{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 3405,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 5.865102639296188e-07,
      "loss": 7.6534,
      "step": 1
    },
    {
      "epoch": 0.0,
      "learning_rate": 2.9325513196480943e-06,
      "loss": 7.5632,
      "step": 5
    },
    {
      "epoch": 0.01,
      "learning_rate": 5.865102639296189e-06,
      "loss": 7.3976,
      "step": 10
    },
    {
      "epoch": 0.01,
      "learning_rate": 8.797653958944282e-06,
      "loss": 7.9288,
      "step": 15
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.1730205278592377e-05,
      "loss": 7.9065,
      "step": 20
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.4662756598240471e-05,
      "loss": 7.6023,
      "step": 25
    },
    {
      "epoch": 0.03,
      "learning_rate": 1.7595307917888564e-05,
      "loss": 7.456,
      "step": 30
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.052785923753666e-05,
      "loss": 7.4133,
      "step": 35
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.3460410557184755e-05,
      "loss": 6.8687,
      "step": 40
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.6392961876832843e-05,
      "loss": 6.8713,
      "step": 45
    },
    {
      "epoch": 0.04,
      "learning_rate": 2.9325513196480942e-05,
      "loss": 6.2954,
      "step": 50
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.2258064516129034e-05,
      "loss": 6.2245,
      "step": 55
    },
    {
      "epoch": 0.05,
      "learning_rate": 3.519061583577713e-05,
      "loss": 5.6886,
      "step": 60
    },
    {
      "epoch": 0.06,
      "learning_rate": 3.8123167155425224e-05,
      "loss": 5.7786,
      "step": 65
    },
    {
      "epoch": 0.06,
      "learning_rate": 4.105571847507332e-05,
      "loss": 5.48,
      "step": 70
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.398826979472141e-05,
      "loss": 5.4462,
      "step": 75
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.692082111436951e-05,
      "loss": 5.3462,
      "step": 80
    },
    {
      "epoch": 0.07,
      "learning_rate": 4.98533724340176e-05,
      "loss": 5.0227,
      "step": 85
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.2785923753665686e-05,
      "loss": 5.1957,
      "step": 90
    },
    {
      "epoch": 0.08,
      "learning_rate": 5.571847507331378e-05,
      "loss": 5.0037,
      "step": 95
    },
    {
      "epoch": 0.09,
      "learning_rate": 5.8651026392961884e-05,
      "loss": 4.9914,
      "step": 100
    },
    {
      "epoch": 0.09,
      "learning_rate": 6.158357771260998e-05,
      "loss": 4.7335,
      "step": 105
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.451612903225807e-05,
      "loss": 4.6138,
      "step": 110
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.744868035190616e-05,
      "loss": 4.9263,
      "step": 115
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.038123167155426e-05,
      "loss": 4.9091,
      "step": 120
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.331378299120235e-05,
      "loss": 4.75,
      "step": 125
    },
    {
      "epoch": 0.11,
      "learning_rate": 7.624633431085045e-05,
      "loss": 4.5122,
      "step": 130
    },
    {
      "epoch": 0.12,
      "learning_rate": 7.917888563049854e-05,
      "loss": 4.5551,
      "step": 135
    },
    {
      "epoch": 0.12,
      "learning_rate": 8.211143695014664e-05,
      "loss": 4.5593,
      "step": 140
    },
    {
      "epoch": 0.13,
      "learning_rate": 8.504398826979473e-05,
      "loss": 4.2632,
      "step": 145
    },
    {
      "epoch": 0.13,
      "learning_rate": 8.797653958944282e-05,
      "loss": 4.4557,
      "step": 150
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.090909090909092e-05,
      "loss": 4.4678,
      "step": 155
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.384164222873902e-05,
      "loss": 4.3127,
      "step": 160
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.677419354838711e-05,
      "loss": 4.0954,
      "step": 165
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.97067448680352e-05,
      "loss": 3.9996,
      "step": 170
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00010263929618768328,
      "loss": 4.1799,
      "step": 175
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00010557184750733137,
      "loss": 3.8218,
      "step": 180
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00010850439882697947,
      "loss": 4.0961,
      "step": 185
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00011143695014662756,
      "loss": 4.1024,
      "step": 190
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00011436950146627568,
      "loss": 3.9372,
      "step": 195
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00011730205278592377,
      "loss": 3.9716,
      "step": 200
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00012023460410557186,
      "loss": 3.7182,
      "step": 205
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00012316715542521996,
      "loss": 3.7256,
      "step": 210
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00012609970674486805,
      "loss": 3.9024,
      "step": 215
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00012903225806451613,
      "loss": 3.7147,
      "step": 220
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00013196480938416422,
      "loss": 3.5153,
      "step": 225
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.0001348973607038123,
      "loss": 3.5249,
      "step": 230
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00013782991202346043,
      "loss": 3.4098,
      "step": 235
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00014076246334310852,
      "loss": 3.5465,
      "step": 240
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001436950146627566,
      "loss": 3.4768,
      "step": 245
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0001466275659824047,
      "loss": 3.4544,
      "step": 250
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00014956011730205278,
      "loss": 3.6389,
      "step": 255
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0001524926686217009,
      "loss": 3.4381,
      "step": 260
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00015542521994134898,
      "loss": 3.4255,
      "step": 265
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00015835777126099707,
      "loss": 3.3577,
      "step": 270
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00016129032258064516,
      "loss": 3.5253,
      "step": 275
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00016422287390029328,
      "loss": 3.3494,
      "step": 280
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.00016715542521994137,
      "loss": 2.9786,
      "step": 285
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00017008797653958945,
      "loss": 3.2221,
      "step": 290
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00017302052785923754,
      "loss": 2.9849,
      "step": 295
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00017595307917888563,
      "loss": 3.3148,
      "step": 300
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00017888563049853372,
      "loss": 3.0488,
      "step": 305
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00018181818181818183,
      "loss": 3.0925,
      "step": 310
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00018475073313782992,
      "loss": 2.873,
      "step": 315
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00018768328445747804,
      "loss": 3.0446,
      "step": 320
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019061583577712613,
      "loss": 3.1887,
      "step": 325
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00019354838709677422,
      "loss": 2.9703,
      "step": 330
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0001964809384164223,
      "loss": 3.0357,
      "step": 335
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0001994134897360704,
      "loss": 3.2415,
      "step": 340
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00019999915896977904,
      "loss": 3.0616,
      "step": 345
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0001999957423087518,
      "loss": 2.833,
      "step": 350
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00019998969754225768,
      "loss": 3.1048,
      "step": 355
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001999810248291663,
      "loss": 2.9998,
      "step": 360
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.00019996972439741538,
      "loss": 2.938,
      "step": 365
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019995579654400483,
      "loss": 2.8948,
      "step": 370
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00019993924163498893,
      "loss": 2.9246,
      "step": 375
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0001999200601054667,
      "loss": 2.829,
      "step": 380
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019989825245957037,
      "loss": 3.0443,
      "step": 385
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00019987381927045233,
      "loss": 2.8411,
      "step": 390
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00019984676118026985,
      "loss": 3.0017,
      "step": 395
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.0001998170789001684,
      "loss": 2.9336,
      "step": 400
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019978477321026282,
      "loss": 2.7251,
      "step": 405
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00019974984495961683,
      "loss": 2.9361,
      "step": 410
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019971229506622083,
      "loss": 2.6362,
      "step": 415
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0001996721245169676,
      "loss": 3.0726,
      "step": 420
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00019962933436762646,
      "loss": 2.6445,
      "step": 425
