{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.12655344351919814,
  "eval_steps": 500,
  "global_step": 45000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 2.8122987448710702e-05,
      "grad_norm": 26.75,
      "learning_rate": 1.999999999648735e-05,
      "loss": 9.4046,
      "step": 10
    },
    {
      "epoch": 5.6245974897421405e-05,
      "grad_norm": 30.25,
      "learning_rate": 1.9999999984344847e-05,
      "loss": 9.1867,
      "step": 20
    },
    {
      "epoch": 8.436896234613211e-05,
      "grad_norm": 30.875,
      "learning_rate": 1.999999996352913e-05,
      "loss": 8.6504,
      "step": 30
    },
    {
      "epoch": 0.00011249194979484281,
      "grad_norm": 31.5,
      "learning_rate": 1.9999999934040198e-05,
      "loss": 9.4269,
      "step": 40
    },
    {
      "epoch": 0.0001406149372435535,
      "grad_norm": 62.5,
      "learning_rate": 1.999999989587805e-05,
      "loss": 9.6804,
      "step": 50
    },
    {
      "epoch": 0.00016873792469226421,
      "grad_norm": 30.375,
      "learning_rate": 1.9999999849042684e-05,
      "loss": 9.0471,
      "step": 60
    },
    {
      "epoch": 0.0001968609121409749,
      "grad_norm": 27.375,
      "learning_rate": 1.9999999793534105e-05,
      "loss": 8.7142,
      "step": 70
    },
    {
      "epoch": 0.00022498389958968562,
      "grad_norm": 38.75,
      "learning_rate": 1.999999972935231e-05,
      "loss": 8.8189,
      "step": 80
    },
    {
      "epoch": 0.0002531068870383963,
      "grad_norm": 31.5,
      "learning_rate": 1.99999996564973e-05,
      "loss": 9.1094,
      "step": 90
    },
    {
      "epoch": 0.000281229874487107,
      "grad_norm": 22.375,
      "learning_rate": 1.9999999574969077e-05,
      "loss": 8.7019,
      "step": 100
    },
    {
      "epoch": 0.00030935286193581774,
      "grad_norm": 29.625,
      "learning_rate": 1.9999999484767634e-05,
      "loss": 9.7579,
      "step": 110
    },
    {
      "epoch": 0.00033747584938452843,
      "grad_norm": 29.75,
      "learning_rate": 1.999999938589298e-05,
      "loss": 8.6248,
      "step": 120
    },
    {
      "epoch": 0.0003655988368332391,
      "grad_norm": 31.875,
      "learning_rate": 1.9999999278345105e-05,
      "loss": 8.2482,
      "step": 130
    },
    {
      "epoch": 0.0003937218242819498,
      "grad_norm": 35.5,
      "learning_rate": 1.999999916212402e-05,
      "loss": 8.621,
      "step": 140
    },
    {
      "epoch": 0.00042184481173066055,
      "grad_norm": 37.75,
      "learning_rate": 1.999999903722972e-05,
      "loss": 9.3644,
      "step": 150
    },
    {
      "epoch": 0.00044996779917937124,
      "grad_norm": 50.75,
      "learning_rate": 1.9999998903662202e-05,
      "loss": 8.8872,
      "step": 160
    },
    {
      "epoch": 0.0004780907866280819,
      "grad_norm": 27.625,
      "learning_rate": 1.9999998761421473e-05,
      "loss": 9.3207,
      "step": 170
    },
    {
      "epoch": 0.0005062137740767926,
      "grad_norm": 45.5,
      "learning_rate": 1.999999861050753e-05,
      "loss": 8.9391,
      "step": 180
    },
    {
      "epoch": 0.0005343367615255033,
      "grad_norm": 41.75,
      "learning_rate": 1.999999845092037e-05,
      "loss": 7.9287,
      "step": 190
    },
    {
      "epoch": 0.000562459748974214,
      "grad_norm": 46.0,
      "learning_rate": 1.9999998282659996e-05,
      "loss": 8.0963,
      "step": 200
    },
    {
      "epoch": 0.0005905827364229248,
      "grad_norm": 33.75,
      "learning_rate": 1.999999810572641e-05,
      "loss": 8.2999,
      "step": 210
    },
    {
      "epoch": 0.0006187057238716355,
      "grad_norm": 24.625,
      "learning_rate": 1.999999792011961e-05,
      "loss": 9.307,
      "step": 220
    },
    {
      "epoch": 0.0006468287113203462,
      "grad_norm": 63.0,
      "learning_rate": 1.9999997725839592e-05,
      "loss": 9.2622,
      "step": 230
    },
    {
      "epoch": 0.0006749516987690569,
      "grad_norm": 27.875,
      "learning_rate": 1.9999997522886364e-05,
      "loss": 8.0718,
      "step": 240
    },
    {
      "epoch": 0.0007030746862177675,
      "grad_norm": 61.5,
      "learning_rate": 1.9999997311259922e-05,
      "loss": 8.9239,
      "step": 250
    },
    {
      "epoch": 0.0007311976736664782,
      "grad_norm": 24.125,
      "learning_rate": 1.9999997090960266e-05,
      "loss": 8.4316,
      "step": 260
    },
    {
      "epoch": 0.0007593206611151889,
      "grad_norm": 36.75,
      "learning_rate": 1.9999996861987396e-05,
      "loss": 7.9428,
      "step": 270
    },
    {
      "epoch": 0.0007874436485638996,
      "grad_norm": 62.75,
      "learning_rate": 1.9999996624341315e-05,
      "loss": 8.4613,
      "step": 280
    },
    {
      "epoch": 0.0008155666360126103,
      "grad_norm": 26.75,
      "learning_rate": 1.9999996378022023e-05,
      "loss": 9.8768,
      "step": 290
    },
    {
      "epoch": 0.0008436896234613211,
      "grad_norm": 26.125,
      "learning_rate": 1.9999996123029514e-05,
      "loss": 8.7369,
      "step": 300
    },
    {
      "epoch": 0.0008718126109100318,
      "grad_norm": 53.5,
      "learning_rate": 1.9999995859363797e-05,
      "loss": 8.1794,
      "step": 310
    },
    {
      "epoch": 0.0008999355983587425,
      "grad_norm": 63.75,
      "learning_rate": 1.9999995587024866e-05,
      "loss": 8.6823,
      "step": 320
    },
    {
      "epoch": 0.0009280585858074532,
      "grad_norm": 47.5,
      "learning_rate": 1.9999995306012724e-05,
      "loss": 8.809,
      "step": 330
    },
    {
      "epoch": 0.0009561815732561639,
      "grad_norm": 27.125,
      "learning_rate": 1.9999995016327368e-05,
      "loss": 8.9519,
      "step": 340
    },
    {
      "epoch": 0.0009843045607048745,
      "grad_norm": 46.0,
      "learning_rate": 1.99999947179688e-05,
      "loss": 9.0159,
      "step": 350
    },
    {
      "epoch": 0.0010124275481535852,
      "grad_norm": 44.5,
      "learning_rate": 1.9999994410937027e-05,
      "loss": 8.0437,
      "step": 360
    },
    {
      "epoch": 0.001040550535602296,
      "grad_norm": 58.25,
      "learning_rate": 1.9999994095232038e-05,
      "loss": 7.9795,
      "step": 370
    },
    {
      "epoch": 0.0010686735230510066,
      "grad_norm": 46.75,
      "learning_rate": 1.9999993770853842e-05,
      "loss": 8.9127,
      "step": 380
    },
    {
      "epoch": 0.0010967965104997173,
      "grad_norm": 43.25,
      "learning_rate": 1.9999993437802436e-05,
      "loss": 8.5466,
      "step": 390
    },
    {
      "epoch": 0.001124919497948428,
      "grad_norm": 22.375,
      "learning_rate": 1.9999993096077818e-05,
      "loss": 8.608,
      "step": 400
    },
    {
      "epoch": 0.0011530424853971387,
      "grad_norm": 32.5,
      "learning_rate": 1.999999274567999e-05,
      "loss": 8.3159,
      "step": 410
    },
    {
      "epoch": 0.0011811654728458496,
      "grad_norm": 36.0,
      "learning_rate": 1.9999992386608954e-05,
      "loss": 8.198,
      "step": 420
    },
    {
      "epoch": 0.0012092884602945603,
      "grad_norm": 35.75,
      "learning_rate": 1.9999992018864708e-05,
      "loss": 8.959,
      "step": 430
    },
    {
      "epoch": 0.001237411447743271,
      "grad_norm": 28.25,
      "learning_rate": 1.9999991642447254e-05,
      "loss": 8.2303,
      "step": 440
    },
    {
      "epoch": 0.0012655344351919816,
      "grad_norm": 27.375,
      "learning_rate": 1.999999125735659e-05,
      "loss": 8.5069,
      "step": 450
    },
    {
      "epoch": 0.0012936574226406923,
      "grad_norm": 39.75,
      "learning_rate": 1.999999086359272e-05,
      "loss": 7.9837,
      "step": 460
    },
    {
      "epoch": 0.001321780410089403,
      "grad_norm": 52.0,
      "learning_rate": 1.999999046115564e-05,
      "loss": 8.5483,
      "step": 470
    },
    {
      "epoch": 0.0013499033975381137,
      "grad_norm": 41.75,
      "learning_rate": 1.9999990050045354e-05,
      "loss": 8.9188,
      "step": 480
    },
    {
      "epoch": 0.0013780263849868244,
      "grad_norm": 27.375,
      "learning_rate": 1.9999989630261863e-05,
      "loss": 8.6146,
      "step": 490
    },
    {
      "epoch": 0.001406149372435535,
      "grad_norm": 29.0,
      "learning_rate": 1.9999989201805164e-05,
      "loss": 8.4098,
      "step": 500
    },
    {
      "epoch": 0.0014342723598842458,
      "grad_norm": 24.25,
      "learning_rate": 1.999998876467526e-05,
      "loss": 8.5872,
      "step": 510
    },
    {
      "epoch": 0.0014623953473329565,
      "grad_norm": 26.375,
      "learning_rate": 1.9999988318872145e-05,
      "loss": 8.7304,
      "step": 520
    },
    {
      "epoch": 0.0014905183347816672,
      "grad_norm": 22.75,
      "learning_rate": 1.999998786439583e-05,
      "loss": 8.2715,
      "step": 530
    },
    {
      "epoch": 0.0015186413222303778,
      "grad_norm": 32.25,
      "learning_rate": 1.9999987401246306e-05,
      "loss": 8.472,
      "step": 540
    },
    {
      "epoch": 0.0015467643096790885,
      "grad_norm": 21.75,
      "learning_rate": 1.999998692942358e-05,
      "loss": 7.9425,
      "step": 550
    },
    {
      "epoch": 0.0015748872971277992,
      "grad_norm": 37.75,
      "learning_rate": 1.9999986448927652e-05,
      "loss": 9.0206,
      "step": 560
    },
    {
      "epoch": 0.00160301028457651,
      "grad_norm": 39.5,
      "learning_rate": 1.9999985959758515e-05,
      "loss": 8.0803,
      "step": 570
    },
    {
      "epoch": 0.0016311332720252206,
      "grad_norm": 37.25,
      "learning_rate": 1.9999985461916178e-05,
      "loss": 8.3754,
      "step": 580
    },
    {
      "epoch": 0.0016592562594739315,
      "grad_norm": 24.25,
      "learning_rate": 1.9999984955400636e-05,
      "loss": 9.2416,
      "step": 590
    },
    {
      "epoch": 0.0016873792469226422,
      "grad_norm": 42.5,
      "learning_rate": 1.9999984440211894e-05,
      "loss": 8.3339,
      "step": 600
    },
    {
      "epoch": 0.0017155022343713529,
      "grad_norm": 35.25,
      "learning_rate": 1.9999983916349948e-05,
      "loss": 8.1455,
      "step": 610
    },
    {
      "epoch": 0.0017436252218200636,
      "grad_norm": 51.0,
      "learning_rate": 1.99999833838148e-05,
      "loss": 7.6625,
      "step": 620
    },
    {
      "epoch": 0.0017717482092687743,
      "grad_norm": 34.5,
      "learning_rate": 1.999998284260645e-05,
      "loss": 8.1291,
      "step": 630
    },
    {
      "epoch": 0.001799871196717485,
      "grad_norm": 56.5,
      "learning_rate": 1.9999982292724902e-05,
      "loss": 8.5348,
      "step": 640
    },
    {
      "epoch": 0.0018279941841661956,
      "grad_norm": 27.5,
      "learning_rate": 1.9999981734170154e-05,
      "loss": 7.475,
      "step": 650
    },
    {
      "epoch": 0.0018561171716149063,
      "grad_norm": 25.0,
      "learning_rate": 1.9999981166942204e-05,
      "loss": 9.086,
      "step": 660
    },
    {
      "epoch": 0.001884240159063617,
      "grad_norm": 47.25,
      "learning_rate": 1.9999980591041054e-05,
      "loss": 8.1769,
      "step": 670
    },
    {
      "epoch": 0.0019123631465123277,
      "grad_norm": 44.5,
      "learning_rate": 1.999998000646671e-05,
      "loss": 8.5343,
      "step": 680
    },
    {
      "epoch": 0.0019404861339610384,
      "grad_norm": 26.375,
      "learning_rate": 1.9999979413219163e-05,
      "loss": 8.3249,
      "step": 690
    },
    {
      "epoch": 0.001968609121409749,
      "grad_norm": 29.5,
      "learning_rate": 1.999997881129842e-05,
      "loss": 8.9132,
      "step": 700
    },
    {
      "epoch": 0.0019967321088584598,
      "grad_norm": 31.25,
      "learning_rate": 1.999997820070448e-05,
      "loss": 8.0182,
      "step": 710
    },
    {
      "epoch": 0.0020248550963071705,
      "grad_norm": 29.625,
      "learning_rate": 1.9999977581437342e-05,
      "loss": 8.1667,
      "step": 720
    },
    {
      "epoch": 0.002052978083755881,
      "grad_norm": 27.5,
      "learning_rate": 1.9999976953497012e-05,
      "loss": 8.2431,
      "step": 730
    },
    {
      "epoch": 0.002081101071204592,
      "grad_norm": 68.5,
      "learning_rate": 1.999997631688348e-05,
      "loss": 9.0525,
      "step": 740
    },
    {
      "epoch": 0.0021092240586533025,
      "grad_norm": 30.375,
      "learning_rate": 1.9999975671596757e-05,
      "loss": 7.8491,
      "step": 750
    },
    {
      "epoch": 0.002137347046102013,
      "grad_norm": 31.5,
      "learning_rate": 1.9999975017636835e-05,
      "loss": 8.1601,
      "step": 760
    },
    {
      "epoch": 0.002165470033550724,
      "grad_norm": 38.25,
      "learning_rate": 1.9999974355003726e-05,
      "loss": 8.5655,
      "step": 770
    },
    {
      "epoch": 0.0021935930209994346,
      "grad_norm": 51.25,
      "learning_rate": 1.999997368369742e-05,
      "loss": 8.4149,
      "step": 780
    },
    {
      "epoch": 0.0022217160084481453,
      "grad_norm": 33.25,
      "learning_rate": 1.999997300371792e-05,
      "loss": 8.2946,
      "step": 790
    },
    {
      "epoch": 0.002249838995896856,
      "grad_norm": 34.0,
      "learning_rate": 1.999997231506523e-05,
      "loss": 7.6409,
      "step": 800
    },
    {
      "epoch": 0.0022779619833455667,
      "grad_norm": 27.25,
      "learning_rate": 1.999997161773935e-05,
      "loss": 8.7731,
      "step": 810
    },
    {
      "epoch": 0.0023060849707942773,
      "grad_norm": 25.625,
      "learning_rate": 1.9999970911740277e-05,
      "loss": 8.0884,
      "step": 820
    },
    {
      "epoch": 0.002334207958242988,
      "grad_norm": 45.0,
      "learning_rate": 1.9999970197068012e-05,
      "loss": 8.8899,
      "step": 830
    },
    {
      "epoch": 0.002362330945691699,
      "grad_norm": 42.0,
      "learning_rate": 1.999996947372256e-05,
      "loss": 8.7895,
      "step": 840
    },
    {
      "epoch": 0.00239045393314041,
      "grad_norm": 33.5,
      "learning_rate": 1.9999968741703917e-05,
      "loss": 8.2149,
      "step": 850
    },
    {
      "epoch": 0.0024185769205891205,
      "grad_norm": 30.75,
      "learning_rate": 1.9999968001012083e-05,
      "loss": 8.2464,
      "step": 860
    },
    {
      "epoch": 0.0024466999080378312,
      "grad_norm": 37.75,
      "learning_rate": 1.9999967251647066e-05,
      "loss": 8.2323,
      "step": 870
    },
    {
      "epoch": 0.002474822895486542,
      "grad_norm": 34.5,
      "learning_rate": 1.999996649360886e-05,
      "loss": 8.3075,
      "step": 880
    },
    {
      "epoch": 0.0025029458829352526,
      "grad_norm": 42.75,
      "learning_rate": 1.9999965726897467e-05,
      "loss": 7.8954,
      "step": 890
    },
    {
      "epoch": 0.0025310688703839633,
      "grad_norm": 22.75,
      "learning_rate": 1.999996495151289e-05,
      "loss": 8.7934,
      "step": 900
    },
    {
      "epoch": 0.002559191857832674,
      "grad_norm": 28.5,
      "learning_rate": 1.9999964167455127e-05,
      "loss": 7.965,
      "step": 910
    },
    {
      "epoch": 0.0025873148452813847,
      "grad_norm": 25.25,
      "learning_rate": 1.999996337472418e-05,
      "loss": 8.4655,
      "step": 920
    },
    {
      "epoch": 0.0026154378327300954,
      "grad_norm": 28.375,
      "learning_rate": 1.9999962573320045e-05,
      "loss": 7.8498,
      "step": 930
    },
    {
      "epoch": 0.002643560820178806,
      "grad_norm": 45.25,
      "learning_rate": 1.9999961763242732e-05,
      "loss": 7.3685,
      "step": 940
    },
    {
      "epoch": 0.0026716838076275167,
      "grad_norm": 39.25,
      "learning_rate": 1.9999960944492235e-05,
      "loss": 8.991,
      "step": 950
    },
    {
      "epoch": 0.0026998067950762274,
      "grad_norm": 33.5,
      "learning_rate": 1.9999960117068555e-05,
      "loss": 8.6423,
      "step": 960
    },
    {
      "epoch": 0.002727929782524938,
      "grad_norm": 37.0,
      "learning_rate": 1.9999959280971697e-05,
      "loss": 8.3628,
      "step": 970
    },
    {
      "epoch": 0.002756052769973649,
      "grad_norm": 29.25,
      "learning_rate": 1.9999958436201656e-05,
      "loss": 8.3249,
      "step": 980
    },
    {
      "epoch": 0.0027841757574223595,
      "grad_norm": 29.125,
      "learning_rate": 1.9999957582758435e-05,
      "loss": 8.2939,
      "step": 990
    },
    {
      "epoch": 0.00281229874487107,
      "grad_norm": 40.5,
      "learning_rate": 1.9999956720642036e-05,
      "loss": 8.7847,
      "step": 1000
    },
    {
      "epoch": 0.002840421732319781,
      "grad_norm": 25.0,
      "learning_rate": 1.999995584985246e-05,
      "loss": 8.669,
      "step": 1010
    },
    {
      "epoch": 0.0028685447197684916,
      "grad_norm": 37.0,
      "learning_rate": 1.9999954970389705e-05,
      "loss": 8.0318,
      "step": 1020
    },
    {
      "epoch": 0.0028966677072172022,
      "grad_norm": 27.125,
      "learning_rate": 1.9999954082253773e-05,
      "loss": 8.5188,
      "step": 1030
    },
    {
      "epoch": 0.002924790694665913,
      "grad_norm": 32.5,
      "learning_rate": 1.9999953185444667e-05,
      "loss": 9.3141,
      "step": 1040
    },
    {
      "epoch": 0.0029529136821146236,
      "grad_norm": 46.75,
      "learning_rate": 1.9999952279962387e-05,
      "loss": 8.1453,
      "step": 1050
    },
    {
      "epoch": 0.0029810366695633343,
      "grad_norm": 53.5,
      "learning_rate": 1.999995136580693e-05,
      "loss": 8.3799,
      "step": 1060
    },
    {
      "epoch": 0.003009159657012045,
      "grad_norm": 43.75,
      "learning_rate": 1.99999504429783e-05,
      "loss": 7.6179,
      "step": 1070
    },
    {
      "epoch": 0.0030372826444607557,
      "grad_norm": 44.5,
      "learning_rate": 1.99999495114765e-05,
      "loss": 7.5147,
      "step": 1080
    },
    {
      "epoch": 0.0030654056319094664,
      "grad_norm": 32.75,
      "learning_rate": 1.999994857130153e-05,
      "loss": 8.6252,
      "step": 1090
    },
    {
      "epoch": 0.003093528619358177,
      "grad_norm": 32.5,
      "learning_rate": 1.9999947622453383e-05,
      "loss": 8.4846,
      "step": 1100
    },
    {
      "epoch": 0.0031216516068068877,
      "grad_norm": 31.375,
      "learning_rate": 1.9999946664932065e-05,
      "loss": 8.3329,
      "step": 1110
    },
    {
      "epoch": 0.0031497745942555984,
      "grad_norm": 31.5,
      "learning_rate": 1.9999945698737585e-05,
      "loss": 7.5005,
      "step": 1120
    },
    {
      "epoch": 0.003177897581704309,
      "grad_norm": 29.0,
      "learning_rate": 1.9999944723869934e-05,
      "loss": 8.5725,
      "step": 1130
    },
    {
      "epoch": 0.00320602056915302,
      "grad_norm": 42.75,
      "learning_rate": 1.999994374032911e-05,
      "loss": 9.1367,
      "step": 1140
    },
    {
      "epoch": 0.0032341435566017305,
      "grad_norm": 34.75,
      "learning_rate": 1.9999942748115125e-05,
      "loss": 8.0121,
      "step": 1150
    },
    {
      "epoch": 0.003262266544050441,
      "grad_norm": 27.0,
      "learning_rate": 1.999994174722797e-05,
      "loss": 8.2077,
      "step": 1160
    },
    {
      "epoch": 0.003290389531499152,
      "grad_norm": 30.75,
      "learning_rate": 1.9999940737667652e-05,
      "loss": 9.0612,
      "step": 1170
    },
    {
      "epoch": 0.003318512518947863,
      "grad_norm": 36.75,
      "learning_rate": 1.999993971943417e-05,
      "loss": 8.223,
      "step": 1180
    },
    {
      "epoch": 0.0033466355063965737,
      "grad_norm": 39.0,
      "learning_rate": 1.9999938692527526e-05,
      "loss": 7.9979,
      "step": 1190
    },
    {
      "epoch": 0.0033747584938452844,
      "grad_norm": 33.0,
      "learning_rate": 1.9999937656947718e-05,
      "loss": 8.5902,
      "step": 1200
    },
    {
      "epoch": 0.003402881481293995,
      "grad_norm": 21.75,
      "learning_rate": 1.9999936612694747e-05,
      "loss": 7.9853,
      "step": 1210
    },
    {
      "epoch": 0.0034310044687427058,
      "grad_norm": 28.375,
      "learning_rate": 1.999993555976862e-05,
      "loss": 8.129,
      "step": 1220
    },
    {
      "epoch": 0.0034591274561914164,
      "grad_norm": 34.0,
      "learning_rate": 1.999993449816933e-05,
      "loss": 7.971,
      "step": 1230
    },
    {
      "epoch": 0.003487250443640127,
      "grad_norm": 26.875,
      "learning_rate": 1.999993342789688e-05,
      "loss": 8.9445,
      "step": 1240
    },
    {
      "epoch": 0.003515373431088838,
      "grad_norm": 30.75,
      "learning_rate": 1.9999932348951275e-05,
      "loss": 7.3252,
      "step": 1250
    },
    {
      "epoch": 0.0035434964185375485,
      "grad_norm": 34.0,
      "learning_rate": 1.999993126133251e-05,
      "loss": 8.2313,
      "step": 1260
    },
    {
      "epoch": 0.003571619405986259,
      "grad_norm": 40.5,
      "learning_rate": 1.9999930165040592e-05,
      "loss": 8.6545,
      "step": 1270
