000004e-06,
      "loss": 0.0,
      "step": 25350
    },
    {
      "epoch": 8.128,
      "grad_norm": 0.00021317804687393338,
      "learning_rate": 3.7440000000000005e-06,
      "loss": 0.0,
      "step": 25400
    },
    {
      "epoch": 8.144,
      "grad_norm": 0.002043492139335667,
      "learning_rate": 3.712e-06,
      "loss": 0.0,
      "step": 25450
    },
    {
      "epoch": 8.16,
      "grad_norm": 0.0007697758477107576,
      "learning_rate": 3.6800000000000003e-06,
      "loss": 0.0,
      "step": 25500
    },
    {
      "epoch": 8.176,
      "grad_norm": 0.00042121951583966874,
      "learning_rate": 3.6480000000000005e-06,
      "loss": 0.0,
      "step": 25550
    },
    {
      "epoch": 8.192,
      "grad_norm": 0.0010275855739816497,
      "learning_rate": 3.616e-06,
      "loss": 0.0,
      "step": 25600
    },
    {
      "epoch": 8.208,
      "grad_norm": 0.0004073857184752387,
      "learning_rate": 3.5840000000000003e-06,
      "loss": 0.0,
      "step": 25650
    },
    {
      "epoch": 8.224,
      "grad_norm": 2.6395644739553403e-05,
      "learning_rate": 3.5520000000000005e-06,
      "loss": 0.0,
      "step": 25700
    },
    {
      "epoch": 8.24,
      "grad_norm": 8.350450297173982e-05,
      "learning_rate": 3.52e-06,
      "loss": 0.0,
      "step": 25750
    },
    {
      "epoch": 8.256,
      "grad_norm": 0.0005017647122814175,
      "learning_rate": 3.4880000000000003e-06,
      "loss": 0.0,
      "step": 25800
    },
    {
      "epoch": 8.272,
      "grad_norm": 0.0002957195671807615,
      "learning_rate": 3.4560000000000005e-06,
      "loss": 0.0,
      "step": 25850
    },
    {
      "epoch": 8.288,
      "grad_norm": 0.0007477102897762827,
      "learning_rate": 3.424e-06,
      "loss": 0.0,
      "step": 25900
    },
    {
      "epoch": 8.304,
      "grad_norm": 0.00021069394537475973,
      "learning_rate": 3.3920000000000003e-06,
      "loss": 0.0,
      "step": 25950
    },
    {
      "epoch": 8.32,
      "grad_norm": 0.0007237320727261912,
      "learning_rate": 3.3600000000000004e-06,
      "loss": 0.0,
      "step": 26000
    },
    {
      "epoch": 8.336,
      "grad_norm": 0.000516590777468066,
      "learning_rate": 3.328e-06,
      "loss": 0.0,
      "step": 26050
    },
    {
      "epoch": 8.352,
      "grad_norm": 0.0005010114184220883,
      "learning_rate": 3.2960000000000003e-06,
      "loss": 0.0,
      "step": 26100
    },
    {
      "epoch": 8.368,
      "grad_norm": 5.28964766080006e-06,
      "learning_rate": 3.2640000000000004e-06,
      "loss": 0.0,
      "step": 26150
    },
    {
      "epoch": 8.384,
      "grad_norm": 0.0005157406666104107,
      "learning_rate": 3.232e-06,
      "loss": 0.0,
      "step": 26200
    },
    {
      "epoch": 8.4,
      "grad_norm": 0.0006507550123747543,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.0,
      "step": 26250
    },
    {
      "epoch": 8.416,
      "grad_norm": 0.0007155645493115397,
      "learning_rate": 3.1680000000000004e-06,
      "loss": 0.0,
      "step": 26300
    },
    {
      "epoch": 8.432,
      "grad_norm": 0.00035348868456337083,
      "learning_rate": 3.136e-06,
      "loss": 0.0,
      "step": 26350
    },
    {
      "epoch": 8.448,
      "grad_norm": 0.0005827653476283665,
      "learning_rate": 3.1040000000000003e-06,
      "loss": 0.0,
      "step": 26400
    },
    {
      "epoch": 8.464,
      "grad_norm": 5.392560798091041e-05,
      "learning_rate": 3.072e-06,
      "loss": 0.0,
      "step": 26450
    },
    {
      "epoch": 8.48,
      "grad_norm": 0.0003822684090234525,
      "learning_rate": 3.04e-06,
      "loss": 0.0,
      "step": 26500
    },
    {
      "epoch": 8.496,
      "grad_norm": 0.00047049084334724653,