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00019958392574281558,
      "loss": 2.7365,
      "step": 430
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.0001995358998359724,
      "loss": 2.6591,
      "step": 435
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019948525790932205,
      "loss": 2.7724,
      "step": 440
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00019943200129384444,
      "loss": 2.8148,
      "step": 445
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019937613138923922,
      "loss": 2.7555,
      "step": 450
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00019931764966388895,
      "loss": 2.712,
      "step": 455
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019925655765482046,
      "loss": 2.759,
      "step": 460
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.00019919285696766453,
      "loss": 2.7315,
      "step": 465
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0001991265492766137,
      "loss": 2.6124,
      "step": 470
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019905763632437827,
      "loss": 2.7523,
      "step": 475
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00019898611992214034,
      "loss": 2.6089,
      "step": 480
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00019891200194950643,
      "loss": 2.7065,
      "step": 485
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.000198835284354458,
      "loss": 2.709,
      "step": 490
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001987559691533002,
      "loss": 2.8017,
      "step": 495
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019867405843060896,
      "loss": 2.5728,
      "step": 500
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00019858955433917604,
      "loss": 2.6404,
      "step": 505
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0001985024590999527,
      "loss": 2.5759,
      "step": 510
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0001984127750019911,
      "loss": 2.7597,
      "step": 515
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00019832050440238432,
      "loss": 2.636,
      "step": 520
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00019822564972620427,
      "loss": 2.6216,
      "step": 525
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00019812821346643807,
      "loss": 2.6913,
      "step": 530
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00019802819818392237,
      "loss": 2.535,
      "step": 535
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00019792560650727618,
      "loss": 2.7263,
      "step": 540
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00019782044113283181,
      "loss": 2.5684,
      "step": 545
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00019771270482456385,
      "loss": 2.5042,
      "step": 550
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00019760240041401673,
      "loss": 2.5333,
      "step": 555
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00019748953080023007,
      "loss": 2.549,
      "step": 560
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00019737409894966267,
      "loss": 2.5871,
      "step": 565
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.00019725610789611452,
      "loss": 2.7041,
      "step": 570
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00019713556074064695,
      "loss": 2.5114,
      "step": 575
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00019701246065150123,
      "loss": 2.6314,
      "step": 580
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00019688681086401527,
      "loss": 2.3458,
      "step": 585
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00019675861468053865,
      "loss": 2.5374,
      "step": 590
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.00019662787547034569,
      "loss": 2.7692,
      "step": 595
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00019649459666954704,
      "loss": 2.4264,
      "step": 600
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00019635878178099928,
      "loss": 2.4306,
      "step": 605
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00019622043437421294,
      "loss": 2.4421,
      "step": 610
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00019607955808525853,
      "loss": 2.5787,
      "step": 615
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0001959361566166712,
      "loss": 2.576,
      "step": 620
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.00019579023373735322,
      "loss": 2.4709,
      "step": 625
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00019564179328247506,
      "loss": 2.4937,
      "step": 630
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00019549083915337451,
      "loss": 2.5605,
      "step": 635
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00019533737531745424,
      "loss": 2.4941,
      "step": 640
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00019518140580807744,
      "loss": 2.5547,
      "step": 645
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00019502293472446186,
      "loss": 2.46,
      "step": 650
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.000194861966231572,
      "loss": 2.6948,
      "step": 655
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00019469850456000975,
      "loss": 2.5352,
      "step": 660
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0001945325540059032,
      "loss": 2.492,
      "step": 665
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0001943641189307936,
      "loss": 2.5583,
      "step": 670
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00019419320376152084,
      "loss": 2.6595,
      "step": 675
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00019401981299010712,
      "loss": 2.3485,
      "step": 680
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.0001938439511736388,
      "loss": 2.4808,
      "step": 685
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0001936656229341466,
      "loss": 2.58,
      "step": 690
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00019348483295848442,
      "loss": 2.5846,
      "step": 695
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00019330158599820574,
      "loss": 2.5121,
      "step": 700
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00019311588686943897,
      "loss": 2.5782,
      "step": 705
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00019292774045276086,
      "loss": 2.4354,
      "step": 710
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00019273715169306827,
      "loss": 2.3542,
      "step": 715
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.00019254412559944806,
      "loss": 2.4883,
      "step": 720
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00019234866724504555,
      "loss": 2.2913,
      "step": 725
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00019215078176693112,
      "loss": 2.5519,
      "step": 730
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0001919504743659653,
      "loss": 2.2724,
      "step": 735
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00019174775030666202,
      "loss": 2.5216,
      "step": 740
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0001915426149170502,
      "loss": 2.3688,
      "step": 745
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00019133507358853373,
      "loss": 2.6479,
      "step": 750
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00019112513177574992,
      "loss": 2.2218,
      "step": 755
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00019091279499642593,
      "loss": 2.3275,
      "step": 760
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00019069806883123387,
      "loss": 2.5138,
      "step": 765
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00019048095892364408,
      "loss": 2.4388,
      "step": 770
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00019026147097977693,
      "loss": 2.4809,
      "step": 775
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00019003961076825263,
      "loss": 2.4191,
      "step": 780
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00018981538412003978,
      "loss": 2.3277,
      "step": 785
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00018958879692830216,
      "loss": 2.3928,
      "step": 790
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0001893598551482436,
      "loss": 2.3763,
      "step": 795
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00018912856479695185,
      "loss": 2.3602,
      "step": 800
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00018889493195323997,
      "loss": 2.3453,
      "step": 805
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00018865896275748704,
      "loss": 2.5283,
      "step": 810
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00018842066341147635,
      "loss": 2.4586,
      "step": 815
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00018818004017823276,
      "loss": 2.443,
      "step": 820
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00018793709938185789,
      "loss": 2.3027,
      "step": 825
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.0001876918474073639,
      "loss": 2.3587,
      "step": 830
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00018744429070050578,
      "loss": 2.355,
      "step": 835
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00018719443576761195,
      "loss": 2.5382,
      "step": 840
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00018694228917541313,
      "loss": 2.2777,
      "step": 845
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0001866878575508698,
      "loss": 2.4265,
      "step": 850
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00018643114758099818,
      "loss": 2.4205,
      "step": 855
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00018617216601269423,
      "loss": 2.4819,
      "step": 860
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00018591091965255654,