    },
    {
      "epoch": 0.00359974239343497,
      "grad_norm": 30.875,
      "learning_rate": 1.9999929060075518e-05,
      "loss": 7.8905,
      "step": 1280
    },
    {
      "epoch": 0.0036278653808836806,
      "grad_norm": 27.5,
      "learning_rate": 1.999992794643729e-05,
      "loss": 7.475,
      "step": 1290
    },
    {
      "epoch": 0.0036559883683323913,
      "grad_norm": 30.25,
      "learning_rate": 1.999992682412591e-05,
      "loss": 8.5303,
      "step": 1300
    },
    {
      "epoch": 0.003684111355781102,
      "grad_norm": 31.375,
      "learning_rate": 1.9999925693141375e-05,
      "loss": 8.0424,
      "step": 1310
    },
    {
      "epoch": 0.0037122343432298126,
      "grad_norm": 25.125,
      "learning_rate": 1.9999924553483688e-05,
      "loss": 8.4007,
      "step": 1320
    },
    {
      "epoch": 0.0037403573306785233,
      "grad_norm": 35.5,
      "learning_rate": 1.9999923405152855e-05,
      "loss": 8.7585,
      "step": 1330
    },
    {
      "epoch": 0.003768480318127234,
      "grad_norm": 30.875,
      "learning_rate": 1.999992224814887e-05,
      "loss": 8.2205,
      "step": 1340
    },
    {
      "epoch": 0.0037966033055759447,
      "grad_norm": 36.25,
      "learning_rate": 1.9999921082471738e-05,
      "loss": 7.3258,
      "step": 1350
    },
    {
      "epoch": 0.0038247262930246554,
      "grad_norm": 71.0,
      "learning_rate": 1.9999919908121458e-05,
      "loss": 8.4646,
      "step": 1360
    },
    {
      "epoch": 0.003852849280473366,
      "grad_norm": 23.375,
      "learning_rate": 1.9999918725098032e-05,
      "loss": 8.2898,
      "step": 1370
    },
    {
      "epoch": 0.0038809722679220768,
      "grad_norm": 40.5,
      "learning_rate": 1.9999917533401466e-05,
      "loss": 8.3837,
      "step": 1380
    },
    {
      "epoch": 0.0039090952553707875,
      "grad_norm": 20.75,
      "learning_rate": 1.999991633303175e-05,
      "loss": 8.3823,
      "step": 1390
    },
    {
      "epoch": 0.003937218242819498,
      "grad_norm": 28.5,
      "learning_rate": 1.9999915123988892e-05,
      "loss": 7.7302,
      "step": 1400
    },
    {
      "epoch": 0.003965341230268209,
      "grad_norm": 38.75,
      "learning_rate": 1.9999913906272894e-05,
      "loss": 9.2115,
      "step": 1410
    },
    {
      "epoch": 0.0039934642177169195,
      "grad_norm": 36.75,
      "learning_rate": 1.999991267988375e-05,
      "loss": 8.3871,
      "step": 1420
    },
    {
      "epoch": 0.00402158720516563,
      "grad_norm": 47.0,
      "learning_rate": 1.9999911444821468e-05,
      "loss": 8.1022,
      "step": 1430
    },
    {
      "epoch": 0.004049710192614341,
      "grad_norm": 37.0,
      "learning_rate": 1.9999910201086054e-05,
      "loss": 8.6464,
      "step": 1440
    },
    {
      "epoch": 0.004077833180063052,
      "grad_norm": 35.0,
      "learning_rate": 1.9999908948677494e-05,
      "loss": 8.2938,
      "step": 1450
    },
    {
      "epoch": 0.004105956167511762,
      "grad_norm": 29.0,
      "learning_rate": 1.99999076875958e-05,
      "loss": 7.6254,
      "step": 1460
    },
    {
      "epoch": 0.004134079154960473,
      "grad_norm": 36.0,
      "learning_rate": 1.999990641784097e-05,
      "loss": 7.8324,
      "step": 1470
    },
    {
      "epoch": 0.004162202142409184,
      "grad_norm": 36.0,
      "learning_rate": 1.999990513941301e-05,
      "loss": 7.6072,
      "step": 1480
    },
    {
      "epoch": 0.004190325129857894,
      "grad_norm": 58.0,
      "learning_rate": 1.999990385231191e-05,
      "loss": 8.1061,
      "step": 1490
    },
    {
      "epoch": 0.004218448117306605,
      "grad_norm": 23.5,
      "learning_rate": 1.9999902556537682e-05,
      "loss": 8.0443,
      "step": 1500
    },
    {
      "epoch": 0.004246571104755316,
      "grad_norm": 56.5,
      "learning_rate": 1.999990125209032e-05,
      "loss": 7.6882,
      "step": 1510
    },
    {
      "epoch": 0.004274694092204026,
      "grad_norm": 22.375,
      "learning_rate": 1.9999899938969834e-05,
      "loss": 8.3186,
      "step": 1520
    },
    {
      "epoch": 0.004302817079652737,
      "grad_norm": 37.75,
      "learning_rate": 1.9999898617176214e-05,
      "loss": 8.4319,
      "step": 1530
    },
    {
      "epoch": 0.004330940067101448,
      "grad_norm": 48.75,
      "learning_rate": 1.9999897286709465e-05,
      "loss": 7.6039,
      "step": 1540
    },
    {
      "epoch": 0.0043590630545501585,
      "grad_norm": 32.25,
      "learning_rate": 1.9999895947569593e-05,
      "loss": 8.9144,
      "step": 1550
    },
    {
      "epoch": 0.004387186041998869,
      "grad_norm": 42.0,
      "learning_rate": 1.9999894599756598e-05,
      "loss": 7.8979,
      "step": 1560
    },
    {
      "epoch": 0.00441530902944758,
      "grad_norm": 41.0,
      "learning_rate": 1.9999893243270474e-05,
      "loss": 7.9144,
      "step": 1570
    },
    {
      "epoch": 0.0044434320168962906,
      "grad_norm": 24.75,
      "learning_rate": 1.999989187811123e-05,
      "loss": 7.8997,
      "step": 1580
    },
    {
      "epoch": 0.004471555004345001,
      "grad_norm": 47.75,
      "learning_rate": 1.999989050427886e-05,
      "loss": 8.5669,
      "step": 1590
    },
    {
      "epoch": 0.004499677991793712,
      "grad_norm": 26.75,
      "learning_rate": 1.999988912177337e-05,
      "loss": 8.1317,
      "step": 1600
    },
    {
      "epoch": 0.004527800979242423,
      "grad_norm": 23.625,
      "learning_rate": 1.9999887730594766e-05,
      "loss": 8.7139,
      "step": 1610
    },
    {
      "epoch": 0.004555923966691133,
      "grad_norm": 36.5,
      "learning_rate": 1.999988633074304e-05,
      "loss": 8.8209,
      "step": 1620
    },
    {
      "epoch": 0.004584046954139844,
      "grad_norm": 24.125,
      "learning_rate": 1.9999884922218195e-05,
      "loss": 8.7102,
      "step": 1630
    },
    {
      "epoch": 0.004612169941588555,
      "grad_norm": 32.25,
      "learning_rate": 1.999988350502024e-05,
      "loss": 8.3722,
      "step": 1640
    },
    {
      "epoch": 0.004640292929037265,
      "grad_norm": 23.5,
      "learning_rate": 1.9999882079149166e-05,
      "loss": 8.12,
      "step": 1650
    },
    {
      "epoch": 0.004668415916485976,
      "grad_norm": 38.5,
      "learning_rate": 1.999988064460498e-05,
      "loss": 7.867,
      "step": 1660
    },
    {
      "epoch": 0.004696538903934687,
      "grad_norm": 42.25,
      "learning_rate": 1.999987920138768e-05,
      "loss": 8.528,
      "step": 1670
    },
    {
      "epoch": 0.004724661891383398,
      "grad_norm": 37.5,
      "learning_rate": 1.999987774949727e-05,
      "loss": 8.0916,
      "step": 1680
    },
    {
      "epoch": 0.004752784878832109,
      "grad_norm": 30.625,
      "learning_rate": 1.9999876288933752e-05,
      "loss": 7.4676,
      "step": 1690
    },
    {
      "epoch": 0.00478090786628082,
      "grad_norm": 26.0,
      "learning_rate": 1.9999874819697128e-05,
      "loss": 7.8035,
      "step": 1700
    },
    {
      "epoch": 0.00480903085372953,
      "grad_norm": 22.25,
      "learning_rate": 1.999987334178739e-05,
      "loss": 7.8775,
      "step": 1710
    },
    {
      "epoch": 0.004837153841178241,
      "grad_norm": 25.75,
      "learning_rate": 1.999987185520455e-05,
      "loss": 9.0641,
      "step": 1720
    },
    {
      "epoch": 0.004865276828626952,
      "grad_norm": 67.5,
      "learning_rate": 1.999987035994861e-05,
      "loss": 8.6583,
      "step": 1730
    },
    {
      "epoch": 0.0048933998160756624,
      "grad_norm": 34.5,
      "learning_rate": 1.999986885601956e-05,
      "loss": 7.6526,
      "step": 1740
    },
    {
      "epoch": 0.004921522803524373,
      "grad_norm": 38.0,
      "learning_rate": 1.999986734341741e-05,
      "loss": 8.4408,
      "step": 1750
    },
    {
      "epoch": 0.004949645790973084,
      "grad_norm": 23.875,
      "learning_rate": 1.999986582214216e-05,
      "loss": 8.9989,
      "step": 1760
    },
    {
      "epoch": 0.0049777687784217945,
      "grad_norm": 28.25,
      "learning_rate": 1.9999864292193813e-05,
      "loss": 8.1971,
      "step": 1770
    },
    {
      "epoch": 0.005005891765870505,
      "grad_norm": 29.875,
      "learning_rate": 1.9999862753572366e-05,
      "loss": 7.9196,
      "step": 1780
    },
    {
      "epoch": 0.005034014753319216,
      "grad_norm": 34.75,
      "learning_rate": 1.9999861206277823e-05,
      "loss": 7.9966,
      "step": 1790
    },
    {
      "epoch": 0.005062137740767927,
      "grad_norm": 32.25,
      "learning_rate": 1.999985965031018e-05,
      "loss": 8.5606,
      "step": 1800
    },
    {
      "epoch": 0.005090260728216637,
      "grad_norm": 21.375,
      "learning_rate": 1.999985808566945e-05,
      "loss": 8.1658,
      "step": 1810
    },
    {
      "epoch": 0.005118383715665348,
      "grad_norm": 36.0,
      "learning_rate": 1.9999856512355628e-05,
      "loss": 8.6702,
      "step": 1820
    },
    {
      "epoch": 0.005146506703114059,
      "grad_norm": 38.0,
      "learning_rate": 1.999985493036871e-05,
      "loss": 7.7636,
      "step": 1830
    },
    {
      "epoch": 0.005174629690562769,
      "grad_norm": 27.875,
      "learning_rate": 1.9999853339708702e-05,
      "loss": 8.1655,
      "step": 1840
    },
    {
      "epoch": 0.00520275267801148,
      "grad_norm": 36.25,
      "learning_rate": 1.999985174037561e-05,
      "loss": 8.1779,
      "step": 1850
    },
    {
      "epoch": 0.005230875665460191,
      "grad_norm": 30.75,
      "learning_rate": 1.9999850132369428e-05,
      "loss": 8.1862,
      "step": 1860
    },
    {
      "epoch": 0.005258998652908901,
      "grad_norm": 23.875,
      "learning_rate": 1.999984851569016e-05,
      "loss": 8.5442,
      "step": 1870
    },
    {
      "epoch": 0.005287121640357612,
      "grad_norm": 24.125,
      "learning_rate": 1.999984689033781e-05,
      "loss": 8.0296,
      "step": 1880
    },
    {
      "epoch": 0.005315244627806323,
      "grad_norm": 44.75,
      "learning_rate": 1.9999845256312374e-05,
      "loss": 7.9315,
      "step": 1890
    },
    {
      "epoch": 0.0053433676152550335,
      "grad_norm": 28.25,
      "learning_rate": 1.9999843613613858e-05,
      "loss": 8.4237,
      "step": 1900
    },
    {
      "epoch": 0.005371490602703744,
      "grad_norm": 29.375,
      "learning_rate": 1.9999841962242262e-05,
      "loss": 9.3751,
      "step": 1910
    },
    {
      "epoch": 0.005399613590152455,
      "grad_norm": 40.0,
      "learning_rate": 1.9999840302197592e-05,
      "loss": 8.5878,
      "step": 1920
    },
    {
      "epoch": 0.0054277365776011655,
      "grad_norm": 29.375,
      "learning_rate": 1.999983863347984e-05,
      "loss": 7.7537,
      "step": 1930
    },
    {
      "epoch": 0.005455859565049876,
      "grad_norm": 27.625,
      "learning_rate": 1.9999836956089015e-05,
      "loss": 8.2422,
      "step": 1940
    },
    {
      "epoch": 0.005483982552498587,
      "grad_norm": 45.25,
      "learning_rate": 1.9999835270025115e-05,
      "loss": 8.5899,
      "step": 1950
    },
    {
      "epoch": 0.005512105539947298,
      "grad_norm": 24.75,
      "learning_rate": 1.999983357528814e-05,
      "loss": 8.1926,
      "step": 1960
    },
    {
      "epoch": 0.005540228527396008,
      "grad_norm": 52.0,
      "learning_rate": 1.9999831871878093e-05,
      "loss": 8.2475,
      "step": 1970
    },
    {
      "epoch": 0.005568351514844719,
      "grad_norm": 33.5,
      "learning_rate": 1.9999830159794978e-05,
      "loss": 8.4705,
      "step": 1980
    },
    {
      "epoch": 0.00559647450229343,
      "grad_norm": 33.0,
      "learning_rate": 1.9999828439038797e-05,
      "loss": 7.8861,
      "step": 1990
    },
    {
      "epoch": 0.00562459748974214,
      "grad_norm": 35.75,
      "learning_rate": 1.9999826709609545e-05,
      "loss": 8.3294,
      "step": 2000
    },
    {
      "epoch": 0.005652720477190851,
      "grad_norm": 24.25,
      "learning_rate": 1.999982497150723e-05,
      "loss": 7.7923,
      "step": 2010
    },
    {
      "epoch": 0.005680843464639562,
      "grad_norm": 31.875,
      "learning_rate": 1.9999823224731855e-05,
      "loss": 7.5634,
      "step": 2020
    },
    {
      "epoch": 0.005708966452088272,
      "grad_norm": 48.0,
      "learning_rate": 1.9999821469283414e-05,
      "loss": 8.2047,
      "step": 2030
    },
    {
      "epoch": 0.005737089439536983,
      "grad_norm": 26.875,
      "learning_rate": 1.999981970516191e-05,
      "loss": 7.9333,
      "step": 2040
    },
    {
      "epoch": 0.005765212426985694,
      "grad_norm": 35.75,
      "learning_rate": 1.999981793236735e-05,
      "loss": 8.4141,
      "step": 2050
    },
    {
      "epoch": 0.0057933354144344045,
      "grad_norm": 36.5,
      "learning_rate": 1.999981615089973e-05,
      "loss": 8.2208,
      "step": 2060
    },
    {
      "epoch": 0.005821458401883115,
      "grad_norm": 37.0,
      "learning_rate": 1.9999814360759057e-05,
      "loss": 8.2442,
      "step": 2070
    },
    {
      "epoch": 0.005849581389331826,
      "grad_norm": 33.75,
      "learning_rate": 1.9999812561945327e-05,
      "loss": 8.892,
      "step": 2080
    },
    {
      "epoch": 0.0058777043767805365,
      "grad_norm": 28.375,
      "learning_rate": 1.9999810754458547e-05,
      "loss": 8.0526,
      "step": 2090
    },
    {
      "epoch": 0.005905827364229247,
      "grad_norm": 26.75,
      "learning_rate": 1.999980893829871e-05,
      "loss": 8.3751,
      "step": 2100
    },
    {
      "epoch": 0.005933950351677958,
      "grad_norm": 32.0,
      "learning_rate": 1.9999807113465827e-05,
      "loss": 7.9085,
      "step": 2110
    },
    {
      "epoch": 0.005962073339126669,
      "grad_norm": 41.5,
      "learning_rate": 1.9999805279959895e-05,
      "loss": 8.9577,
      "step": 2120
    },
    {
      "epoch": 0.005990196326575379,
      "grad_norm": 67.5,
      "learning_rate": 1.9999803437780918e-05,
      "loss": 8.204,
      "step": 2130
    },
    {
      "epoch": 0.00601831931402409,
      "grad_norm": 29.125,
      "learning_rate": 1.9999801586928893e-05,
      "loss": 7.8166,
      "step": 2140
    },
    {
      "epoch": 0.006046442301472801,
      "grad_norm": 47.5,
      "learning_rate": 1.9999799727403826e-05,
      "loss": 8.6401,
      "step": 2150
    },
    {
      "epoch": 0.006074565288921511,
      "grad_norm": 24.25,
      "learning_rate": 1.9999797859205715e-05,
      "loss": 8.569,
      "step": 2160
    },
    {
      "epoch": 0.006102688276370222,
      "grad_norm": 20.375,
      "learning_rate": 1.999979598233457e-05,
      "loss": 8.4517,
      "step": 2170
    },
    {
      "epoch": 0.006130811263818933,
      "grad_norm": 36.5,
      "learning_rate": 1.999979409679038e-05,
      "loss": 8.8763,
      "step": 2180
    },
    {
      "epoch": 0.0061589342512676434,
      "grad_norm": 32.75,
      "learning_rate": 1.9999792202573153e-05,
      "loss": 8.3364,
      "step": 2190
    },
    {
      "epoch": 0.006187057238716354,
      "grad_norm": 33.5,
      "learning_rate": 1.9999790299682897e-05,
      "loss": 7.9852,
      "step": 2200
    },
    {
      "epoch": 0.006215180226165065,
      "grad_norm": 34.75,
      "learning_rate": 1.9999788388119602e-05,
      "loss": 7.6193,
      "step": 2210
    },
    {
      "epoch": 0.0062433032136137755,
      "grad_norm": 32.0,
      "learning_rate": 1.9999786467883273e-05,
      "loss": 8.3854,
      "step": 2220
    },
    {
      "epoch": 0.006271426201062486,
      "grad_norm": 28.625,
      "learning_rate": 1.9999784538973916e-05,
      "loss": 8.6284,
      "step": 2230
    },
    {
      "epoch": 0.006299549188511197,
      "grad_norm": 23.875,
      "learning_rate": 1.999978260139153e-05,
      "loss": 7.3057,
      "step": 2240
    },
    {
      "epoch": 0.0063276721759599076,
      "grad_norm": 55.75,
      "learning_rate": 1.999978065513612e-05,
      "loss": 8.1596,
      "step": 2250
    },
    {
      "epoch": 0.006355795163408618,
      "grad_norm": 27.0,
      "learning_rate": 1.9999778700207678e-05,
      "loss": 8.3685,
      "step": 2260
    },
    {
      "epoch": 0.006383918150857329,
      "grad_norm": 36.25,
      "learning_rate": 1.999977673660622e-05,
      "loss": 7.9203,
      "step": 2270
    },
    {
      "epoch": 0.00641204113830604,
      "grad_norm": 35.0,
      "learning_rate": 1.9999774764331737e-05,
      "loss": 7.9711,
      "step": 2280
    },
    {
      "epoch": 0.00644016412575475,
      "grad_norm": 32.25,
      "learning_rate": 1.9999772783384233e-05,
      "loss": 8.959,
      "step": 2290
    },
    {
      "epoch": 0.006468287113203461,
      "grad_norm": 23.375,
      "learning_rate": 1.999977079376371e-05,
      "loss": 8.0193,
      "step": 2300
    },
    {
      "epoch": 0.006496410100652172,
      "grad_norm": 27.75,
      "learning_rate": 1.9999768795470173e-05,
      "loss": 8.1791,
      "step": 2310
    },
    {
      "epoch": 0.006524533088100882,
      "grad_norm": 22.875,
      "learning_rate": 1.9999766788503616e-05,
      "loss": 8.3491,
      "step": 2320
    },
    {
      "epoch": 0.006552656075549593,
      "grad_norm": 43.25,
      "learning_rate": 1.9999764772864048e-05,
      "loss": 8.3152,
      "step": 2330
    },
    {
      "epoch": 0.006580779062998304,
      "grad_norm": 39.5,
      "learning_rate": 1.999976274855147e-05,
      "loss": 8.4961,
      "step": 2340
    },
    {
      "epoch": 0.0066089020504470145,
      "grad_norm": 42.0,
      "learning_rate": 1.999976071556588e-05,
      "loss": 7.7311,
      "step": 2350
    },
    {
      "epoch": 0.006637025037895726,
      "grad_norm": 25.75,
      "learning_rate": 1.9999758673907286e-05,
      "loss": 8.254,
      "step": 2360
    },
    {
      "epoch": 0.006665148025344437,
      "grad_norm": 27.75,
      "learning_rate": 1.9999756623575682e-05,
      "loss": 8.4002,
      "step": 2370
    },
    {
      "epoch": 0.006693271012793147,
      "grad_norm": 29.125,
      "learning_rate": 1.9999754564571073e-05,
      "loss": 8.8437,
      "step": 2380
    },
    {
      "epoch": 0.006721394000241858,
      "grad_norm": 21.125,
      "learning_rate": 1.9999752496893464e-05,
      "loss": 7.9154,
      "step": 2390
    },
    {
      "epoch": 0.006749516987690569,
      "grad_norm": 35.75,
      "learning_rate": 1.999975042054285e-05,
      "loss": 8.1271,
      "step": 2400
    },
    {
      "epoch": 0.0067776399751392795,
      "grad_norm": 23.0,
      "learning_rate": 1.9999748335519243e-05,
      "loss": 8.7894,
      "step": 2410
    },
    {
      "epoch": 0.00680576296258799,
      "grad_norm": 30.0,
      "learning_rate": 1.9999746241822635e-05,
      "loss": 7.7237,
      "step": 2420
    },
    {
      "epoch": 0.006833885950036701,
      "grad_norm": 27.375,
      "learning_rate": 1.9999744139453032e-05,
      "loss": 7.8372,
      "step": 2430
    },
    {
      "epoch": 0.0068620089374854115,
      "grad_norm": 25.875,
      "learning_rate": 1.9999742028410436e-05,
      "loss": 8.0883,
      "step": 2440
    },
    {
      "epoch": 0.006890131924934122,
      "grad_norm": 33.5,
      "learning_rate": 1.9999739908694846e-05,
      "loss": 8.5254,
      "step": 2450
    },
    {
      "epoch": 0.006918254912382833,
      "grad_norm": 54.0,
      "learning_rate": 1.999973778030627e-05,
      "loss": 8.7582,
      "step": 2460
    },
    {
      "epoch": 0.006946377899831544,
      "grad_norm": 20.875,
      "learning_rate": 1.99997356432447e-05,
      "loss": 8.2299,
      "step": 2470
    },
    {
      "epoch": 0.006974500887280254,
      "grad_norm": 35.25,
      "learning_rate": 1.9999733497510146e-05,
      "loss": 8.6918,
      "step": 2480
    },
    {
      "epoch": 0.007002623874728965,
      "grad_norm": 27.125,
      "learning_rate": 1.999973134310261e-05,
      "loss": 9.4593,
      "step": 2490
    },
    {
      "epoch": 0.007030746862177676,
      "grad_norm": 23.0,
      "learning_rate": 1.9999729180022088e-05,
      "loss": 8.6678,
      "step": 2500
    },
    {
      "epoch": 0.007058869849626386,
      "grad_norm": 37.5,
      "learning_rate": 1.999972700826859e-05,
      "loss": 8.4213,
      "step": 2510
    },
    {
      "epoch": 0.007086992837075097,
      "grad_norm": 31.75,
      "learning_rate": 1.9999724827842108e-05,
      "loss": 7.9444,
      "step": 2520
    },
    {
      "epoch": 0.007115115824523808,
      "grad_norm": 27.125,
      "learning_rate": 1.9999722638742654e-05,
      "loss": 8.1803,
      "step": 2530
    },
    {
      "epoch": 0.007143238811972518,
      "grad_norm": 31.5,
      "learning_rate": 1.999972044097022e-05,
      "loss": 8.2442,
      "step": 2540
    },
    {
      "epoch": 0.007171361799421229,
      "grad_norm": 25.125,
      "learning_rate": 1.9999718234524816e-05,
