{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 16.3265306122449,
  "eval_steps": 500,
  "global_step": 4000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.04081632653061224,
      "grad_norm": 7.114395618438721,
      "learning_rate": 9.981632653061225e-06,
      "loss": 0.7362,
      "step": 10
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 11.572301864624023,
      "learning_rate": 9.961224489795919e-06,
      "loss": 0.8729,
      "step": 20
    },
    {
      "epoch": 0.12244897959183673,
      "grad_norm": 9.383491516113281,
      "learning_rate": 9.940816326530614e-06,
      "loss": 0.773,
      "step": 30
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 7.83120059967041,
      "learning_rate": 9.920408163265307e-06,
      "loss": 0.7817,
      "step": 40
    },
    {
      "epoch": 0.20408163265306123,
      "grad_norm": 10.92087173461914,
      "learning_rate": 9.9e-06,
      "loss": 0.6256,
      "step": 50
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 3.8826725482940674,
      "learning_rate": 9.879591836734695e-06,
      "loss": 0.5759,
      "step": 60
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 11.15483283996582,
      "learning_rate": 9.859183673469388e-06,
      "loss": 0.7333,
      "step": 70
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 11.470726013183594,
      "learning_rate": 9.838775510204083e-06,
      "loss": 0.5943,
      "step": 80
    },
    {
      "epoch": 0.3673469387755102,
      "grad_norm": 13.159674644470215,
      "learning_rate": 9.818367346938777e-06,
      "loss": 0.7804,
      "step": 90
    },
    {
      "epoch": 0.40816326530612246,
      "grad_norm": 9.58558464050293,
      "learning_rate": 9.79795918367347e-06,
      "loss": 0.6491,
      "step": 100
    },
    {
      "epoch": 0.4489795918367347,
      "grad_norm": 9.653897285461426,
      "learning_rate": 9.777551020408163e-06,
      "loss": 0.5919,
      "step": 110
    },
    {
      "epoch": 0.4897959183673469,
      "grad_norm": 8.117432594299316,
      "learning_rate": 9.757142857142858e-06,
      "loss": 0.4571,
      "step": 120
    },
    {
      "epoch": 0.5306122448979592,
      "grad_norm": 6.9328460693359375,
      "learning_rate": 9.736734693877551e-06,
      "loss": 0.6597,
      "step": 130
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 7.962501049041748,
      "learning_rate": 9.716326530612246e-06,
      "loss": 0.5132,
      "step": 140
    },
    {
      "epoch": 0.6122448979591837,
      "grad_norm": 10.508763313293457,
      "learning_rate": 9.69591836734694e-06,
      "loss": 0.6893,
      "step": 150
    },
    {
      "epoch": 0.6530612244897959,
      "grad_norm": 7.637253761291504,
      "learning_rate": 9.675510204081635e-06,
      "loss": 0.6142,
      "step": 160
    },
    {
      "epoch": 0.6938775510204082,
      "grad_norm": 10.0332670211792,
      "learning_rate": 9.655102040816328e-06,
      "loss": 0.582,
      "step": 170
    },
    {
      "epoch": 0.7346938775510204,
      "grad_norm": 8.150875091552734,
      "learning_rate": 9.634693877551021e-06,
      "loss": 0.477,
      "step": 180
    },
    {
      "epoch": 0.7755102040816326,
      "grad_norm": 10.330913543701172,
      "learning_rate": 9.614285714285714e-06,
      "loss": 0.5916,
      "step": 190
    },
    {
      "epoch": 0.8163265306122449,
      "grad_norm": 11.654999732971191,
      "learning_rate": 9.593877551020408e-06,
      "loss": 0.6236,
      "step": 200
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 8.048078536987305,
      "learning_rate": 9.573469387755103e-06,
      "loss": 0.6142,
      "step": 210
    },
    {
      "epoch": 0.8979591836734694,
      "grad_norm": 9.869592666625977,
      "learning_rate": 9.553061224489798e-06,
      "loss": 0.625,
      "step": 220
    },
    {
      "epoch": 0.9387755102040817,
      "grad_norm": 8.321409225463867,
      "learning_rate": 9.532653061224491e-06,
      "loss": 0.5767,
      "step": 230
    },
    {
      "epoch": 0.9795918367346939,
      "grad_norm": 7.6769256591796875,
      "learning_rate": 9.512244897959184e-06,
      "loss": 0.5134,
      "step": 240
    },
    {
      "epoch": 1.0204081632653061,
      "grad_norm": 9.609123229980469,
      "learning_rate": 9.491836734693877e-06,
      "loss": 0.5868,
      "step": 250
    },
    {
      "epoch": 1.0612244897959184,
      "grad_norm": 9.19683837890625,
      "learning_rate": 9.471428571428572e-06,
      "loss": 0.5215,
      "step": 260
    },
    {
      "epoch": 1.1020408163265305,
      "grad_norm": 7.328164577484131,
      "learning_rate": 9.451020408163266e-06,
      "loss": 0.5422,
      "step": 270
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 6.913904190063477,
      "learning_rate": 9.430612244897959e-06,
      "loss": 0.5214,
      "step": 280
    },
    {
      "epoch": 1.183673469387755,
      "grad_norm": 9.28811264038086,
      "learning_rate": 9.410204081632654e-06,
      "loss": 0.5319,
      "step": 290
    },
    {
      "epoch": 1.2244897959183674,
      "grad_norm": 9.132966041564941,
      "learning_rate": 9.389795918367349e-06,
      "loss": 0.5581,
      "step": 300
    },
    {
      "epoch": 1.2653061224489797,
      "grad_norm": 6.9722065925598145,
      "learning_rate": 9.369387755102042e-06,
      "loss": 0.431,
      "step": 310
    },
    {
      "epoch": 1.306122448979592,
      "grad_norm": 5.06177282333374,
      "learning_rate": 9.348979591836736e-06,
      "loss": 0.4583,
      "step": 320
    },
    {
      "epoch": 1.346938775510204,
      "grad_norm": 7.732840538024902,
      "learning_rate": 9.328571428571429e-06,
      "loss": 0.4194,
      "step": 330
    },
    {
      "epoch": 1.3877551020408163,
      "grad_norm": 8.94101333618164,
      "learning_rate": 9.308163265306122e-06,
      "loss": 0.4519,
      "step": 340
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 7.5437750816345215,
      "learning_rate": 9.287755102040817e-06,
      "loss": 0.5095,
      "step": 350
    },
    {
      "epoch": 1.469387755102041,
      "grad_norm": 5.702700138092041,
      "learning_rate": 9.26734693877551e-06,
      "loss": 0.3936,
      "step": 360
    },
    {
      "epoch": 1.510204081632653,
      "grad_norm": 9.153871536254883,
      "learning_rate": 9.246938775510205e-06,
      "loss": 0.4566,
      "step": 370
    },
    {
      "epoch": 1.5510204081632653,
      "grad_norm": 13.249794006347656,
      "learning_rate": 9.226530612244899e-06,
      "loss": 0.5216,
      "step": 380
    },
    {
      "epoch": 1.5918367346938775,
      "grad_norm": 7.065913200378418,
      "learning_rate": 9.206122448979594e-06,
      "loss": 0.4562,
      "step": 390
    },
    {
      "epoch": 1.6326530612244898,
      "grad_norm": 7.559301853179932,
      "learning_rate": 9.185714285714287e-06,
      "loss": 0.3883,
      "step": 400
    },
    {
      "epoch": 1.6734693877551021,
      "grad_norm": 12.103629112243652,
      "learning_rate": 9.16530612244898e-06,
      "loss": 0.4149,
      "step": 410
    },
    {
      "epoch": 1.7142857142857144,
      "grad_norm": 7.9720072746276855,
      "learning_rate": 9.144897959183673e-06,
      "loss": 0.4718,
      "step": 420
    },
    {
      "epoch": 1.7551020408163265,
      "grad_norm": 4.845782279968262,
      "learning_rate": 9.124489795918368e-06,
      "loss": 0.4304,
      "step": 430
    },
    {
      "epoch": 1.7959183673469388,
      "grad_norm": 6.954368591308594,
      "learning_rate": 9.104081632653062e-06,
      "loss": 0.3436,
      "step": 440
    },
    {
      "epoch": 1.836734693877551,
      "grad_norm": 4.751299858093262,
      "learning_rate": 9.083673469387757e-06,
      "loss": 0.4366,
      "step": 450
    },
    {
      "epoch": 1.8775510204081631,
      "grad_norm": 6.507364273071289,
      "learning_rate": 9.06326530612245e-06,
      "loss": 0.5794,
      "step": 460
    },
    {
      "epoch": 1.9183673469387754,
      "grad_norm": 8.891802787780762,
      "learning_rate": 9.042857142857143e-06,
      "loss": 0.4616,
      "step": 470
    },
    {
      "epoch": 1.9591836734693877,
      "grad_norm": 10.056327819824219,
      "learning_rate": 9.022448979591838e-06,
      "loss": 0.4946,
      "step": 480
    },
    {
      "epoch": 2.0,
      "grad_norm": 7.899660110473633,
      "learning_rate": 9.002040816326531e-06,
      "loss": 0.4437,
      "step": 490
    },
    {
      "epoch": 2.0408163265306123,
      "grad_norm": 6.761326313018799,
      "learning_rate": 8.981632653061225e-06,
      "loss": 0.4303,
      "step": 500
    },
    {
      "epoch": 2.0816326530612246,