      "loss": 2.2673,
      "step": 865
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00018564741536670728,
      "loss": 2.5598,
      "step": 870
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00018538166008061192,
      "loss": 2.1621,
      "step": 875
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.000185113660778897,
      "loss": 2.4109,
      "step": 880
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00018484342450516671,
      "loss": 2.4062,
      "step": 885
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00018457095836181778,
      "loss": 2.3055,
      "step": 890
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00018429626950985267,
      "loss": 2.4536,
      "step": 895
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.00018401936516869148,
      "loss": 2.3542,
      "step": 900
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00018374025261598225,
      "loss": 2.2079,
      "step": 905
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00018345893918740952,
      "loss": 2.3577,
      "step": 910
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.00018317543227650172,
      "loss": 2.5193,
      "step": 915
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0001828897393344367,
      "loss": 2.4154,
      "step": 920
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.000182601867869846,
      "loss": 2.2691,
      "step": 925
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00018231182544861741,
      "loss": 2.3977,
      "step": 930
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00018201961969369624,
      "loss": 2.3604,
      "step": 935
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00018172525828488495,
      "loss": 2.4388,
      "step": 940
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.00018142874895864112,
      "loss": 2.4306,
      "step": 945
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0001811300995078744,
      "loss": 2.4967,
      "step": 950
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00018082931778174154,
      "loss": 2.3996,
      "step": 955
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001805264116854401,
      "loss": 2.3246,
      "step": 960
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001802213891800007,
      "loss": 2.3288,
      "step": 965
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0001799142582820778,
      "loss": 2.4168,
      "step": 970
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00017960502706373903,
      "loss": 2.4643,
      "step": 975
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00017929370365225297,
      "loss": 2.3672,
      "step": 980
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00017898029622987556,
      "loss": 2.4626,
      "step": 985
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.00017866481303363514,
      "loss": 2.1713,
      "step": 990
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00017834726235511585,
      "loss": 2.3403,
      "step": 995
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00017802765254023981,
      "loss": 2.3151,
      "step": 1000
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00017770599198904763,
      "loss": 2.3607,
      "step": 1005
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.00017738228915547784,
      "loss": 2.3208,
      "step": 1010
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0001770565525471445,
      "loss": 2.4054,
      "step": 1015
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00017672879072511374,
      "loss": 2.2819,
      "step": 1020
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00017639901230367868,
      "loss": 2.1317,
      "step": 1025
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.00017606722595013308,
      "loss": 2.4226,
      "step": 1030
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0001757334403845435,
      "loss": 2.339,
      "step": 1035
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00017539766437952014,
      "loss": 2.2972,
      "step": 1040
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.0001750599067599863,
      "loss": 2.2524,
      "step": 1045
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00017472017640294643,
      "loss": 2.2363,
      "step": 1050
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00017437848223725267,
      "loss": 2.2668,
      "step": 1055
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00017403483324337056,
      "loss": 2.1602,
      "step": 1060
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.0001736892384531426,
      "loss": 2.4429,
      "step": 1065
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00017334170694955102,
      "loss": 2.4132,
      "step": 1070
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00017299224786647925,
      "loss": 2.4951,
      "step": 1075
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00017264087038847156,
      "loss": 2.2886,
      "step": 1080
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00017228758375049185,
      "loss": 2.446,
      "step": 1085
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00017193239723768092,
      "loss": 2.3134,
      "step": 1090
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.00017157532018511237,
      "loss": 2.2684,
      "step": 1095
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.00017121636197754735,
      "loss": 2.2554,
      "step": 1100
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0001708555320491878,
      "loss": 2.3473,
      "step": 1105
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00017049283988342855,
      "loss": 2.2431,
      "step": 1110
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0001701282950126081,
      "loss": 2.2468,
      "step": 1115
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00016976190701775806,
      "loss": 2.2984,
      "step": 1120
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00016939368552835137,
      "loss": 2.3279,
      "step": 1125
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00016902364022204912,
      "loss": 2.2503,
      "step": 1130
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00016865178082444638,
      "loss": 2.0986,
      "step": 1135
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00016827811710881635,
      "loss": 2.3541,
      "step": 1140
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0001679026588958538,
      "loss": 2.2958,
      "step": 1145
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0001675254160534166,
      "loss": 2.278,
      "step": 1150
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00016714639849626668,
      "loss": 2.2633,
      "step": 1155
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00016676561618580933,
      "loss": 2.3342,
      "step": 1160
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00016638307912983136,
      "loss": 2.266,
      "step": 1165
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00016599879738223814,
      "loss": 2.2961,
      "step": 1170
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00016561278104278927,
      "loss": 2.3334,
      "step": 1175
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.0001652250402568332,
      "loss": 2.2961,
      "step": 1180
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00016483558521504068,
      "loss": 2.118,
      "step": 1185
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00016444442615313668,
      "loss": 2.4154,
      "step": 1190
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.0001640515733516317,
      "loss": 2.3285,
      "step": 1195
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0001636570371355512,
      "loss": 2.2754,
      "step": 1200
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0001632608278741646,
      "loss": 2.2198,
      "step": 1205
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00016286295598071253,
      "loss": 2.1981,
      "step": 1210
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00016246343191213315,
      "loss": 2.1861,
      "step": 1215
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00016206226616878749,
      "loss": 2.094,
      "step": 1220
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00016165946929418322,
      "loss": 2.1678,
      "step": 1225
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.00016125505187469775,
      "loss": 2.2353,
      "step": 1230
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.0001608490245393,
      "loss": 2.1292,
      "step": 1235
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00016044139795927086,
      "loss": 2.2069,
      "step": 1240
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00016003218284792298,
      "loss": 2.2202,
      "step": 1245
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.0001596213899603189,
      "loss": 2.2953,
      "step": 1250
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0001592090300929887,
      "loss": 2.3847,
      "step": 1255
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00015879511408364605,
      "loss": 2.3227,
      "step": 1260
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.0001583796528109033,
      "loss": 2.3445,
      "step": 1265
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00015796265719398593,
      "loss": 2.2701,
      "step": 1270
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.000157544138192445,
      "loss": 2.2308,
      "step": 1275
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00015712410680586973,
      "loss": 2.3682,
      "step": 1280
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00015670257407359792,
      "loss": 2.1626,
      "step": 1285
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00015627955107442605,
      "loss": 2.1219,
      "step": 1290
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00015585504892631807,
      "loss": 2.2555,
      "step": 1295
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00015542907878611314,