      "loss": 8.5528,
      "step": 2550
    },
    {
      "epoch": 0.00719948478686994,
      "grad_norm": 48.25,
      "learning_rate": 1.9999716019406438e-05,
      "loss": 8.6968,
      "step": 2560
    },
    {
      "epoch": 0.0072276077743186505,
      "grad_norm": 38.5,
      "learning_rate": 1.9999713795615093e-05,
      "loss": 7.8031,
      "step": 2570
    },
    {
      "epoch": 0.007255730761767361,
      "grad_norm": 27.875,
      "learning_rate": 1.999971156315078e-05,
      "loss": 7.8565,
      "step": 2580
    },
    {
      "epoch": 0.007283853749216072,
      "grad_norm": 26.75,
      "learning_rate": 1.9999709322013502e-05,
      "loss": 7.6584,
      "step": 2590
    },
    {
      "epoch": 0.0073119767366647825,
      "grad_norm": 37.5,
      "learning_rate": 1.9999707072203264e-05,
      "loss": 8.1999,
      "step": 2600
    },
    {
      "epoch": 0.007340099724113493,
      "grad_norm": 35.75,
      "learning_rate": 1.9999704813720058e-05,
      "loss": 8.0539,
      "step": 2610
    },
    {
      "epoch": 0.007368222711562204,
      "grad_norm": 49.5,
      "learning_rate": 1.9999702546563896e-05,
      "loss": 7.6773,
      "step": 2620
    },
    {
      "epoch": 0.007396345699010915,
      "grad_norm": 36.75,
      "learning_rate": 1.9999700270734777e-05,
      "loss": 8.2563,
      "step": 2630
    },
    {
      "epoch": 0.007424468686459625,
      "grad_norm": 22.75,
      "learning_rate": 1.99996979862327e-05,
      "loss": 8.3656,
      "step": 2640
    },
    {
      "epoch": 0.007452591673908336,
      "grad_norm": 23.0,
      "learning_rate": 1.999969569305767e-05,
      "loss": 8.2729,
      "step": 2650
    },
    {
      "epoch": 0.007480714661357047,
      "grad_norm": 25.875,
      "learning_rate": 1.999969339120969e-05,
      "loss": 8.248,
      "step": 2660
    },
    {
      "epoch": 0.007508837648805757,
      "grad_norm": 38.75,
      "learning_rate": 1.9999691080688758e-05,
      "loss": 8.5252,
      "step": 2670
    },
    {
      "epoch": 0.007536960636254468,
      "grad_norm": 25.0,
      "learning_rate": 1.9999688761494876e-05,
      "loss": 8.3807,
      "step": 2680
    },
    {
      "epoch": 0.007565083623703179,
      "grad_norm": 55.75,
      "learning_rate": 1.9999686433628054e-05,
      "loss": 7.4805,
      "step": 2690
    },
    {
      "epoch": 0.007593206611151889,
      "grad_norm": 34.75,
      "learning_rate": 1.999968409708829e-05,
      "loss": 8.0859,
      "step": 2700
    },
    {
      "epoch": 0.0076213295986006,
      "grad_norm": 29.625,
      "learning_rate": 1.9999681751875578e-05,
      "loss": 7.2083,
      "step": 2710
    },
    {
      "epoch": 0.007649452586049311,
      "grad_norm": 55.0,
      "learning_rate": 1.999967939798993e-05,
      "loss": 8.823,
      "step": 2720
    },
    {
      "epoch": 0.0076775755734980215,
      "grad_norm": 20.75,
      "learning_rate": 1.999967703543134e-05,
      "loss": 8.5313,
      "step": 2730
    },
    {
      "epoch": 0.007705698560946732,
      "grad_norm": 40.0,
      "learning_rate": 1.9999674664199818e-05,
      "loss": 7.3648,
      "step": 2740
    },
    {
      "epoch": 0.007733821548395443,
      "grad_norm": 27.0,
      "learning_rate": 1.9999672284295365e-05,
      "loss": 7.8972,
      "step": 2750
    },
    {
      "epoch": 0.0077619445358441536,
      "grad_norm": 31.5,
      "learning_rate": 1.9999669895717978e-05,
      "loss": 7.856,
      "step": 2760
    },
    {
      "epoch": 0.007790067523292864,
      "grad_norm": 33.75,
      "learning_rate": 1.9999667498467662e-05,
      "loss": 8.2645,
      "step": 2770
    },
    {
      "epoch": 0.007818190510741575,
      "grad_norm": 31.625,
      "learning_rate": 1.9999665092544416e-05,
      "loss": 8.112,
      "step": 2780
    },
    {
      "epoch": 0.007846313498190286,
      "grad_norm": 24.625,
      "learning_rate": 1.9999662677948248e-05,
      "loss": 8.0376,
      "step": 2790
    },
    {
      "epoch": 0.007874436485638996,
      "grad_norm": 25.75,
      "learning_rate": 1.9999660254679156e-05,
      "loss": 7.6112,
      "step": 2800
    },
    {
      "epoch": 0.007902559473087708,
      "grad_norm": 45.25,
      "learning_rate": 1.9999657822737145e-05,
      "loss": 7.7822,
      "step": 2810
    },
    {
      "epoch": 0.007930682460536418,
      "grad_norm": 31.375,
      "learning_rate": 1.9999655382122212e-05,
      "loss": 7.888,
      "step": 2820
    },
    {
      "epoch": 0.00795880544798513,
      "grad_norm": 42.0,
      "learning_rate": 1.9999652932834366e-05,
      "loss": 8.043,
      "step": 2830
    },
    {
      "epoch": 0.007986928435433839,
      "grad_norm": 43.25,
      "learning_rate": 1.9999650474873603e-05,
      "loss": 8.4205,
      "step": 2840
    },
    {
      "epoch": 0.00801505142288255,
      "grad_norm": 29.75,
      "learning_rate": 1.999964800823993e-05,
      "loss": 8.0209,
      "step": 2850
    },
    {
      "epoch": 0.00804317441033126,
      "grad_norm": 27.125,
      "learning_rate": 1.9999645532933344e-05,
      "loss": 8.923,
      "step": 2860
    },
    {
      "epoch": 0.008071297397779972,
      "grad_norm": 24.125,
      "learning_rate": 1.999964304895385e-05,
      "loss": 7.5883,
      "step": 2870
    },
    {
      "epoch": 0.008099420385228682,
      "grad_norm": 41.0,
      "learning_rate": 1.9999640556301448e-05,
      "loss": 7.8274,
      "step": 2880
    },
    {
      "epoch": 0.008127543372677393,
      "grad_norm": 35.0,
      "learning_rate": 1.9999638054976146e-05,
      "loss": 8.3555,
      "step": 2890
    },
    {
      "epoch": 0.008155666360126103,
      "grad_norm": 32.75,
      "learning_rate": 1.9999635544977942e-05,
      "loss": 7.4039,
      "step": 2900
    },
    {
      "epoch": 0.008183789347574815,
      "grad_norm": 28.5,
      "learning_rate": 1.999963302630684e-05,
      "loss": 7.7538,
      "step": 2910
    },
    {
      "epoch": 0.008211912335023525,
      "grad_norm": 29.875,
      "learning_rate": 1.9999630498962833e-05,
      "loss": 9.2657,
      "step": 2920
    },
    {
      "epoch": 0.008240035322472236,
      "grad_norm": 38.0,
      "learning_rate": 1.9999627962945938e-05,
      "loss": 8.4342,
      "step": 2930
    },
    {
      "epoch": 0.008268158309920946,
      "grad_norm": 26.0,
      "learning_rate": 1.9999625418256147e-05,
      "loss": 8.1592,
      "step": 2940
    },
    {
      "epoch": 0.008296281297369658,
      "grad_norm": 28.875,
      "learning_rate": 1.9999622864893468e-05,
      "loss": 7.8994,
      "step": 2950
    },
    {
      "epoch": 0.008324404284818367,
      "grad_norm": 22.875,
      "learning_rate": 1.99996203028579e-05,
      "loss": 7.6056,
      "step": 2960
    },
    {
      "epoch": 0.008352527272267079,
      "grad_norm": 34.25,
      "learning_rate": 1.9999617732149442e-05,
      "loss": 8.0552,
      "step": 2970
    },
    {
      "epoch": 0.008380650259715789,
      "grad_norm": 31.125,
      "learning_rate": 1.9999615152768103e-05,
      "loss": 7.8963,
      "step": 2980
    },
    {
      "epoch": 0.0084087732471645,
      "grad_norm": 18.75,
      "learning_rate": 1.999961256471388e-05,
      "loss": 8.4423,
      "step": 2990
    },
    {
      "epoch": 0.00843689623461321,
      "grad_norm": 27.625,
      "learning_rate": 1.999960996798678e-05,
      "loss": 8.0885,
      "step": 3000
    },
    {
      "epoch": 0.008465019222061922,
      "grad_norm": 29.5,
      "learning_rate": 1.9999607362586805e-05,
      "loss": 7.9177,
      "step": 3010
    },
    {
      "epoch": 0.008493142209510631,
      "grad_norm": 48.0,
      "learning_rate": 1.999960474851395e-05,
      "loss": 8.5548,
      "step": 3020
    },
    {
      "epoch": 0.008521265196959343,
      "grad_norm": 30.125,
      "learning_rate": 1.9999602125768227e-05,
      "loss": 8.1429,
      "step": 3030
    },
    {
      "epoch": 0.008549388184408053,
      "grad_norm": 25.0,
      "learning_rate": 1.999959949434963e-05,
      "loss": 7.3102,
      "step": 3040
    },
    {
      "epoch": 0.008577511171856764,
      "grad_norm": 26.75,
      "learning_rate": 1.9999596854258162e-05,
      "loss": 7.824,
      "step": 3050
    },
    {
      "epoch": 0.008605634159305474,
      "grad_norm": 50.5,
      "learning_rate": 1.9999594205493836e-05,
      "loss": 7.7445,
      "step": 3060
    },
    {
      "epoch": 0.008633757146754186,
      "grad_norm": 32.75,
      "learning_rate": 1.9999591548056642e-05,
      "loss": 9.1142,
      "step": 3070
    },
    {
      "epoch": 0.008661880134202896,
      "grad_norm": 47.75,
      "learning_rate": 1.9999588881946583e-05,
      "loss": 7.5972,
      "step": 3080
    },
    {
      "epoch": 0.008690003121651607,
      "grad_norm": 20.875,
      "learning_rate": 1.999958620716367e-05,
      "loss": 7.804,
      "step": 3090
    },
    {
      "epoch": 0.008718126109100317,
      "grad_norm": 23.5,
      "learning_rate": 1.99995835237079e-05,
      "loss": 7.8338,
      "step": 3100
    },
    {
      "epoch": 0.008746249096549029,
      "grad_norm": 27.0,
      "learning_rate": 1.9999580831579274e-05,
      "loss": 7.747,
      "step": 3110
    },
    {
      "epoch": 0.008774372083997738,
      "grad_norm": 41.5,
      "learning_rate": 1.9999578130777798e-05,
      "loss": 7.3338,
      "step": 3120
    },
    {
      "epoch": 0.00880249507144645,
      "grad_norm": 25.125,
      "learning_rate": 1.9999575421303467e-05,
      "loss": 8.548,
      "step": 3130
    },
    {
      "epoch": 0.00883061805889516,
      "grad_norm": 24.5,
      "learning_rate": 1.9999572703156294e-05,
      "loss": 8.2418,
      "step": 3140
    },
    {
      "epoch": 0.008858741046343871,
      "grad_norm": 36.75,
      "learning_rate": 1.9999569976336277e-05,
      "loss": 8.0847,
      "step": 3150
    },
    {
      "epoch": 0.008886864033792581,
      "grad_norm": 20.25,
      "learning_rate": 1.999956724084341e-05,
      "loss": 8.2338,
      "step": 3160
    },
    {
      "epoch": 0.008914987021241293,
      "grad_norm": 25.375,
      "learning_rate": 1.9999564496677707e-05,
      "loss": 8.462,
      "step": 3170
    },
    {
      "epoch": 0.008943110008690002,
      "grad_norm": 31.5,
      "learning_rate": 1.9999561743839168e-05,
      "loss": 7.7826,
      "step": 3180
    },
    {
      "epoch": 0.008971232996138714,
      "grad_norm": 17.375,
      "learning_rate": 1.9999558982327795e-05,
      "loss": 8.2453,
      "step": 3190
    },
    {
      "epoch": 0.008999355983587424,
      "grad_norm": 23.0,
      "learning_rate": 1.9999556212143583e-05,
      "loss": 7.9897,
      "step": 3200
    },
    {
      "epoch": 0.009027478971036135,
      "grad_norm": 35.75,
      "learning_rate": 1.9999553433286543e-05,
      "loss": 7.1996,
      "step": 3210
    },
    {
      "epoch": 0.009055601958484845,
      "grad_norm": 24.875,
      "learning_rate": 1.9999550645756676e-05,
      "loss": 8.0288,
      "step": 3220
    },
    {
      "epoch": 0.009083724945933557,
      "grad_norm": 31.5,
      "learning_rate": 1.999954784955398e-05,
      "loss": 8.2478,
      "step": 3230
    },
    {
      "epoch": 0.009111847933382267,
      "grad_norm": 30.5,
      "learning_rate": 1.9999545044678464e-05,
      "loss": 7.9006,
      "step": 3240
    },
    {
      "epoch": 0.009139970920830978,
      "grad_norm": 39.75,
      "learning_rate": 1.9999542231130126e-05,
      "loss": 8.0676,
      "step": 3250
    },
    {
      "epoch": 0.009168093908279688,
      "grad_norm": 32.0,
      "learning_rate": 1.999953940890897e-05,
      "loss": 7.9471,
      "step": 3260
    },
    {
      "epoch": 0.0091962168957284,
      "grad_norm": 38.0,
      "learning_rate": 1.9999536578014998e-05,
      "loss": 7.2531,
      "step": 3270
    },
    {
      "epoch": 0.00922433988317711,
      "grad_norm": 30.375,
      "learning_rate": 1.999953373844821e-05,
      "loss": 7.6782,
      "step": 3280
    },
    {
      "epoch": 0.009252462870625821,
      "grad_norm": 30.375,
      "learning_rate": 1.999953089020861e-05,
      "loss": 7.6688,
      "step": 3290
    },
    {
      "epoch": 0.00928058585807453,
      "grad_norm": 35.75,
      "learning_rate": 1.999952803329621e-05,
      "loss": 7.3213,
      "step": 3300
    },
    {
      "epoch": 0.009308708845523242,
      "grad_norm": 24.0,
      "learning_rate": 1.9999525167710994e-05,
      "loss": 7.8643,
      "step": 3310
    },
    {
      "epoch": 0.009336831832971952,
      "grad_norm": 24.25,
      "learning_rate": 1.9999522293452976e-05,
      "loss": 8.3795,
      "step": 3320
    },
    {
      "epoch": 0.009364954820420664,
      "grad_norm": 27.125,
      "learning_rate": 1.999951941052216e-05,
      "loss": 8.7058,
      "step": 3330
    },
    {
      "epoch": 0.009393077807869374,
      "grad_norm": 24.375,
      "learning_rate": 1.999951651891854e-05,
      "loss": 8.6353,
      "step": 3340
    },
    {
      "epoch": 0.009421200795318085,
      "grad_norm": 26.75,
      "learning_rate": 1.9999513618642127e-05,
      "loss": 8.2442,
      "step": 3350
    },
    {
      "epoch": 0.009449323782766797,
      "grad_norm": 26.875,
      "learning_rate": 1.9999510709692922e-05,
      "loss": 8.1887,
      "step": 3360
    },
    {
      "epoch": 0.009477446770215506,
      "grad_norm": 29.125,
      "learning_rate": 1.9999507792070924e-05,
      "loss": 8.5246,
      "step": 3370
    },
    {
      "epoch": 0.009505569757664218,
      "grad_norm": 23.625,
      "learning_rate": 1.9999504865776138e-05,
      "loss": 8.0467,
      "step": 3380
    },
    {
      "epoch": 0.009533692745112928,
      "grad_norm": 44.25,
      "learning_rate": 1.9999501930808565e-05,
      "loss": 7.8232,
      "step": 3390
    },
    {
      "epoch": 0.00956181573256164,
      "grad_norm": 28.375,
      "learning_rate": 1.999949898716821e-05,
      "loss": 7.5936,
      "step": 3400
    },
    {
      "epoch": 0.00958993872001035,
      "grad_norm": 37.25,
      "learning_rate": 1.9999496034855073e-05,
      "loss": 7.6965,
      "step": 3410
    },
    {
      "epoch": 0.00961806170745906,
      "grad_norm": 22.0,
      "learning_rate": 1.9999493073869155e-05,
      "loss": 8.2151,
      "step": 3420
    },
    {
      "epoch": 0.00964618469490777,
      "grad_norm": 23.5,
      "learning_rate": 1.9999490104210463e-05,
      "loss": 8.0461,
      "step": 3430
    },
    {
      "epoch": 0.009674307682356482,
      "grad_norm": 42.25,
      "learning_rate": 1.9999487125879e-05,
      "loss": 9.0034,
      "step": 3440
    },
    {
      "epoch": 0.009702430669805192,
      "grad_norm": 24.75,
      "learning_rate": 1.9999484138874762e-05,
      "loss": 8.3002,
      "step": 3450
    },
    {
      "epoch": 0.009730553657253904,
      "grad_norm": 38.0,
      "learning_rate": 1.999948114319776e-05,
      "loss": 7.8161,
      "step": 3460
    },
    {
      "epoch": 0.009758676644702613,
      "grad_norm": 33.25,
      "learning_rate": 1.999947813884799e-05,
      "loss": 8.0935,
      "step": 3470
    },
    {
      "epoch": 0.009786799632151325,
      "grad_norm": 27.25,
      "learning_rate": 1.999947512582546e-05,
      "loss": 7.4854,
      "step": 3480
    },
    {
      "epoch": 0.009814922619600035,
      "grad_norm": 23.625,
      "learning_rate": 1.9999472104130165e-05,
      "loss": 7.6675,
      "step": 3490
    },
    {
      "epoch": 0.009843045607048746,
      "grad_norm": 35.75,
      "learning_rate": 1.9999469073762115e-05,
      "loss": 7.3555,
      "step": 3500
    },
    {
      "epoch": 0.009871168594497456,
      "grad_norm": 34.25,
      "learning_rate": 1.9999466034721308e-05,
      "loss": 7.9388,
      "step": 3510
    },
    {
      "epoch": 0.009899291581946168,
      "grad_norm": 30.375,
      "learning_rate": 1.999946298700775e-05,
      "loss": 8.803,
      "step": 3520
    },
    {
      "epoch": 0.009927414569394877,
      "grad_norm": 40.5,
      "learning_rate": 1.999945993062144e-05,
      "loss": 7.7273,
      "step": 3530
    },
    {
      "epoch": 0.009955537556843589,
      "grad_norm": 22.25,
      "learning_rate": 1.9999456865562386e-05,
      "loss": 7.9366,
      "step": 3540
    },
    {
      "epoch": 0.009983660544292299,
      "grad_norm": 24.625,
      "learning_rate": 1.999945379183059e-05,
      "loss": 7.5814,
      "step": 3550
    },
    {
      "epoch": 0.01001178353174101,
      "grad_norm": 33.0,
      "learning_rate": 1.999945070942605e-05,
      "loss": 8.4469,
      "step": 3560
    },
    {
      "epoch": 0.01003990651918972,
      "grad_norm": 26.625,
      "learning_rate": 1.9999447618348765e-05,
      "loss": 8.1064,
      "step": 3570
    },
    {
      "epoch": 0.010068029506638432,
      "grad_norm": 49.25,
      "learning_rate": 1.999944451859875e-05,
      "loss": 7.3223,
      "step": 3580
    },
    {
      "epoch": 0.010096152494087142,
      "grad_norm": 26.875,
      "learning_rate": 1.9999441410176e-05,
      "loss": 7.3847,
      "step": 3590
    },
    {
      "epoch": 0.010124275481535853,
      "grad_norm": 38.25,
      "learning_rate": 1.9999438293080512e-05,
      "loss": 7.5861,
      "step": 3600
    },
    {
      "epoch": 0.010152398468984563,
      "grad_norm": 32.25,
      "learning_rate": 1.9999435167312303e-05,
      "loss": 8.5341,
      "step": 3610
    },
    {
      "epoch": 0.010180521456433275,
      "grad_norm": 50.25,
      "learning_rate": 1.9999432032871368e-05,
      "loss": 7.4763,
      "step": 3620
    },
    {
      "epoch": 0.010208644443881984,
      "grad_norm": 33.25,
      "learning_rate": 1.999942888975771e-05,
      "loss": 7.9193,
      "step": 3630
    },
    {
      "epoch": 0.010236767431330696,
      "grad_norm": 54.25,
      "learning_rate": 1.9999425737971328e-05,
      "loss": 8.376,
      "step": 3640
    },
    {
      "epoch": 0.010264890418779406,
      "grad_norm": 63.25,
      "learning_rate": 1.999942257751223e-05,
      "loss": 7.7004,
      "step": 3650
    },
    {
      "epoch": 0.010293013406228117,
      "grad_norm": 28.375,
      "learning_rate": 1.999941940838042e-05,
      "loss": 8.0735,
      "step": 3660
    },
    {
      "epoch": 0.010321136393676827,
      "grad_norm": 25.25,
      "learning_rate": 1.9999416230575894e-05,
      "loss": 7.6901,
      "step": 3670
    },
    {
      "epoch": 0.010349259381125539,
      "grad_norm": 28.75,
      "learning_rate": 1.9999413044098662e-05,
      "loss": 8.1958,
      "step": 3680
    },
    {
      "epoch": 0.010377382368574248,
      "grad_norm": 22.0,
      "learning_rate": 1.999940984894872e-05,
      "loss": 8.2711,
      "step": 3690
    },
    {
      "epoch": 0.01040550535602296,
      "grad_norm": 26.375,
      "learning_rate": 1.999940664512608e-05,
      "loss": 8.2227,
      "step": 3700
    },
    {
      "epoch": 0.01043362834347167,
      "grad_norm": 36.5,
      "learning_rate": 1.9999403432630733e-05,
      "loss": 7.9065,
      "step": 3710
    },
    {
      "epoch": 0.010461751330920381,
      "grad_norm": 28.375,
      "learning_rate": 1.999940021146269e-05,
      "loss": 7.9618,
      "step": 3720
    },
    {
      "epoch": 0.010489874318369091,
      "grad_norm": 32.25,
      "learning_rate": 1.9999396981621952e-05,
      "loss": 8.2882,
      "step": 3730
    },
    {
      "epoch": 0.010517997305817803,
      "grad_norm": 28.25,
      "learning_rate": 1.999939374310852e-05,
      "loss": 7.9705,
      "step": 3740
    },
    {
      "epoch": 0.010546120293266513,
      "grad_norm": 25.5,
      "learning_rate": 1.99993904959224e-05,
      "loss": 7.4949,
      "step": 3750
    },
    {
      "epoch": 0.010574243280715224,
      "grad_norm": 33.75,
      "learning_rate": 1.9999387240063596e-05,
      "loss": 7.865,
      "step": 3760
    },
    {
      "epoch": 0.010602366268163934,
      "grad_norm": 26.625,
      "learning_rate": 1.99993839755321e-05,
      "loss": 7.4999,
      "step": 3770
    },
    {
      "epoch": 0.010630489255612646,
      "grad_norm": 24.125,
      "learning_rate": 1.999938070232793e-05,
      "loss": 8.2593,
      "step": 3780
    },
    {
      "epoch": 0.010658612243061355,
      "grad_norm": 42.0,
      "learning_rate": 1.999937742045108e-05,
      "loss": 7.7764,
      "step": 3790
    },
    {
      "epoch": 0.010686735230510067,
      "grad_norm": 47.25,
      "learning_rate": 1.999937412990155e-05,
      "loss": 7.5523,
      "step": 3800
    },
    {
      "epoch": 0.010714858217958777,
      "grad_norm": 26.125,
      "learning_rate": 1.9999370830679355e-05,
      "loss": 8.2505,
      "step": 3810
    },
    {
      "epoch": 0.010742981205407488,
      "grad_norm": 37.25,
      "learning_rate": 1.9999367522784486e-05,
      "loss": 7.9532,
      "step": 3820
    },
    {
      "epoch": 0.010771104192856198,
      "grad_norm": 37.25,