      "grad_norm": 8.639615058898926,
      "learning_rate": 8.96122448979592e-06,
      "loss": 0.3267,
      "step": 510
    },
    {
      "epoch": 2.122448979591837,
      "grad_norm": 7.710758209228516,
      "learning_rate": 8.940816326530613e-06,
      "loss": 0.3559,
      "step": 520
    },
    {
      "epoch": 2.163265306122449,
      "grad_norm": 6.812905311584473,
      "learning_rate": 8.920408163265308e-06,
      "loss": 0.4761,
      "step": 530
    },
    {
      "epoch": 2.204081632653061,
      "grad_norm": 7.2431511878967285,
      "learning_rate": 8.900000000000001e-06,
      "loss": 0.405,
      "step": 540
    },
    {
      "epoch": 2.2448979591836733,
      "grad_norm": 7.230724811553955,
      "learning_rate": 8.879591836734694e-06,
      "loss": 0.3638,
      "step": 550
    },
    {
      "epoch": 2.2857142857142856,
      "grad_norm": 9.520208358764648,
      "learning_rate": 8.859183673469388e-06,
      "loss": 0.3473,
      "step": 560
    },
    {
      "epoch": 2.326530612244898,
      "grad_norm": 7.048585414886475,
      "learning_rate": 8.838775510204083e-06,
      "loss": 0.3652,
      "step": 570
    },
    {
      "epoch": 2.36734693877551,
      "grad_norm": 6.979404449462891,
      "learning_rate": 8.818367346938776e-06,
      "loss": 0.3855,
      "step": 580
    },
    {
      "epoch": 2.4081632653061225,
      "grad_norm": 3.765305280685425,
      "learning_rate": 8.797959183673471e-06,
      "loss": 0.3452,
      "step": 590
    },
    {
      "epoch": 2.4489795918367347,
      "grad_norm": 10.533697128295898,
      "learning_rate": 8.777551020408164e-06,
      "loss": 0.3874,
      "step": 600
    },
    {
      "epoch": 2.489795918367347,
      "grad_norm": 8.108145713806152,
      "learning_rate": 8.757142857142858e-06,
      "loss": 0.3695,
      "step": 610
    },
    {
      "epoch": 2.5306122448979593,
      "grad_norm": 7.947360992431641,
      "learning_rate": 8.736734693877552e-06,
      "loss": 0.408,
      "step": 620
    },
    {
      "epoch": 2.571428571428571,
      "grad_norm": 7.8081374168396,
      "learning_rate": 8.716326530612246e-06,
      "loss": 0.4059,
      "step": 630
    },
    {
      "epoch": 2.612244897959184,
      "grad_norm": 8.579155921936035,
      "learning_rate": 8.695918367346939e-06,
      "loss": 0.3934,
      "step": 640
    },
    {
      "epoch": 2.6530612244897958,
      "grad_norm": 6.4387712478637695,
      "learning_rate": 8.675510204081632e-06,
      "loss": 0.4256,
      "step": 650
    },
    {
      "epoch": 2.693877551020408,
      "grad_norm": 8.415692329406738,
      "learning_rate": 8.655102040816327e-06,
      "loss": 0.3453,
      "step": 660
    },
    {
      "epoch": 2.7346938775510203,
      "grad_norm": 8.50904369354248,
      "learning_rate": 8.63469387755102e-06,
      "loss": 0.4766,
      "step": 670
    },
    {
      "epoch": 2.7755102040816326,
      "grad_norm": 4.662519931793213,
      "learning_rate": 8.614285714285716e-06,
      "loss": 0.3973,
      "step": 680
    },
    {
      "epoch": 2.816326530612245,
      "grad_norm": 6.288435935974121,
      "learning_rate": 8.593877551020409e-06,
      "loss": 0.4408,
      "step": 690
    },
    {
      "epoch": 2.857142857142857,
      "grad_norm": 6.625838279724121,
      "learning_rate": 8.573469387755102e-06,
      "loss": 0.2908,
      "step": 700
    },
    {
      "epoch": 2.8979591836734695,
      "grad_norm": 8.510032653808594,
      "learning_rate": 8.553061224489797e-06,
      "loss": 0.3813,
      "step": 710
    },
    {
      "epoch": 2.938775510204082,
      "grad_norm": 11.82463264465332,
      "learning_rate": 8.53265306122449e-06,
      "loss": 0.4352,
      "step": 720
    },
    {
      "epoch": 2.979591836734694,
      "grad_norm": 8.821819305419922,
      "learning_rate": 8.512244897959184e-06,
      "loss": 0.4318,
      "step": 730
    },
    {
      "epoch": 3.020408163265306,
      "grad_norm": 8.010713577270508,
      "learning_rate": 8.491836734693879e-06,
      "loss": 0.2323,
      "step": 740
    },
    {
      "epoch": 3.061224489795918,
      "grad_norm": 9.03991985321045,
      "learning_rate": 8.471428571428572e-06,
      "loss": 0.3603,
      "step": 750
    },
    {
      "epoch": 3.1020408163265305,
      "grad_norm": 10.94204044342041,
      "learning_rate": 8.451020408163267e-06,
      "loss": 0.3576,
      "step": 760
    },
    {
      "epoch": 3.142857142857143,
      "grad_norm": 7.89410924911499,
      "learning_rate": 8.43061224489796e-06,
      "loss": 0.2851,
      "step": 770
    },
    {
      "epoch": 3.183673469387755,
      "grad_norm": 6.53656005859375,
      "learning_rate": 8.410204081632653e-06,
      "loss": 0.318,
      "step": 780
    },
    {
      "epoch": 3.2244897959183674,
      "grad_norm": 6.487284183502197,
      "learning_rate": 8.389795918367347e-06,
      "loss": 0.317,
      "step": 790
    },
    {
      "epoch": 3.2653061224489797,
      "grad_norm": 6.947931289672852,
      "learning_rate": 8.369387755102042e-06,
      "loss": 0.2879,
      "step": 800
    },
    {
      "epoch": 3.306122448979592,
      "grad_norm": 4.166048526763916,
      "learning_rate": 8.348979591836735e-06,
      "loss": 0.3392,
      "step": 810
    },
    {
      "epoch": 3.3469387755102042,
      "grad_norm": 9.974846839904785,
      "learning_rate": 8.32857142857143e-06,
      "loss": 0.3663,
      "step": 820
    },
    {
      "epoch": 3.387755102040816,
      "grad_norm": 9.668428421020508,
      "learning_rate": 8.308163265306123e-06,
      "loss": 0.3212,
      "step": 830
    },
    {
      "epoch": 3.4285714285714284,
      "grad_norm": 11.81507396697998,
      "learning_rate": 8.287755102040816e-06,
      "loss": 0.3241,
      "step": 840
    },
    {
      "epoch": 3.4693877551020407,
      "grad_norm": 13.690321922302246,
      "learning_rate": 8.267346938775511e-06,
      "loss": 0.4535,
      "step": 850
    },
    {
      "epoch": 3.510204081632653,
      "grad_norm": 11.042778968811035,
      "learning_rate": 8.246938775510205e-06,
      "loss": 0.3826,
      "step": 860
    },
    {
      "epoch": 3.5510204081632653,
      "grad_norm": 8.57719612121582,
      "learning_rate": 8.226530612244898e-06,
      "loss": 0.3905,
      "step": 870
    },
    {
      "epoch": 3.5918367346938775,
      "grad_norm": 7.843425750732422,
      "learning_rate": 8.206122448979591e-06,
      "loss": 0.3125,
      "step": 880
    },
    {
      "epoch": 3.63265306122449,
      "grad_norm": 5.9236931800842285,
      "learning_rate": 8.185714285714286e-06,
      "loss": 0.3512,
      "step": 890
    },
    {
      "epoch": 3.673469387755102,
      "grad_norm": 8.213603973388672,
      "learning_rate": 8.165306122448981e-06,
      "loss": 0.4094,
      "step": 900
    },
    {
      "epoch": 3.7142857142857144,
      "grad_norm": 3.8083949089050293,
      "learning_rate": 8.144897959183674e-06,
      "loss": 0.2751,
      "step": 910
    },
    {
      "epoch": 3.7551020408163263,
      "grad_norm": 12.339240074157715,
      "learning_rate": 8.124489795918368e-06,
      "loss": 0.3296,
      "step": 920
    },
    {
      "epoch": 3.795918367346939,
      "grad_norm": 9.532052040100098,
      "learning_rate": 8.104081632653061e-06,
      "loss": 0.3033,
      "step": 930
    },
    {
      "epoch": 3.836734693877551,
      "grad_norm": 6.307032108306885,
      "learning_rate": 8.083673469387756e-06,
      "loss": 0.3765,
      "step": 940
    },
    {
      "epoch": 3.877551020408163,
      "grad_norm": 7.3003010749816895,
      "learning_rate": 8.06326530612245e-06,
      "loss": 0.2161,
      "step": 950
    },
    {
      "epoch": 3.9183673469387754,
      "grad_norm": 7.6572699546813965,
      "learning_rate": 8.042857142857143e-06,
      "loss": 0.2886,
      "step": 960
    },
    {
      "epoch": 3.9591836734693877,
      "grad_norm": 6.745776176452637,
      "learning_rate": 8.022448979591838e-06,
      "loss": 0.3376,
      "step": 970
    },
    {
      "epoch": 4.0,
      "grad_norm": 10.482270240783691,
      "learning_rate": 8.002040816326533e-06,
      "loss": 0.2657,
      "step": 980
    },
    {
      "epoch": 4.040816326530612,
      "grad_norm": 6.213717460632324,
      "learning_rate": 7.981632653061226e-06,
      "loss": 0.2596,
      "step": 990
    },
    {
      "epoch": 4.081632653061225,
      "grad_norm": 10.256094932556152,
      "learning_rate": 7.961224489795919e-06,
      "loss": 0.2338,
      "step": 1000
    },
    {
      "epoch": 4.122448979591836,
      "grad_norm": 8.640167236328125,
      "learning_rate": 7.940816326530612e-06,