      "loss": 2.1176,
      "step": 1300
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00015500165184923248,
      "loss": 2.393,
      "step": 1305
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.00015457277934938509,
      "loss": 2.158,
      "step": 1310
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00015414247255827253,
      "loss": 2.2454,
      "step": 1315
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.00015371074278529257,
      "loss": 2.3419,
      "step": 1320
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00015327760137724212,
      "loss": 2.1038,
      "step": 1325
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00015284305971801883,
      "loss": 2.2645,
      "step": 1330
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00015240712922832215,
      "loss": 2.2823,
      "step": 1335
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00015196982136535288,
      "loss": 2.2989,
      "step": 1340
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00015153114762251222,
      "loss": 2.1625,
      "step": 1345
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00015109111952909968,
      "loss": 2.0568,
      "step": 1350
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00015064974865001008,
      "loss": 2.2737,
      "step": 1355
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00015020704658542946,
      "loss": 2.2047,
      "step": 1360
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00014976302497053036,
      "loss": 2.4024,
      "step": 1365
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00014931769547516603,
      "loss": 2.154,
      "step": 1370
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00014887106980356358,
      "loss": 2.2493,
      "step": 1375
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0001484231596940165,
      "loss": 2.2429,
      "step": 1380
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00014797397691857612,
      "loss": 2.2164,
      "step": 1385
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.0001475235332827421,
      "loss": 2.2448,
      "step": 1390
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0001470718406251523,
      "loss": 2.3309,
      "step": 1395
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.0001466189108172717,
      "loss": 2.1357,
      "step": 1400
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00014616475576308005,
      "loss": 2.3023,
      "step": 1405
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00014570938739875944,
      "loss": 2.3552,
      "step": 1410
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.0001452528176923803,
      "loss": 2.2292,
      "step": 1415
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00014479505864358692,
      "loss": 2.1244,
      "step": 1420
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00014433612228328214,
      "loss": 2.3995,
      "step": 1425
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.0001438760206733111,
      "loss": 2.254,
      "step": 1430
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00014341476590614424,
      "loss": 2.1234,
      "step": 1435
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00014295237010455945,
      "loss": 2.2274,
      "step": 1440
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.0001424888454213235,
      "loss": 2.1644,
      "step": 1445
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00014202420403887258,
      "loss": 2.2781,
      "step": 1450
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00014155845816899224,
      "loss": 2.149,
      "step": 1455
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00014109162005249633,
      "loss": 2.2153,
      "step": 1460
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.0001406237019589053,
      "loss": 2.0983,
      "step": 1465
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00014015471618612378,
      "loss": 2.3075,
      "step": 1470
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00013968467506011723,
      "loss": 2.2857,
      "step": 1475
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.00013921359093458825,
      "loss": 2.2647,
      "step": 1480
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.0001387414761906516,
      "loss": 2.2328,
      "step": 1485
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.000138268343236509,
      "loss": 2.1563,
      "step": 1490
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00013779420450712288,
      "loss": 2.3384,
      "step": 1495
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00013731907246388977,
      "loss": 2.333,
      "step": 1500
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.0001368429595943124,
      "loss": 2.1938,
      "step": 1505
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00013636587841167207,
      "loss": 2.2841,
      "step": 1510
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00013588784145469914,
      "loss": 2.2626,
      "step": 1515
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00013540886128724403,
      "loss": 2.1829,
      "step": 1520
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.0001349289504979467,
      "loss": 2.2862,
      "step": 1525
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.0001344481216999058,
      "loss": 2.2542,
      "step": 1530
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00013396638753034734,
      "loss": 2.3202,
      "step": 1535
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00013348376065029245,
      "loss": 2.2086,
      "step": 1540
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0001330002537442246,
      "loss": 2.0535,
      "step": 1545
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00013251587951975625,
      "loss": 2.3245,
      "step": 1550
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00013203065070729495,
      "loss": 2.272,
      "step": 1555
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.00013154458005970856,
      "loss": 2.2416,
      "step": 1560
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00013105768035199034,
      "loss": 2.2201,
      "step": 1565
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0001305699643809229,
      "loss": 2.2501,
      "step": 1570
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00013008144496474213,
      "loss": 2.3365,
      "step": 1575
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00012959213494280017,
      "loss": 2.3172,
      "step": 1580
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00012910204717522805,
      "loss": 2.2071,
      "step": 1585
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.0001286111945425975,
      "loss": 2.0843,
      "step": 1590
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.0001281195899455827,
      "loss": 2.1805,
      "step": 1595
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00012762724630462112,
      "loss": 2.2205,
      "step": 1600
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00012713417655957376,
      "loss": 2.3792,
      "step": 1605
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00012664039366938533,
      "loss": 2.2761,
      "step": 1610
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00012614591061174351,
      "loss": 2.2718,
      "step": 1615
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00012565074038273788,
      "loss": 2.3257,
      "step": 1620
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00012515489599651848,
      "loss": 2.108,
      "step": 1625
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.0001246583904849535,
      "loss": 2.2152,
      "step": 1630
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00012416123689728707,
      "loss": 2.3599,
      "step": 1635
    },
    {
      "epoch": 1.44,
      "learning_rate": 0.00012366344829979611,
      "loss": 2.1716,
      "step": 1640
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.000123165037775447,
      "loss": 2.244,
      "step": 1645
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.0001226660184235517,
      "loss": 2.0568,
      "step": 1650
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00012216640335942352,
      "loss": 2.1132,
      "step": 1655
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00012166620571403231,
      "loss": 2.0462,
      "step": 1660
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.0001211654386336595,
      "loss": 2.1835,
      "step": 1665
    },
    {
      "epoch": 1.47,
      "learning_rate": 0.00012066411527955243,
      "loss": 2.1925,
      "step": 1670
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00012016224882757862,
      "loss": 2.2295,
      "step": 1675
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00011965985246787927,
      "loss": 2.0637,
      "step": 1680
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.0001191569394045228,
      "loss": 2.2134,
      "step": 1685
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00011865352285515765,
      "loss": 2.0748,
      "step": 1690
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00011814961605066504,
      "loss": 2.2397,
      "step": 1695
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00011764523223481112,
      "loss": 2.2193,
      "step": 1700
    },
    {
      "epoch": 1.5,
      "learning_rate": 0.00011714038466389891,
      "loss": 2.1098,
      "step": 1705
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00011663508660641999,
      "loss": 2.2415,
      "step": 1710
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00011612935134270566,
      "loss": 2.1913,
      "step": 1715
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00011562319216457795,
      "loss": 2.0807,
      "step": 1720
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00011511662237500032,
      "loss": 2.111,
      "step": 1725
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.00011460965528772792,