      "learning_rate": 1.9999364206216946e-05,
      "loss": 8.2539,
      "step": 3830
    },
    {
      "epoch": 0.01079922718030491,
      "grad_norm": 27.125,
      "learning_rate": 1.9999360880976747e-05,
      "loss": 7.9396,
      "step": 3840
    },
    {
      "epoch": 0.01082735016775362,
      "grad_norm": 24.625,
      "learning_rate": 1.9999357547063886e-05,
      "loss": 8.4424,
      "step": 3850
    },
    {
      "epoch": 0.010855473155202331,
      "grad_norm": 25.875,
      "learning_rate": 1.9999354204478366e-05,
      "loss": 8.6787,
      "step": 3860
    },
    {
      "epoch": 0.010883596142651041,
      "grad_norm": 34.0,
      "learning_rate": 1.9999350853220188e-05,
      "loss": 8.3079,
      "step": 3870
    },
    {
      "epoch": 0.010911719130099752,
      "grad_norm": 26.25,
      "learning_rate": 1.999934749328936e-05,
      "loss": 8.3721,
      "step": 3880
    },
    {
      "epoch": 0.010939842117548462,
      "grad_norm": 33.0,
      "learning_rate": 1.9999344124685883e-05,
      "loss": 8.3725,
      "step": 3890
    },
    {
      "epoch": 0.010967965104997174,
      "grad_norm": 25.375,
      "learning_rate": 1.9999340747409756e-05,
      "loss": 8.0926,
      "step": 3900
    },
    {
      "epoch": 0.010996088092445884,
      "grad_norm": 21.375,
      "learning_rate": 1.9999337361460988e-05,
      "loss": 7.9782,
      "step": 3910
    },
    {
      "epoch": 0.011024211079894595,
      "grad_norm": 42.5,
      "learning_rate": 1.9999333966839578e-05,
      "loss": 8.305,
      "step": 3920
    },
    {
      "epoch": 0.011052334067343305,
      "grad_norm": 54.25,
      "learning_rate": 1.999933056354553e-05,
      "loss": 7.9586,
      "step": 3930
    },
    {
      "epoch": 0.011080457054792017,
      "grad_norm": 31.25,
      "learning_rate": 1.999932715157885e-05,
      "loss": 8.1783,
      "step": 3940
    },
    {
      "epoch": 0.011108580042240726,
      "grad_norm": 26.875,
      "learning_rate": 1.9999323730939536e-05,
      "loss": 8.2139,
      "step": 3950
    },
    {
      "epoch": 0.011136703029689438,
      "grad_norm": 31.375,
      "learning_rate": 1.999932030162759e-05,
      "loss": 8.6769,
      "step": 3960
    },
    {
      "epoch": 0.011164826017138148,
      "grad_norm": 31.25,
      "learning_rate": 1.999931686364302e-05,
      "loss": 8.1601,
      "step": 3970
    },
    {
      "epoch": 0.01119294900458686,
      "grad_norm": 63.0,
      "learning_rate": 1.999931341698583e-05,
      "loss": 8.4853,
      "step": 3980
    },
    {
      "epoch": 0.01122107199203557,
      "grad_norm": 40.5,
      "learning_rate": 1.999930996165602e-05,
      "loss": 8.6766,
      "step": 3990
    },
    {
      "epoch": 0.01124919497948428,
      "grad_norm": 30.75,
      "learning_rate": 1.9999306497653584e-05,
      "loss": 8.3272,
      "step": 4000
    },
    {
      "epoch": 0.01127731796693299,
      "grad_norm": 30.75,
      "learning_rate": 1.9999303024978542e-05,
      "loss": 8.9331,
      "step": 4010
    },
    {
      "epoch": 0.011305440954381702,
      "grad_norm": 42.5,
      "learning_rate": 1.9999299543630886e-05,
      "loss": 7.7461,
      "step": 4020
    },
    {
      "epoch": 0.011333563941830412,
      "grad_norm": 23.75,
      "learning_rate": 1.9999296053610625e-05,
      "loss": 8.6209,
      "step": 4030
    },
    {
      "epoch": 0.011361686929279123,
      "grad_norm": 41.0,
      "learning_rate": 1.9999292554917757e-05,
      "loss": 7.3432,
      "step": 4040
    },
    {
      "epoch": 0.011389809916727835,
      "grad_norm": 47.5,
      "learning_rate": 1.9999289047552287e-05,
      "loss": 8.6455,
      "step": 4050
    },
    {
      "epoch": 0.011417932904176545,
      "grad_norm": 30.0,
      "learning_rate": 1.9999285531514217e-05,
      "loss": 8.5096,
      "step": 4060
    },
    {
      "epoch": 0.011446055891625256,
      "grad_norm": 23.125,
      "learning_rate": 1.9999282006803552e-05,
      "loss": 8.3673,
      "step": 4070
    },
    {
      "epoch": 0.011474178879073966,
      "grad_norm": 23.875,
      "learning_rate": 1.9999278473420294e-05,
      "loss": 7.9102,
      "step": 4080
    },
    {
      "epoch": 0.011502301866522678,
      "grad_norm": 45.0,
      "learning_rate": 1.9999274931364448e-05,
      "loss": 8.6135,
      "step": 4090
    },
    {
      "epoch": 0.011530424853971388,
      "grad_norm": 28.0,
      "learning_rate": 1.999927138063601e-05,
      "loss": 7.6448,
      "step": 4100
    },
    {
      "epoch": 0.0115585478414201,
      "grad_norm": 21.5,
      "learning_rate": 1.9999267821234993e-05,
      "loss": 7.6886,
      "step": 4110
    },
    {
      "epoch": 0.011586670828868809,
      "grad_norm": 27.0,
      "learning_rate": 1.9999264253161395e-05,
      "loss": 8.2981,
      "step": 4120
    },
    {
      "epoch": 0.01161479381631752,
      "grad_norm": 46.5,
      "learning_rate": 1.9999260676415217e-05,
      "loss": 8.2034,
      "step": 4130
    },
    {
      "epoch": 0.01164291680376623,
      "grad_norm": 27.875,
      "learning_rate": 1.9999257090996467e-05,
      "loss": 8.5675,
      "step": 4140
    },
    {
      "epoch": 0.011671039791214942,
      "grad_norm": 38.0,
      "learning_rate": 1.9999253496905148e-05,
      "loss": 7.6995,
      "step": 4150
    },
    {
      "epoch": 0.011699162778663652,
      "grad_norm": 26.0,
      "learning_rate": 1.9999249894141254e-05,
      "loss": 7.38,
      "step": 4160
    },
    {
      "epoch": 0.011727285766112363,
      "grad_norm": 49.5,
      "learning_rate": 1.9999246282704804e-05,
      "loss": 8.2433,
      "step": 4170
    },
    {
      "epoch": 0.011755408753561073,
      "grad_norm": 22.5,
      "learning_rate": 1.9999242662595787e-05,
      "loss": 8.2355,
      "step": 4180
    },
    {
      "epoch": 0.011783531741009785,
      "grad_norm": 32.75,
      "learning_rate": 1.999923903381421e-05,
      "loss": 7.953,
      "step": 4190
    },
    {
      "epoch": 0.011811654728458494,
      "grad_norm": 29.75,
      "learning_rate": 1.9999235396360078e-05,
      "loss": 8.5966,
      "step": 4200
    },
    {
      "epoch": 0.011839777715907206,
      "grad_norm": 38.25,
      "learning_rate": 1.9999231750233393e-05,
      "loss": 8.3424,
      "step": 4210
    },
    {
      "epoch": 0.011867900703355916,
      "grad_norm": 28.125,
      "learning_rate": 1.9999228095434162e-05,
      "loss": 8.7131,
      "step": 4220
    },
    {
      "epoch": 0.011896023690804627,
      "grad_norm": 30.25,
      "learning_rate": 1.9999224431962384e-05,
      "loss": 7.8723,
      "step": 4230
    },
    {
      "epoch": 0.011924146678253337,
      "grad_norm": 25.875,
      "learning_rate": 1.9999220759818062e-05,
      "loss": 7.92,
      "step": 4240
    },
    {
      "epoch": 0.011952269665702049,
      "grad_norm": 40.5,
      "learning_rate": 1.99992170790012e-05,
      "loss": 8.0392,
      "step": 4250
    },
    {
      "epoch": 0.011980392653150759,
      "grad_norm": 26.125,
      "learning_rate": 1.9999213389511803e-05,
      "loss": 7.944,
      "step": 4260
    },
    {
      "epoch": 0.01200851564059947,
      "grad_norm": 31.75,
      "learning_rate": 1.999920969134987e-05,
      "loss": 8.143,
      "step": 4270
    },
    {
      "epoch": 0.01203663862804818,
      "grad_norm": 36.25,
      "learning_rate": 1.9999205984515407e-05,
      "loss": 7.5141,
      "step": 4280
    },
    {
      "epoch": 0.012064761615496892,
      "grad_norm": 29.0,
      "learning_rate": 1.999920226900842e-05,
      "loss": 7.7645,
      "step": 4290
    },
    {
      "epoch": 0.012092884602945601,
      "grad_norm": 29.75,
      "learning_rate": 1.9999198544828906e-05,
      "loss": 8.5484,
      "step": 4300
    },
    {
      "epoch": 0.012121007590394313,
      "grad_norm": 25.0,
      "learning_rate": 1.9999194811976873e-05,
      "loss": 8.0031,
      "step": 4310
    },
    {
      "epoch": 0.012149130577843023,
      "grad_norm": 35.25,
      "learning_rate": 1.9999191070452323e-05,
      "loss": 7.8425,
      "step": 4320
    },
    {
      "epoch": 0.012177253565291734,
      "grad_norm": 30.5,
      "learning_rate": 1.999918732025526e-05,
      "loss": 8.5762,
      "step": 4330
    },
    {
      "epoch": 0.012205376552740444,
      "grad_norm": 23.375,
      "learning_rate": 1.9999183561385685e-05,
      "loss": 7.4557,
      "step": 4340
    },
    {
      "epoch": 0.012233499540189156,
      "grad_norm": 43.75,
      "learning_rate": 1.9999179793843604e-05,
      "loss": 8.876,
      "step": 4350
    },
    {
      "epoch": 0.012261622527637865,
      "grad_norm": 26.0,
      "learning_rate": 1.999917601762902e-05,
      "loss": 7.8134,
      "step": 4360
    },
    {
      "epoch": 0.012289745515086577,
      "grad_norm": 55.5,
      "learning_rate": 1.999917223274193e-05,
      "loss": 8.4313,
      "step": 4370
    },
    {
      "epoch": 0.012317868502535287,
      "grad_norm": 22.25,
      "learning_rate": 1.9999168439182344e-05,
      "loss": 7.781,
      "step": 4380
    },
    {
      "epoch": 0.012345991489983998,
      "grad_norm": 25.875,
      "learning_rate": 1.9999164636950265e-05,
      "loss": 8.3306,
      "step": 4390
    },
    {
      "epoch": 0.012374114477432708,
      "grad_norm": 37.5,
      "learning_rate": 1.9999160826045694e-05,
      "loss": 8.1507,
      "step": 4400
    },
    {
      "epoch": 0.01240223746488142,
      "grad_norm": 23.875,
      "learning_rate": 1.9999157006468637e-05,
      "loss": 8.6167,
      "step": 4410
    },
    {
      "epoch": 0.01243036045233013,
      "grad_norm": 20.875,
      "learning_rate": 1.9999153178219094e-05,
      "loss": 8.5502,
      "step": 4420
    },
    {
      "epoch": 0.012458483439778841,
      "grad_norm": 58.0,
      "learning_rate": 1.999914934129707e-05,
      "loss": 8.4065,
      "step": 4430
    },
    {
      "epoch": 0.012486606427227551,
      "grad_norm": 21.25,
      "learning_rate": 1.999914549570257e-05,
      "loss": 7.2662,
      "step": 4440
    },
    {
      "epoch": 0.012514729414676263,
      "grad_norm": 44.75,
      "learning_rate": 1.9999141641435593e-05,
      "loss": 7.3791,
      "step": 4450
    },
    {
      "epoch": 0.012542852402124972,
      "grad_norm": 33.25,
      "learning_rate": 1.9999137778496145e-05,
      "loss": 7.6404,
      "step": 4460
    },
    {
      "epoch": 0.012570975389573684,
      "grad_norm": 32.5,
      "learning_rate": 1.999913390688423e-05,
      "loss": 7.3878,
      "step": 4470
    },
    {
      "epoch": 0.012599098377022394,
      "grad_norm": 27.5,
      "learning_rate": 1.9999130026599852e-05,
      "loss": 8.4735,
      "step": 4480
    },
    {
      "epoch": 0.012627221364471105,
      "grad_norm": 76.5,
      "learning_rate": 1.999912613764301e-05,
      "loss": 8.2022,
      "step": 4490
    },
    {
      "epoch": 0.012655344351919815,
      "grad_norm": 24.0,
      "learning_rate": 1.9999122240013714e-05,
      "loss": 8.0111,
      "step": 4500
    },
    {
      "epoch": 0.012683467339368527,
      "grad_norm": 30.375,
      "learning_rate": 1.9999118333711963e-05,
      "loss": 7.6184,
      "step": 4510
    },
    {
      "epoch": 0.012711590326817237,
      "grad_norm": 28.125,
      "learning_rate": 1.9999114418737756e-05,
      "loss": 8.358,
      "step": 4520
    },
    {
      "epoch": 0.012739713314265948,
      "grad_norm": 57.5,
      "learning_rate": 1.9999110495091107e-05,
      "loss": 7.808,
      "step": 4530
    },
    {
      "epoch": 0.012767836301714658,
      "grad_norm": 24.5,
      "learning_rate": 1.999910656277201e-05,
      "loss": 8.7598,
      "step": 4540
    },
    {
      "epoch": 0.01279595928916337,
      "grad_norm": 39.5,
      "learning_rate": 1.9999102621780478e-05,
      "loss": 8.854,
      "step": 4550
    },
    {
      "epoch": 0.01282408227661208,
      "grad_norm": 47.75,
      "learning_rate": 1.9999098672116504e-05,
      "loss": 7.4798,
      "step": 4560
    },
    {
      "epoch": 0.01285220526406079,
      "grad_norm": 26.375,
      "learning_rate": 1.99990947137801e-05,
      "loss": 7.779,
      "step": 4570
    },
    {
      "epoch": 0.0128803282515095,
      "grad_norm": 89.5,
      "learning_rate": 1.999909074677126e-05,
      "loss": 8.3723,
      "step": 4580
    },
    {
      "epoch": 0.012908451238958212,
      "grad_norm": 41.5,
      "learning_rate": 1.9999086771089997e-05,
      "loss": 8.441,
      "step": 4590
    },
    {
      "epoch": 0.012936574226406922,
      "grad_norm": 30.125,
      "learning_rate": 1.9999082786736308e-05,
      "loss": 8.4291,
      "step": 4600
    },
    {
      "epoch": 0.012964697213855634,
      "grad_norm": 37.75,
      "learning_rate": 1.99990787937102e-05,
      "loss": 7.896,
      "step": 4610
    },
    {
      "epoch": 0.012992820201304343,
      "grad_norm": 37.75,
      "learning_rate": 1.9999074792011678e-05,
      "loss": 7.9664,
      "step": 4620
    },
    {
      "epoch": 0.013020943188753055,
      "grad_norm": 37.25,
      "learning_rate": 1.999907078164074e-05,
      "loss": 8.7379,
      "step": 4630
    },
    {
      "epoch": 0.013049066176201765,
      "grad_norm": 31.5,
      "learning_rate": 1.9999066762597393e-05,
      "loss": 8.4004,
      "step": 4640
    },
    {
      "epoch": 0.013077189163650476,
      "grad_norm": 21.375,
      "learning_rate": 1.9999062734881637e-05,
      "loss": 7.1223,
      "step": 4650
    },
    {
      "epoch": 0.013105312151099186,
      "grad_norm": 24.125,
      "learning_rate": 1.9999058698493483e-05,
      "loss": 7.5563,
      "step": 4660
    },
    {
      "epoch": 0.013133435138547898,
      "grad_norm": 31.125,
      "learning_rate": 1.9999054653432926e-05,
      "loss": 8.6648,
      "step": 4670
    },
    {
      "epoch": 0.013161558125996608,
      "grad_norm": 32.0,
      "learning_rate": 1.9999050599699977e-05,
      "loss": 8.3401,
      "step": 4680
    },
    {
      "epoch": 0.013189681113445319,
      "grad_norm": 23.125,
      "learning_rate": 1.9999046537294634e-05,
      "loss": 7.662,
      "step": 4690
    },
    {
      "epoch": 0.013217804100894029,
      "grad_norm": 31.625,
      "learning_rate": 1.9999042466216903e-05,
      "loss": 7.9025,
      "step": 4700
    },
    {
      "epoch": 0.01324592708834274,
      "grad_norm": 29.375,
      "learning_rate": 1.9999038386466785e-05,
      "loss": 7.3699,
      "step": 4710
    },
    {
      "epoch": 0.013274050075791452,
      "grad_norm": 45.0,
      "learning_rate": 1.9999034298044286e-05,
      "loss": 8.7693,
      "step": 4720
    },
    {
      "epoch": 0.013302173063240162,
      "grad_norm": 25.25,
      "learning_rate": 1.999903020094941e-05,
      "loss": 7.4744,
      "step": 4730
    },
    {
      "epoch": 0.013330296050688873,
      "grad_norm": 27.5,
      "learning_rate": 1.999902609518216e-05,
      "loss": 8.1181,
      "step": 4740
    },
    {
      "epoch": 0.013358419038137583,
      "grad_norm": 29.5,
      "learning_rate": 1.9999021980742536e-05,
      "loss": 7.7701,
      "step": 4750
    },
    {
      "epoch": 0.013386542025586295,
      "grad_norm": 34.0,
      "learning_rate": 1.999901785763055e-05,
      "loss": 8.7794,
      "step": 4760
    },
    {
      "epoch": 0.013414665013035005,
      "grad_norm": 24.0,
      "learning_rate": 1.9999013725846197e-05,
      "loss": 7.8328,
      "step": 4770
    },
    {
      "epoch": 0.013442788000483716,
      "grad_norm": 37.75,
      "learning_rate": 1.999900958538948e-05,
      "loss": 8.4463,
      "step": 4780
    },
    {
      "epoch": 0.013470910987932426,
      "grad_norm": 24.25,
      "learning_rate": 1.9999005436260416e-05,
      "loss": 7.8103,
      "step": 4790
    },
    {
      "epoch": 0.013499033975381138,
      "grad_norm": 38.75,
      "learning_rate": 1.999900127845899e-05,
      "loss": 8.541,
      "step": 4800
    },
    {
      "epoch": 0.013527156962829847,
      "grad_norm": 34.25,
      "learning_rate": 1.9998997111985218e-05,
      "loss": 8.0218,
      "step": 4810
    },
    {
      "epoch": 0.013555279950278559,
      "grad_norm": 25.25,
      "learning_rate": 1.99989929368391e-05,
      "loss": 8.3862,
      "step": 4820
    },
    {
      "epoch": 0.013583402937727269,
      "grad_norm": 22.5,
      "learning_rate": 1.9998988753020642e-05,
      "loss": 8.297,
      "step": 4830
    },
    {
      "epoch": 0.01361152592517598,
      "grad_norm": 29.375,
      "learning_rate": 1.9998984560529842e-05,
      "loss": 7.6039,
      "step": 4840
    },
    {
      "epoch": 0.01363964891262469,
      "grad_norm": 25.5,
      "learning_rate": 1.999898035936671e-05,
      "loss": 7.9024,
      "step": 4850
    },
    {
      "epoch": 0.013667771900073402,
      "grad_norm": 36.25,
      "learning_rate": 1.9998976149531243e-05,
      "loss": 8.1757,
      "step": 4860
    },
    {
      "epoch": 0.013695894887522111,
      "grad_norm": 26.0,
      "learning_rate": 1.9998971931023454e-05,
      "loss": 8.3246,
      "step": 4870
    },
    {
      "epoch": 0.013724017874970823,
      "grad_norm": 21.75,
      "learning_rate": 1.9998967703843338e-05,
      "loss": 8.3151,
      "step": 4880
    },
    {
      "epoch": 0.013752140862419533,
      "grad_norm": 23.25,
      "learning_rate": 1.9998963467990902e-05,
      "loss": 8.3641,
      "step": 4890
    },
    {
      "epoch": 0.013780263849868244,
      "grad_norm": 26.5,
      "learning_rate": 1.9998959223466147e-05,
      "loss": 7.4782,
      "step": 4900
    },
    {
      "epoch": 0.013808386837316954,
      "grad_norm": 26.375,
      "learning_rate": 1.9998954970269082e-05,
      "loss": 8.3574,
      "step": 4910
    },
    {
      "epoch": 0.013836509824765666,
      "grad_norm": 37.5,
      "learning_rate": 1.9998950708399708e-05,
      "loss": 8.1733,
      "step": 4920
    },
    {
      "epoch": 0.013864632812214376,
      "grad_norm": 40.25,
      "learning_rate": 1.9998946437858027e-05,
      "loss": 8.5876,
      "step": 4930
    },
    {
      "epoch": 0.013892755799663087,
      "grad_norm": 20.875,
      "learning_rate": 1.9998942158644047e-05,
      "loss": 8.8059,
      "step": 4940
    },
    {
      "epoch": 0.013920878787111797,
      "grad_norm": 48.0,
      "learning_rate": 1.9998937870757765e-05,
      "loss": 8.2536,
      "step": 4950
    },
    {
      "epoch": 0.013949001774560509,
      "grad_norm": 22.5,
      "learning_rate": 1.999893357419919e-05,
      "loss": 7.8894,
      "step": 4960
    },
    {
      "epoch": 0.013977124762009218,
      "grad_norm": 35.25,
      "learning_rate": 1.9998929268968325e-05,
      "loss": 8.1565,
      "step": 4970
    },
    {
      "epoch": 0.01400524774945793,
      "grad_norm": 27.875,
      "learning_rate": 1.9998924955065174e-05,
      "loss": 7.1035,
      "step": 4980
    },
    {
      "epoch": 0.01403337073690664,
      "grad_norm": 32.25,
      "learning_rate": 1.9998920632489734e-05,
      "loss": 8.3748,
      "step": 4990
    },
    {
      "epoch": 0.014061493724355351,
      "grad_norm": 25.625,
      "learning_rate": 1.9998916301242023e-05,
      "loss": 7.5022,
      "step": 5000
    },
    {
      "epoch": 0.014089616711804061,
      "grad_norm": 22.5,
      "learning_rate": 1.9998911961322028e-05,
      "loss": 8.234,
      "step": 5010
    },
    {
      "epoch": 0.014117739699252773,
      "grad_norm": 46.25,
      "learning_rate": 1.9998907612729768e-05,
      "loss": 7.8939,
      "step": 5020
    },
    {
      "epoch": 0.014145862686701483,
      "grad_norm": 39.5,
      "learning_rate": 1.999890325546524e-05,
      "loss": 8.5412,
      "step": 5030
    },
    {
      "epoch": 0.014173985674150194,
      "grad_norm": 44.25,
      "learning_rate": 1.999889888952844e-05,
      "loss": 7.7492,
      "step": 5040
    },
    {
      "epoch": 0.014202108661598904,
      "grad_norm": 47.5,
      "learning_rate": 1.999889451491939e-05,
      "loss": 8.2012,
      "step": 5050
    },
    {
      "epoch": 0.014230231649047615,
      "grad_norm": 26.0,
      "learning_rate": 1.9998890131638073e-05,
      "loss": 7.9138,
      "step": 5060
    },
    {
      "epoch": 0.014258354636496325,
      "grad_norm": 27.75,
      "learning_rate": 1.999888573968451e-05,
      "loss": 7.5371,
      "step": 5070
    },
    {
      "epoch": 0.014286477623945037,
      "grad_norm": 24.875,
      "learning_rate": 1.9998881339058694e-05,
      "loss": 8.8557,
      "step": 5080
    },
    {
      "epoch": 0.014314600611393747,
      "grad_norm": 45.75,
      "learning_rate": 1.9998876929760633e-05,
      "loss": 8.1463,
      "step": 5090
    },
    {
      "epoch": 0.014342723598842458,
      "grad_norm": 30.0,
      "learning_rate": 1.9998872511790333e-05,
      "loss": 8.3442,
      "step": 5100
    },
    {
      "epoch": 0.014370846586291168,