      "loss": 0.2936,
      "step": 1010
    },
    {
      "epoch": 4.163265306122449,
      "grad_norm": 6.750300407409668,
      "learning_rate": 7.920408163265306e-06,
      "loss": 0.3511,
      "step": 1020
    },
    {
      "epoch": 4.204081632653061,
      "grad_norm": 9.488007545471191,
      "learning_rate": 7.9e-06,
      "loss": 0.2913,
      "step": 1030
    },
    {
      "epoch": 4.244897959183674,
      "grad_norm": 4.2671003341674805,
      "learning_rate": 7.879591836734694e-06,
      "loss": 0.2768,
      "step": 1040
    },
    {
      "epoch": 4.285714285714286,
      "grad_norm": 5.375782489776611,
      "learning_rate": 7.859183673469389e-06,
      "loss": 0.274,
      "step": 1050
    },
    {
      "epoch": 4.326530612244898,
      "grad_norm": 10.316515922546387,
      "learning_rate": 7.838775510204082e-06,
      "loss": 0.2373,
      "step": 1060
    },
    {
      "epoch": 4.36734693877551,
      "grad_norm": 6.733712673187256,
      "learning_rate": 7.818367346938777e-06,
      "loss": 0.32,
      "step": 1070
    },
    {
      "epoch": 4.408163265306122,
      "grad_norm": 6.432683944702148,
      "learning_rate": 7.79795918367347e-06,
      "loss": 0.3149,
      "step": 1080
    },
    {
      "epoch": 4.448979591836735,
      "grad_norm": 8.520115852355957,
      "learning_rate": 7.777551020408164e-06,
      "loss": 0.3369,
      "step": 1090
    },
    {
      "epoch": 4.489795918367347,
      "grad_norm": 3.52677845954895,
      "learning_rate": 7.757142857142857e-06,
      "loss": 0.2584,
      "step": 1100
    },
    {
      "epoch": 4.530612244897959,
      "grad_norm": 9.642569541931152,
      "learning_rate": 7.736734693877552e-06,
      "loss": 0.3076,
      "step": 1110
    },
    {
      "epoch": 4.571428571428571,
      "grad_norm": 11.233070373535156,
      "learning_rate": 7.716326530612245e-06,
      "loss": 0.3263,
      "step": 1120
    },
    {
      "epoch": 4.612244897959184,
      "grad_norm": 7.919038772583008,
      "learning_rate": 7.69591836734694e-06,
      "loss": 0.3152,
      "step": 1130
    },
    {
      "epoch": 4.653061224489796,
      "grad_norm": 7.116144180297852,
      "learning_rate": 7.675510204081633e-06,
      "loss": 0.2809,
      "step": 1140
    },
    {
      "epoch": 4.6938775510204085,
      "grad_norm": 6.757585525512695,
      "learning_rate": 7.655102040816327e-06,
      "loss": 0.2642,
      "step": 1150
    },
    {
      "epoch": 4.73469387755102,
      "grad_norm": 9.032824516296387,
      "learning_rate": 7.634693877551022e-06,
      "loss": 0.3546,
      "step": 1160
    },
    {
      "epoch": 4.775510204081632,
      "grad_norm": 7.837385654449463,
      "learning_rate": 7.614285714285715e-06,
      "loss": 0.3196,
      "step": 1170
    },
    {
      "epoch": 4.816326530612245,
      "grad_norm": 5.3134846687316895,
      "learning_rate": 7.593877551020409e-06,
      "loss": 0.2068,
      "step": 1180
    },
    {
      "epoch": 4.857142857142857,
      "grad_norm": 8.036781311035156,
      "learning_rate": 7.573469387755102e-06,
      "loss": 0.24,
      "step": 1190
    },
    {
      "epoch": 4.8979591836734695,
      "grad_norm": 12.13339900970459,
      "learning_rate": 7.5530612244897965e-06,
      "loss": 0.2795,
      "step": 1200
    },
    {
      "epoch": 4.938775510204081,
      "grad_norm": 5.253749370574951,
      "learning_rate": 7.532653061224491e-06,
      "loss": 0.2667,
      "step": 1210
    },
    {
      "epoch": 4.979591836734694,
      "grad_norm": 7.921318531036377,
      "learning_rate": 7.512244897959185e-06,
      "loss": 0.3218,
      "step": 1220
    },
    {
      "epoch": 5.020408163265306,
      "grad_norm": 9.056379318237305,
      "learning_rate": 7.491836734693878e-06,
      "loss": 0.2324,
      "step": 1230
    },
    {
      "epoch": 5.061224489795919,
      "grad_norm": 5.0674920082092285,
      "learning_rate": 7.471428571428571e-06,
      "loss": 0.3063,
      "step": 1240
    },
    {
      "epoch": 5.1020408163265305,
      "grad_norm": 12.247352600097656,
      "learning_rate": 7.451020408163266e-06,
      "loss": 0.3469,
      "step": 1250
    },
    {
      "epoch": 5.142857142857143,
      "grad_norm": 7.278679847717285,
      "learning_rate": 7.43061224489796e-06,
      "loss": 0.2851,
      "step": 1260
    },
    {
      "epoch": 5.183673469387755,
      "grad_norm": 11.491589546203613,
      "learning_rate": 7.410204081632654e-06,
      "loss": 0.245,
      "step": 1270
    },
    {
      "epoch": 5.224489795918367,
      "grad_norm": 8.652173042297363,
      "learning_rate": 7.389795918367347e-06,
      "loss": 0.2328,
      "step": 1280
    },
    {
      "epoch": 5.26530612244898,
      "grad_norm": 7.953402519226074,
      "learning_rate": 7.369387755102041e-06,
      "loss": 0.2784,
      "step": 1290
    },
    {
      "epoch": 5.3061224489795915,
      "grad_norm": 6.98084831237793,
      "learning_rate": 7.348979591836736e-06,
      "loss": 0.2379,
      "step": 1300
    },
    {
      "epoch": 5.346938775510204,
      "grad_norm": 7.400093078613281,
      "learning_rate": 7.328571428571429e-06,
      "loss": 0.2057,
      "step": 1310
    },
    {
      "epoch": 5.387755102040816,
      "grad_norm": 4.131850719451904,
      "learning_rate": 7.308163265306123e-06,
      "loss": 0.2792,
      "step": 1320
    },
    {
      "epoch": 5.428571428571429,
      "grad_norm": 10.25373363494873,
      "learning_rate": 7.287755102040817e-06,
      "loss": 0.2642,
      "step": 1330
    },
    {
      "epoch": 5.469387755102041,
      "grad_norm": 5.801217079162598,
      "learning_rate": 7.267346938775512e-06,
      "loss": 0.2301,
      "step": 1340
    },
    {
      "epoch": 5.510204081632653,
      "grad_norm": 3.3520116806030273,
      "learning_rate": 7.246938775510205e-06,
      "loss": 0.2728,
      "step": 1350
    },
    {
      "epoch": 5.551020408163265,
      "grad_norm": 5.335054874420166,
      "learning_rate": 7.226530612244898e-06,
      "loss": 0.2812,
      "step": 1360
    },
    {
      "epoch": 5.591836734693878,
      "grad_norm": 4.713162899017334,
      "learning_rate": 7.206122448979592e-06,
      "loss": 0.221,
      "step": 1370
    },
    {
      "epoch": 5.63265306122449,
      "grad_norm": 8.641826629638672,
      "learning_rate": 7.185714285714286e-06,
      "loss": 0.2204,
      "step": 1380
    },
    {
      "epoch": 5.673469387755102,
      "grad_norm": 4.286067485809326,
      "learning_rate": 7.165306122448981e-06,
      "loss": 0.2446,
      "step": 1390
    },
    {
      "epoch": 5.714285714285714,
      "grad_norm": 7.071314811706543,
      "learning_rate": 7.144897959183674e-06,
      "loss": 0.1788,
      "step": 1400
    },
    {
      "epoch": 5.755102040816326,
      "grad_norm": 9.792560577392578,
      "learning_rate": 7.124489795918368e-06,
      "loss": 0.272,
      "step": 1410
    },
    {
      "epoch": 5.795918367346939,
      "grad_norm": 6.726722240447998,
      "learning_rate": 7.104081632653061e-06,
      "loss": 0.18,
      "step": 1420
    },
    {
      "epoch": 5.836734693877551,
      "grad_norm": 7.378534317016602,
      "learning_rate": 7.083673469387755e-06,
      "loss": 0.2308,
      "step": 1430
    },
    {
      "epoch": 5.877551020408164,
      "grad_norm": 10.169758796691895,
      "learning_rate": 7.0632653061224495e-06,
      "loss": 0.296,
      "step": 1440
    },
    {
      "epoch": 5.918367346938775,
      "grad_norm": 5.440324306488037,
      "learning_rate": 7.042857142857144e-06,
      "loss": 0.2725,
      "step": 1450
    },
    {
      "epoch": 5.959183673469388,
      "grad_norm": 8.175464630126953,
      "learning_rate": 7.022448979591837e-06,
      "loss": 0.2251,
      "step": 1460
    },
    {
      "epoch": 6.0,
      "grad_norm": 6.7653398513793945,
      "learning_rate": 7.002040816326531e-06,
      "loss": 0.3411,
      "step": 1470
    },
    {
      "epoch": 6.040816326530612,
      "grad_norm": 6.0152716636657715,
      "learning_rate": 6.981632653061225e-06,
      "loss": 0.2071,
      "step": 1480
    },
    {
      "epoch": 6.081632653061225,
      "grad_norm": 8.066520690917969,
      "learning_rate": 6.961224489795919e-06,
      "loss": 0.1639,
      "step": 1490
    },
    {
      "epoch": 6.122448979591836,
      "grad_norm": 4.027646541595459,
      "learning_rate": 6.940816326530613e-06,
      "loss": 0.2069,
      "step": 1500
    },
    {
      "epoch": 6.163265306122449,
      "grad_norm": 5.61140251159668,
      "learning_rate": 6.920408163265307e-06,
      "loss": 0.2413,
      "step": 1510