      "learning_rate": 3.0080000000000003e-06,
      "loss": 0.0,
      "step": 26550
    },
    {
      "epoch": 8.512,
      "grad_norm": 0.0004919742737354103,
      "learning_rate": 2.976e-06,
      "loss": 0.0,
      "step": 26600
    },
    {
      "epoch": 8.528,
      "grad_norm": 0.0002140207608021743,
      "learning_rate": 2.944e-06,
      "loss": 0.0,
      "step": 26650
    },
    {
      "epoch": 8.544,
      "grad_norm": 3.5859741698396775e-05,
      "learning_rate": 2.9120000000000002e-06,
      "loss": 0.0,
      "step": 26700
    },
    {
      "epoch": 8.56,
      "grad_norm": 2.8239349881814044e-05,
      "learning_rate": 2.88e-06,
      "loss": 0.0,
      "step": 26750
    },
    {
      "epoch": 8.576,
      "grad_norm": 3.6122702859208877e-06,
      "learning_rate": 2.848e-06,
      "loss": 0.0,
      "step": 26800
    },
    {
      "epoch": 8.592,
      "grad_norm": 0.0026791316330236677,
      "learning_rate": 2.8160000000000002e-06,
      "loss": 0.0,
      "step": 26850
    },
    {
      "epoch": 8.608,
      "grad_norm": 0.00020946549940804708,
      "learning_rate": 2.784e-06,
      "loss": 0.0,
      "step": 26900
    },
    {
      "epoch": 8.624,
      "grad_norm": 0.0003340978536385738,
      "learning_rate": 2.752e-06,
      "loss": 0.0,
      "step": 26950
    },
    {
      "epoch": 8.64,
      "grad_norm": 0.0006912123574857489,
      "learning_rate": 2.7200000000000002e-06,
      "loss": 0.0,
      "step": 27000
    },
    {
      "epoch": 8.656,
      "grad_norm": 1.1918299463922867e-05,
      "learning_rate": 2.688e-06,
      "loss": 0.0,
      "step": 27050
    },
    {
      "epoch": 8.672,
      "grad_norm": 0.0005004993104376737,
      "learning_rate": 2.656e-06,
      "loss": 0.0,
      "step": 27100
    },
    {
      "epoch": 8.688,
      "grad_norm": 0.0006647739241711589,
      "learning_rate": 2.6240000000000006e-06,
      "loss": 0.0,
      "step": 27150
    },
    {
      "epoch": 8.704,
      "grad_norm": 0.0005349888973660498,
      "learning_rate": 2.592e-06,
      "loss": 0.0,
      "step": 27200
    },
    {
      "epoch": 8.72,
      "grad_norm": 0.0001071714356645616,
      "learning_rate": 2.56e-06,
      "loss": 0.0,
      "step": 27250
    },
    {
      "epoch": 8.736,
      "grad_norm": 0.0004991064010764234,
      "learning_rate": 2.5280000000000006e-06,
      "loss": 0.0,
      "step": 27300
    },
    {
      "epoch": 8.752,
      "grad_norm": 0.00022779807704196374,
      "learning_rate": 2.496e-06,
      "loss": 0.0,
      "step": 27350
    },
    {
      "epoch": 8.768,
      "grad_norm": 0.0012819212872404967,
      "learning_rate": 2.4640000000000005e-06,
      "loss": 0.0,
      "step": 27400
    },
    {
      "epoch": 8.784,
      "grad_norm": 2.4458772638304895e-06,
      "learning_rate": 2.432e-06,
      "loss": 0.0,
      "step": 27450
    },
    {
      "epoch": 8.8,
      "grad_norm": 4.978203463715246e-05,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.0,
      "step": 27500
    },
    {
      "epoch": 8.816,
      "grad_norm": 0.0006102147175976708,
      "learning_rate": 2.3680000000000005e-06,
      "loss": 0.0,
      "step": 27550
    },
    {
      "epoch": 8.832,
      "grad_norm": 0.00054039008837191,
      "learning_rate": 2.336e-06,
      "loss": 0.0,
      "step": 27600
    },
    {
      "epoch": 8.848,
      "grad_norm": 0.00044893481288572814,
      "learning_rate": 2.3040000000000003e-06,
      "loss": 0.0,
      "step": 27650
    },
    {
      "epoch": 8.864,
      "grad_norm": 0.00046239574733305844,
      "learning_rate": 2.2720000000000004e-06,
      "loss": 0.0,
      "step": 27700
    },
    {
      "epoch": 8.88,