      "grad_norm": 33.25,
      "learning_rate": 1.9998868085147792e-05,
      "loss": 7.7405,
      "step": 5110
    },
    {
      "epoch": 0.01439896957373988,
      "grad_norm": 24.625,
      "learning_rate": 1.999886364983302e-05,
      "loss": 8.1353,
      "step": 5120
    },
    {
      "epoch": 0.01442709256118859,
      "grad_norm": 30.875,
      "learning_rate": 1.999885920584602e-05,
      "loss": 8.4558,
      "step": 5130
    },
    {
      "epoch": 0.014455215548637301,
      "grad_norm": 26.0,
      "learning_rate": 1.9998854753186786e-05,
      "loss": 8.3118,
      "step": 5140
    },
    {
      "epoch": 0.01448333853608601,
      "grad_norm": 49.25,
      "learning_rate": 1.9998850291855335e-05,
      "loss": 7.6215,
      "step": 5150
    },
    {
      "epoch": 0.014511461523534722,
      "grad_norm": 24.625,
      "learning_rate": 1.9998845821851666e-05,
      "loss": 7.5545,
      "step": 5160
    },
    {
      "epoch": 0.014539584510983432,
      "grad_norm": 27.25,
      "learning_rate": 1.9998841343175783e-05,
      "loss": 7.6474,
      "step": 5170
    },
    {
      "epoch": 0.014567707498432144,
      "grad_norm": 32.75,
      "learning_rate": 1.9998836855827684e-05,
      "loss": 8.2626,
      "step": 5180
    },
    {
      "epoch": 0.014595830485880854,
      "grad_norm": 28.625,
      "learning_rate": 1.999883235980738e-05,
      "loss": 8.6014,
      "step": 5190
    },
    {
      "epoch": 0.014623953473329565,
      "grad_norm": 53.25,
      "learning_rate": 1.999882785511488e-05,
      "loss": 8.1841,
      "step": 5200
    },
    {
      "epoch": 0.014652076460778275,
      "grad_norm": 23.25,
      "learning_rate": 1.9998823341750176e-05,
      "loss": 7.6087,
      "step": 5210
    },
    {
      "epoch": 0.014680199448226986,
      "grad_norm": 24.0,
      "learning_rate": 1.9998818819713277e-05,
      "loss": 8.3995,
      "step": 5220
    },
    {
      "epoch": 0.014708322435675696,
      "grad_norm": 45.75,
      "learning_rate": 1.9998814289004186e-05,
      "loss": 7.9991,
      "step": 5230
    },
    {
      "epoch": 0.014736445423124408,
      "grad_norm": 22.375,
      "learning_rate": 1.9998809749622912e-05,
      "loss": 8.1138,
      "step": 5240
    },
    {
      "epoch": 0.014764568410573118,
      "grad_norm": 31.0,
      "learning_rate": 1.999880520156945e-05,
      "loss": 8.8926,
      "step": 5250
    },
    {
      "epoch": 0.01479269139802183,
      "grad_norm": 31.0,
      "learning_rate": 1.999880064484381e-05,
      "loss": 8.3993,
      "step": 5260
    },
    {
      "epoch": 0.014820814385470539,
      "grad_norm": 24.125,
      "learning_rate": 1.9998796079445993e-05,
      "loss": 7.889,
      "step": 5270
    },
    {
      "epoch": 0.01484893737291925,
      "grad_norm": 39.75,
      "learning_rate": 1.999879150537601e-05,
      "loss": 7.8774,
      "step": 5280
    },
    {
      "epoch": 0.01487706036036796,
      "grad_norm": 26.125,
      "learning_rate": 1.9998786922633854e-05,
      "loss": 8.0017,
      "step": 5290
    },
    {
      "epoch": 0.014905183347816672,
      "grad_norm": 24.125,
      "learning_rate": 1.999878233121954e-05,
      "loss": 7.0155,
      "step": 5300
    },
    {
      "epoch": 0.014933306335265382,
      "grad_norm": 29.625,
      "learning_rate": 1.9998777731133063e-05,
      "loss": 7.2782,
      "step": 5310
    },
    {
      "epoch": 0.014961429322714093,
      "grad_norm": 59.0,
      "learning_rate": 1.999877312237443e-05,
      "loss": 8.1134,
      "step": 5320
    },
    {
      "epoch": 0.014989552310162803,
      "grad_norm": 46.5,
      "learning_rate": 1.9998768504943643e-05,
      "loss": 8.6425,
      "step": 5330
    },
    {
      "epoch": 0.015017675297611515,
      "grad_norm": 34.0,
      "learning_rate": 1.9998763878840713e-05,
      "loss": 8.0684,
      "step": 5340
    },
    {
      "epoch": 0.015045798285060225,
      "grad_norm": 28.875,
      "learning_rate": 1.999875924406564e-05,
      "loss": 7.882,
      "step": 5350
    },
    {
      "epoch": 0.015073921272508936,
      "grad_norm": 40.25,
      "learning_rate": 1.9998754600618425e-05,
      "loss": 8.1564,
      "step": 5360
    },
    {
      "epoch": 0.015102044259957646,
      "grad_norm": 24.875,
      "learning_rate": 1.999874994849907e-05,
      "loss": 8.5689,
      "step": 5370
    },
    {
      "epoch": 0.015130167247406357,
      "grad_norm": 30.25,
      "learning_rate": 1.999874528770759e-05,
      "loss": 8.4312,
      "step": 5380
    },
    {
      "epoch": 0.015158290234855069,
      "grad_norm": 25.625,
      "learning_rate": 1.9998740618243983e-05,
      "loss": 8.3391,
      "step": 5390
    },
    {
      "epoch": 0.015186413222303779,
      "grad_norm": 24.875,
      "learning_rate": 1.9998735940108252e-05,
      "loss": 8.267,
      "step": 5400
    },
    {
      "epoch": 0.01521453620975249,
      "grad_norm": 28.125,
      "learning_rate": 1.9998731253300402e-05,
      "loss": 8.5453,
      "step": 5410
    },
    {
      "epoch": 0.0152426591972012,
      "grad_norm": 49.25,
      "learning_rate": 1.9998726557820432e-05,
      "loss": 7.5432,
      "step": 5420
    },
    {
      "epoch": 0.015270782184649912,
      "grad_norm": 34.0,
      "learning_rate": 1.9998721853668356e-05,
      "loss": 8.2142,
      "step": 5430
    },
    {
      "epoch": 0.015298905172098622,
      "grad_norm": 40.5,
      "learning_rate": 1.9998717140844173e-05,
      "loss": 8.1221,
      "step": 5440
    },
    {
      "epoch": 0.015327028159547333,
      "grad_norm": 23.375,
      "learning_rate": 1.9998712419347884e-05,
      "loss": 7.7446,
      "step": 5450
    },
    {
      "epoch": 0.015355151146996043,
      "grad_norm": 30.125,
      "learning_rate": 1.99987076891795e-05,
      "loss": 7.8589,
      "step": 5460
    },
    {
      "epoch": 0.015383274134444755,
      "grad_norm": 22.0,
      "learning_rate": 1.999870295033902e-05,
      "loss": 7.7178,
      "step": 5470
    },
    {
      "epoch": 0.015411397121893464,
      "grad_norm": 23.875,
      "learning_rate": 1.9998698202826444e-05,
      "loss": 7.9956,
      "step": 5480
    },
    {
      "epoch": 0.015439520109342176,
      "grad_norm": 39.0,
      "learning_rate": 1.9998693446641787e-05,
      "loss": 8.2401,
      "step": 5490
    },
    {
      "epoch": 0.015467643096790886,
      "grad_norm": 27.625,
      "learning_rate": 1.999868868178505e-05,
      "loss": 7.4811,
      "step": 5500
    },
    {
      "epoch": 0.015495766084239597,
      "grad_norm": 26.5,
      "learning_rate": 1.999868390825623e-05,
      "loss": 8.83,
      "step": 5510
    },
    {
      "epoch": 0.015523889071688307,
      "grad_norm": 25.625,
      "learning_rate": 1.9998679126055336e-05,
      "loss": 7.7178,
      "step": 5520
    },
    {
      "epoch": 0.015552012059137019,
      "grad_norm": 31.125,
      "learning_rate": 1.9998674335182373e-05,
      "loss": 8.4372,
      "step": 5530
    },
    {
      "epoch": 0.015580135046585728,
      "grad_norm": 21.5,
      "learning_rate": 1.9998669535637345e-05,
      "loss": 8.2238,
      "step": 5540
    },
    {
      "epoch": 0.01560825803403444,
      "grad_norm": 25.125,
      "learning_rate": 1.9998664727420257e-05,
      "loss": 7.3885,
      "step": 5550
    },
    {
      "epoch": 0.01563638102148315,
      "grad_norm": 31.125,
      "learning_rate": 1.9998659910531108e-05,
      "loss": 8.2133,
      "step": 5560
    },
    {
      "epoch": 0.01566450400893186,
      "grad_norm": 36.0,
      "learning_rate": 1.9998655084969906e-05,
      "loss": 8.048,
      "step": 5570
    },
    {
      "epoch": 0.015692626996380573,
      "grad_norm": 25.875,
      "learning_rate": 1.9998650250736655e-05,
      "loss": 8.9737,
      "step": 5580
    },
    {
      "epoch": 0.01572074998382928,
      "grad_norm": 31.0,
      "learning_rate": 1.999864540783136e-05,
      "loss": 8.1481,
      "step": 5590
    },
    {
      "epoch": 0.015748872971277993,
      "grad_norm": 35.75,
      "learning_rate": 1.9998640556254022e-05,
      "loss": 7.7248,
      "step": 5600
    },
    {
      "epoch": 0.015776995958726704,
      "grad_norm": 27.125,
      "learning_rate": 1.9998635696004653e-05,
      "loss": 8.5561,
      "step": 5610
    },
    {
      "epoch": 0.015805118946175416,
      "grad_norm": 43.0,
      "learning_rate": 1.999863082708325e-05,
      "loss": 8.6236,
      "step": 5620
    },
    {
      "epoch": 0.015833241933624124,
      "grad_norm": 28.75,
      "learning_rate": 1.9998625949489816e-05,
      "loss": 7.8709,
      "step": 5630
    },
    {
      "epoch": 0.015861364921072835,
      "grad_norm": 38.0,
      "learning_rate": 1.9998621063224358e-05,
      "loss": 7.9756,
      "step": 5640
    },
    {
      "epoch": 0.015889487908521547,
      "grad_norm": 22.25,
      "learning_rate": 1.9998616168286882e-05,
      "loss": 8.3834,
      "step": 5650
    },
    {
      "epoch": 0.01591761089597026,
      "grad_norm": 27.25,
      "learning_rate": 1.999861126467739e-05,
      "loss": 8.5246,
      "step": 5660
    },
    {
      "epoch": 0.015945733883418967,
      "grad_norm": 72.5,
      "learning_rate": 1.9998606352395886e-05,
      "loss": 7.8261,
      "step": 5670
    },
    {
      "epoch": 0.015973856870867678,
      "grad_norm": 21.125,
      "learning_rate": 1.999860143144238e-05,
      "loss": 7.8429,
      "step": 5680
    },
    {
      "epoch": 0.01600197985831639,
      "grad_norm": 33.25,
      "learning_rate": 1.9998596501816868e-05,
      "loss": 7.8281,
      "step": 5690
    },
    {
      "epoch": 0.0160301028457651,
      "grad_norm": 21.875,
      "learning_rate": 1.9998591563519355e-05,
      "loss": 8.1761,
      "step": 5700
    },
    {
      "epoch": 0.01605822583321381,
      "grad_norm": 21.75,
      "learning_rate": 1.999858661654985e-05,
      "loss": 7.5968,
      "step": 5710
    },
    {
      "epoch": 0.01608634882066252,
      "grad_norm": 38.75,
      "learning_rate": 1.9998581660908357e-05,
      "loss": 8.1945,
      "step": 5720
    },
    {
      "epoch": 0.016114471808111232,
      "grad_norm": 21.5,
      "learning_rate": 1.9998576696594874e-05,
      "loss": 8.4048,
      "step": 5730
    },
    {
      "epoch": 0.016142594795559944,
      "grad_norm": 25.75,
      "learning_rate": 1.9998571723609418e-05,
      "loss": 8.0875,
      "step": 5740
    },
    {
      "epoch": 0.016170717783008652,
      "grad_norm": 26.875,
      "learning_rate": 1.999856674195198e-05,
      "loss": 8.2936,
      "step": 5750
    },
    {
      "epoch": 0.016198840770457364,
      "grad_norm": 31.625,
      "learning_rate": 1.999856175162257e-05,
      "loss": 7.3004,
      "step": 5760
    },
    {
      "epoch": 0.016226963757906075,
      "grad_norm": 37.0,
      "learning_rate": 1.999855675262119e-05,
      "loss": 7.6768,
      "step": 5770
    },
    {
      "epoch": 0.016255086745354787,
      "grad_norm": 28.75,
      "learning_rate": 1.9998551744947847e-05,
      "loss": 7.3202,
      "step": 5780
    },
    {
      "epoch": 0.016283209732803495,
      "grad_norm": 24.75,
      "learning_rate": 1.9998546728602546e-05,
      "loss": 7.2178,
      "step": 5790
    },
    {
      "epoch": 0.016311332720252206,
      "grad_norm": 26.25,
      "learning_rate": 1.999854170358529e-05,
      "loss": 8.1874,
      "step": 5800
    },
    {
      "epoch": 0.016339455707700918,
      "grad_norm": 28.25,
      "learning_rate": 1.999853666989608e-05,
      "loss": 8.938,
      "step": 5810
    },
    {
      "epoch": 0.01636757869514963,
      "grad_norm": 34.5,
      "learning_rate": 1.999853162753493e-05,
      "loss": 8.9892,
      "step": 5820
    },
    {
      "epoch": 0.016395701682598338,
      "grad_norm": 30.375,
      "learning_rate": 1.999852657650183e-05,
      "loss": 8.0376,
      "step": 5830
    },
    {
      "epoch": 0.01642382467004705,
      "grad_norm": 23.125,
      "learning_rate": 1.99985215167968e-05,
      "loss": 8.7777,
      "step": 5840
    },
    {
      "epoch": 0.01645194765749576,
      "grad_norm": 22.5,
      "learning_rate": 1.999851644841983e-05,
      "loss": 8.1604,
      "step": 5850
    },
    {
      "epoch": 0.016480070644944472,
      "grad_norm": 27.0,
      "learning_rate": 1.9998511371370935e-05,
      "loss": 7.5964,
      "step": 5860
    },
    {
      "epoch": 0.01650819363239318,
      "grad_norm": 31.75,
      "learning_rate": 1.9998506285650116e-05,
      "loss": 8.5186,
      "step": 5870
    },
    {
      "epoch": 0.016536316619841892,
      "grad_norm": 45.0,
      "learning_rate": 1.9998501191257374e-05,
      "loss": 8.4827,
      "step": 5880
    },
    {
      "epoch": 0.016564439607290603,
      "grad_norm": 51.75,
      "learning_rate": 1.999849608819272e-05,
      "loss": 8.1855,
      "step": 5890
    },
    {
      "epoch": 0.016592562594739315,
      "grad_norm": 34.5,
      "learning_rate": 1.999849097645615e-05,
      "loss": 7.9797,
      "step": 5900
    },
    {
      "epoch": 0.016620685582188023,
      "grad_norm": 35.75,
      "learning_rate": 1.999848585604768e-05,
      "loss": 8.892,
      "step": 5910
    },
    {
      "epoch": 0.016648808569636735,
      "grad_norm": 25.75,
      "learning_rate": 1.99984807269673e-05,
      "loss": 8.8391,
      "step": 5920
    },
    {
      "epoch": 0.016676931557085446,
      "grad_norm": 21.625,
      "learning_rate": 1.9998475589215027e-05,
      "loss": 8.8296,
      "step": 5930
    },
    {
      "epoch": 0.016705054544534158,
      "grad_norm": 33.75,
      "learning_rate": 1.9998470442790857e-05,
      "loss": 8.1961,
      "step": 5940
    },
    {
      "epoch": 0.016733177531982866,
      "grad_norm": 25.875,
      "learning_rate": 1.99984652876948e-05,
      "loss": 8.8623,
      "step": 5950
    },
    {
      "epoch": 0.016761300519431577,
      "grad_norm": 41.5,
      "learning_rate": 1.9998460123926862e-05,
      "loss": 8.1579,
      "step": 5960
    },
    {
      "epoch": 0.01678942350688029,
      "grad_norm": 30.875,
      "learning_rate": 1.999845495148704e-05,
      "loss": 7.9915,
      "step": 5970
    },
    {
      "epoch": 0.016817546494329,
      "grad_norm": 36.75,
      "learning_rate": 1.9998449770375343e-05,
      "loss": 8.344,
      "step": 5980
    },
    {
      "epoch": 0.01684566948177771,
      "grad_norm": 33.25,
      "learning_rate": 1.9998444580591776e-05,
      "loss": 8.2671,
      "step": 5990
    },
    {
      "epoch": 0.01687379246922642,
      "grad_norm": 24.125,
      "learning_rate": 1.9998439382136344e-05,
      "loss": 8.0181,
      "step": 6000
    },
    {
      "epoch": 0.01690191545667513,
      "grad_norm": 39.5,
      "learning_rate": 1.9998434175009046e-05,
      "loss": 8.2912,
      "step": 6010
    },
    {
      "epoch": 0.016930038444123843,
      "grad_norm": 30.125,
      "learning_rate": 1.9998428959209893e-05,
      "loss": 8.5638,
      "step": 6020
    },
    {
      "epoch": 0.01695816143157255,
      "grad_norm": 44.75,
      "learning_rate": 1.9998423734738887e-05,
      "loss": 7.8268,
      "step": 6030
    },
    {
      "epoch": 0.016986284419021263,
      "grad_norm": 29.0,
      "learning_rate": 1.999841850159603e-05,
      "loss": 7.3282,
      "step": 6040
    },
    {
      "epoch": 0.017014407406469974,
      "grad_norm": 21.625,
      "learning_rate": 1.999841325978133e-05,
      "loss": 7.9607,
      "step": 6050
    },
    {
      "epoch": 0.017042530393918686,
      "grad_norm": 35.75,
      "learning_rate": 1.999840800929479e-05,
      "loss": 8.2691,
      "step": 6060
    },
    {
      "epoch": 0.017070653381367398,
      "grad_norm": 45.25,
      "learning_rate": 1.9998402750136417e-05,
      "loss": 7.5812,
      "step": 6070
    },
    {
      "epoch": 0.017098776368816106,
      "grad_norm": 27.125,
      "learning_rate": 1.9998397482306215e-05,
      "loss": 7.8885,
      "step": 6080
    },
    {
      "epoch": 0.017126899356264817,
      "grad_norm": 25.75,
      "learning_rate": 1.9998392205804184e-05,
      "loss": 8.3677,
      "step": 6090
    },
    {
      "epoch": 0.01715502234371353,
      "grad_norm": 40.5,
      "learning_rate": 1.9998386920630332e-05,
      "loss": 8.3508,
      "step": 6100
    },
    {
      "epoch": 0.01718314533116224,
      "grad_norm": 32.25,
      "learning_rate": 1.999838162678467e-05,
      "loss": 7.9316,
      "step": 6110
    },
    {
      "epoch": 0.01721126831861095,
      "grad_norm": 26.0,
      "learning_rate": 1.999837632426719e-05,
      "loss": 9.0578,
      "step": 6120
    },
    {
      "epoch": 0.01723939130605966,
      "grad_norm": 30.0,
      "learning_rate": 1.9998371013077903e-05,
      "loss": 8.3462,
      "step": 6130
    },
    {
      "epoch": 0.01726751429350837,
      "grad_norm": 23.25,
      "learning_rate": 1.999836569321681e-05,
      "loss": 8.1155,
      "step": 6140
    },
    {
      "epoch": 0.017295637280957083,
      "grad_norm": 37.5,
      "learning_rate": 1.9998360364683925e-05,
      "loss": 7.1628,
      "step": 6150
    },
    {
      "epoch": 0.01732376026840579,
      "grad_norm": 31.375,
      "learning_rate": 1.9998355027479245e-05,
      "loss": 8.3626,
      "step": 6160
    },
    {
      "epoch": 0.017351883255854503,
      "grad_norm": 28.625,
      "learning_rate": 1.9998349681602773e-05,
      "loss": 8.069,
      "step": 6170
    },
    {
      "epoch": 0.017380006243303214,
      "grad_norm": 41.75,
      "learning_rate": 1.999834432705452e-05,
      "loss": 8.4142,
      "step": 6180
    },
    {
      "epoch": 0.017408129230751926,
      "grad_norm": 33.5,
      "learning_rate": 1.9998338963834486e-05,
      "loss": 8.0817,
      "step": 6190
    },
    {
      "epoch": 0.017436252218200634,
      "grad_norm": 22.125,
      "learning_rate": 1.999833359194268e-05,
      "loss": 7.8722,
      "step": 6200
    },
    {
      "epoch": 0.017464375205649346,
      "grad_norm": 27.5,
      "learning_rate": 1.9998328211379103e-05,
      "loss": 7.586,
      "step": 6210
    },
    {
      "epoch": 0.017492498193098057,
      "grad_norm": 35.0,
      "learning_rate": 1.9998322822143756e-05,
      "loss": 8.5163,
      "step": 6220
    },
    {
      "epoch": 0.01752062118054677,
      "grad_norm": 27.375,
      "learning_rate": 1.999831742423665e-05,
      "loss": 8.1256,
      "step": 6230
    },
    {
      "epoch": 0.017548744167995477,
      "grad_norm": 40.25,
      "learning_rate": 1.999831201765779e-05,
      "loss": 7.4541,
      "step": 6240
    },
    {
      "epoch": 0.017576867155444188,
      "grad_norm": 30.5,
      "learning_rate": 1.9998306602407177e-05,
      "loss": 8.0544,
      "step": 6250
    },
    {
      "epoch": 0.0176049901428929,
      "grad_norm": 28.375,
      "learning_rate": 1.9998301178484815e-05,
      "loss": 8.1091,
      "step": 6260
    },
    {
      "epoch": 0.01763311313034161,
      "grad_norm": 20.875,
      "learning_rate": 1.9998295745890716e-05,
      "loss": 8.4898,
      "step": 6270
    },
    {
      "epoch": 0.01766123611779032,
      "grad_norm": 34.0,
      "learning_rate": 1.9998290304624873e-05,
      "loss": 8.0686,
      "step": 6280
    },
    {
      "epoch": 0.01768935910523903,
      "grad_norm": 30.125,
      "learning_rate": 1.9998284854687304e-05,
      "loss": 9.1744,
      "step": 6290
    },
    {
      "epoch": 0.017717482092687743,
      "grad_norm": 31.25,
      "learning_rate": 1.9998279396078e-05,
      "loss": 7.9447,
      "step": 6300
    },
    {
      "epoch": 0.017745605080136454,
      "grad_norm": 28.375,
      "learning_rate": 1.9998273928796977e-05,
      "loss": 7.4735,
      "step": 6310
    },
    {
      "epoch": 0.017773728067585162,
      "grad_norm": 29.125,
      "learning_rate": 1.9998268452844238e-05,
      "loss": 8.118,
      "step": 6320
    },
    {
      "epoch": 0.017801851055033874,
      "grad_norm": 27.25,
      "learning_rate": 1.9998262968219778e-05,
      "loss": 7.8616,
      "step": 6330
    },
    {
      "epoch": 0.017829974042482585,
      "grad_norm": 48.25,
      "learning_rate": 1.9998257474923615e-05,
      "loss": 8.1934,
      "step": 6340
    },
    {
      "epoch": 0.017858097029931297,
      "grad_norm": 27.875,
      "learning_rate": 1.9998251972955746e-05,
      "loss": 7.6326,
      "step": 6350
    },
    {
      "epoch": 0.017886220017380005,
      "grad_norm": 27.875,
      "learning_rate": 1.9998246462316177e-05,
      "loss": 7.7852,
      "step": 6360
    },
    {
      "epoch": 0.017914343004828717,
      "grad_norm": 33.0,
      "learning_rate": 1.9998240943004915e-05,
      "loss": 7.738,
      "step": 6370
    },
    {
      "epoch": 0.017942465992277428,
      "grad_norm": 51.75,
      "learning_rate": 1.9998235415021963e-05,
      "loss": 7.6015,
      "step": 6380
    },
    {