    },
    {
      "epoch": 6.204081632653061,
      "grad_norm": 9.809159278869629,
      "learning_rate": 6.9e-06,
      "loss": 0.26,
      "step": 1520
    },
    {
      "epoch": 6.244897959183674,
      "grad_norm": 6.755568504333496,
      "learning_rate": 6.879591836734695e-06,
      "loss": 0.1965,
      "step": 1530
    },
    {
      "epoch": 6.285714285714286,
      "grad_norm": 4.21774435043335,
      "learning_rate": 6.859183673469388e-06,
      "loss": 0.1959,
      "step": 1540
    },
    {
      "epoch": 6.326530612244898,
      "grad_norm": 5.166352272033691,
      "learning_rate": 6.838775510204082e-06,
      "loss": 0.1941,
      "step": 1550
    },
    {
      "epoch": 6.36734693877551,
      "grad_norm": 10.346336364746094,
      "learning_rate": 6.818367346938776e-06,
      "loss": 0.2278,
      "step": 1560
    },
    {
      "epoch": 6.408163265306122,
      "grad_norm": 7.703672409057617,
      "learning_rate": 6.797959183673471e-06,
      "loss": 0.2854,
      "step": 1570
    },
    {
      "epoch": 6.448979591836735,
      "grad_norm": 9.366389274597168,
      "learning_rate": 6.777551020408164e-06,
      "loss": 0.2229,
      "step": 1580
    },
    {
      "epoch": 6.489795918367347,
      "grad_norm": 7.013561248779297,
      "learning_rate": 6.757142857142858e-06,
      "loss": 0.2732,
      "step": 1590
    },
    {
      "epoch": 6.530612244897959,
      "grad_norm": 5.82119083404541,
      "learning_rate": 6.736734693877551e-06,
      "loss": 0.2389,
      "step": 1600
    },
    {
      "epoch": 6.571428571428571,
      "grad_norm": 8.112947463989258,
      "learning_rate": 6.716326530612245e-06,
      "loss": 0.193,
      "step": 1610
    },
    {
      "epoch": 6.612244897959184,
      "grad_norm": 8.238506317138672,
      "learning_rate": 6.6959183673469396e-06,
      "loss": 0.2344,
      "step": 1620
    },
    {
      "epoch": 6.653061224489796,
      "grad_norm": 6.531697750091553,
      "learning_rate": 6.675510204081634e-06,
      "loss": 0.2146,
      "step": 1630
    },
    {
      "epoch": 6.6938775510204085,
      "grad_norm": 6.525634288787842,
      "learning_rate": 6.655102040816327e-06,
      "loss": 0.1993,
      "step": 1640
    },
    {
      "epoch": 6.73469387755102,
      "grad_norm": 8.847986221313477,
      "learning_rate": 6.63469387755102e-06,
      "loss": 0.2884,
      "step": 1650
    },
    {
      "epoch": 6.775510204081632,
      "grad_norm": 9.006918907165527,
      "learning_rate": 6.614285714285715e-06,
      "loss": 0.2173,
      "step": 1660
    },
    {
      "epoch": 6.816326530612245,
      "grad_norm": 9.229476928710938,
      "learning_rate": 6.593877551020409e-06,
      "loss": 0.201,
      "step": 1670
    },
    {
      "epoch": 6.857142857142857,
      "grad_norm": 5.367541313171387,
      "learning_rate": 6.573469387755103e-06,
      "loss": 0.2505,
      "step": 1680
    },
    {
      "epoch": 6.8979591836734695,
      "grad_norm": 7.771108150482178,
      "learning_rate": 6.553061224489796e-06,
      "loss": 0.2225,
      "step": 1690
    },
    {
      "epoch": 6.938775510204081,
      "grad_norm": 5.306410789489746,
      "learning_rate": 6.53265306122449e-06,
      "loss": 0.2549,
      "step": 1700
    },
    {
      "epoch": 6.979591836734694,
      "grad_norm": 4.359670162200928,
      "learning_rate": 6.512244897959185e-06,
      "loss": 0.2585,
      "step": 1710
    },
    {
      "epoch": 7.020408163265306,
      "grad_norm": 4.528923034667969,
      "learning_rate": 6.491836734693878e-06,
      "loss": 0.2489,
      "step": 1720
    },
    {
      "epoch": 7.061224489795919,
      "grad_norm": 4.760287761688232,
      "learning_rate": 6.4714285714285715e-06,
      "loss": 0.1955,
      "step": 1730
    },
    {
      "epoch": 7.1020408163265305,
      "grad_norm": 9.205543518066406,
      "learning_rate": 6.451020408163266e-06,
      "loss": 0.2283,
      "step": 1740
    },
    {
      "epoch": 7.142857142857143,
      "grad_norm": 5.910974025726318,
      "learning_rate": 6.43061224489796e-06,
      "loss": 0.2308,
      "step": 1750
    },
    {
      "epoch": 7.183673469387755,
      "grad_norm": 5.042090892791748,
      "learning_rate": 6.410204081632654e-06,
      "loss": 0.1883,
      "step": 1760
    },
    {
      "epoch": 7.224489795918367,
      "grad_norm": 5.0842742919921875,
      "learning_rate": 6.389795918367347e-06,
      "loss": 0.1942,
      "step": 1770
    },
    {
      "epoch": 7.26530612244898,
      "grad_norm": 9.219313621520996,
      "learning_rate": 6.369387755102041e-06,
      "loss": 0.1946,
      "step": 1780
    },
    {
      "epoch": 7.3061224489795915,
      "grad_norm": 7.5656208992004395,
      "learning_rate": 6.348979591836735e-06,
      "loss": 0.2016,
      "step": 1790
    },
    {
      "epoch": 7.346938775510204,
      "grad_norm": 5.8213653564453125,
      "learning_rate": 6.3285714285714296e-06,
      "loss": 0.2808,
      "step": 1800
    },
    {
      "epoch": 7.387755102040816,
      "grad_norm": 8.004778861999512,
      "learning_rate": 6.308163265306123e-06,
      "loss": 0.2393,
      "step": 1810
    },
    {
      "epoch": 7.428571428571429,
      "grad_norm": 7.5001139640808105,
      "learning_rate": 6.287755102040817e-06,
      "loss": 0.2007,
      "step": 1820
    },
    {
      "epoch": 7.469387755102041,
      "grad_norm": 9.618229866027832,
      "learning_rate": 6.26734693877551e-06,
      "loss": 0.2464,
      "step": 1830
    },
    {
      "epoch": 7.510204081632653,
      "grad_norm": 7.257756233215332,
      "learning_rate": 6.246938775510205e-06,
      "loss": 0.1692,
      "step": 1840
    },
    {
      "epoch": 7.551020408163265,
      "grad_norm": 7.658279895782471,
      "learning_rate": 6.2265306122448985e-06,
      "loss": 0.2367,
      "step": 1850
    },
    {
      "epoch": 7.591836734693878,
      "grad_norm": 6.590469837188721,
      "learning_rate": 6.206122448979593e-06,
      "loss": 0.2508,
      "step": 1860
    },
    {
      "epoch": 7.63265306122449,
      "grad_norm": 8.601705551147461,
      "learning_rate": 6.185714285714286e-06,
      "loss": 0.2168,
      "step": 1870
    },
    {
      "epoch": 7.673469387755102,
      "grad_norm": 6.144942283630371,
      "learning_rate": 6.16530612244898e-06,
      "loss": 0.2308,
      "step": 1880
    },
    {
      "epoch": 7.714285714285714,
      "grad_norm": 3.256690502166748,
      "learning_rate": 6.144897959183674e-06,
      "loss": 0.1791,
      "step": 1890
    },
    {
      "epoch": 7.755102040816326,
      "grad_norm": 6.810645580291748,
      "learning_rate": 6.124489795918368e-06,
      "loss": 0.2335,
      "step": 1900
    },
    {
      "epoch": 7.795918367346939,
      "grad_norm": 3.4259018898010254,
      "learning_rate": 6.1040816326530616e-06,
      "loss": 0.212,
      "step": 1910
    },
    {
      "epoch": 7.836734693877551,
      "grad_norm": 8.353039741516113,
      "learning_rate": 6.083673469387756e-06,
      "loss": 0.2312,
      "step": 1920
    },
    {
      "epoch": 7.877551020408164,
      "grad_norm": 5.548733711242676,
      "learning_rate": 6.06326530612245e-06,
      "loss": 0.2666,
      "step": 1930
    },
    {
      "epoch": 7.918367346938775,
      "grad_norm": 8.386053085327148,
      "learning_rate": 6.042857142857144e-06,
      "loss": 0.2543,
      "step": 1940
    },
    {
      "epoch": 7.959183673469388,
      "grad_norm": 7.863219261169434,
      "learning_rate": 6.022448979591837e-06,
      "loss": 0.2171,
      "step": 1950
    },
    {
      "epoch": 8.0,
      "grad_norm": 5.328917503356934,
      "learning_rate": 6.0020408163265305e-06,
      "loss": 0.1592,
      "step": 1960
    },
    {
      "epoch": 8.040816326530612,
      "grad_norm": 11.456307411193848,
      "learning_rate": 5.981632653061225e-06,
      "loss": 0.2345,
      "step": 1970
    },
    {
      "epoch": 8.081632653061224,
      "grad_norm": 3.9219276905059814,
      "learning_rate": 5.96122448979592e-06,
      "loss": 0.228,
      "step": 1980
    },
    {
      "epoch": 8.122448979591837,
      "grad_norm": 7.155372142791748,
      "learning_rate": 5.940816326530613e-06,
      "loss": 0.1879,
      "step": 1990
    },
    {
      "epoch": 8.16326530612245,
      "grad_norm": 8.530526161193848,
      "learning_rate": 5.920408163265306e-06,
      "loss": 0.1808,
      "step": 2000
    },
    {
      "epoch": 8.204081632653061,
      "grad_norm": 10.660292625427246,
      "learning_rate": 5.902040816326531e-06,
      "loss": 0.2545,
      "step": 2010
    },
    {
      "epoch": 8.244897959183673,