      "loss": 2.1528,
      "step": 1730
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00011410230422695778,
      "loss": 2.1324,
      "step": 1735
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.00011359458252697864,
      "loss": 2.0786,
      "step": 1740
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.00011308650353182036,
      "loss": 2.1816,
      "step": 1745
    },
    {
      "epoch": 1.54,
      "learning_rate": 0.0001125780805949034,
      "loss": 2.252,
      "step": 1750
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00011206932707868762,
      "loss": 2.1158,
      "step": 1755
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.00011156025635432135,
      "loss": 2.1911,
      "step": 1760
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00011105088180128976,
      "loss": 2.2904,
      "step": 1765
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00011054121680706328,
      "loss": 2.1534,
      "step": 1770
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.00011003127476674591,
      "loss": 2.1485,
      "step": 1775
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00010952106908272284,
      "loss": 2.1895,
      "step": 1780
    },
    {
      "epoch": 1.57,
      "learning_rate": 0.00010901061316430847,
      "loss": 2.1565,
      "step": 1785
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00010849992042739394,
      "loss": 2.0069,
      "step": 1790
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.00010798900429409444,
      "loss": 2.2347,
      "step": 1795
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.00010747787819239656,
      "loss": 2.3373,
      "step": 1800
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.00010696655555580524,
      "loss": 2.0499,
      "step": 1805
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.00010645504982299082,
      "loss": 2.2529,
      "step": 1810
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00010594337443743585,
      "loss": 2.1572,
      "step": 1815
    },
    {
      "epoch": 1.6,
      "learning_rate": 0.00010543154284708163,
      "loss": 2.137,
      "step": 1820
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00010491956850397495,
      "loss": 2.1686,
      "step": 1825
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00010440746486391439,
      "loss": 2.1664,
      "step": 1830
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.00010389524538609681,
      "loss": 2.2125,
      "step": 1835
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.00010338292353276341,
      "loss": 2.1918,
      "step": 1840
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00010287051276884618,
      "loss": 2.2608,
      "step": 1845
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00010235802656161384,
      "loss": 2.1754,
      "step": 1850
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00010184547838031782,
      "loss": 2.2016,
      "step": 1855
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00010133288169583848,
      "loss": 2.0677,
      "step": 1860
    },
    {
      "epoch": 1.64,
      "learning_rate": 0.00010082024998033092,
      "loss": 2.1323,
      "step": 1865
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.00010030759670687093,
      "loss": 2.1701,
      "step": 1870
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.979493534910085e-05,
      "loss": 2.1928,
      "step": 1875
    },
    {
      "epoch": 1.66,
      "learning_rate": 9.928227938087553e-05,
      "loss": 2.2373,
      "step": 1880
    },
    {
      "epoch": 1.66,
      "learning_rate": 9.87696422759082e-05,
      "loss": 2.1827,
      "step": 1885
    },
    {
      "epoch": 1.67,
      "learning_rate": 9.825703750741626e-05,
      "loss": 2.1034,
      "step": 1890
    },
    {
      "epoch": 1.67,
      "learning_rate": 9.774447854776725e-05,
      "loss": 2.2465,
      "step": 1895
    },
    {
      "epoch": 1.67,
      "learning_rate": 9.723197886812481e-05,
      "loss": 2.3259,
      "step": 1900
    },
    {
      "epoch": 1.68,
      "learning_rate": 9.671955193809452e-05,
      "loss": 2.1901,
      "step": 1905
    },
    {
      "epoch": 1.68,
      "learning_rate": 9.620721122536998e-05,
      "loss": 2.0353,
      "step": 1910
    },
    {
      "epoch": 1.69,
      "learning_rate": 9.569497019537872e-05,
      "loss": 2.2914,
      "step": 1915
    },
    {
      "epoch": 1.69,
      "learning_rate": 9.518284231092848e-05,
      "loss": 2.0966,
      "step": 1920
    },
    {
      "epoch": 1.7,
      "learning_rate": 9.467084103185329e-05,
      "loss": 2.0779,
      "step": 1925
    },
    {
      "epoch": 1.7,
      "learning_rate": 9.41589798146597e-05,
      "loss": 2.1169,
      "step": 1930
    },
    {
      "epoch": 1.7,
      "learning_rate": 9.36472721121731e-05,
      "loss": 2.3157,
      "step": 1935
    },
    {
      "epoch": 1.71,
      "learning_rate": 9.313573137318419e-05,
      "loss": 2.1449,
      "step": 1940
    },
    {
      "epoch": 1.71,
      "learning_rate": 9.262437104209555e-05,
      "loss": 2.1209,
      "step": 1945
    },
    {
      "epoch": 1.72,
      "learning_rate": 9.21132045585682e-05,
      "loss": 2.1572,
      "step": 1950
    },
    {
      "epoch": 1.72,
      "learning_rate": 9.160224535716842e-05,
      "loss": 2.1399,
      "step": 1955
    },
    {
      "epoch": 1.73,
      "learning_rate": 9.109150686701476e-05,
      "loss": 2.1079,
      "step": 1960
    },
    {
      "epoch": 1.73,
      "learning_rate": 9.058100251142483e-05,
      "loss": 2.0942,
      "step": 1965
    },
    {
      "epoch": 1.74,
      "learning_rate": 9.007074570756278e-05,
      "loss": 1.9655,
      "step": 1970
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.956074986608661e-05,
      "loss": 2.0805,
      "step": 1975
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.905102839079559e-05,
      "loss": 2.2586,
      "step": 1980
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.854159467827807e-05,
      "loss": 2.0632,
      "step": 1985
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.803246211755931e-05,
      "loss": 2.2206,
      "step": 1990
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.752364408974974e-05,
      "loss": 2.3172,
      "step": 1995
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.701515396769311e-05,
      "loss": 2.0313,
      "step": 2000
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.650700511561514e-05,
      "loss": 2.1009,
      "step": 2005
    },
    {
      "epoch": 1.77,
      "learning_rate": 8.599921088877225e-05,
      "loss": 2.074,
      "step": 2010
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.549178463310046e-05,
      "loss": 2.0483,
      "step": 2015
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.498473968486476e-05,
      "loss": 2.1001,
      "step": 2020
    },
    {
      "epoch": 1.78,
      "learning_rate": 8.447808937030856e-05,
      "loss": 2.2748,
      "step": 2025
    },
    {
      "epoch": 1.79,
      "learning_rate": 8.39718470053034e-05,
      "loss": 2.1422,
      "step": 2030
    },
    {
      "epoch": 1.79,
      "learning_rate": 8.346602589499912e-05,
      "loss": 2.204,
      "step": 2035
    },
    {
      "epoch": 1.8,
      "learning_rate": 8.296063933347388e-05,
      "loss": 2.235,
      "step": 2040
    },
    {
      "epoch": 1.8,
      "learning_rate": 8.245570060338512e-05,
      "loss": 2.0509,
      "step": 2045
    },
    {
      "epoch": 1.81,
      "learning_rate": 8.195122297562023e-05,
      "loss": 2.219,
      "step": 2050
    },
    {
      "epoch": 1.81,
      "learning_rate": 8.144721970894786e-05,
      "loss": 2.0662,
      "step": 2055
    },
    {
      "epoch": 1.81,
      "learning_rate": 8.094370404966942e-05,
      "loss": 2.0488,
      "step": 2060
    },
    {
      "epoch": 1.82,
      "learning_rate": 8.044068923127087e-05,
      "loss": 2.0054,
      "step": 2065
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.993818847407506e-05,
      "loss": 2.1268,
      "step": 2070
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.943621498489415e-05,
      "loss": 2.0403,
      "step": 2075
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.893478195668259e-05,
      "loss": 1.9364,
      "step": 2080
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.843390256819034e-05,
      "loss": 2.1122,
      "step": 2085
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.793358998361642e-05,
      "loss": 2.0691,
      "step": 2090
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.743385735226309e-05,
      "loss": 2.1533,
      "step": 2095
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.693471780819019e-05,
      "loss": 2.0395,
      "step": 2100
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.643618446986988e-05,
      "loss": 2.2597,
      "step": 2105
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.593827043984198e-05,
      "loss": 2.07,
      "step": 2110
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.544098880436946e-05,
      "loss": 2.0882,
      "step": 2115
    },
    {
      "epoch": 1.87,
      "learning_rate": 7.494435263309466e-05,
      "loss": 2.0713,
      "step": 2120
    },
    {
      "epoch": 1.87,
      "learning_rate": 7.44483749786957e-05,
      "loss": 2.1475,
      "step": 2125
    },
    {
      "epoch": 1.88,
      "learning_rate": 7.395306887654341e-05,
      "loss": 1.9508,
      "step": 2130
    },
    {
      "epoch": 1.88,
      "learning_rate": 7.345844734435884e-05,
      "loss": 2.1495,
      "step": 2135
    },
    {
      "epoch": 1.89,
      "learning_rate": 7.296452338187094e-05,
      "loss": 2.1613,
      "step": 2140
    },
    {
      "epoch": 1.89,
      "learning_rate": 7.247130997047511e-05,
      "loss": 2.164,
      "step": 2145
    },
    {
      "epoch": 1.89,
      "learning_rate": 7.197882007289189e-05,
      "loss": 2.003,
      "step": 2150
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.148706663282629e-05,