      "epoch": 0.01797058897972614,
      "grad_norm": 27.875,
      "learning_rate": 1.9998229878367322e-05,
      "loss": 8.0608,
      "step": 6390
    },
    {
      "epoch": 0.017998711967174848,
      "grad_norm": 42.0,
      "learning_rate": 1.9998224333041002e-05,
      "loss": 7.8947,
      "step": 6400
    },
    {
      "epoch": 0.01802683495462356,
      "grad_norm": 45.0,
      "learning_rate": 1.999821877904301e-05,
      "loss": 7.9311,
      "step": 6410
    },
    {
      "epoch": 0.01805495794207227,
      "grad_norm": 24.5,
      "learning_rate": 1.9998213216373347e-05,
      "loss": 8.3355,
      "step": 6420
    },
    {
      "epoch": 0.018083080929520982,
      "grad_norm": 35.75,
      "learning_rate": 1.9998207645032015e-05,
      "loss": 8.7402,
      "step": 6430
    },
    {
      "epoch": 0.01811120391696969,
      "grad_norm": 24.25,
      "learning_rate": 1.999820206501902e-05,
      "loss": 8.3129,
      "step": 6440
    },
    {
      "epoch": 0.018139326904418402,
      "grad_norm": 32.25,
      "learning_rate": 1.9998196476334372e-05,
      "loss": 7.8271,
      "step": 6450
    },
    {
      "epoch": 0.018167449891867114,
      "grad_norm": 45.75,
      "learning_rate": 1.9998190878978074e-05,
      "loss": 8.3484,
      "step": 6460
    },
    {
      "epoch": 0.018195572879315825,
      "grad_norm": 24.25,
      "learning_rate": 1.9998185272950126e-05,
      "loss": 8.5476,
      "step": 6470
    },
    {
      "epoch": 0.018223695866764533,
      "grad_norm": 24.0,
      "learning_rate": 1.999817965825054e-05,
      "loss": 8.7645,
      "step": 6480
    },
    {
      "epoch": 0.018251818854213245,
      "grad_norm": 30.625,
      "learning_rate": 1.9998174034879316e-05,
      "loss": 7.4244,
      "step": 6490
    },
    {
      "epoch": 0.018279941841661956,
      "grad_norm": 23.75,
      "learning_rate": 1.999816840283646e-05,
      "loss": 8.3478,
      "step": 6500
    },
    {
      "epoch": 0.018308064829110668,
      "grad_norm": 23.75,
      "learning_rate": 1.9998162762121977e-05,
      "loss": 7.7965,
      "step": 6510
    },
    {
      "epoch": 0.018336187816559376,
      "grad_norm": 21.0,
      "learning_rate": 1.999815711273587e-05,
      "loss": 8.0261,
      "step": 6520
    },
    {
      "epoch": 0.018364310804008088,
      "grad_norm": 40.75,
      "learning_rate": 1.999815145467815e-05,
      "loss": 7.5809,
      "step": 6530
    },
    {
      "epoch": 0.0183924337914568,
      "grad_norm": 22.5,
      "learning_rate": 1.9998145787948812e-05,
      "loss": 8.1214,
      "step": 6540
    },
    {
      "epoch": 0.01842055677890551,
      "grad_norm": 24.5,
      "learning_rate": 1.9998140112547873e-05,
      "loss": 7.7434,
      "step": 6550
    },
    {
      "epoch": 0.01844867976635422,
      "grad_norm": 33.0,
      "learning_rate": 1.9998134428475327e-05,
      "loss": 8.7913,
      "step": 6560
    },
    {
      "epoch": 0.01847680275380293,
      "grad_norm": 52.0,
      "learning_rate": 1.9998128735731182e-05,
      "loss": 8.5184,
      "step": 6570
    },
    {
      "epoch": 0.018504925741251642,
      "grad_norm": 33.75,
      "learning_rate": 1.999812303431545e-05,
      "loss": 8.0932,
      "step": 6580
    },
    {
      "epoch": 0.018533048728700353,
      "grad_norm": 40.75,
      "learning_rate": 1.9998117324228126e-05,
      "loss": 8.5118,
      "step": 6590
    },
    {
      "epoch": 0.01856117171614906,
      "grad_norm": 26.5,
      "learning_rate": 1.9998111605469218e-05,
      "loss": 8.0939,
      "step": 6600
    },
    {
      "epoch": 0.018589294703597773,
      "grad_norm": 41.0,
      "learning_rate": 1.999810587803874e-05,
      "loss": 8.1472,
      "step": 6610
    },
    {
      "epoch": 0.018617417691046485,
      "grad_norm": 20.875,
      "learning_rate": 1.999810014193668e-05,
      "loss": 8.2657,
      "step": 6620
    },
    {
      "epoch": 0.018645540678495196,
      "grad_norm": 39.5,
      "learning_rate": 1.9998094397163057e-05,
      "loss": 7.6033,
      "step": 6630
    },
    {
      "epoch": 0.018673663665943904,
      "grad_norm": 29.875,
      "learning_rate": 1.999808864371787e-05,
      "loss": 8.4298,
      "step": 6640
    },
    {
      "epoch": 0.018701786653392616,
      "grad_norm": 38.75,
      "learning_rate": 1.9998082881601126e-05,
      "loss": 8.4281,
      "step": 6650
    },
    {
      "epoch": 0.018729909640841327,
      "grad_norm": 23.875,
      "learning_rate": 1.999807711081283e-05,
      "loss": 8.0242,
      "step": 6660
    },
    {
      "epoch": 0.01875803262829004,
      "grad_norm": 22.75,
      "learning_rate": 1.9998071331352985e-05,
      "loss": 7.201,
      "step": 6670
    },
    {
      "epoch": 0.018786155615738747,
      "grad_norm": 19.75,
      "learning_rate": 1.9998065543221597e-05,
      "loss": 7.4634,
      "step": 6680
    },
    {
      "epoch": 0.01881427860318746,
      "grad_norm": 26.125,
      "learning_rate": 1.9998059746418673e-05,
      "loss": 7.6623,
      "step": 6690
    },
    {
      "epoch": 0.01884240159063617,
      "grad_norm": 45.0,
      "learning_rate": 1.9998053940944214e-05,
      "loss": 7.7956,
      "step": 6700
    },
    {
      "epoch": 0.01887052457808488,
      "grad_norm": 29.25,
      "learning_rate": 1.9998048126798225e-05,
      "loss": 7.3149,
      "step": 6710
    },
    {
      "epoch": 0.018898647565533593,
      "grad_norm": 39.0,
      "learning_rate": 1.999804230398072e-05,
      "loss": 7.3412,
      "step": 6720
    },
    {
      "epoch": 0.0189267705529823,
      "grad_norm": 35.5,
      "learning_rate": 1.999803647249169e-05,
      "loss": 8.4152,
      "step": 6730
    },
    {
      "epoch": 0.018954893540431013,
      "grad_norm": 39.25,
      "learning_rate": 1.9998030632331155e-05,
      "loss": 8.6922,
      "step": 6740
    },
    {
      "epoch": 0.018983016527879724,
      "grad_norm": 44.5,
      "learning_rate": 1.9998024783499106e-05,
      "loss": 7.5313,
      "step": 6750
    },
    {
      "epoch": 0.019011139515328436,
      "grad_norm": 44.0,
      "learning_rate": 1.9998018925995557e-05,
      "loss": 7.4298,
      "step": 6760
    },
    {
      "epoch": 0.019039262502777144,
      "grad_norm": 28.125,
      "learning_rate": 1.9998013059820512e-05,
      "loss": 7.4348,
      "step": 6770
    },
    {
      "epoch": 0.019067385490225856,
      "grad_norm": 32.0,
      "learning_rate": 1.9998007184973974e-05,
      "loss": 7.9927,
      "step": 6780
    },
    {
      "epoch": 0.019095508477674567,
      "grad_norm": 34.25,
      "learning_rate": 1.9998001301455946e-05,
      "loss": 7.9237,
      "step": 6790
    },
    {
      "epoch": 0.01912363146512328,
      "grad_norm": 33.25,
      "learning_rate": 1.999799540926644e-05,
      "loss": 7.6304,
      "step": 6800
    },
    {
      "epoch": 0.019151754452571987,
      "grad_norm": 31.5,
      "learning_rate": 1.9997989508405452e-05,
      "loss": 7.4487,
      "step": 6810
    },
    {
      "epoch": 0.0191798774400207,
      "grad_norm": 26.25,
      "learning_rate": 1.9997983598872996e-05,
      "loss": 8.1076,
      "step": 6820
    },
    {
      "epoch": 0.01920800042746941,
      "grad_norm": 34.25,
      "learning_rate": 1.999797768066907e-05,
      "loss": 8.3876,
      "step": 6830
    },
    {
      "epoch": 0.01923612341491812,
      "grad_norm": 25.125,
      "learning_rate": 1.999797175379369e-05,
      "loss": 7.6084,
      "step": 6840
    },
    {
      "epoch": 0.01926424640236683,
      "grad_norm": 35.5,
      "learning_rate": 1.9997965818246846e-05,
      "loss": 7.9332,
      "step": 6850
    },
    {
      "epoch": 0.01929236938981554,
      "grad_norm": 25.5,
      "learning_rate": 1.9997959874028553e-05,
      "loss": 8.2781,
      "step": 6860
    },
    {
      "epoch": 0.019320492377264253,
      "grad_norm": 37.5,
      "learning_rate": 1.999795392113881e-05,
      "loss": 7.4716,
      "step": 6870
    },
    {
      "epoch": 0.019348615364712964,
      "grad_norm": 30.625,
      "learning_rate": 1.9997947959577635e-05,
      "loss": 8.3292,
      "step": 6880
    },
    {
      "epoch": 0.019376738352161672,
      "grad_norm": 29.125,
      "learning_rate": 1.999794198934502e-05,
      "loss": 8.7477,
      "step": 6890
    },
    {
      "epoch": 0.019404861339610384,
      "grad_norm": 28.625,
      "learning_rate": 1.9997936010440974e-05,
      "loss": 8.3882,
      "step": 6900
    },
    {
      "epoch": 0.019432984327059095,
      "grad_norm": 24.875,
      "learning_rate": 1.99979300228655e-05,
      "loss": 7.798,
      "step": 6910
    },
    {
      "epoch": 0.019461107314507807,
      "grad_norm": 33.0,
      "learning_rate": 1.9997924026618606e-05,
      "loss": 8.0021,
      "step": 6920
    },
    {
      "epoch": 0.019489230301956515,
      "grad_norm": 25.25,
      "learning_rate": 1.9997918021700303e-05,
      "loss": 8.1944,
      "step": 6930
    },
    {
      "epoch": 0.019517353289405227,
      "grad_norm": 23.75,
      "learning_rate": 1.9997912008110584e-05,
      "loss": 7.5742,
      "step": 6940
    },
    {
      "epoch": 0.019545476276853938,
      "grad_norm": 28.125,
      "learning_rate": 1.999790598584946e-05,
      "loss": 8.1905,
      "step": 6950
    },
    {
      "epoch": 0.01957359926430265,
      "grad_norm": 26.375,
      "learning_rate": 1.9997899954916943e-05,
      "loss": 7.5328,
      "step": 6960
    },
    {
      "epoch": 0.019601722251751358,
      "grad_norm": 26.375,
      "learning_rate": 1.9997893915313023e-05,
      "loss": 7.9268,
      "step": 6970
    },
    {
      "epoch": 0.01962984523920007,
      "grad_norm": 28.625,
      "learning_rate": 1.9997887867037722e-05,
      "loss": 8.7345,
      "step": 6980
    },
    {
      "epoch": 0.01965796822664878,
      "grad_norm": 32.0,
      "learning_rate": 1.9997881810091033e-05,
      "loss": 7.8559,
      "step": 6990
    },
    {
      "epoch": 0.019686091214097493,
      "grad_norm": 32.75,
      "learning_rate": 1.9997875744472967e-05,
      "loss": 7.7247,
      "step": 7000
    },
    {
      "epoch": 0.0197142142015462,
      "grad_norm": 32.25,
      "learning_rate": 1.999786967018353e-05,
      "loss": 8.5207,
      "step": 7010
    },
    {
      "epoch": 0.019742337188994912,
      "grad_norm": 21.125,
      "learning_rate": 1.999786358722272e-05,
      "loss": 7.7105,
      "step": 7020
    },
    {
      "epoch": 0.019770460176443624,
      "grad_norm": 26.5,
      "learning_rate": 1.999785749559055e-05,
      "loss": 8.5356,
      "step": 7030
    },
    {
      "epoch": 0.019798583163892335,
      "grad_norm": 24.875,
      "learning_rate": 1.9997851395287023e-05,
      "loss": 7.246,
      "step": 7040
    },
    {
      "epoch": 0.019826706151341043,
      "grad_norm": 21.125,
      "learning_rate": 1.9997845286312142e-05,
      "loss": 7.8412,
      "step": 7050
    },
    {
      "epoch": 0.019854829138789755,
      "grad_norm": 23.0,
      "learning_rate": 1.9997839168665917e-05,
      "loss": 8.609,
      "step": 7060
    },
    {
      "epoch": 0.019882952126238466,
      "grad_norm": 26.375,
      "learning_rate": 1.9997833042348348e-05,
      "loss": 9.1348,
      "step": 7070
    },
    {
      "epoch": 0.019911075113687178,
      "grad_norm": 28.625,
      "learning_rate": 1.9997826907359445e-05,
      "loss": 8.5175,
      "step": 7080
    },
    {
      "epoch": 0.019939198101135886,
      "grad_norm": 28.5,
      "learning_rate": 1.9997820763699208e-05,
      "loss": 7.8184,
      "step": 7090
    },
    {
      "epoch": 0.019967321088584598,
      "grad_norm": 27.75,
      "learning_rate": 1.999781461136765e-05,
      "loss": 8.0497,
      "step": 7100
    },
    {
      "epoch": 0.01999544407603331,
      "grad_norm": 25.625,
      "learning_rate": 1.999780845036477e-05,
      "loss": 8.7238,
      "step": 7110
    },
    {
      "epoch": 0.02002356706348202,
      "grad_norm": 21.75,
      "learning_rate": 1.9997802280690573e-05,
      "loss": 8.3382,
      "step": 7120
    },
    {
      "epoch": 0.02005169005093073,
      "grad_norm": 31.125,
      "learning_rate": 1.999779610234507e-05,
      "loss": 7.534,
      "step": 7130
    },
    {
      "epoch": 0.02007981303837944,
      "grad_norm": 26.625,
      "learning_rate": 1.9997789915328262e-05,
      "loss": 8.1322,
      "step": 7140
    },
    {
      "epoch": 0.020107936025828152,
      "grad_norm": 27.625,
      "learning_rate": 1.9997783719640155e-05,
      "loss": 7.6091,
      "step": 7150
    },
    {
      "epoch": 0.020136059013276864,
      "grad_norm": 26.25,
      "learning_rate": 1.9997777515280754e-05,
      "loss": 8.5611,
      "step": 7160
    },
    {
      "epoch": 0.02016418200072557,
      "grad_norm": 24.5,
      "learning_rate": 1.9997771302250064e-05,
      "loss": 8.6052,
      "step": 7170
    },
    {
      "epoch": 0.020192304988174283,
      "grad_norm": 29.875,
      "learning_rate": 1.9997765080548094e-05,
      "loss": 8.4787,
      "step": 7180
    },
    {
      "epoch": 0.020220427975622995,
      "grad_norm": 24.75,
      "learning_rate": 1.9997758850174845e-05,
      "loss": 7.5623,
      "step": 7190
    },
    {
      "epoch": 0.020248550963071706,
      "grad_norm": 46.0,
      "learning_rate": 1.9997752611130323e-05,
      "loss": 8.9582,
      "step": 7200
    },
    {
      "epoch": 0.020276673950520414,
      "grad_norm": 58.25,
      "learning_rate": 1.999774636341454e-05,
      "loss": 8.7094,
      "step": 7210
    },
    {
      "epoch": 0.020304796937969126,
      "grad_norm": 20.375,
      "learning_rate": 1.999774010702749e-05,
      "loss": 7.7926,
      "step": 7220
    },
    {
      "epoch": 0.020332919925417837,
      "grad_norm": 28.0,
      "learning_rate": 1.9997733841969188e-05,
      "loss": 7.9416,
      "step": 7230
    },
    {
      "epoch": 0.02036104291286655,
      "grad_norm": 25.875,
      "learning_rate": 1.9997727568239632e-05,
      "loss": 7.8061,
      "step": 7240
    },
    {
      "epoch": 0.020389165900315257,
      "grad_norm": 19.375,
      "learning_rate": 1.9997721285838835e-05,
      "loss": 8.3291,
      "step": 7250
    },
    {
      "epoch": 0.02041728888776397,
      "grad_norm": 36.75,
      "learning_rate": 1.9997714994766798e-05,
      "loss": 7.8051,
      "step": 7260
    },
    {
      "epoch": 0.02044541187521268,
      "grad_norm": 40.5,
      "learning_rate": 1.999770869502353e-05,
      "loss": 7.5696,
      "step": 7270
    },
    {
      "epoch": 0.020473534862661392,
      "grad_norm": 22.25,
      "learning_rate": 1.9997702386609024e-05,
      "loss": 8.1361,
      "step": 7280
    },
    {
      "epoch": 0.0205016578501101,
      "grad_norm": 26.5,
      "learning_rate": 1.9997696069523305e-05,
      "loss": 8.2315,
      "step": 7290
    },
    {
      "epoch": 0.02052978083755881,
      "grad_norm": 35.75,
      "learning_rate": 1.9997689743766363e-05,
      "loss": 8.7212,
      "step": 7300
    },
    {
      "epoch": 0.020557903825007523,
      "grad_norm": 49.25,
      "learning_rate": 1.9997683409338213e-05,
      "loss": 8.2517,
      "step": 7310
    },
    {
      "epoch": 0.020586026812456235,
      "grad_norm": 33.0,
      "learning_rate": 1.9997677066238858e-05,
      "loss": 8.6049,
      "step": 7320
    },
    {
      "epoch": 0.020614149799904943,
      "grad_norm": 25.75,
      "learning_rate": 1.9997670714468298e-05,
      "loss": 7.8715,
      "step": 7330
    },
    {
      "epoch": 0.020642272787353654,
      "grad_norm": 21.0,
      "learning_rate": 1.9997664354026543e-05,
      "loss": 8.0771,
      "step": 7340
    },
    {
      "epoch": 0.020670395774802366,
      "grad_norm": 28.875,
      "learning_rate": 1.99976579849136e-05,
      "loss": 7.4014,
      "step": 7350
    },
    {
      "epoch": 0.020698518762251077,
      "grad_norm": 18.375,
      "learning_rate": 1.9997651607129472e-05,
      "loss": 7.5094,
      "step": 7360
    },
    {
      "epoch": 0.020726641749699785,
      "grad_norm": 25.25,
      "learning_rate": 1.9997645220674166e-05,
      "loss": 8.8834,
      "step": 7370
    },
    {
      "epoch": 0.020754764737148497,
      "grad_norm": 24.0,
      "learning_rate": 1.9997638825547686e-05,
      "loss": 8.7092,
      "step": 7380
    },
    {
      "epoch": 0.02078288772459721,
      "grad_norm": 35.75,
      "learning_rate": 1.9997632421750038e-05,
      "loss": 7.8499,
      "step": 7390
    },
    {
      "epoch": 0.02081101071204592,
      "grad_norm": 24.625,
      "learning_rate": 1.9997626009281226e-05,
      "loss": 7.8206,
      "step": 7400
    },
    {
      "epoch": 0.02083913369949463,
      "grad_norm": 23.875,
      "learning_rate": 1.999761958814126e-05,
      "loss": 7.5156,
      "step": 7410
    },
    {
      "epoch": 0.02086725668694334,
      "grad_norm": 37.0,
      "learning_rate": 1.9997613158330145e-05,
      "loss": 8.3361,
      "step": 7420
    },
    {
      "epoch": 0.02089537967439205,
      "grad_norm": 24.625,
      "learning_rate": 1.999760671984788e-05,
      "loss": 7.9261,
      "step": 7430
    },
    {
      "epoch": 0.020923502661840763,
      "grad_norm": 46.5,
      "learning_rate": 1.9997600272694478e-05,
      "loss": 7.7139,
      "step": 7440
    },
    {
      "epoch": 0.020951625649289474,
      "grad_norm": 21.875,
      "learning_rate": 1.999759381686994e-05,
      "loss": 8.4161,
      "step": 7450
    },
    {
      "epoch": 0.020979748636738182,
      "grad_norm": 24.0,
      "learning_rate": 1.9997587352374276e-05,
      "loss": 7.7741,
      "step": 7460
    },
    {
      "epoch": 0.021007871624186894,
      "grad_norm": 31.75,
      "learning_rate": 1.9997580879207485e-05,
      "loss": 8.3733,
      "step": 7470
    },
    {
      "epoch": 0.021035994611635606,
      "grad_norm": 27.375,
      "learning_rate": 1.9997574397369578e-05,
      "loss": 7.8929,
      "step": 7480
    },
    {
      "epoch": 0.021064117599084317,
      "grad_norm": 52.5,
      "learning_rate": 1.999756790686056e-05,
      "loss": 8.6431,
      "step": 7490
    },
    {
      "epoch": 0.021092240586533025,
      "grad_norm": 29.75,
      "learning_rate": 1.9997561407680437e-05,
      "loss": 7.8175,
      "step": 7500
    },
    {
      "epoch": 0.021120363573981737,
      "grad_norm": 28.875,
      "learning_rate": 1.9997554899829213e-05,
      "loss": 8.2436,
      "step": 7510
    },
    {
      "epoch": 0.02114848656143045,
      "grad_norm": 22.375,
      "learning_rate": 1.9997548383306892e-05,
      "loss": 7.2461,
      "step": 7520
    },
    {
      "epoch": 0.02117660954887916,
      "grad_norm": 30.125,
      "learning_rate": 1.9997541858113485e-05,
      "loss": 7.7224,
      "step": 7530
    },
    {
      "epoch": 0.021204732536327868,
      "grad_norm": 83.5,
      "learning_rate": 1.999753532424899e-05,
      "loss": 8.6864,
      "step": 7540
    },
    {
      "epoch": 0.02123285552377658,
      "grad_norm": 32.5,
      "learning_rate": 1.999752878171342e-05,
      "loss": 7.7229,
      "step": 7550
    },
    {
      "epoch": 0.02126097851122529,
      "grad_norm": 37.25,
      "learning_rate": 1.9997522230506776e-05,
      "loss": 7.869,
      "step": 7560
    },
    {
      "epoch": 0.021289101498674003,
      "grad_norm": 71.5,
      "learning_rate": 1.999751567062907e-05,
      "loss": 8.0629,
      "step": 7570
    },
    {
      "epoch": 0.02131722448612271,
      "grad_norm": 34.0,
      "learning_rate": 1.99975091020803e-05,
      "loss": 8.4305,
      "step": 7580
    },
    {
      "epoch": 0.021345347473571422,
      "grad_norm": 28.5,
      "learning_rate": 1.999750252486047e-05,
      "loss": 8.7654,
      "step": 7590
    },
    {
      "epoch": 0.021373470461020134,
      "grad_norm": 24.0,
      "learning_rate": 1.9997495938969598e-05,
      "loss": 8.2996,
      "step": 7600
    },
    {
      "epoch": 0.021401593448468845,
      "grad_norm": 25.0,
      "learning_rate": 1.999748934440768e-05,
      "loss": 8.3266,
      "step": 7610
    },
    {
      "epoch": 0.021429716435917553,
      "grad_norm": 37.0,
      "learning_rate": 1.9997482741174724e-05,
      "loss": 8.342,
      "step": 7620
    },
    {
      "epoch": 0.021457839423366265,
      "grad_norm": 34.75,
      "learning_rate": 1.9997476129270733e-05,
      "loss": 8.2805,
      "step": 7630
    },
    {
      "epoch": 0.021485962410814977,
      "grad_norm": 21.875,
      "learning_rate": 1.9997469508695718e-05,
      "loss": 7.6063,
      "step": 7640
    },
    {
      "epoch": 0.021514085398263688,
      "grad_norm": 21.5,
      "learning_rate": 1.999746287944968e-05,
      "loss": 7.3674,
      "step": 7650
    },
    {
      "epoch": 0.021542208385712396,
      "grad_norm": 35.75,
      "learning_rate": 1.999745624153263e-05,
      "loss": 8.2359,