      "grad_norm": 2.4285929203033447,
      "learning_rate": 5.881632653061225e-06,
      "loss": 0.1697,
      "step": 2020
    },
    {
      "epoch": 8.285714285714286,
      "grad_norm": 7.390512943267822,
      "learning_rate": 5.861224489795919e-06,
      "loss": 0.1868,
      "step": 2030
    },
    {
      "epoch": 8.326530612244898,
      "grad_norm": 6.263368129730225,
      "learning_rate": 5.840816326530613e-06,
      "loss": 0.2508,
      "step": 2040
    },
    {
      "epoch": 8.36734693877551,
      "grad_norm": 0.5456896424293518,
      "learning_rate": 5.820408163265306e-06,
      "loss": 0.1737,
      "step": 2050
    },
    {
      "epoch": 8.408163265306122,
      "grad_norm": 9.859989166259766,
      "learning_rate": 5.8e-06,
      "loss": 0.2154,
      "step": 2060
    },
    {
      "epoch": 8.448979591836734,
      "grad_norm": 7.002127647399902,
      "learning_rate": 5.7795918367346945e-06,
      "loss": 0.178,
      "step": 2070
    },
    {
      "epoch": 8.489795918367347,
      "grad_norm": 11.321815490722656,
      "learning_rate": 5.759183673469389e-06,
      "loss": 0.2147,
      "step": 2080
    },
    {
      "epoch": 8.53061224489796,
      "grad_norm": 4.573301315307617,
      "learning_rate": 5.738775510204082e-06,
      "loss": 0.2011,
      "step": 2090
    },
    {
      "epoch": 8.571428571428571,
      "grad_norm": 5.418411731719971,
      "learning_rate": 5.718367346938776e-06,
      "loss": 0.1873,
      "step": 2100
    },
    {
      "epoch": 8.612244897959183,
      "grad_norm": 5.087385177612305,
      "learning_rate": 5.697959183673469e-06,
      "loss": 0.2519,
      "step": 2110
    },
    {
      "epoch": 8.653061224489797,
      "grad_norm": 5.11081600189209,
      "learning_rate": 5.677551020408164e-06,
      "loss": 0.1733,
      "step": 2120
    },
    {
      "epoch": 8.693877551020408,
      "grad_norm": 9.11174201965332,
      "learning_rate": 5.6571428571428576e-06,
      "loss": 0.2496,
      "step": 2130
    },
    {
      "epoch": 8.73469387755102,
      "grad_norm": 7.031070709228516,
      "learning_rate": 5.636734693877552e-06,
      "loss": 0.2239,
      "step": 2140
    },
    {
      "epoch": 8.775510204081632,
      "grad_norm": 7.756526947021484,
      "learning_rate": 5.616326530612245e-06,
      "loss": 0.1974,
      "step": 2150
    },
    {
      "epoch": 8.816326530612244,
      "grad_norm": 7.697716236114502,
      "learning_rate": 5.59591836734694e-06,
      "loss": 0.258,
      "step": 2160
    },
    {
      "epoch": 8.857142857142858,
      "grad_norm": 8.356952667236328,
      "learning_rate": 5.575510204081633e-06,
      "loss": 0.1936,
      "step": 2170
    },
    {
      "epoch": 8.89795918367347,
      "grad_norm": 7.454901695251465,
      "learning_rate": 5.555102040816327e-06,
      "loss": 0.1504,
      "step": 2180
    },
    {
      "epoch": 8.938775510204081,
      "grad_norm": 4.652848720550537,
      "learning_rate": 5.534693877551021e-06,
      "loss": 0.1793,
      "step": 2190
    },
    {
      "epoch": 8.979591836734693,
      "grad_norm": 6.558859348297119,
      "learning_rate": 5.514285714285714e-06,
      "loss": 0.1974,
      "step": 2200
    },
    {
      "epoch": 9.020408163265307,
      "grad_norm": 7.869504928588867,
      "learning_rate": 5.493877551020409e-06,
      "loss": 0.2128,
      "step": 2210
    },
    {
      "epoch": 9.061224489795919,
      "grad_norm": 8.646539688110352,
      "learning_rate": 5.473469387755103e-06,
      "loss": 0.1755,
      "step": 2220
    },
    {
      "epoch": 9.10204081632653,
      "grad_norm": 7.58027458190918,
      "learning_rate": 5.453061224489796e-06,
      "loss": 0.1701,
      "step": 2230
    },
    {
      "epoch": 9.142857142857142,
      "grad_norm": 8.236135482788086,
      "learning_rate": 5.4326530612244895e-06,
      "loss": 0.219,
      "step": 2240
    },
    {
      "epoch": 9.183673469387756,
      "grad_norm": 6.239899635314941,
      "learning_rate": 5.4122448979591845e-06,
      "loss": 0.1846,
      "step": 2250
    },
    {
      "epoch": 9.224489795918368,
      "grad_norm": 5.651657581329346,
      "learning_rate": 5.391836734693879e-06,
      "loss": 0.237,
      "step": 2260
    },
    {
      "epoch": 9.26530612244898,
      "grad_norm": 6.521792411804199,
      "learning_rate": 5.371428571428572e-06,
      "loss": 0.2158,
      "step": 2270
    },
    {
      "epoch": 9.306122448979592,
      "grad_norm": 12.538895606994629,
      "learning_rate": 5.351020408163265e-06,
      "loss": 0.2103,
      "step": 2280
    },
    {
      "epoch": 9.346938775510203,
      "grad_norm": 6.341105937957764,
      "learning_rate": 5.330612244897959e-06,
      "loss": 0.1468,
      "step": 2290
    },
    {
      "epoch": 9.387755102040817,
      "grad_norm": 8.830240249633789,
      "learning_rate": 5.310204081632654e-06,
      "loss": 0.165,
      "step": 2300
    },
    {
      "epoch": 9.428571428571429,
      "grad_norm": 7.01543664932251,
      "learning_rate": 5.2897959183673476e-06,
      "loss": 0.1742,
      "step": 2310
    },
    {
      "epoch": 9.46938775510204,
      "grad_norm": 7.823669910430908,
      "learning_rate": 5.269387755102041e-06,
      "loss": 0.2414,
      "step": 2320
    },
    {
      "epoch": 9.510204081632653,
      "grad_norm": 7.9082770347595215,
      "learning_rate": 5.248979591836735e-06,
      "loss": 0.1961,
      "step": 2330
    },
    {
      "epoch": 9.551020408163264,
      "grad_norm": 8.552570343017578,
      "learning_rate": 5.22857142857143e-06,
      "loss": 0.1862,
      "step": 2340
    },
    {
      "epoch": 9.591836734693878,
      "grad_norm": 10.67483901977539,
      "learning_rate": 5.208163265306123e-06,
      "loss": 0.1773,
      "step": 2350
    },
    {
      "epoch": 9.63265306122449,
      "grad_norm": 5.287507057189941,
      "learning_rate": 5.1877551020408165e-06,
      "loss": 0.2165,
      "step": 2360
    },
    {
      "epoch": 9.673469387755102,
      "grad_norm": 3.9037272930145264,
      "learning_rate": 5.167346938775511e-06,
      "loss": 0.1598,
      "step": 2370
    },
    {
      "epoch": 9.714285714285714,
      "grad_norm": 2.465554714202881,
      "learning_rate": 5.146938775510204e-06,
      "loss": 0.1398,
      "step": 2380
    },
    {
      "epoch": 9.755102040816327,
      "grad_norm": 4.5391950607299805,
      "learning_rate": 5.126530612244899e-06,
      "loss": 0.1642,
      "step": 2390
    },
    {
      "epoch": 9.795918367346939,
      "grad_norm": 9.52216911315918,
      "learning_rate": 5.106122448979592e-06,
      "loss": 0.2317,
      "step": 2400
    },
    {
      "epoch": 9.83673469387755,
      "grad_norm": 7.413267135620117,
      "learning_rate": 5.085714285714286e-06,
      "loss": 0.2339,
      "step": 2410
    },
    {
      "epoch": 9.877551020408163,
      "grad_norm": 11.318732261657715,
      "learning_rate": 5.0653061224489795e-06,
      "loss": 0.2225,
      "step": 2420
    },
    {
      "epoch": 9.918367346938776,
      "grad_norm": 11.242213249206543,
      "learning_rate": 5.0448979591836745e-06,
      "loss": 0.3078,
      "step": 2430
    },
    {
      "epoch": 9.959183673469388,
      "grad_norm": 9.637420654296875,
      "learning_rate": 5.024489795918368e-06,
      "loss": 0.1856,
      "step": 2440
    },
    {
      "epoch": 10.0,
      "grad_norm": 10.665837287902832,
      "learning_rate": 5.004081632653062e-06,
      "loss": 0.2125,
      "step": 2450
    },
    {
      "epoch": 10.040816326530612,
      "grad_norm": 5.397568702697754,
      "learning_rate": 4.983673469387755e-06,
      "loss": 0.1722,
      "step": 2460
    },
    {
      "epoch": 10.081632653061224,
      "grad_norm": 5.377045631408691,
      "learning_rate": 4.963265306122449e-06,
      "loss": 0.1958,
      "step": 2470
    },
    {
      "epoch": 10.122448979591837,
      "grad_norm": 7.449840068817139,
      "learning_rate": 4.9428571428571435e-06,
      "loss": 0.1832,
      "step": 2480
    },
    {
      "epoch": 10.16326530612245,
      "grad_norm": 7.062196731567383,
      "learning_rate": 4.9224489795918376e-06,
      "loss": 0.1913,
      "step": 2490
    },
    {
      "epoch": 10.204081632653061,
      "grad_norm": 6.084071636199951,
      "learning_rate": 4.902040816326531e-06,
      "loss": 0.1187,
      "step": 2500
    },
    {
      "epoch": 10.244897959183673,
      "grad_norm": 4.809638023376465,
      "learning_rate": 4.881632653061225e-06,
      "loss": 0.1405,
      "step": 2510
    },
    {
      "epoch": 10.285714285714286,
      "grad_norm": 4.271588325500488,