      "loss": 2.169,
      "step": 2155
    },
    {
      "epoch": 1.9,
      "learning_rate": 7.09960625746277e-05,
      "loss": 2.2235,
      "step": 2160
    },
    {
      "epoch": 1.91,
      "learning_rate": 7.050582080294996e-05,
      "loss": 2.0062,
      "step": 2165
    },
    {
      "epoch": 1.91,
      "learning_rate": 7.00163542024125e-05,
      "loss": 2.2334,
      "step": 2170
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.952767563726154e-05,
      "loss": 2.0814,
      "step": 2175
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.903979795103199e-05,
      "loss": 2.0644,
      "step": 2180
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.855273396620998e-05,
      "loss": 2.1427,
      "step": 2185
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.806649648389567e-05,
      "loss": 2.2253,
      "step": 2190
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.758109828346709e-05,
      "loss": 2.0299,
      "step": 2195
    },
    {
      "epoch": 1.94,
      "learning_rate": 6.709655212224398e-05,
      "loss": 2.0782,
      "step": 2200
    },
    {
      "epoch": 1.94,
      "learning_rate": 6.661287073515275e-05,
      "loss": 1.9985,
      "step": 2205
    },
    {
      "epoch": 1.95,
      "learning_rate": 6.613006683439169e-05,
      "loss": 2.0108,
      "step": 2210
    },
    {
      "epoch": 1.95,
      "learning_rate": 6.564815310909667e-05,
      "loss": 2.2069,
      "step": 2215
    },
    {
      "epoch": 1.96,
      "learning_rate": 6.516714222500794e-05,
      "loss": 1.9994,
      "step": 2220
    },
    {
      "epoch": 1.96,
      "learning_rate": 6.468704682413711e-05,
      "loss": 2.1872,
      "step": 2225
    },
    {
      "epoch": 1.96,
      "learning_rate": 6.420787952443491e-05,
      "loss": 2.0854,
      "step": 2230
    },
    {
      "epoch": 1.97,
      "learning_rate": 6.372965291945953e-05,
      "loss": 2.2533,
      "step": 2235
    },
    {
      "epoch": 1.97,
      "learning_rate": 6.325237957804563e-05,
      "loss": 2.0482,
      "step": 2240
    },
    {
      "epoch": 1.98,
      "learning_rate": 6.277607204397408e-05,
      "loss": 2.147,
      "step": 2245
    },
    {
      "epoch": 1.98,
      "learning_rate": 6.230074283564222e-05,
      "loss": 2.0424,
      "step": 2250
    },
    {
      "epoch": 1.99,
      "learning_rate": 6.182640444573487e-05,
      "loss": 2.175,
      "step": 2255
    },
    {
      "epoch": 1.99,
      "learning_rate": 6.1353069340896e-05,
      "loss": 2.2443,
      "step": 2260
    },
    {
      "epoch": 2.0,
      "learning_rate": 6.0880749961400964e-05,
      "loss": 2.0684,
      "step": 2265
    },
    {
      "epoch": 2.0,
      "learning_rate": 6.040945872082978e-05,
      "loss": 2.1364,
      "step": 2270
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.993920800574069e-05,
      "loss": 2.1039,
      "step": 2275
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.947001017534465e-05,
      "loss": 2.115,
      "step": 2280
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.900187756118055e-05,
      "loss": 2.1173,
      "step": 2285
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.853482246679104e-05,
      "loss": 2.1462,
      "step": 2290
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.806885716739923e-05,
      "loss": 1.9374,
      "step": 2295
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.760399390958606e-05,
      "loss": 2.2031,
      "step": 2300
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.714024491096841e-05,
      "loss": 2.056,
      "step": 2305
    },
    {
      "epoch": 2.04,
      "learning_rate": 5.6677622359878036e-05,
      "loss": 2.1539,
      "step": 2310
    },
    {
      "epoch": 2.04,
      "learning_rate": 5.6216138415041105e-05,
      "loss": 2.0291,
      "step": 2315
    },
    {
      "epoch": 2.04,
      "learning_rate": 5.575580520525885e-05,
      "loss": 2.067,
      "step": 2320
    },
    {
      "epoch": 2.05,
      "learning_rate": 5.529663482908864e-05,
      "loss": 2.0761,
      "step": 2325
    },
    {
      "epoch": 2.05,
      "learning_rate": 5.483863935452605e-05,
      "loss": 1.9784,
      "step": 2330
    },
    {
      "epoch": 2.06,
      "learning_rate": 5.4381830818687705e-05,
      "loss": 2.1079,
      "step": 2335
    },
    {
      "epoch": 2.06,
      "learning_rate": 5.3926221227494845e-05,
      "loss": 2.1377,
      "step": 2340
    },
    {
      "epoch": 2.07,
      "learning_rate": 5.3471822555357896e-05,
      "loss": 2.1082,
      "step": 2345
    },
    {
      "epoch": 2.07,
      "learning_rate": 5.301864674486168e-05,
      "loss": 1.9832,
      "step": 2350
    },
    {
      "epoch": 2.07,
      "learning_rate": 5.2566705706451566e-05,
      "loss": 2.1141,
      "step": 2355
    },
    {
      "epoch": 2.08,
      "learning_rate": 5.2116011318120405e-05,
      "loss": 1.9963,
      "step": 2360
    },
    {
      "epoch": 2.08,
      "learning_rate": 5.1666575425096396e-05,
      "loss": 2.0701,
      "step": 2365
    },
    {
      "epoch": 2.09,
      "learning_rate": 5.121840983953175e-05,
      "loss": 2.0663,
      "step": 2370
    },
    {
      "epoch": 2.09,
      "learning_rate": 5.07715263401922e-05,
      "loss": 2.1082,
      "step": 2375
    },
    {
      "epoch": 2.1,
      "learning_rate": 5.032593667214752e-05,
      "loss": 2.0038,
      "step": 2380
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.988165254646277e-05,
      "loss": 1.9715,
      "step": 2385
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.943868563989046e-05,
      "loss": 2.0917,
      "step": 2390
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.899704759456377e-05,
      "loss": 2.0824,
      "step": 2395
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.855675001769052e-05,
      "loss": 2.1039,
      "step": 2400
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.811780448124812e-05,
      "loss": 2.0707,
      "step": 2405
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.7680222521679405e-05,
      "loss": 2.0922,
      "step": 2410
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.724401563958938e-05,
      "loss": 1.9923,
      "step": 2415
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.68091952994431e-05,
      "loss": 2.0032,
      "step": 2420
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.637577292926428e-05,
      "loss": 2.0474,
      "step": 2425
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.594375992033494e-05,
      "loss": 2.2291,
      "step": 2430
    },
    {
      "epoch": 2.15,
      "learning_rate": 4.551316762689602e-05,
      "loss": 2.1552,
      "step": 2435
    },
    {
      "epoch": 2.15,
      "learning_rate": 4.5084007365848936e-05,
      "loss": 2.18,
      "step": 2440
    },
    {
      "epoch": 2.15,
      "learning_rate": 4.465629041645819e-05,
      "loss": 2.1891,
      "step": 2445
    },
    {
      "epoch": 2.16,
      "learning_rate": 4.423002802005494e-05,
      "loss": 2.0704,
      "step": 2450
    },
    {
      "epoch": 2.16,
      "learning_rate": 4.3805231379741564e-05,
      "loss": 2.257,
      "step": 2455
    },
    {
      "epoch": 2.17,
      "learning_rate": 4.338191166009713e-05,
      "loss": 2.0064,
      "step": 2460
    },
    {
      "epoch": 2.17,
      "learning_rate": 4.296007998688405e-05,
      "loss": 2.2268,
      "step": 2465
    },
    {
      "epoch": 2.18,
      "learning_rate": 4.2539747446755565e-05,
      "loss": 2.0656,
      "step": 2470
    },
    {
      "epoch": 2.18,
      "learning_rate": 4.2120925086964534e-05,
      "loss": 2.0588,
      "step": 2475
    },
    {
      "epoch": 2.19,
      "learning_rate": 4.170362391507295e-05,
      "loss": 2.1166,
      "step": 2480
    },
    {
      "epoch": 2.19,
      "learning_rate": 4.1287854898662705e-05,
      "loss": 2.0328,
      "step": 2485
    },
    {
      "epoch": 2.19,
      "learning_rate": 4.0873628965047315e-05,
      "loss": 2.0747,
      "step": 2490
    },
    {
      "epoch": 2.2,
      "learning_rate": 4.046095700098466e-05,
      "loss": 1.9675,
      "step": 2495
    },
    {
      "epoch": 2.2,
      "learning_rate": 4.0049849852391e-05,
      "loss": 2.2333,
      "step": 2500
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.9640318324055815e-05,
      "loss": 2.1956,
      "step": 2505
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.923237317935787e-05,
      "loss": 2.0067,
      "step": 2510
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.882602513998234e-05,
      "loss": 1.997,
      "step": 2515
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.84212848856389e-05,
      "loss": 2.0763,
      "step": 2520
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.801816305378124e-05,
      "loss": 2.0233,
      "step": 2525
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.761667023932738e-05,
      "loss": 2.1154,
      "step": 2530
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.7216816994381175e-05,
      "loss": 2.0546,
      "step": 2535
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.6818613827955084e-05,
      "loss": 2.1157,
      "step": 2540
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.6422071205693866e-05,
      "loss": 2.0575,
      "step": 2545
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.60271995495996e-05,
      "loss": 2.1667,
      "step": 2550
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.563400923775774e-05,
      "loss": 2.1915,
      "step": 2555
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.524251060406434e-05,
      "loss": 2.0662,
      "step": 2560
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.485271393795453e-05,
      "loss": 2.0043,
      "step": 2565
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.446462948413193e-05,
      "loss": 2.1659,
      "step": 2570
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.407826744229957e-05,