      "grad_norm": 0.00037309598762979125,
      "learning_rate": 2.24e-06,
      "loss": 0.0,
      "step": 27750
    },
    {
      "epoch": 8.896,
      "grad_norm": 0.0004490570977280375,
      "learning_rate": 2.2080000000000003e-06,
      "loss": 0.0,
      "step": 27800
    },
    {
      "epoch": 8.912,
      "grad_norm": 5.971637937791403e-06,
      "learning_rate": 2.176e-06,
      "loss": 0.0,
      "step": 27850
    },
    {
      "epoch": 8.928,
      "grad_norm": 0.00019607249904638697,
      "learning_rate": 2.144e-06,
      "loss": 0.0,
      "step": 27900
    },
    {
      "epoch": 8.943999999999999,
      "grad_norm": 2.173728894982944e-05,
      "learning_rate": 2.1120000000000003e-06,
      "loss": 0.0,
      "step": 27950
    },
    {
      "epoch": 8.96,
      "grad_norm": 0.0005814522831536409,
      "learning_rate": 2.08e-06,
      "loss": 0.0,
      "step": 28000
    },
    {
      "epoch": 8.975999999999999,
      "grad_norm": 1.1979971851172046e-05,
      "learning_rate": 2.048e-06,
      "loss": 0.0,
      "step": 28050
    },
    {
      "epoch": 8.992,
      "grad_norm": 0.001145707081894591,
      "learning_rate": 2.0160000000000003e-06,
      "loss": 0.0,
      "step": 28100
    },
    {
      "epoch": 9.008,
      "grad_norm": 0.00015980546781926578,
      "learning_rate": 1.984e-06,
      "loss": 0.0,
      "step": 28150
    },
    {
      "epoch": 9.024,
      "grad_norm": 0.00038132672038477974,
      "learning_rate": 1.952e-06,
      "loss": 0.0,
      "step": 28200
    },
    {
      "epoch": 9.04,
      "grad_norm": 0.00029638386517636414,
      "learning_rate": 1.9200000000000003e-06,
      "loss": 0.0,
      "step": 28250
    },
    {
      "epoch": 9.056,
      "grad_norm": 9.84668351885671e-05,
      "learning_rate": 1.8880000000000002e-06,
      "loss": 0.0,
      "step": 28300
    },
    {
      "epoch": 9.072,
      "grad_norm": 0.0006899543451120302,
      "learning_rate": 1.856e-06,
      "loss": 0.0,
      "step": 28350
    },
    {
      "epoch": 9.088,
      "grad_norm": 0.00023321439846025538,
      "learning_rate": 1.8240000000000002e-06,
      "loss": 0.0,
      "step": 28400
    },
    {
      "epoch": 9.104,
      "grad_norm": 1.2255045251781342e-06,
      "learning_rate": 1.7920000000000002e-06,
      "loss": 0.0,
      "step": 28450
    },
    {
      "epoch": 9.12,
      "grad_norm": 0.000224750950400635,
      "learning_rate": 1.76e-06,
      "loss": 0.0,
      "step": 28500
    },
    {
      "epoch": 9.136,
      "grad_norm": 1.504019715073742e-06,
      "learning_rate": 1.7280000000000002e-06,
      "loss": 0.0,
      "step": 28550
    },
    {
      "epoch": 9.152,
      "grad_norm": 0.0004469699493177947,
      "learning_rate": 1.6960000000000002e-06,
      "loss": 0.0,
      "step": 28600
    },
    {
      "epoch": 9.168,
      "grad_norm": 0.0008992359796643962,
      "learning_rate": 1.664e-06,
      "loss": 0.0,
      "step": 28650
    },
    {
      "epoch": 9.184,
      "grad_norm": 1.1807481020292317e-06,
      "learning_rate": 1.6320000000000002e-06,
      "loss": 0.0,
      "step": 28700
    },
    {
      "epoch": 9.2,
      "grad_norm": 0.0006720244284168616,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.0,
      "step": 28750
    },
    {
      "epoch": 9.216,
      "grad_norm": 1.954402796502112e-06,
      "learning_rate": 1.568e-06,
      "loss": 0.0,
      "step": 28800
    },
    {
      "epoch": 9.232,
      "grad_norm": 0.00045854257866828515,
      "learning_rate": 1.536e-06,
      "loss": 0.0,
      "step": 28850
    },
    {
      "epoch": 9.248,
      "grad_norm": 1.2499022623047775e-06,
      "learning_rate": 1.5040000000000001e-06,
      "loss": 0.0,
      "step": 28900
    },
    {