      "learning_rate": 4.861224489795919e-06,
      "loss": 0.1357,
      "step": 2520
    },
    {
      "epoch": 10.326530612244898,
      "grad_norm": 8.498083114624023,
      "learning_rate": 4.840816326530612e-06,
      "loss": 0.2201,
      "step": 2530
    },
    {
      "epoch": 10.36734693877551,
      "grad_norm": 7.665253162384033,
      "learning_rate": 4.8204081632653065e-06,
      "loss": 0.1959,
      "step": 2540
    },
    {
      "epoch": 10.408163265306122,
      "grad_norm": 5.3332743644714355,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.1956,
      "step": 2550
    },
    {
      "epoch": 10.448979591836734,
      "grad_norm": 4.948959827423096,
      "learning_rate": 4.779591836734695e-06,
      "loss": 0.2092,
      "step": 2560
    },
    {
      "epoch": 10.489795918367347,
      "grad_norm": 7.226635456085205,
      "learning_rate": 4.759183673469388e-06,
      "loss": 0.1584,
      "step": 2570
    },
    {
      "epoch": 10.53061224489796,
      "grad_norm": 4.212419509887695,
      "learning_rate": 4.738775510204082e-06,
      "loss": 0.1476,
      "step": 2580
    },
    {
      "epoch": 10.571428571428571,
      "grad_norm": 7.545459270477295,
      "learning_rate": 4.7183673469387754e-06,
      "loss": 0.2103,
      "step": 2590
    },
    {
      "epoch": 10.612244897959183,
      "grad_norm": 1.853903889656067,
      "learning_rate": 4.69795918367347e-06,
      "loss": 0.2105,
      "step": 2600
    },
    {
      "epoch": 10.653061224489797,
      "grad_norm": 4.872255802154541,
      "learning_rate": 4.677551020408164e-06,
      "loss": 0.1979,
      "step": 2610
    },
    {
      "epoch": 10.693877551020408,
      "grad_norm": 8.121048927307129,
      "learning_rate": 4.657142857142857e-06,
      "loss": 0.1904,
      "step": 2620
    },
    {
      "epoch": 10.73469387755102,
      "grad_norm": 6.789455890655518,
      "learning_rate": 4.636734693877551e-06,
      "loss": 0.231,
      "step": 2630
    },
    {
      "epoch": 10.775510204081632,
      "grad_norm": 6.744252681732178,
      "learning_rate": 4.616326530612245e-06,
      "loss": 0.1028,
      "step": 2640
    },
    {
      "epoch": 10.816326530612244,
      "grad_norm": 1.3881915807724,
      "learning_rate": 4.595918367346939e-06,
      "loss": 0.1611,
      "step": 2650
    },
    {
      "epoch": 10.857142857142858,
      "grad_norm": 4.106182098388672,
      "learning_rate": 4.575510204081633e-06,
      "loss": 0.1721,
      "step": 2660
    },
    {
      "epoch": 10.89795918367347,
      "grad_norm": 6.046643257141113,
      "learning_rate": 4.555102040816327e-06,
      "loss": 0.2162,
      "step": 2670
    },
    {
      "epoch": 10.938775510204081,
      "grad_norm": 2.952366352081299,
      "learning_rate": 4.534693877551021e-06,
      "loss": 0.1941,
      "step": 2680
    },
    {
      "epoch": 10.979591836734693,
      "grad_norm": 5.679470539093018,
      "learning_rate": 4.514285714285714e-06,
      "loss": 0.1322,
      "step": 2690
    },
    {
      "epoch": 11.020408163265307,
      "grad_norm": 4.781472206115723,
      "learning_rate": 4.493877551020408e-06,
      "loss": 0.1756,
      "step": 2700
    },
    {
      "epoch": 11.061224489795919,
      "grad_norm": 4.453323841094971,
      "learning_rate": 4.473469387755102e-06,
      "loss": 0.1652,
      "step": 2710
    },
    {
      "epoch": 11.10204081632653,
      "grad_norm": 6.313192367553711,
      "learning_rate": 4.4530612244897965e-06,
      "loss": 0.2071,
      "step": 2720
    },
    {
      "epoch": 11.142857142857142,
      "grad_norm": 3.620354413986206,
      "learning_rate": 4.43265306122449e-06,
      "loss": 0.2166,
      "step": 2730
    },
    {
      "epoch": 11.183673469387756,
      "grad_norm": 5.309145450592041,
      "learning_rate": 4.412244897959184e-06,
      "loss": 0.1787,
      "step": 2740
    },
    {
      "epoch": 11.224489795918368,
      "grad_norm": 5.886495590209961,
      "learning_rate": 4.391836734693878e-06,
      "loss": 0.1699,
      "step": 2750
    },
    {
      "epoch": 11.26530612244898,
      "grad_norm": 6.391637325286865,
      "learning_rate": 4.371428571428572e-06,
      "loss": 0.1928,
      "step": 2760
    },
    {
      "epoch": 11.306122448979592,
      "grad_norm": 9.88926887512207,
      "learning_rate": 4.3510204081632654e-06,
      "loss": 0.2344,
      "step": 2770
    },
    {
      "epoch": 11.346938775510203,
      "grad_norm": 4.097028732299805,
      "learning_rate": 4.3306122448979596e-06,
      "loss": 0.1457,
      "step": 2780
    },
    {
      "epoch": 11.387755102040817,
      "grad_norm": 6.217812538146973,
      "learning_rate": 4.310204081632654e-06,
      "loss": 0.1773,
      "step": 2790
    },
    {
      "epoch": 11.428571428571429,
      "grad_norm": 6.636881351470947,
      "learning_rate": 4.289795918367347e-06,
      "loss": 0.1977,
      "step": 2800
    },
    {
      "epoch": 11.46938775510204,
      "grad_norm": 4.406877517700195,
      "learning_rate": 4.269387755102041e-06,
      "loss": 0.1331,
      "step": 2810
    },
    {
      "epoch": 11.510204081632653,
      "grad_norm": 7.739985466003418,
      "learning_rate": 4.248979591836735e-06,
      "loss": 0.1624,
      "step": 2820
    },
    {
      "epoch": 11.551020408163264,
      "grad_norm": 8.672730445861816,
      "learning_rate": 4.228571428571429e-06,
      "loss": 0.1435,
      "step": 2830
    },
    {
      "epoch": 11.591836734693878,
      "grad_norm": 8.240572929382324,
      "learning_rate": 4.208163265306123e-06,
      "loss": 0.1864,
      "step": 2840
    },
    {
      "epoch": 11.63265306122449,
      "grad_norm": 6.257174968719482,
      "learning_rate": 4.187755102040817e-06,
      "loss": 0.2008,
      "step": 2850
    },
    {
      "epoch": 11.673469387755102,
      "grad_norm": 7.619112014770508,
      "learning_rate": 4.167346938775511e-06,
      "loss": 0.1529,
      "step": 2860
    },
    {
      "epoch": 11.714285714285714,
      "grad_norm": 3.644754648208618,
      "learning_rate": 4.146938775510204e-06,
      "loss": 0.1434,
      "step": 2870
    },
    {
      "epoch": 11.755102040816327,
      "grad_norm": 3.984260082244873,
      "learning_rate": 4.126530612244898e-06,
      "loss": 0.1327,
      "step": 2880
    },
    {
      "epoch": 11.795918367346939,
      "grad_norm": 7.851540565490723,
      "learning_rate": 4.106122448979592e-06,
      "loss": 0.1709,
      "step": 2890
    },
    {
      "epoch": 11.83673469387755,
      "grad_norm": 2.9527790546417236,
      "learning_rate": 4.0857142857142865e-06,
      "loss": 0.1487,
      "step": 2900
    },
    {
      "epoch": 11.877551020408163,
      "grad_norm": 3.5094189643859863,
      "learning_rate": 4.06530612244898e-06,
      "loss": 0.1733,
      "step": 2910
    },
    {
      "epoch": 11.918367346938776,
      "grad_norm": 8.27623176574707,
      "learning_rate": 4.044897959183674e-06,
      "loss": 0.2207,
      "step": 2920
    },
    {
      "epoch": 11.959183673469388,
      "grad_norm": 6.333447456359863,
      "learning_rate": 4.024489795918368e-06,
      "loss": 0.1912,
      "step": 2930
    },
    {
      "epoch": 12.0,
      "grad_norm": 3.803018569946289,
      "learning_rate": 4.004081632653062e-06,
      "loss": 0.1969,
      "step": 2940
    },
    {
      "epoch": 12.040816326530612,
      "grad_norm": 7.993055820465088,
      "learning_rate": 3.9836734693877555e-06,
      "loss": 0.1708,
      "step": 2950
    },
    {
      "epoch": 12.081632653061224,
      "grad_norm": 3.1507952213287354,
      "learning_rate": 3.963265306122449e-06,
      "loss": 0.0929,
      "step": 2960
    },
    {
      "epoch": 12.122448979591837,
      "grad_norm": 3.0786514282226562,
      "learning_rate": 3.942857142857143e-06,
      "loss": 0.1392,
      "step": 2970
    },
    {
      "epoch": 12.16326530612245,
      "grad_norm": 12.369327545166016,
      "learning_rate": 3.922448979591837e-06,
      "loss": 0.1887,
      "step": 2980
    },
    {
      "epoch": 12.204081632653061,
      "grad_norm": 11.310943603515625,
      "learning_rate": 3.902040816326531e-06,
      "loss": 0.2074,
      "step": 2990
    },
    {
      "epoch": 12.244897959183673,
      "grad_norm": 8.117951393127441,
      "learning_rate": 3.881632653061224e-06,
      "loss": 0.2066,
      "step": 3000
    },
    {
      "epoch": 12.285714285714286,
      "grad_norm": 0.911245584487915,
      "learning_rate": 3.8612244897959185e-06,
      "loss": 0.1504,
      "step": 3010
    },
    {
      "epoch": 12.326530612244898,
      "grad_norm": 7.755861282348633,