      "loss": 2.2091,
      "step": 2575
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.3693637966891755e-05,
      "loss": 2.122,
      "step": 2580
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.331075116680715e-05,
      "loss": 2.1096,
      "step": 2585
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.292961710514317e-05,
      "loss": 2.1768,
      "step": 2590
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.255024579893138e-05,
      "loss": 2.0958,
      "step": 2595
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.217264721887433e-05,
      "loss": 1.9104,
      "step": 2600
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.179683128908352e-05,
      "loss": 2.125,
      "step": 2605
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.1422807886818463e-05,
      "loss": 2.2961,
      "step": 2610
    },
    {
      "epoch": 2.3,
      "learning_rate": 3.105058684222723e-05,
      "loss": 2.1653,
      "step": 2615
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.0680177938087886e-05,
      "loss": 2.1683,
      "step": 2620
    },
    {
      "epoch": 2.31,
      "learning_rate": 3.0311590909551625e-05,
      "loss": 1.9607,
      "step": 2625
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.9944835443886743e-05,
      "loss": 2.133,
      "step": 2630
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.957992118022408e-05,
      "loss": 2.0095,
      "step": 2635
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.9216857709303713e-05,
      "loss": 2.0514,
      "step": 2640
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.8855654573222825e-05,
      "loss": 1.9734,
      "step": 2645
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.8496321265184977e-05,
      "loss": 2.1773,
      "step": 2650
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.8138867229250588e-05,
      "loss": 1.9544,
      "step": 2655
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.7783301860088705e-05,
      "loss": 2.1265,
      "step": 2660
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.742963450273016e-05,
      "loss": 2.1215,
      "step": 2665
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.7077874452321783e-05,
      "loss": 2.1721,
      "step": 2670
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.6728030953882356e-05,
      "loss": 2.1026,
      "step": 2675
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.6380113202059463e-05,
      "loss": 2.1411,
      "step": 2680
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.6034130340887895e-05,
      "loss": 2.038,
      "step": 2685
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.5690091463549326e-05,
      "loss": 2.0145,
      "step": 2690
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.5348005612133207e-05,
      "loss": 1.9861,
      "step": 2695
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.5007881777399354e-05,
      "loss": 2.0184,
      "step": 2700
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.4669728898541456e-05,
      "loss": 1.9674,
      "step": 2705
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.4333555862952197e-05,
      "loss": 1.9125,
      "step": 2710
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.3999371505989686e-05,
      "loss": 2.1081,
      "step": 2715
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.3667184610745252e-05,
      "loss": 2.0656,
      "step": 2720
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.333700390781256e-05,
      "loss": 2.0374,
      "step": 2725
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.300883807505819e-05,
      "loss": 2.0221,
      "step": 2730
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.2682695737393567e-05,
      "loss": 2.2395,
      "step": 2735
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.2358585466548267e-05,
      "loss": 2.153,
      "step": 2740
    },
    {
      "epoch": 2.42,
      "learning_rate": 2.2036515780844667e-05,
      "loss": 2.0518,
      "step": 2745
    },
    {
      "epoch": 2.42,
      "learning_rate": 2.1716495144974213e-05,
      "loss": 2.0957,
      "step": 2750
    },
    {
      "epoch": 2.43,
      "learning_rate": 2.139853196977485e-05,
      "loss": 2.0684,
      "step": 2755
    },
    {
      "epoch": 2.43,
      "learning_rate": 2.1082634612009954e-05,
      "loss": 2.0296,
      "step": 2760
    },
    {
      "epoch": 2.44,
      "learning_rate": 2.07688113741488e-05,
      "loss": 2.0239,
      "step": 2765
    },
    {
      "epoch": 2.44,
      "learning_rate": 2.0457070504148158e-05,
      "loss": 2.0583,
      "step": 2770
    },
    {
      "epoch": 2.44,
      "learning_rate": 2.0147420195235745e-05,
      "loss": 2.2235,
      "step": 2775
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.9839868585694764e-05,
      "loss": 2.0935,
      "step": 2780
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.9534423758650043e-05,
      "loss": 2.0169,
      "step": 2785
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.923109374185559e-05,
      "loss": 2.1425,
      "step": 2790
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.8929886507483542e-05,
      "loss": 2.0657,
      "step": 2795
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.8630809971914775e-05,
      "loss": 2.0442,
      "step": 2800
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.8333871995530726e-05,
      "loss": 2.0198,
      "step": 2805
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.803908038250689e-05,
      "loss": 2.1117,
      "step": 2810
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.7746442880607627e-05,
      "loss": 2.1318,
      "step": 2815
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.7455967180982592e-05,
      "loss": 2.2086,
      "step": 2820
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.7167660917964556e-05,
      "loss": 2.101,
      "step": 2825
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.6881531668868822e-05,
      "loss": 2.1029,
      "step": 2830
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.6597586953793986e-05,
      "loss": 2.0794,
      "step": 2835
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.6315834235424422e-05,
      "loss": 2.1892,
      "step": 2840
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.6036280918833924e-05,
      "loss": 2.0438,
      "step": 2845
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.5758934351291366e-05,
      "loss": 2.1444,
      "step": 2850
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.5483801822067388e-05,
      "loss": 2.0991,
      "step": 2855
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.521089056224292e-05,
      "loss": 2.1623,
      "step": 2860
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.4940207744519118e-05,
      "loss": 1.9828,
      "step": 2865
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.467176048302875e-05,
      "loss": 2.0164,
      "step": 2870
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.4405555833149398e-05,
      "loss": 1.9968,
      "step": 2875
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.414160079131791e-05,
      "loss": 2.1906,
      "step": 2880
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.3879902294846536e-05,
      "loss": 2.0576,
      "step": 2885
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.362046722174063e-05,
      "loss": 2.1238,
      "step": 2890
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.3363302390517796e-05,
      "loss": 2.0443,
      "step": 2895
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.3108414560028836e-05,
      "loss": 2.1498,
      "step": 2900
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.2855810429279957e-05,
      "loss": 1.9367,
      "step": 2905
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.2605496637256809e-05,
      "loss": 2.0559,
      "step": 2910
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.2357479762749946e-05,
      "loss": 1.9912,
      "step": 2915
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.2111766324181938e-05,
      "loss": 2.0629,
      "step": 2920
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.186836277943606e-05,
      "loss": 2.0062,
      "step": 2925
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.1627275525686543e-05,
      "loss": 2.0624,
      "step": 2930
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.1388510899230465e-05,
      "loss": 2.118,
      "step": 2935
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.115207517532122e-05,
      "loss": 2.0604,
      "step": 2940
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.0917974568003552e-05,
      "loss": 2.0186,
      "step": 2945
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.0686215229950269e-05,
      "loss": 2.2035,
      "step": 2950
    },
    {
      "epoch": 2.6,
      "learning_rate": 1.0456803252300574e-05,
      "loss": 2.0591,
      "step": 2955
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.0229744664499897e-05,
      "loss": 1.9376,
      "step": 2960
    },
    {
      "epoch": 2.61,
      "learning_rate": 1.0005045434141502e-05,
      "loss": 2.2335,
      "step": 2965
    },
    {
      "epoch": 2.62,
      "learning_rate": 9.782711466809536e-06,
      "loss": 2.1306,
      "step": 2970
    },
    {
      "epoch": 2.62,
      "learning_rate": 9.562748605923977e-06,
      "loss": 1.9716,
      "step": 2975
    },
    {
      "epoch": 2.63,
      "learning_rate": 9.345162632586945e-06,
      "loss": 2.0057,
      "step": 2980
    },
    {
      "epoch": 2.63,
      "learning_rate": 9.129959265430766e-06,
      "loss": 2.1155,
      "step": 2985
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.917144160467717e-06,
      "loss": 2.1094,
      "step": 2990
    },
    {
      "epoch": 2.64,
      "learning_rate": 8.706722910941345e-06,
      "loss": 1.9811,
      "step": 2995
    },
    {
      "epoch": 2.64,
      "learning_rate": 8.498701047179491e-06,
      "loss": 2.1029,
      "step": 3000