      "epoch": 9.264,
      "grad_norm": 3.5069474526942418e-06,
      "learning_rate": 1.472e-06,
      "loss": 0.0,
      "step": 28950
    },
    {
      "epoch": 9.28,
      "grad_norm": 0.0004939352589701349,
      "learning_rate": 1.44e-06,
      "loss": 0.0,
      "step": 29000
    },
    {
      "epoch": 9.296,
      "grad_norm": 0.0006403164759103728,
      "learning_rate": 1.4080000000000001e-06,
      "loss": 0.0,
      "step": 29050
    },
    {
      "epoch": 9.312,
      "grad_norm": 0.0004906591431802971,
      "learning_rate": 1.376e-06,
      "loss": 0.0,
      "step": 29100
    },
    {
      "epoch": 9.328,
      "grad_norm": 0.0006355382829934992,
      "learning_rate": 1.344e-06,
      "loss": 0.0,
      "step": 29150
    },
    {
      "epoch": 9.344,
      "grad_norm": 0.0008888097102175601,
      "learning_rate": 1.3120000000000003e-06,
      "loss": 0.0,
      "step": 29200
    },
    {
      "epoch": 9.36,
      "grad_norm": 2.2844333657135604e-06,
      "learning_rate": 1.28e-06,
      "loss": 0.0,
      "step": 29250
    },
    {
      "epoch": 9.376,
      "grad_norm": 0.0007355380038243548,
      "learning_rate": 1.248e-06,
      "loss": 0.0,
      "step": 29300
    },
    {
      "epoch": 9.392,
      "grad_norm": 0.00016113117251488604,
      "learning_rate": 1.216e-06,
      "loss": 0.0,
      "step": 29350
    },
    {
      "epoch": 9.408,
      "grad_norm": 6.144866567676681e-05,
      "learning_rate": 1.1840000000000002e-06,
      "loss": 0.0,
      "step": 29400
    },
    {
      "epoch": 9.424,
      "grad_norm": 1.280262092788024e-06,
      "learning_rate": 1.1520000000000002e-06,
      "loss": 0.0,
      "step": 29450
    },
    {
      "epoch": 9.44,
      "grad_norm": 0.00023403691930978503,
      "learning_rate": 1.12e-06,
      "loss": 0.0,
      "step": 29500
    },
    {
      "epoch": 9.456,
      "grad_norm": 1.0651580903925167e-05,
      "learning_rate": 1.088e-06,
      "loss": 0.0,
      "step": 29550
    },
    {
      "epoch": 9.472,
      "grad_norm": 0.0004401165605022274,
      "learning_rate": 1.0560000000000001e-06,
      "loss": 0.0,
      "step": 29600
    },
    {
      "epoch": 9.488,
      "grad_norm": 0.0007374198247804132,
      "learning_rate": 1.024e-06,
      "loss": 0.0,
      "step": 29650
    },
    {
      "epoch": 9.504,
      "grad_norm": 0.00033734901422662343,
      "learning_rate": 9.92e-07,
      "loss": 0.0,
      "step": 29700
    },
    {
      "epoch": 9.52,
      "grad_norm": 0.0009391331921544294,
      "learning_rate": 9.600000000000001e-07,
      "loss": 0.0,
      "step": 29750
    },
    {
      "epoch": 9.536,
      "grad_norm": 0.0012792312844443715,
      "learning_rate": 9.28e-07,
      "loss": 0.0,
      "step": 29800
    },
    {
      "epoch": 9.552,
      "grad_norm": 0.000600425656969412,
      "learning_rate": 8.960000000000001e-07,
      "loss": 0.0,
      "step": 29850
    },
    {
      "epoch": 9.568,
      "grad_norm": 0.0005356183437125068,
      "learning_rate": 8.640000000000001e-07,
      "loss": 0.0,
      "step": 29900
    },
    {
      "epoch": 9.584,
      "grad_norm": 1.169450332702385e-06,
      "learning_rate": 8.32e-07,
      "loss": 0.0,
      "step": 29950
    },
    {
      "epoch": 9.6,
      "grad_norm": 1.603757236150262e-06,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.0,
      "step": 30000
    },
    {
      "epoch": 9.616,
      "grad_norm": 1.3789867557409609e-06,
      "learning_rate": 7.68e-07,
      "loss": 0.0,
      "step": 30050
    },
    {
      "epoch": 9.632,
      "grad_norm": 0.00022794675409267614,
      "learning_rate": 7.36e-07,
      "loss": 0.0,
      "step": 30100
    },
    {
      "epoch": 9.648,