      "step": 7660
    },
    {
      "epoch": 0.021570331373161108,
      "grad_norm": 35.75,
      "learning_rate": 1.9997449594944568e-05,
      "loss": 8.8457,
      "step": 7670
    },
    {
      "epoch": 0.02159845436060982,
      "grad_norm": 24.5,
      "learning_rate": 1.99974429396855e-05,
      "loss": 8.3223,
      "step": 7680
    },
    {
      "epoch": 0.02162657734805853,
      "grad_norm": 36.5,
      "learning_rate": 1.9997436275755443e-05,
      "loss": 7.7577,
      "step": 7690
    },
    {
      "epoch": 0.02165470033550724,
      "grad_norm": 30.25,
      "learning_rate": 1.999742960315439e-05,
      "loss": 7.8929,
      "step": 7700
    },
    {
      "epoch": 0.02168282332295595,
      "grad_norm": 25.375,
      "learning_rate": 1.9997422921882348e-05,
      "loss": 7.8984,
      "step": 7710
    },
    {
      "epoch": 0.021710946310404662,
      "grad_norm": 32.0,
      "learning_rate": 1.9997416231939325e-05,
      "loss": 7.9924,
      "step": 7720
    },
    {
      "epoch": 0.021739069297853374,
      "grad_norm": 41.25,
      "learning_rate": 1.999740953332533e-05,
      "loss": 8.5479,
      "step": 7730
    },
    {
      "epoch": 0.021767192285302082,
      "grad_norm": 27.625,
      "learning_rate": 1.9997402826040365e-05,
      "loss": 8.4177,
      "step": 7740
    },
    {
      "epoch": 0.021795315272750793,
      "grad_norm": 74.0,
      "learning_rate": 1.999739611008444e-05,
      "loss": 7.9194,
      "step": 7750
    },
    {
      "epoch": 0.021823438260199505,
      "grad_norm": 27.75,
      "learning_rate": 1.9997389385457555e-05,
      "loss": 8.6962,
      "step": 7760
    },
    {
      "epoch": 0.021851561247648216,
      "grad_norm": 31.875,
      "learning_rate": 1.999738265215972e-05,
      "loss": 8.0286,
      "step": 7770
    },
    {
      "epoch": 0.021879684235096925,
      "grad_norm": 23.625,
      "learning_rate": 1.9997375910190937e-05,
      "loss": 8.4991,
      "step": 7780
    },
    {
      "epoch": 0.021907807222545636,
      "grad_norm": 26.25,
      "learning_rate": 1.9997369159551217e-05,
      "loss": 7.4188,
      "step": 7790
    },
    {
      "epoch": 0.021935930209994348,
      "grad_norm": 25.875,
      "learning_rate": 1.999736240024056e-05,
      "loss": 7.9129,
      "step": 7800
    },
    {
      "epoch": 0.02196405319744306,
      "grad_norm": 26.25,
      "learning_rate": 1.9997355632258978e-05,
      "loss": 8.6951,
      "step": 7810
    },
    {
      "epoch": 0.021992176184891767,
      "grad_norm": 38.0,
      "learning_rate": 1.9997348855606473e-05,
      "loss": 8.5205,
      "step": 7820
    },
    {
      "epoch": 0.02202029917234048,
      "grad_norm": 33.25,
      "learning_rate": 1.999734207028305e-05,
      "loss": 8.2526,
      "step": 7830
    },
    {
      "epoch": 0.02204842215978919,
      "grad_norm": 44.75,
      "learning_rate": 1.9997335276288722e-05,
      "loss": 7.9077,
      "step": 7840
    },
    {
      "epoch": 0.022076545147237902,
      "grad_norm": 24.5,
      "learning_rate": 1.9997328473623486e-05,
      "loss": 7.8463,
      "step": 7850
    },
    {
      "epoch": 0.02210466813468661,
      "grad_norm": 20.75,
      "learning_rate": 1.999732166228735e-05,
      "loss": 7.5722,
      "step": 7860
    },
    {
      "epoch": 0.02213279112213532,
      "grad_norm": 30.0,
      "learning_rate": 1.999731484228032e-05,
      "loss": 7.8458,
      "step": 7870
    },
    {
      "epoch": 0.022160914109584033,
      "grad_norm": 23.875,
      "learning_rate": 1.999730801360241e-05,
      "loss": 8.0548,
      "step": 7880
    },
    {
      "epoch": 0.022189037097032745,
      "grad_norm": 31.75,
      "learning_rate": 1.9997301176253616e-05,
      "loss": 8.1572,
      "step": 7890
    },
    {
      "epoch": 0.022217160084481453,
      "grad_norm": 33.5,
      "learning_rate": 1.9997294330233944e-05,
      "loss": 7.3252,
      "step": 7900
    },
    {
      "epoch": 0.022245283071930164,
      "grad_norm": 25.125,
      "learning_rate": 1.999728747554341e-05,
      "loss": 8.5656,
      "step": 7910
    },
    {
      "epoch": 0.022273406059378876,
      "grad_norm": 35.25,
      "learning_rate": 1.9997280612182005e-05,
      "loss": 8.3027,
      "step": 7920
    },
    {
      "epoch": 0.022301529046827587,
      "grad_norm": 21.5,
      "learning_rate": 1.9997273740149748e-05,
      "loss": 8.2364,
      "step": 7930
    },
    {
      "epoch": 0.022329652034276296,
      "grad_norm": 24.625,
      "learning_rate": 1.9997266859446638e-05,
      "loss": 7.4403,
      "step": 7940
    },
    {
      "epoch": 0.022357775021725007,
      "grad_norm": 27.375,
      "learning_rate": 1.9997259970072686e-05,
      "loss": 7.8787,
      "step": 7950
    },
    {
      "epoch": 0.02238589800917372,
      "grad_norm": 22.5,
      "learning_rate": 1.999725307202789e-05,
      "loss": 7.6426,
      "step": 7960
    },
    {
      "epoch": 0.02241402099662243,
      "grad_norm": 56.0,
      "learning_rate": 1.9997246165312264e-05,
      "loss": 7.9893,
      "step": 7970
    },
    {
      "epoch": 0.02244214398407114,
      "grad_norm": 29.25,
      "learning_rate": 1.999723924992581e-05,
      "loss": 8.1582,
      "step": 7980
    },
    {
      "epoch": 0.02247026697151985,
      "grad_norm": 35.0,
      "learning_rate": 1.9997232325868537e-05,
      "loss": 8.1756,
      "step": 7990
    },
    {
      "epoch": 0.02249838995896856,
      "grad_norm": 44.0,
      "learning_rate": 1.9997225393140446e-05,
      "loss": 8.1742,
      "step": 8000
    },
    {
      "epoch": 0.022526512946417273,
      "grad_norm": 24.875,
      "learning_rate": 1.9997218451741548e-05,
      "loss": 8.038,
      "step": 8010
    },
    {
      "epoch": 0.02255463593386598,
      "grad_norm": 26.375,
      "learning_rate": 1.9997211501671844e-05,
      "loss": 8.2978,
      "step": 8020
    },
    {
      "epoch": 0.022582758921314693,
      "grad_norm": 35.75,
      "learning_rate": 1.9997204542931344e-05,
      "loss": 8.0201,
      "step": 8030
    },
    {
      "epoch": 0.022610881908763404,
      "grad_norm": 23.125,
      "learning_rate": 1.9997197575520057e-05,
      "loss": 8.416,
      "step": 8040
    },
    {
      "epoch": 0.022639004896212116,
      "grad_norm": 36.25,
      "learning_rate": 1.9997190599437978e-05,
      "loss": 8.2767,
      "step": 8050
    },
    {
      "epoch": 0.022667127883660824,
      "grad_norm": 45.0,
      "learning_rate": 1.9997183614685124e-05,
      "loss": 8.5677,
      "step": 8060
    },
    {
      "epoch": 0.022695250871109535,
      "grad_norm": 47.25,
      "learning_rate": 1.99971766212615e-05,
      "loss": 8.3419,
      "step": 8070
    },
    {
      "epoch": 0.022723373858558247,
      "grad_norm": 26.0,
      "learning_rate": 1.9997169619167102e-05,
      "loss": 8.1733,
      "step": 8080
    },
    {
      "epoch": 0.02275149684600696,
      "grad_norm": 31.125,
      "learning_rate": 1.999716260840195e-05,
      "loss": 8.9879,
      "step": 8090
    },
    {
      "epoch": 0.02277961983345567,
      "grad_norm": 46.5,
      "learning_rate": 1.999715558896604e-05,
      "loss": 7.4397,
      "step": 8100
    },
    {
      "epoch": 0.022807742820904378,
      "grad_norm": 32.5,
      "learning_rate": 1.9997148560859383e-05,
      "loss": 8.1972,
      "step": 8110
    },
    {
      "epoch": 0.02283586580835309,
      "grad_norm": 42.0,
      "learning_rate": 1.9997141524081982e-05,
      "loss": 7.7304,
      "step": 8120
    },
    {
      "epoch": 0.0228639887958018,
      "grad_norm": 25.125,
      "learning_rate": 1.9997134478633848e-05,
      "loss": 8.4608,
      "step": 8130
    },
    {
      "epoch": 0.022892111783250513,
      "grad_norm": 30.5,
      "learning_rate": 1.999712742451498e-05,
      "loss": 7.5254,
      "step": 8140
    },
    {
      "epoch": 0.02292023477069922,
      "grad_norm": 24.75,
      "learning_rate": 1.9997120361725386e-05,
      "loss": 7.3108,
      "step": 8150
    },
    {
      "epoch": 0.022948357758147932,
      "grad_norm": 39.0,
      "learning_rate": 1.9997113290265076e-05,
      "loss": 8.335,
      "step": 8160
    },
    {
      "epoch": 0.022976480745596644,
      "grad_norm": 33.0,
      "learning_rate": 1.9997106210134056e-05,
      "loss": 8.6027,
      "step": 8170
    },
    {
      "epoch": 0.023004603733045356,
      "grad_norm": 42.5,
      "learning_rate": 1.999709912133233e-05,
      "loss": 7.6828,
      "step": 8180
    },
    {
      "epoch": 0.023032726720494064,
      "grad_norm": 22.125,
      "learning_rate": 1.9997092023859902e-05,
      "loss": 8.1195,
      "step": 8190
    },
    {
      "epoch": 0.023060849707942775,
      "grad_norm": 24.25,
      "learning_rate": 1.999708491771678e-05,
      "loss": 7.8631,
      "step": 8200
    },
    {
      "epoch": 0.023088972695391487,
      "grad_norm": 24.875,
      "learning_rate": 1.9997077802902977e-05,
      "loss": 8.5451,
      "step": 8210
    },
    {
      "epoch": 0.0231170956828402,
      "grad_norm": 29.625,
      "learning_rate": 1.9997070679418486e-05,
      "loss": 7.6865,
      "step": 8220
    },
    {
      "epoch": 0.023145218670288906,
      "grad_norm": 26.875,
      "learning_rate": 1.9997063547263326e-05,
      "loss": 8.3994,
      "step": 8230
    },
    {
      "epoch": 0.023173341657737618,
      "grad_norm": 41.75,
      "learning_rate": 1.9997056406437492e-05,
      "loss": 7.2586,
      "step": 8240
    },
    {
      "epoch": 0.02320146464518633,
      "grad_norm": 30.875,
      "learning_rate": 1.9997049256940996e-05,
      "loss": 7.5855,
      "step": 8250
    },
    {
      "epoch": 0.02322958763263504,
      "grad_norm": 35.25,
      "learning_rate": 1.999704209877385e-05,
      "loss": 8.7402,
      "step": 8260
    },
    {
      "epoch": 0.02325771062008375,
      "grad_norm": 26.625,
      "learning_rate": 1.9997034931936045e-05,
      "loss": 7.9202,
      "step": 8270
    },
    {
      "epoch": 0.02328583360753246,
      "grad_norm": 40.25,
      "learning_rate": 1.99970277564276e-05,
      "loss": 7.8678,
      "step": 8280
    },
    {
      "epoch": 0.023313956594981172,
      "grad_norm": 24.75,
      "learning_rate": 1.9997020572248517e-05,
      "loss": 7.8925,
      "step": 8290
    },
    {
      "epoch": 0.023342079582429884,
      "grad_norm": 30.5,
      "learning_rate": 1.9997013379398804e-05,
      "loss": 7.8366,
      "step": 8300
    },
    {
      "epoch": 0.023370202569878592,
      "grad_norm": 29.5,
      "learning_rate": 1.9997006177878465e-05,
      "loss": 7.7181,
      "step": 8310
    },
    {
      "epoch": 0.023398325557327303,
      "grad_norm": 35.0,
      "learning_rate": 1.9996998967687508e-05,
      "loss": 8.234,
      "step": 8320
    },
    {
      "epoch": 0.023426448544776015,
      "grad_norm": 35.75,
      "learning_rate": 1.9996991748825935e-05,
      "loss": 7.6394,
      "step": 8330
    },
    {
      "epoch": 0.023454571532224727,
      "grad_norm": 41.0,
      "learning_rate": 1.9996984521293757e-05,
      "loss": 7.6562,
      "step": 8340
    },
    {
      "epoch": 0.023482694519673435,
      "grad_norm": 28.5,
      "learning_rate": 1.999697728509098e-05,
      "loss": 7.9189,
      "step": 8350
    },
    {
      "epoch": 0.023510817507122146,
      "grad_norm": 41.5,
      "learning_rate": 1.999697004021761e-05,
      "loss": 7.5574,
      "step": 8360
    },
    {
      "epoch": 0.023538940494570858,
      "grad_norm": 23.5,
      "learning_rate": 1.999696278667365e-05,
      "loss": 7.9631,
      "step": 8370
    },
    {
      "epoch": 0.02356706348201957,
      "grad_norm": 25.5,
      "learning_rate": 1.999695552445911e-05,
      "loss": 7.7609,
      "step": 8380
    },
    {
      "epoch": 0.023595186469468277,
      "grad_norm": 40.25,
      "learning_rate": 1.9996948253573993e-05,
      "loss": 8.5702,
      "step": 8390
    },
    {
      "epoch": 0.02362330945691699,
      "grad_norm": 19.75,
      "learning_rate": 1.9996940974018308e-05,
      "loss": 8.0607,
      "step": 8400
    },
    {
      "epoch": 0.0236514324443657,
      "grad_norm": 30.0,
      "learning_rate": 1.9996933685792062e-05,
      "loss": 8.068,
      "step": 8410
    },
    {
      "epoch": 0.023679555431814412,
      "grad_norm": 28.375,
      "learning_rate": 1.9996926388895258e-05,
      "loss": 8.7895,
      "step": 8420
    },
    {
      "epoch": 0.02370767841926312,
      "grad_norm": 36.5,
      "learning_rate": 1.9996919083327907e-05,
      "loss": 7.1192,
      "step": 8430
    },
    {
      "epoch": 0.02373580140671183,
      "grad_norm": 21.125,
      "learning_rate": 1.9996911769090007e-05,
      "loss": 8.7498,
      "step": 8440
    },
    {
      "epoch": 0.023763924394160543,
      "grad_norm": 20.625,
      "learning_rate": 1.9996904446181577e-05,
      "loss": 8.61,
      "step": 8450
    },
    {
      "epoch": 0.023792047381609255,
      "grad_norm": 36.0,
      "learning_rate": 1.9996897114602613e-05,
      "loss": 8.8897,
      "step": 8460
    },
    {
      "epoch": 0.023820170369057963,
      "grad_norm": 32.25,
      "learning_rate": 1.9996889774353124e-05,
      "loss": 8.3375,
      "step": 8470
    },
    {
      "epoch": 0.023848293356506674,
      "grad_norm": 30.5,
      "learning_rate": 1.999688242543312e-05,
      "loss": 8.7113,
      "step": 8480
    },
    {
      "epoch": 0.023876416343955386,
      "grad_norm": 32.5,
      "learning_rate": 1.99968750678426e-05,
      "loss": 8.244,
      "step": 8490
    },
    {
      "epoch": 0.023904539331404098,
      "grad_norm": 25.875,
      "learning_rate": 1.9996867701581576e-05,
      "loss": 8.7278,
      "step": 8500
    },
    {
      "epoch": 0.023932662318852806,
      "grad_norm": 36.25,
      "learning_rate": 1.9996860326650052e-05,
      "loss": 8.672,
      "step": 8510
    },
    {
      "epoch": 0.023960785306301517,
      "grad_norm": 28.875,
      "learning_rate": 1.999685294304804e-05,
      "loss": 7.6934,
      "step": 8520
    },
    {
      "epoch": 0.02398890829375023,
      "grad_norm": 28.25,
      "learning_rate": 1.9996845550775538e-05,
      "loss": 7.9508,
      "step": 8530
    },
    {
      "epoch": 0.02401703128119894,
      "grad_norm": 32.25,
      "learning_rate": 1.9996838149832558e-05,
      "loss": 8.157,
      "step": 8540
    },
    {
      "epoch": 0.02404515426864765,
      "grad_norm": 46.25,
      "learning_rate": 1.99968307402191e-05,
      "loss": 8.2076,
      "step": 8550
    },
    {
      "epoch": 0.02407327725609636,
      "grad_norm": 28.375,
      "learning_rate": 1.999682332193518e-05,
      "loss": 7.9685,
      "step": 8560
    },
    {
      "epoch": 0.02410140024354507,
      "grad_norm": 54.25,
      "learning_rate": 1.99968158949808e-05,
      "loss": 7.7684,
      "step": 8570
    },
    {
      "epoch": 0.024129523230993783,
      "grad_norm": 29.375,
      "learning_rate": 1.9996808459355966e-05,
      "loss": 7.0636,
      "step": 8580
    },
    {
      "epoch": 0.02415764621844249,
      "grad_norm": 29.75,
      "learning_rate": 1.9996801015060684e-05,
      "loss": 7.6273,
      "step": 8590
    },
    {
      "epoch": 0.024185769205891203,
      "grad_norm": 22.5,
      "learning_rate": 1.999679356209496e-05,
      "loss": 7.8155,
      "step": 8600
    },
    {
      "epoch": 0.024213892193339914,
      "grad_norm": 34.25,
      "learning_rate": 1.9996786100458802e-05,
      "loss": 7.5556,
      "step": 8610
    },
    {
      "epoch": 0.024242015180788626,
      "grad_norm": 65.0,
      "learning_rate": 1.9996778630152216e-05,
      "loss": 8.121,
      "step": 8620
    },
    {
      "epoch": 0.024270138168237334,
      "grad_norm": 27.375,
      "learning_rate": 1.9996771151175207e-05,
      "loss": 8.1539,
      "step": 8630
    },
    {
      "epoch": 0.024298261155686045,
      "grad_norm": 30.375,
      "learning_rate": 1.9996763663527786e-05,
      "loss": 8.2234,
      "step": 8640
    },
    {
      "epoch": 0.024326384143134757,
      "grad_norm": 43.75,
      "learning_rate": 1.9996756167209953e-05,
      "loss": 7.5178,
      "step": 8650
    },
    {
      "epoch": 0.02435450713058347,
      "grad_norm": 23.875,
      "learning_rate": 1.9996748662221718e-05,
      "loss": 7.7804,
      "step": 8660
    },
    {
      "epoch": 0.024382630118032177,
      "grad_norm": 34.75,
      "learning_rate": 1.999674114856309e-05,
      "loss": 7.6625,
      "step": 8670
    },
    {
      "epoch": 0.024410753105480888,
      "grad_norm": 33.75,
      "learning_rate": 1.9996733626234072e-05,
      "loss": 7.7876,
      "step": 8680
    },
    {
      "epoch": 0.0244388760929296,
      "grad_norm": 25.125,
      "learning_rate": 1.9996726095234668e-05,
      "loss": 7.5176,
      "step": 8690
    },
    {
      "epoch": 0.02446699908037831,
      "grad_norm": 35.5,
      "learning_rate": 1.9996718555564892e-05,
      "loss": 7.7748,
      "step": 8700
    },
    {
      "epoch": 0.02449512206782702,
      "grad_norm": 36.75,
      "learning_rate": 1.999671100722475e-05,
      "loss": 7.4708,
      "step": 8710
    },
    {
      "epoch": 0.02452324505527573,
      "grad_norm": 36.25,
      "learning_rate": 1.9996703450214237e-05,
      "loss": 7.8013,
      "step": 8720
    },
    {
      "epoch": 0.024551368042724443,
      "grad_norm": 44.5,
      "learning_rate": 1.999669588453337e-05,
      "loss": 8.7622,
      "step": 8730
    },
    {
      "epoch": 0.024579491030173154,
      "grad_norm": 47.5,
      "learning_rate": 1.9996688310182156e-05,
      "loss": 8.0576,
      "step": 8740
    },
    {
      "epoch": 0.024607614017621866,
      "grad_norm": 24.125,
      "learning_rate": 1.9996680727160594e-05,
      "loss": 8.4018,
      "step": 8750
    },
    {
      "epoch": 0.024635737005070574,
      "grad_norm": 24.75,
      "learning_rate": 1.9996673135468697e-05,
      "loss": 7.7353,
      "step": 8760
    },
    {
      "epoch": 0.024663859992519285,
      "grad_norm": 23.0,
      "learning_rate": 1.9996665535106473e-05,
      "loss": 8.1296,
      "step": 8770
    },
    {
      "epoch": 0.024691982979967997,
      "grad_norm": 33.5,
      "learning_rate": 1.9996657926073922e-05,
      "loss": 8.8128,
      "step": 8780
    },
    {
      "epoch": 0.02472010596741671,
      "grad_norm": 42.0,
      "learning_rate": 1.9996650308371056e-05,
      "loss": 8.0906,
      "step": 8790
    },
    {
      "epoch": 0.024748228954865416,
      "grad_norm": 41.5,
      "learning_rate": 1.999664268199788e-05,
      "loss": 7.3259,
      "step": 8800
    },
    {
      "epoch": 0.024776351942314128,
      "grad_norm": 27.25,
      "learning_rate": 1.9996635046954397e-05,
      "loss": 8.8595,
      "step": 8810
    },
    {
      "epoch": 0.02480447492976284,
      "grad_norm": 20.25,
      "learning_rate": 1.999662740324062e-05,
      "loss": 7.5234,
      "step": 8820
    },
    {
      "epoch": 0.02483259791721155,
      "grad_norm": 21.375,
      "learning_rate": 1.999661975085655e-05,
      "loss": 8.2014,
      "step": 8830
    },
    {
      "epoch": 0.02486072090466026,
      "grad_norm": 57.25,
      "learning_rate": 1.99966120898022e-05,
      "loss": 8.3155,
      "step": 8840
    },
    {
      "epoch": 0.02488884389210897,
      "grad_norm": 37.75,
      "learning_rate": 1.9996604420077573e-05,
      "loss": 7.9681,
      "step": 8850
    },
    {
      "epoch": 0.024916966879557682,
      "grad_norm": 35.5,
      "learning_rate": 1.999659674168267e-05,
      "loss": 8.3407,
      "step": 8860
    },
    {
      "epoch": 0.024945089867006394,
      "grad_norm": 28.75,
      "learning_rate": 1.999658905461751e-05,
      "loss": 7.9579,
      "step": 8870
    },
    {
      "epoch": 0.024973212854455102,
      "grad_norm": 22.125,
      "learning_rate": 1.9996581358882085e-05,
      "loss": 7.7966,
      "step": 8880
    },
    {
      "epoch": 0.025001335841903814,
      "grad_norm": 56.0,
      "learning_rate": 1.9996573654476416e-05,
      "loss": 7.926,
      "step": 8890
    },
    {
      "epoch": 0.025029458829352525,
      "grad_norm": 25.25,
      "learning_rate": 1.99965659414005e-05,
      "loss": 8.0135,
      "step": 8900
    },
    {
      "epoch": 0.025057581816801237,
      "grad_norm": 23.75,
      "learning_rate": 1.999655821965435e-05,
      "loss": 7.7444,
      "step": 8910
    },
    {
      "epoch": 0.025085704804249945,
      "grad_norm": 24.125,
      "learning_rate": 1.9996550489237964e-05,
      "loss": 7.7179,
      "step": 8920
    },
    {
      "epoch": 0.025113827791698656,
      "grad_norm": 26.375,
      "learning_rate": 1.9996542750151357e-05,
      "loss": 7.9628,
      "step": 8930
    },
    {
      "epoch": 0.025141950779147368,
      "grad_norm": 40.0,
      "learning_rate": 1.9996535002394535e-05,
      "loss": 8.035,