    },
    {
      "epoch": 2.65,
      "learning_rate": 8.293084036448895e-06,
      "loss": 2.0438,
      "step": 3005
    },
    {
      "epoch": 2.65,
      "learning_rate": 8.089877282811542e-06,
      "loss": 1.9135,
      "step": 3010
    },
    {
      "epoch": 2.66,
      "learning_rate": 7.889086126982658e-06,
      "loss": 2.0472,
      "step": 3015
    },
    {
      "epoch": 2.66,
      "learning_rate": 7.690715846190233e-06,
      "loss": 2.0515,
      "step": 3020
    },
    {
      "epoch": 2.67,
      "learning_rate": 7.4947716540364475e-06,
      "loss": 2.0981,
      "step": 3025
    },
    {
      "epoch": 2.67,
      "learning_rate": 7.301258700360603e-06,
      "loss": 1.9314,
      "step": 3030
    },
    {
      "epoch": 2.67,
      "learning_rate": 7.110182071103766e-06,
      "loss": 2.1015,
      "step": 3035
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.921546788175081e-06,
      "loss": 2.0897,
      "step": 3040
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.735357809319809e-06,
      "loss": 1.9404,
      "step": 3045
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.551620027989036e-06,
      "loss": 2.0471,
      "step": 3050
    },
    {
      "epoch": 2.69,
      "learning_rate": 6.370338273211052e-06,
      "loss": 2.1991,
      "step": 3055
    },
    {
      "epoch": 2.7,
      "learning_rate": 6.191517309464412e-06,
      "loss": 2.1049,
      "step": 3060
    },
    {
      "epoch": 2.7,
      "learning_rate": 6.015161836552763e-06,
      "loss": 2.102,
      "step": 3065
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.84127648948124e-06,
      "loss": 2.106,
      "step": 3070
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.669865838334754e-06,
      "loss": 2.0623,
      "step": 3075
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.500934388157808e-06,
      "loss": 2.0537,
      "step": 3080
    },
    {
      "epoch": 2.72,
      "learning_rate": 5.334486578836118e-06,
      "loss": 2.0713,
      "step": 3085
    },
    {
      "epoch": 2.72,
      "learning_rate": 5.1705267849799126e-06,
      "loss": 2.1706,
      "step": 3090
    },
    {
      "epoch": 2.73,
      "learning_rate": 5.009059315808972e-06,
      "loss": 2.0049,
      "step": 3095
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.850088415039378e-06,
      "loss": 2.1148,
      "step": 3100
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.693618260771937e-06,
      "loss": 2.034,
      "step": 3105
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.539652965382435e-06,
      "loss": 2.0008,
      "step": 3110
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.388196575413483e-06,
      "loss": 2.0006,
      "step": 3115
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.2392530714682745e-06,
      "loss": 2.2408,
      "step": 3120
    },
    {
      "epoch": 2.75,
      "learning_rate": 4.092826368105795e-06,
      "loss": 1.8795,
      "step": 3125
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.948920313738114e-06,
      "loss": 2.1633,
      "step": 3130
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.8075386905291554e-06,
      "loss": 2.1717,
      "step": 3135
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.6686852142952868e-06,
      "loss": 2.0487,
      "step": 3140
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.532363534407712e-06,
      "loss": 2.0741,
      "step": 3145
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.398577233696454e-06,
      "loss": 2.0168,
      "step": 3150
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.2673298283563246e-06,
      "loss": 2.0445,
      "step": 3155
    },
    {
      "epoch": 2.78,
      "learning_rate": 3.1386247678544365e-06,
      "loss": 2.1638,
      "step": 3160
    },
    {
      "epoch": 2.79,
      "learning_rate": 3.012465434839529e-06,
      "loss": 2.135,
      "step": 3165
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.8888551450531266e-06,
      "loss": 1.9357,
      "step": 3170
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.7677971472423013e-06,
      "loss": 2.028,
      "step": 3175
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.6492946230743943e-06,
      "loss": 2.2156,
      "step": 3180
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.5333506870533374e-06,
      "loss": 1.9134,
      "step": 3185
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.4199683864377763e-06,
      "loss": 2.0354,
      "step": 3190
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.3091507011610557e-06,
      "loss": 2.0408,
      "step": 3195
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.200900543752793e-06,
      "loss": 2.2062,
      "step": 3200
    },
    {
      "epoch": 2.82,
      "learning_rate": 2.0952207592624505e-06,
      "loss": 1.9886,
      "step": 3205
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.9921141251844633e-06,
      "loss": 2.1352,
      "step": 3210
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.8915833513852955e-06,
      "loss": 2.1835,
      "step": 3215
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.793631080032232e-06,
      "loss": 1.9694,
      "step": 3220
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.6982598855238564e-06,
      "loss": 2.1328,
      "step": 3225
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.6054722744224593e-06,
      "loss": 2.1902,
      "step": 3230
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.5152706853881704e-06,
      "loss": 1.9398,
      "step": 3235
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.4276574891148087e-06,
      "loss": 2.1243,
      "step": 3240
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.3426349882676325e-06,
      "loss": 2.0591,
      "step": 3245
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.260205417422755e-06,
      "loss": 2.0554,
      "step": 3250
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.1803709430084798e-06,
      "loss": 2.0492,
      "step": 3255
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.1031336632483347e-06,
      "loss": 1.9757,
      "step": 3260
    },
    {
      "epoch": 2.88,
      "learning_rate": 1.0284956081059172e-06,
      "loss": 2.1138,
      "step": 3265
    },
    {
      "epoch": 2.88,
      "learning_rate": 9.564587392315473e-07,
      "loss": 1.9727,
      "step": 3270
    },
    {
      "epoch": 2.89,
      "learning_rate": 8.870249499107419e-07,
      "loss": 2.0748,
      "step": 3275
    },
    {
      "epoch": 2.89,
      "learning_rate": 8.201960650144003e-07,
      "loss": 2.1319,
      "step": 3280
    },
    {
      "epoch": 2.89,
      "learning_rate": 7.559738409508855e-07,
      "loss": 2.2263,
      "step": 3285
    },
    {
      "epoch": 2.9,
      "learning_rate": 6.943599656198507e-07,
      "loss": 2.0016,
      "step": 3290
    },
    {
      "epoch": 2.9,
      "learning_rate": 6.353560583678641e-07,
      "loss": 2.1376,
      "step": 3295
    },
    {
      "epoch": 2.91,
      "learning_rate": 5.789636699458534e-07,
      "loss": 1.9401,
      "step": 3300
    },
    {
      "epoch": 2.91,
      "learning_rate": 5.251842824683716e-07,
      "loss": 2.0978,
      "step": 3305
    },
    {
      "epoch": 2.92,
      "learning_rate": 4.740193093746292e-07,
      "loss": 2.0932,
      "step": 3310
    },
    {
      "epoch": 2.92,
      "learning_rate": 4.2547009539132264e-07,
      "loss": 2.1511,
      "step": 3315
    },
    {
      "epoch": 2.93,
      "learning_rate": 3.7953791649734115e-07,
      "loss": 2.0815,
      "step": 3320
    },
    {
      "epoch": 2.93,
      "learning_rate": 3.362239798901712e-07,
      "loss": 2.0977,
      "step": 3325
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.9552942395423277e-07,
      "loss": 2.0939,
      "step": 3330
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.5745531823090366e-07,
      "loss": 2.0009,
      "step": 3335
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.2200266339045262e-07,
      "loss": 2.064,
      "step": 3340
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.8917239120569418e-07,
      "loss": 2.0858,
      "step": 3345
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.589653645275635e-07,
      "loss": 2.1642,
      "step": 3350
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.313823772623679e-07,
      "loss": 2.062,
      "step": 3355
    },
    {
      "epoch": 2.96,
      "learning_rate": 1.064241543509814e-07,
      "loss": 2.254,
      "step": 3360
    },
    {
      "epoch": 2.96,
      "learning_rate": 8.40913517497377e-08,
      "loss": 2.08,
      "step": 3365
    },
    {
      "epoch": 2.97,
      "learning_rate": 6.438455641324392e-08,
      "loss": 2.0593,
      "step": 3370
    },
    {
      "epoch": 2.97,
      "learning_rate": 4.730428627892636e-08,
      "loss": 1.963,
      "step": 3375
    },
    {
      "epoch": 2.98,
      "learning_rate": 3.2850990253396885e-08,
      "loss": 2.0726,
      "step": 3380
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.102504820069573e-08,
      "loss": 2.0944,
      "step": 3385
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.1826770932288345e-08,
      "loss": 2.207,
      "step": 3390
    },
    {
      "epoch": 2.99,
      "learning_rate": 5.256400198894174e-09,
      "loss": 2.1563,
      "step": 3395
    },
    {
      "epoch": 3.0,
      "learning_rate": 1.3141086841250706e-09,
      "loss": 2.0391,
      "step": 3400
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0,
      "loss": 1.9875,
      "step": 3405
    },
    {
      "epoch": 3.0,
      "step": 3405,
      "total_flos": 1.620805940412416e+16,
      "train_loss": 2.4601817576370575,
      "train_runtime": 30304.9667,
      "train_samples_per_second": 0.898,
      "train_steps_per_second": 0.112
    }
  ],
  "logging_steps": 5,
  "max_steps": 3405,
  "num_train_epochs": 3,
  "save_steps": 100,
  "total_flos": 1.620805940412416e+16,
  "trial_name": null,
  "trial_params": null
}