      "learning_rate": 3.840816326530613e-06,
      "loss": 0.187,
      "step": 3020
    },
    {
      "epoch": 12.36734693877551,
      "grad_norm": 11.341306686401367,
      "learning_rate": 3.820408163265306e-06,
      "loss": 0.1385,
      "step": 3030
    },
    {
      "epoch": 12.408163265306122,
      "grad_norm": 6.276268005371094,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 0.1776,
      "step": 3040
    },
    {
      "epoch": 12.448979591836734,
      "grad_norm": 3.063488006591797,
      "learning_rate": 3.779591836734694e-06,
      "loss": 0.1903,
      "step": 3050
    },
    {
      "epoch": 12.489795918367347,
      "grad_norm": 6.759077548980713,
      "learning_rate": 3.7591836734693883e-06,
      "loss": 0.1504,
      "step": 3060
    },
    {
      "epoch": 12.53061224489796,
      "grad_norm": 9.553186416625977,
      "learning_rate": 3.738775510204082e-06,
      "loss": 0.1763,
      "step": 3070
    },
    {
      "epoch": 12.571428571428571,
      "grad_norm": 7.130514621734619,
      "learning_rate": 3.718367346938776e-06,
      "loss": 0.1464,
      "step": 3080
    },
    {
      "epoch": 12.612244897959183,
      "grad_norm": 6.302764415740967,
      "learning_rate": 3.69795918367347e-06,
      "loss": 0.1141,
      "step": 3090
    },
    {
      "epoch": 12.653061224489797,
      "grad_norm": 7.9323835372924805,
      "learning_rate": 3.677551020408164e-06,
      "loss": 0.2175,
      "step": 3100
    },
    {
      "epoch": 12.693877551020408,
      "grad_norm": 5.5850324630737305,
      "learning_rate": 3.6571428571428576e-06,
      "loss": 0.1602,
      "step": 3110
    },
    {
      "epoch": 12.73469387755102,
      "grad_norm": 7.268565654754639,
      "learning_rate": 3.636734693877551e-06,
      "loss": 0.181,
      "step": 3120
    },
    {
      "epoch": 12.775510204081632,
      "grad_norm": 5.973303318023682,
      "learning_rate": 3.616326530612245e-06,
      "loss": 0.1024,
      "step": 3130
    },
    {
      "epoch": 12.816326530612244,
      "grad_norm": 3.3456904888153076,
      "learning_rate": 3.5959183673469387e-06,
      "loss": 0.1756,
      "step": 3140
    },
    {
      "epoch": 12.857142857142858,
      "grad_norm": 7.198270797729492,
      "learning_rate": 3.575510204081633e-06,
      "loss": 0.156,
      "step": 3150
    },
    {
      "epoch": 12.89795918367347,
      "grad_norm": 3.496737003326416,
      "learning_rate": 3.5551020408163266e-06,
      "loss": 0.1868,
      "step": 3160
    },
    {
      "epoch": 12.938775510204081,
      "grad_norm": 9.160460472106934,
      "learning_rate": 3.5346938775510207e-06,
      "loss": 0.242,
      "step": 3170
    },
    {
      "epoch": 12.979591836734693,
      "grad_norm": 9.402756690979004,
      "learning_rate": 3.5142857142857144e-06,
      "loss": 0.1947,
      "step": 3180
    },
    {
      "epoch": 13.020408163265307,
      "grad_norm": 4.506322860717773,
      "learning_rate": 3.4938775510204085e-06,
      "loss": 0.1468,
      "step": 3190
    },
    {
      "epoch": 13.061224489795919,
      "grad_norm": 6.5064921379089355,
      "learning_rate": 3.4734693877551022e-06,
      "loss": 0.1914,
      "step": 3200
    },
    {
      "epoch": 13.10204081632653,
      "grad_norm": 5.771793842315674,
      "learning_rate": 3.453061224489796e-06,
      "loss": 0.1401,
      "step": 3210
    },
    {
      "epoch": 13.142857142857142,
      "grad_norm": 7.380532264709473,
      "learning_rate": 3.43265306122449e-06,
      "loss": 0.1248,
      "step": 3220
    },
    {
      "epoch": 13.183673469387756,
      "grad_norm": 3.246613025665283,
      "learning_rate": 3.4122448979591838e-06,
      "loss": 0.1464,
      "step": 3230
    },
    {
      "epoch": 13.224489795918368,
      "grad_norm": 7.83432674407959,
      "learning_rate": 3.391836734693878e-06,
      "loss": 0.1243,
      "step": 3240
    },
    {
      "epoch": 13.26530612244898,
      "grad_norm": 3.1438093185424805,
      "learning_rate": 3.3714285714285716e-06,
      "loss": 0.1624,
      "step": 3250
    },
    {
      "epoch": 13.306122448979592,
      "grad_norm": 11.76649284362793,
      "learning_rate": 3.3510204081632657e-06,
      "loss": 0.1864,
      "step": 3260
    },
    {
      "epoch": 13.346938775510203,
      "grad_norm": 9.060452461242676,
      "learning_rate": 3.3306122448979594e-06,
      "loss": 0.2151,
      "step": 3270
    },
    {
      "epoch": 13.387755102040817,
      "grad_norm": 4.446900367736816,
      "learning_rate": 3.310204081632653e-06,
      "loss": 0.1426,
      "step": 3280
    },
    {
      "epoch": 13.428571428571429,
      "grad_norm": 5.351714611053467,
      "learning_rate": 3.2897959183673472e-06,
      "loss": 0.1786,
      "step": 3290
    },
    {
      "epoch": 13.46938775510204,
      "grad_norm": 5.846502304077148,
      "learning_rate": 3.269387755102041e-06,
      "loss": 0.1612,
      "step": 3300
    },
    {
      "epoch": 13.510204081632653,
      "grad_norm": 1.2500088214874268,
      "learning_rate": 3.248979591836735e-06,
      "loss": 0.1413,
      "step": 3310
    },
    {
      "epoch": 13.551020408163264,
      "grad_norm": 8.840521812438965,
      "learning_rate": 3.2285714285714288e-06,
      "loss": 0.1483,
      "step": 3320
    },
    {
      "epoch": 13.591836734693878,
      "grad_norm": 6.333540916442871,
      "learning_rate": 3.208163265306123e-06,
      "loss": 0.1949,
      "step": 3330
    },
    {
      "epoch": 13.63265306122449,
      "grad_norm": 7.405364990234375,
      "learning_rate": 3.1877551020408166e-06,
      "loss": 0.1546,
      "step": 3340
    },
    {
      "epoch": 13.673469387755102,
      "grad_norm": 4.707162857055664,
      "learning_rate": 3.1673469387755107e-06,
      "loss": 0.1809,
      "step": 3350
    },
    {
      "epoch": 13.714285714285714,
      "grad_norm": 9.862110137939453,
      "learning_rate": 3.1469387755102044e-06,
      "loss": 0.1563,
      "step": 3360
    },
    {
      "epoch": 13.755102040816327,
      "grad_norm": 7.040682792663574,
      "learning_rate": 3.126530612244898e-06,
      "loss": 0.1717,
      "step": 3370
    },
    {
      "epoch": 13.795918367346939,
      "grad_norm": 8.885647773742676,
      "learning_rate": 3.1061224489795922e-06,
      "loss": 0.1468,
      "step": 3380
    },
    {
      "epoch": 13.83673469387755,
      "grad_norm": 5.706595420837402,
      "learning_rate": 3.085714285714286e-06,
      "loss": 0.1334,
      "step": 3390
    },
    {
      "epoch": 13.877551020408163,
      "grad_norm": 6.5305352210998535,
      "learning_rate": 3.06530612244898e-06,
      "loss": 0.1253,
      "step": 3400
    },
    {
      "epoch": 13.918367346938776,
      "grad_norm": 8.774666786193848,
      "learning_rate": 3.0448979591836738e-06,
      "loss": 0.2265,
      "step": 3410
    },
    {
      "epoch": 13.959183673469388,
      "grad_norm": 7.207103729248047,
      "learning_rate": 3.024489795918368e-06,
      "loss": 0.1446,
      "step": 3420
    },
    {
      "epoch": 14.0,
      "grad_norm": 7.917428493499756,
      "learning_rate": 3.0040816326530616e-06,
      "loss": 0.1569,
      "step": 3430
    },
    {
      "epoch": 14.040816326530612,
      "grad_norm": 6.160830974578857,
      "learning_rate": 2.9836734693877557e-06,
      "loss": 0.1126,
      "step": 3440
    },
    {
      "epoch": 14.081632653061224,
      "grad_norm": 6.8933186531066895,
      "learning_rate": 2.9632653061224494e-06,
      "loss": 0.1646,
      "step": 3450
    },
    {
      "epoch": 14.122448979591837,
      "grad_norm": 6.392195701599121,
      "learning_rate": 2.9428571428571427e-06,
      "loss": 0.1732,
      "step": 3460
    },
    {
      "epoch": 14.16326530612245,
      "grad_norm": 4.3666462898254395,
      "learning_rate": 2.9224489795918372e-06,
      "loss": 0.1358,
      "step": 3470
    },
    {
      "epoch": 14.204081632653061,
      "grad_norm": 4.311633586883545,
      "learning_rate": 2.9020408163265305e-06,
      "loss": 0.1994,
      "step": 3480
    },
    {
      "epoch": 14.244897959183673,
      "grad_norm": 7.359527587890625,
      "learning_rate": 2.881632653061225e-06,
      "loss": 0.2524,
      "step": 3490
    },
    {
      "epoch": 14.285714285714286,
      "grad_norm": 7.441856861114502,
      "learning_rate": 2.8612244897959183e-06,
      "loss": 0.1356,
      "step": 3500
    },
    {
      "epoch": 14.326530612244898,
      "grad_norm": 7.622951507568359,
      "learning_rate": 2.8408163265306125e-06,
      "loss": 0.1732,
      "step": 3510
    },
    {
      "epoch": 14.36734693877551,