      "grad_norm": 4.063516211647283e-06,
      "learning_rate": 7.040000000000001e-07,
      "loss": 0.0,
      "step": 30150
    },
    {
      "epoch": 9.664,
      "grad_norm": 0.00020120298713090787,
      "learning_rate": 6.72e-07,
      "loss": 0.0,
      "step": 30200
    },
    {
      "epoch": 9.68,
      "grad_norm": 0.0009682823219180043,
      "learning_rate": 6.4e-07,
      "loss": 0.0,
      "step": 30250
    },
    {
      "epoch": 9.696,
      "grad_norm": 1.653173775427169e-06,
      "learning_rate": 6.08e-07,
      "loss": 0.0,
      "step": 30300
    },
    {
      "epoch": 9.712,
      "grad_norm": 0.0010948577216183535,
      "learning_rate": 5.760000000000001e-07,
      "loss": 0.0,
      "step": 30350
    },
    {
      "epoch": 9.728,
      "grad_norm": 1.579158744151985e-06,
      "learning_rate": 5.44e-07,
      "loss": 0.0,
      "step": 30400
    },
    {
      "epoch": 9.744,
      "grad_norm": 0.0005095525509816707,
      "learning_rate": 5.12e-07,
      "loss": 0.0,
      "step": 30450
    },
    {
      "epoch": 9.76,
      "grad_norm": 5.6971938558435215e-05,
      "learning_rate": 4.800000000000001e-07,
      "loss": 0.0,
      "step": 30500
    },
    {
      "epoch": 9.776,
      "grad_norm": 0.0003876663750785411,
      "learning_rate": 4.4800000000000004e-07,
      "loss": 0.0,
      "step": 30550
    },
    {
      "epoch": 9.792,
      "grad_norm": 0.0004980878635902623,
      "learning_rate": 4.16e-07,
      "loss": 0.0,
      "step": 30600
    },
    {
      "epoch": 9.808,
      "grad_norm": 1.1188634027454468e-06,
      "learning_rate": 3.84e-07,
      "loss": 0.0,
      "step": 30650
    },
    {
      "epoch": 9.824,
      "grad_norm": 0.000647302404932787,
      "learning_rate": 3.5200000000000003e-07,
      "loss": 0.0,
      "step": 30700
    },
    {
      "epoch": 9.84,
      "grad_norm": 0.00032616120339746537,
      "learning_rate": 3.2e-07,
      "loss": 0.0,
      "step": 30750
    },
    {
      "epoch": 9.856,
      "grad_norm": 0.0008723718200811839,
      "learning_rate": 2.8800000000000004e-07,
      "loss": 0.0,
      "step": 30800
    },
    {
      "epoch": 9.872,
      "grad_norm": 0.000471858358002294,
      "learning_rate": 2.56e-07,
      "loss": 0.0,
      "step": 30850
    },
    {
      "epoch": 9.888,
      "grad_norm": 1.3544616120540576e-06,
      "learning_rate": 2.2400000000000002e-07,
      "loss": 0.0,
      "step": 30900
    },
    {
      "epoch": 9.904,
      "grad_norm": 8.617191444341816e-05,
      "learning_rate": 1.92e-07,
      "loss": 0.0,
      "step": 30950
    },
    {
      "epoch": 9.92,
      "grad_norm": 0.0004407695022914176,
      "learning_rate": 1.6e-07,
      "loss": 0.0,
      "step": 31000
    },
    {
      "epoch": 9.936,
      "grad_norm": 0.00020840921158746508,
      "learning_rate": 1.28e-07,
      "loss": 0.0,
      "step": 31050
    },
    {
      "epoch": 9.952,
      "grad_norm": 0.000421999513896553,
      "learning_rate": 9.6e-08,
      "loss": 0.0,
      "step": 31100
    },
    {
      "epoch": 9.968,
      "grad_norm": 1.128787548845069e-06,
      "learning_rate": 6.4e-08,
      "loss": 0.0,
      "step": 31150
    },
    {
      "epoch": 9.984,
      "grad_norm": 7.0422523529939706e-06,
      "learning_rate": 3.2e-08,
      "loss": 0.0,
      "step": 31200
    },
    {
      "epoch": 10.0,
      "grad_norm": 1.438006921888056e-06,
      "learning_rate": 0.0,
      "loss": 0.0,
      "step": 31250
    }
  ],
  "logging_steps": 50,
  "max_steps": 31250,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 10,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8.285894220342886e+16,
  "train_batch_size": 6,
  "trial_name": null,
  "trial_params": null
}