      "grad_norm": 4.243079662322998,
      "learning_rate": 2.820408163265306e-06,
      "loss": 0.1582,
      "step": 3520
    },
    {
      "epoch": 14.408163265306122,
      "grad_norm": 7.601721286773682,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.17,
      "step": 3530
    },
    {
      "epoch": 14.448979591836734,
      "grad_norm": 8.382099151611328,
      "learning_rate": 2.779591836734694e-06,
      "loss": 0.1305,
      "step": 3540
    },
    {
      "epoch": 14.489795918367347,
      "grad_norm": 6.1426591873168945,
      "learning_rate": 2.7591836734693877e-06,
      "loss": 0.2164,
      "step": 3550
    },
    {
      "epoch": 14.53061224489796,
      "grad_norm": 3.3495421409606934,
      "learning_rate": 2.738775510204082e-06,
      "loss": 0.2136,
      "step": 3560
    },
    {
      "epoch": 14.571428571428571,
      "grad_norm": 5.564877986907959,
      "learning_rate": 2.7183673469387755e-06,
      "loss": 0.1569,
      "step": 3570
    },
    {
      "epoch": 14.612244897959183,
      "grad_norm": 2.329392910003662,
      "learning_rate": 2.6979591836734697e-06,
      "loss": 0.2099,
      "step": 3580
    },
    {
      "epoch": 14.653061224489797,
      "grad_norm": 4.251242160797119,
      "learning_rate": 2.6775510204081634e-06,
      "loss": 0.1837,
      "step": 3590
    },
    {
      "epoch": 14.693877551020408,
      "grad_norm": 3.6388752460479736,
      "learning_rate": 2.6571428571428575e-06,
      "loss": 0.1558,
      "step": 3600
    },
    {
      "epoch": 14.73469387755102,
      "grad_norm": 1.2794233560562134,
      "learning_rate": 2.636734693877551e-06,
      "loss": 0.1416,
      "step": 3610
    },
    {
      "epoch": 14.775510204081632,
      "grad_norm": 8.014337539672852,
      "learning_rate": 2.616326530612245e-06,
      "loss": 0.2179,
      "step": 3620
    },
    {
      "epoch": 14.816326530612244,
      "grad_norm": 8.611292839050293,
      "learning_rate": 2.595918367346939e-06,
      "loss": 0.1655,
      "step": 3630
    },
    {
      "epoch": 14.857142857142858,
      "grad_norm": 5.971241474151611,
      "learning_rate": 2.5755102040816327e-06,
      "loss": 0.1575,
      "step": 3640
    },
    {
      "epoch": 14.89795918367347,
      "grad_norm": 8.207197189331055,
      "learning_rate": 2.555102040816327e-06,
      "loss": 0.1448,
      "step": 3650
    },
    {
      "epoch": 14.938775510204081,
      "grad_norm": 7.183290958404541,
      "learning_rate": 2.5346938775510205e-06,
      "loss": 0.1438,
      "step": 3660
    },
    {
      "epoch": 14.979591836734693,
      "grad_norm": 5.2017669677734375,
      "learning_rate": 2.5142857142857147e-06,
      "loss": 0.1582,
      "step": 3670
    },
    {
      "epoch": 15.020408163265307,
      "grad_norm": 5.843130588531494,
      "learning_rate": 2.4938775510204084e-06,
      "loss": 0.1632,
      "step": 3680
    },
    {
      "epoch": 15.061224489795919,
      "grad_norm": 4.228123188018799,
      "learning_rate": 2.473469387755102e-06,
      "loss": 0.1471,
      "step": 3690
    },
    {
      "epoch": 15.10204081632653,
      "grad_norm": 7.808995723724365,
      "learning_rate": 2.453061224489796e-06,
      "loss": 0.1754,
      "step": 3700
    },
    {
      "epoch": 15.142857142857142,
      "grad_norm": 6.6370849609375,
      "learning_rate": 2.43265306122449e-06,
      "loss": 0.1385,
      "step": 3710
    },
    {
      "epoch": 15.183673469387756,
      "grad_norm": 10.751446723937988,
      "learning_rate": 2.412244897959184e-06,
      "loss": 0.2018,
      "step": 3720
    },
    {
      "epoch": 15.224489795918368,
      "grad_norm": 2.6137239933013916,
      "learning_rate": 2.3918367346938777e-06,
      "loss": 0.1027,
      "step": 3730
    },
    {
      "epoch": 15.26530612244898,
      "grad_norm": 8.975826263427734,
      "learning_rate": 2.371428571428572e-06,
      "loss": 0.1591,
      "step": 3740
    },
    {
      "epoch": 15.306122448979592,
      "grad_norm": 5.2156758308410645,
      "learning_rate": 2.3510204081632655e-06,
      "loss": 0.2063,
      "step": 3750
    },
    {
      "epoch": 15.346938775510203,
      "grad_norm": 1.501234531402588,
      "learning_rate": 2.3306122448979592e-06,
      "loss": 0.1375,
      "step": 3760
    },
    {
      "epoch": 15.387755102040817,
      "grad_norm": 8.037463188171387,
      "learning_rate": 2.3102040816326534e-06,
      "loss": 0.1607,
      "step": 3770
    },
    {
      "epoch": 15.428571428571429,
      "grad_norm": 3.0848886966705322,
      "learning_rate": 2.289795918367347e-06,
      "loss": 0.1313,
      "step": 3780
    },
    {
      "epoch": 15.46938775510204,
      "grad_norm": 7.0728254318237305,
      "learning_rate": 2.269387755102041e-06,
      "loss": 0.1805,
      "step": 3790
    },
    {
      "epoch": 15.510204081632653,
      "grad_norm": 4.4785003662109375,
      "learning_rate": 2.248979591836735e-06,
      "loss": 0.1491,
      "step": 3800
    },
    {
      "epoch": 15.551020408163264,
      "grad_norm": 6.812148094177246,
      "learning_rate": 2.228571428571429e-06,
      "loss": 0.1597,
      "step": 3810
    },
    {
      "epoch": 15.591836734693878,
      "grad_norm": 6.794986248016357,
      "learning_rate": 2.2081632653061227e-06,
      "loss": 0.1942,
      "step": 3820
    },
    {
      "epoch": 15.63265306122449,
      "grad_norm": 8.72322940826416,
      "learning_rate": 2.1877551020408164e-06,
      "loss": 0.1595,
      "step": 3830
    },
    {
      "epoch": 15.673469387755102,
      "grad_norm": 6.645898818969727,
      "learning_rate": 2.16734693877551e-06,
      "loss": 0.1789,
      "step": 3840
    },
    {
      "epoch": 15.714285714285714,
      "grad_norm": 2.4683022499084473,
      "learning_rate": 2.1469387755102042e-06,
      "loss": 0.1489,
      "step": 3850
    },
    {
      "epoch": 15.755102040816327,
      "grad_norm": 3.773542881011963,
      "learning_rate": 2.126530612244898e-06,
      "loss": 0.1196,
      "step": 3860
    },
    {
      "epoch": 15.795918367346939,
      "grad_norm": 12.576253890991211,
      "learning_rate": 2.106122448979592e-06,
      "loss": 0.1655,
      "step": 3870
    },
    {
      "epoch": 15.83673469387755,
      "grad_norm": 9.4144287109375,
      "learning_rate": 2.0857142857142858e-06,
      "loss": 0.1791,
      "step": 3880
    },
    {
      "epoch": 15.877551020408163,
      "grad_norm": 6.735049724578857,
      "learning_rate": 2.06530612244898e-06,
      "loss": 0.1278,
      "step": 3890
    },
    {
      "epoch": 15.918367346938776,
      "grad_norm": 5.6283183097839355,
      "learning_rate": 2.0448979591836736e-06,
      "loss": 0.1513,
      "step": 3900
    },
    {
      "epoch": 15.959183673469388,
      "grad_norm": 3.5016186237335205,
      "learning_rate": 2.0244897959183677e-06,
      "loss": 0.0998,
      "step": 3910
    },
    {
      "epoch": 16.0,
      "grad_norm": 9.131093978881836,
      "learning_rate": 2.0040816326530614e-06,
      "loss": 0.1838,
      "step": 3920
    },
    {
      "epoch": 16.040816326530614,
      "grad_norm": 9.901714324951172,
      "learning_rate": 1.983673469387755e-06,
      "loss": 0.19,
      "step": 3930
    },
    {
      "epoch": 16.081632653061224,
      "grad_norm": 8.307121276855469,
      "learning_rate": 1.9632653061224492e-06,
      "loss": 0.153,
      "step": 3940
    },
    {
      "epoch": 16.122448979591837,
      "grad_norm": 5.754699230194092,
      "learning_rate": 1.942857142857143e-06,
      "loss": 0.1497,
      "step": 3950
    },
    {
      "epoch": 16.163265306122447,
      "grad_norm": 9.883703231811523,
      "learning_rate": 1.922448979591837e-06,
      "loss": 0.171,
      "step": 3960
    },
    {
      "epoch": 16.20408163265306,
      "grad_norm": 4.327215194702148,
      "learning_rate": 1.9020408163265308e-06,
      "loss": 0.1696,
      "step": 3970
    },
    {
      "epoch": 16.244897959183675,
      "grad_norm": 5.210219860076904,
      "learning_rate": 1.8816326530612247e-06,
      "loss": 0.16,
      "step": 3980
    },
    {
      "epoch": 16.285714285714285,
      "grad_norm": 4.966328144073486,
      "learning_rate": 1.8612244897959186e-06,
      "loss": 0.1657,
      "step": 3990
    },
    {
      "epoch": 16.3265306122449,
      "grad_norm": 4.904468059539795,
      "learning_rate": 1.8408163265306123e-06,
      "loss": 0.1356,
      "step": 4000
    }
  ],
  "logging_steps": 10,
  "max_steps": 4900,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 20,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 30,
  "trial_name": null,
  "trial_params": null
}