      "step": 8940
    },
    {
      "epoch": 0.02517007376659608,
      "grad_norm": 18.25,
      "learning_rate": 1.99965272459675e-05,
      "loss": 8.2915,
      "step": 8950
    },
    {
      "epoch": 0.025198196754044788,
      "grad_norm": 49.5,
      "learning_rate": 1.9996519480870263e-05,
      "loss": 8.2397,
      "step": 8960
    },
    {
      "epoch": 0.0252263197414935,
      "grad_norm": 37.75,
      "learning_rate": 1.9996511707102827e-05,
      "loss": 8.1704,
      "step": 8970
    },
    {
      "epoch": 0.02525444272894221,
      "grad_norm": 34.75,
      "learning_rate": 1.9996503924665203e-05,
      "loss": 7.9778,
      "step": 8980
    },
    {
      "epoch": 0.025282565716390922,
      "grad_norm": 46.75,
      "learning_rate": 1.9996496133557394e-05,
      "loss": 7.8852,
      "step": 8990
    },
    {
      "epoch": 0.02531068870383963,
      "grad_norm": 41.75,
      "learning_rate": 1.9996488333779412e-05,
      "loss": 8.014,
      "step": 9000
    },
    {
      "epoch": 0.025338811691288342,
      "grad_norm": 45.0,
      "learning_rate": 1.9996480525331256e-05,
      "loss": 8.242,
      "step": 9010
    },
    {
      "epoch": 0.025366934678737053,
      "grad_norm": 58.5,
      "learning_rate": 1.999647270821294e-05,
      "loss": 8.2374,
      "step": 9020
    },
    {
      "epoch": 0.025395057666185765,
      "grad_norm": 37.0,
      "learning_rate": 1.9996464882424463e-05,
      "loss": 7.7369,
      "step": 9030
    },
    {
      "epoch": 0.025423180653634473,
      "grad_norm": 17.375,
      "learning_rate": 1.999645704796584e-05,
      "loss": 8.3109,
      "step": 9040
    },
    {
      "epoch": 0.025451303641083185,
      "grad_norm": 20.75,
      "learning_rate": 1.9996449204837078e-05,
      "loss": 7.214,
      "step": 9050
    },
    {
      "epoch": 0.025479426628531896,
      "grad_norm": 36.25,
      "learning_rate": 1.999644135303817e-05,
      "loss": 8.0174,
      "step": 9060
    },
    {
      "epoch": 0.025507549615980608,
      "grad_norm": 41.5,
      "learning_rate": 1.999643349256914e-05,
      "loss": 8.0048,
      "step": 9070
    },
    {
      "epoch": 0.025535672603429316,
      "grad_norm": 58.75,
      "learning_rate": 1.9996425623429987e-05,
      "loss": 8.5075,
      "step": 9080
    },
    {
      "epoch": 0.025563795590878027,
      "grad_norm": 27.75,
      "learning_rate": 1.999641774562072e-05,
      "loss": 8.4168,
      "step": 9090
    },
    {
      "epoch": 0.02559191857832674,
      "grad_norm": 28.125,
      "learning_rate": 1.9996409859141343e-05,
      "loss": 7.6584,
      "step": 9100
    },
    {
      "epoch": 0.02562004156577545,
      "grad_norm": 23.25,
      "learning_rate": 1.9996401963991863e-05,
      "loss": 7.0018,
      "step": 9110
    },
    {
      "epoch": 0.02564816455322416,
      "grad_norm": 37.25,
      "learning_rate": 1.999639406017229e-05,
      "loss": 7.9149,
      "step": 9120
    },
    {
      "epoch": 0.02567628754067287,
      "grad_norm": 30.75,
      "learning_rate": 1.9996386147682626e-05,
      "loss": 8.3805,
      "step": 9130
    },
    {
      "epoch": 0.02570441052812158,
      "grad_norm": 30.875,
      "learning_rate": 1.9996378226522882e-05,
      "loss": 8.4078,
      "step": 9140
    },
    {
      "epoch": 0.025732533515570293,
      "grad_norm": 39.25,
      "learning_rate": 1.9996370296693065e-05,
      "loss": 8.5351,
      "step": 9150
    },
    {
      "epoch": 0.025760656503019,
      "grad_norm": 27.75,
      "learning_rate": 1.9996362358193177e-05,
      "loss": 7.8481,
      "step": 9160
    },
    {
      "epoch": 0.025788779490467713,
      "grad_norm": 32.75,
      "learning_rate": 1.9996354411023234e-05,
      "loss": 8.544,
      "step": 9170
    },
    {
      "epoch": 0.025816902477916424,
      "grad_norm": 28.875,
      "learning_rate": 1.9996346455183233e-05,
      "loss": 8.299,
      "step": 9180
    },
    {
      "epoch": 0.025845025465365136,
      "grad_norm": 21.625,
      "learning_rate": 1.9996338490673187e-05,
      "loss": 7.5148,
      "step": 9190
    },
    {
      "epoch": 0.025873148452813844,
      "grad_norm": 20.0,
      "learning_rate": 1.99963305174931e-05,
      "loss": 8.2518,
      "step": 9200
    },
    {
      "epoch": 0.025901271440262556,
      "grad_norm": 36.0,
      "learning_rate": 1.9996322535642982e-05,
      "loss": 8.2257,
      "step": 9210
    },
    {
      "epoch": 0.025929394427711267,
      "grad_norm": 53.0,
      "learning_rate": 1.9996314545122834e-05,
      "loss": 7.8572,
      "step": 9220
    },
    {
      "epoch": 0.02595751741515998,
      "grad_norm": 25.25,
      "learning_rate": 1.999630654593267e-05,
      "loss": 7.4187,
      "step": 9230
    },
    {
      "epoch": 0.025985640402608687,
      "grad_norm": 32.5,
      "learning_rate": 1.9996298538072497e-05,
      "loss": 8.5596,
      "step": 9240
    },
    {
      "epoch": 0.0260137633900574,
      "grad_norm": 26.5,
      "learning_rate": 1.9996290521542315e-05,
      "loss": 7.6479,
      "step": 9250
    },
    {
      "epoch": 0.02604188637750611,
      "grad_norm": 31.625,
      "learning_rate": 1.9996282496342135e-05,
      "loss": 7.8732,
      "step": 9260
    },
    {
      "epoch": 0.02607000936495482,
      "grad_norm": 30.5,
      "learning_rate": 1.9996274462471962e-05,
      "loss": 8.1521,
      "step": 9270
    },
    {
      "epoch": 0.02609813235240353,
      "grad_norm": 28.25,
      "learning_rate": 1.9996266419931807e-05,
      "loss": 7.7561,
      "step": 9280
    },
    {
      "epoch": 0.02612625533985224,
      "grad_norm": 23.125,
      "learning_rate": 1.9996258368721674e-05,
      "loss": 7.6299,
      "step": 9290
    },
    {
      "epoch": 0.026154378327300953,
      "grad_norm": 35.75,
      "learning_rate": 1.999625030884157e-05,
      "loss": 7.7965,
      "step": 9300
    },
    {
      "epoch": 0.026182501314749664,
      "grad_norm": 35.75,
      "learning_rate": 1.9996242240291502e-05,
      "loss": 8.018,
      "step": 9310
    },
    {
      "epoch": 0.026210624302198372,
      "grad_norm": 32.5,
      "learning_rate": 1.9996234163071477e-05,
      "loss": 7.7824,
      "step": 9320
    },
    {
      "epoch": 0.026238747289647084,
      "grad_norm": 55.75,
      "learning_rate": 1.9996226077181508e-05,
      "loss": 8.4256,
      "step": 9330
    },
    {
      "epoch": 0.026266870277095795,
      "grad_norm": 27.625,
      "learning_rate": 1.999621798262159e-05,
      "loss": 7.7526,
      "step": 9340
    },
    {
      "epoch": 0.026294993264544507,
      "grad_norm": 37.75,
      "learning_rate": 1.999620987939174e-05,
      "loss": 8.1458,
      "step": 9350
    },
    {
      "epoch": 0.026323116251993215,
      "grad_norm": 37.5,
      "learning_rate": 1.9996201767491962e-05,
      "loss": 7.4293,
      "step": 9360
    },
    {
      "epoch": 0.026351239239441927,
      "grad_norm": 29.375,
      "learning_rate": 1.9996193646922258e-05,
      "loss": 7.9999,
      "step": 9370
    },
    {
      "epoch": 0.026379362226890638,
      "grad_norm": 23.125,
      "learning_rate": 1.9996185517682643e-05,
      "loss": 8.0223,
      "step": 9380
    },
    {
      "epoch": 0.02640748521433935,
      "grad_norm": 27.125,
      "learning_rate": 1.9996177379773125e-05,
      "loss": 8.8174,
      "step": 9390
    },
    {
      "epoch": 0.026435608201788058,
      "grad_norm": 31.625,
      "learning_rate": 1.9996169233193702e-05,
      "loss": 7.409,
      "step": 9400
    },
    {
      "epoch": 0.02646373118923677,
      "grad_norm": 55.25,
      "learning_rate": 1.999616107794439e-05,
      "loss": 8.1505,
      "step": 9410
    },
    {
      "epoch": 0.02649185417668548,
      "grad_norm": 62.5,
      "learning_rate": 1.9996152914025185e-05,
      "loss": 7.9507,
      "step": 9420
    },
    {
      "epoch": 0.026519977164134192,
      "grad_norm": 20.75,
      "learning_rate": 1.9996144741436104e-05,
      "loss": 7.954,
      "step": 9430
    },
    {
      "epoch": 0.026548100151582904,
      "grad_norm": 37.75,
      "learning_rate": 1.9996136560177154e-05,
      "loss": 7.9875,
      "step": 9440
    },
    {
      "epoch": 0.026576223139031612,
      "grad_norm": 48.5,
      "learning_rate": 1.999612837024834e-05,
      "loss": 7.7239,
      "step": 9450
    },
    {
      "epoch": 0.026604346126480324,
      "grad_norm": 42.5,
      "learning_rate": 1.9996120171649666e-05,
      "loss": 8.4069,
      "step": 9460
    },
    {
      "epoch": 0.026632469113929035,
      "grad_norm": 29.125,
      "learning_rate": 1.999611196438114e-05,
      "loss": 8.0084,
      "step": 9470
    },
    {
      "epoch": 0.026660592101377747,
      "grad_norm": 24.125,
      "learning_rate": 1.9996103748442772e-05,
      "loss": 8.2494,
      "step": 9480
    },
    {
      "epoch": 0.026688715088826455,
      "grad_norm": 37.75,
      "learning_rate": 1.999609552383457e-05,
      "loss": 8.019,
      "step": 9490
    },
    {
      "epoch": 0.026716838076275166,
      "grad_norm": 40.0,
      "learning_rate": 1.9996087290556534e-05,
      "loss": 8.1527,
      "step": 9500
    },
    {
      "epoch": 0.026744961063723878,
      "grad_norm": 29.5,
      "learning_rate": 1.999607904860868e-05,
      "loss": 8.5709,
      "step": 9510
    },
    {
      "epoch": 0.02677308405117259,
      "grad_norm": 30.125,
      "learning_rate": 1.9996070797991008e-05,
      "loss": 7.8723,
      "step": 9520
    },
    {
      "epoch": 0.026801207038621298,
      "grad_norm": 37.0,
      "learning_rate": 1.9996062538703528e-05,
      "loss": 8.2198,
      "step": 9530
    },
    {
      "epoch": 0.02682933002607001,
      "grad_norm": 49.75,
      "learning_rate": 1.999605427074625e-05,
      "loss": 7.6952,
      "step": 9540
    },
    {
      "epoch": 0.02685745301351872,
      "grad_norm": 38.5,
      "learning_rate": 1.9996045994119178e-05,
      "loss": 8.5775,
      "step": 9550
    },
    {
      "epoch": 0.026885576000967432,
      "grad_norm": 24.0,
      "learning_rate": 1.999603770882232e-05,
      "loss": 8.0717,
      "step": 9560
    },
    {
      "epoch": 0.02691369898841614,
      "grad_norm": 27.625,
      "learning_rate": 1.9996029414855683e-05,
      "loss": 7.923,
      "step": 9570
    },
    {
      "epoch": 0.026941821975864852,
      "grad_norm": 24.5,
      "learning_rate": 1.9996021112219272e-05,
      "loss": 8.0142,
      "step": 9580
    },
    {
      "epoch": 0.026969944963313564,
      "grad_norm": 32.25,
      "learning_rate": 1.99960128009131e-05,
      "loss": 8.5648,
      "step": 9590
    },
    {
      "epoch": 0.026998067950762275,
      "grad_norm": 44.75,
      "learning_rate": 1.999600448093717e-05,
      "loss": 7.87,
      "step": 9600
    },
    {
      "epoch": 0.027026190938210983,
      "grad_norm": 25.875,
      "learning_rate": 1.9995996152291488e-05,
      "loss": 8.4132,
      "step": 9610
    },
    {
      "epoch": 0.027054313925659695,
      "grad_norm": 38.25,
      "learning_rate": 1.9995987814976065e-05,
      "loss": 7.3158,
      "step": 9620
    },
    {
      "epoch": 0.027082436913108406,
      "grad_norm": 22.0,
      "learning_rate": 1.9995979468990903e-05,
      "loss": 7.7528,
      "step": 9630
    },
    {
      "epoch": 0.027110559900557118,
      "grad_norm": 38.5,
      "learning_rate": 1.9995971114336018e-05,
      "loss": 8.3966,
      "step": 9640
    },
    {
      "epoch": 0.027138682888005826,
      "grad_norm": 22.75,
      "learning_rate": 1.9995962751011408e-05,
      "loss": 8.4306,
      "step": 9650
    },
    {
      "epoch": 0.027166805875454537,
      "grad_norm": 22.5,
      "learning_rate": 1.9995954379017083e-05,
      "loss": 7.5641,
      "step": 9660
    },
    {
      "epoch": 0.02719492886290325,
      "grad_norm": 23.875,
      "learning_rate": 1.999594599835305e-05,
      "loss": 8.2257,
      "step": 9670
    },
    {
      "epoch": 0.02722305185035196,
      "grad_norm": 31.375,
      "learning_rate": 1.9995937609019322e-05,
      "loss": 7.6467,
      "step": 9680
    },
    {
      "epoch": 0.02725117483780067,
      "grad_norm": 20.0,
      "learning_rate": 1.99959292110159e-05,
      "loss": 8.2393,
      "step": 9690
    },
    {
      "epoch": 0.02727929782524938,
      "grad_norm": 24.0,
      "learning_rate": 1.9995920804342792e-05,
      "loss": 7.9213,
      "step": 9700
    },
    {
      "epoch": 0.027307420812698092,
      "grad_norm": 26.5,
      "learning_rate": 1.999591238900001e-05,
      "loss": 7.9248,
      "step": 9710
    },
    {
      "epoch": 0.027335543800146803,
      "grad_norm": 34.0,
      "learning_rate": 1.9995903964987555e-05,
      "loss": 8.5032,
      "step": 9720
    },
    {
      "epoch": 0.02736366678759551,
      "grad_norm": 40.0,
      "learning_rate": 1.9995895532305434e-05,
      "loss": 7.9645,
      "step": 9730
    },
    {
      "epoch": 0.027391789775044223,
      "grad_norm": 24.625,
      "learning_rate": 1.999588709095366e-05,
      "loss": 7.7626,
      "step": 9740
    },
    {
      "epoch": 0.027419912762492935,
      "grad_norm": 25.0,
      "learning_rate": 1.999587864093224e-05,
      "loss": 7.5728,
      "step": 9750
    },
    {
      "epoch": 0.027448035749941646,
      "grad_norm": 22.0,
      "learning_rate": 1.9995870182241176e-05,
      "loss": 7.8912,
      "step": 9760
    },
    {
      "epoch": 0.027476158737390354,
      "grad_norm": 37.75,
      "learning_rate": 1.999586171488048e-05,
      "loss": 8.215,
      "step": 9770
    },
    {
      "epoch": 0.027504281724839066,
      "grad_norm": 34.75,
      "learning_rate": 1.9995853238850156e-05,
      "loss": 8.0155,
      "step": 9780
    },
    {
      "epoch": 0.027532404712287777,
      "grad_norm": 30.875,
      "learning_rate": 1.9995844754150214e-05,
      "loss": 8.2868,
      "step": 9790
    },
    {
      "epoch": 0.02756052769973649,
      "grad_norm": 34.75,
      "learning_rate": 1.999583626078066e-05,
      "loss": 8.0389,
      "step": 9800
    },
    {
      "epoch": 0.027588650687185197,
      "grad_norm": 47.25,
      "learning_rate": 1.9995827758741502e-05,
      "loss": 7.7408,
      "step": 9810
    },
    {
      "epoch": 0.02761677367463391,
      "grad_norm": 47.0,
      "learning_rate": 1.9995819248032746e-05,
      "loss": 8.6171,
      "step": 9820
    },
    {
      "epoch": 0.02764489666208262,
      "grad_norm": 26.875,
      "learning_rate": 1.99958107286544e-05,
      "loss": 7.8723,
      "step": 9830
    },
    {
      "epoch": 0.02767301964953133,
      "grad_norm": 35.25,
      "learning_rate": 1.9995802200606473e-05,
      "loss": 7.1007,
      "step": 9840
    },
    {
      "epoch": 0.02770114263698004,
      "grad_norm": 25.5,
      "learning_rate": 1.999579366388897e-05,
      "loss": 7.7648,
      "step": 9850
    },
    {
      "epoch": 0.02772926562442875,
      "grad_norm": 30.0,
      "learning_rate": 1.9995785118501903e-05,
      "loss": 8.0552,
      "step": 9860
    },
    {
      "epoch": 0.027757388611877463,
      "grad_norm": 37.5,
      "learning_rate": 1.9995776564445274e-05,
      "loss": 7.8143,
      "step": 9870
    },
    {
      "epoch": 0.027785511599326174,
      "grad_norm": 31.375,
      "learning_rate": 1.9995768001719092e-05,
      "loss": 8.0577,
      "step": 9880
    },
    {
      "epoch": 0.027813634586774882,
      "grad_norm": 27.75,
      "learning_rate": 1.9995759430323366e-05,
      "loss": 7.821,
      "step": 9890
    },
    {
      "epoch": 0.027841757574223594,
      "grad_norm": 32.0,
      "learning_rate": 1.9995750850258102e-05,
      "loss": 8.5424,
      "step": 9900
    },
    {
      "epoch": 0.027869880561672306,
      "grad_norm": 21.25,
      "learning_rate": 1.9995742261523307e-05,
      "loss": 7.82,
      "step": 9910
    },
    {
      "epoch": 0.027898003549121017,
      "grad_norm": 35.75,
      "learning_rate": 1.999573366411899e-05,
      "loss": 8.0116,
      "step": 9920
    },
    {
      "epoch": 0.027926126536569725,
      "grad_norm": 35.75,
      "learning_rate": 1.999572505804516e-05,
      "loss": 7.9151,
      "step": 9930
    },
    {
      "epoch": 0.027954249524018437,
      "grad_norm": 31.625,
      "learning_rate": 1.9995716443301818e-05,
      "loss": 7.7329,
      "step": 9940
    },
    {
      "epoch": 0.02798237251146715,
      "grad_norm": 51.75,
      "learning_rate": 1.9995707819888975e-05,
      "loss": 7.9913,
      "step": 9950
    },
    {
      "epoch": 0.02801049549891586,
      "grad_norm": 29.5,
      "learning_rate": 1.9995699187806642e-05,
      "loss": 7.9975,
      "step": 9960
    },
    {
      "epoch": 0.028038618486364568,
      "grad_norm": 40.5,
      "learning_rate": 1.9995690547054825e-05,
      "loss": 8.0908,
      "step": 9970
    },
    {
      "epoch": 0.02806674147381328,
      "grad_norm": 32.5,
      "learning_rate": 1.9995681897633527e-05,
      "loss": 8.2361,
      "step": 9980
    },
    {
      "epoch": 0.02809486446126199,
      "grad_norm": 20.375,
      "learning_rate": 1.999567323954276e-05,
      "loss": 7.896,
      "step": 9990
    },
    {
      "epoch": 0.028122987448710703,
      "grad_norm": 51.25,
      "learning_rate": 1.999566457278253e-05,
      "loss": 8.4663,
      "step": 10000
    },
    {
      "epoch": 0.02815111043615941,
      "grad_norm": 30.875,
      "learning_rate": 1.9995655897352843e-05,
      "loss": 8.1154,
      "step": 10010
    },
    {
      "epoch": 0.028179233423608122,
      "grad_norm": 28.75,
      "learning_rate": 1.999564721325371e-05,
      "loss": 7.8825,
      "step": 10020
    },
    {
      "epoch": 0.028207356411056834,
      "grad_norm": 39.5,
      "learning_rate": 1.999563852048514e-05,
      "loss": 8.1331,
      "step": 10030
    },
    {
      "epoch": 0.028235479398505545,
      "grad_norm": 30.75,
      "learning_rate": 1.9995629819047133e-05,
      "loss": 8.1916,
      "step": 10040
    },
    {
      "epoch": 0.028263602385954253,
      "grad_norm": 52.0,
      "learning_rate": 1.99956211089397e-05,
      "loss": 7.3916,
      "step": 10050
    },
    {
      "epoch": 0.028291725373402965,
      "grad_norm": 30.125,
      "learning_rate": 1.9995612390162852e-05,
      "loss": 7.9241,
      "step": 10060
    },
    {
      "epoch": 0.028319848360851677,
      "grad_norm": 42.75,
      "learning_rate": 1.999560366271659e-05,
      "loss": 8.5944,
      "step": 10070
    },
    {
      "epoch": 0.028347971348300388,
      "grad_norm": 24.75,
      "learning_rate": 1.9995594926600928e-05,
      "loss": 7.6087,
      "step": 10080
    },
    {
      "epoch": 0.0283760943357491,
      "grad_norm": 33.25,
      "learning_rate": 1.999558618181587e-05,
      "loss": 7.81,
      "step": 10090
    },
    {
      "epoch": 0.028404217323197808,
      "grad_norm": 25.875,
      "learning_rate": 1.9995577428361427e-05,
      "loss": 8.3325,
      "step": 10100
    },
    {
      "epoch": 0.02843234031064652,
      "grad_norm": 73.5,
      "learning_rate": 1.99955686662376e-05,
      "loss": 8.3205,
      "step": 10110
    },
    {
      "epoch": 0.02846046329809523,
      "grad_norm": 22.125,
      "learning_rate": 1.9995559895444404e-05,
      "loss": 8.3681,
      "step": 10120
    },
    {
      "epoch": 0.028488586285543942,
      "grad_norm": 31.75,
      "learning_rate": 1.9995551115981844e-05,
      "loss": 8.1783,
      "step": 10130
    },
    {
      "epoch": 0.02851670927299265,
      "grad_norm": 34.5,
      "learning_rate": 1.9995542327849927e-05,
      "loss": 8.5446,
      "step": 10140
    },
    {
      "epoch": 0.028544832260441362,
      "grad_norm": 37.5,
      "learning_rate": 1.999553353104866e-05,
      "loss": 7.9473,
      "step": 10150
    },
    {
      "epoch": 0.028572955247890074,
      "grad_norm": 45.75,
      "learning_rate": 1.999552472557805e-05,
      "loss": 7.891,
      "step": 10160
    },
    {
      "epoch": 0.028601078235338785,
      "grad_norm": 38.0,
      "learning_rate": 1.9995515911438107e-05,
      "loss": 8.4031,
      "step": 10170
    },
    {
      "epoch": 0.028629201222787493,
      "grad_norm": 27.125,
      "learning_rate": 1.9995507088628836e-05,
      "loss": 7.2783,
      "step": 10180
    },
    {
      "epoch": 0.028657324210236205,
      "grad_norm": 32.5,
      "learning_rate": 1.9995498257150248e-05,
      "loss": 7.7798,
      "step": 10190
    },
    {
      "epoch": 0.028685447197684916,
      "grad_norm": 46.5,
      "learning_rate": 1.9995489417002348e-05,
      "loss": 7.7,
      "step": 10200
    },
    {
      "epoch": 0.028713570185133628,
      "grad_norm": 24.625,
      "learning_rate": 1.9995480568185145e-05,
      "loss": 7.6215,
      "step": 10210
    },
    {
      "epoch": 0.028741693172582336,
      "grad_norm": 25.0,
      "learning_rate": 1.9995471710698647e-05,
      "loss": 7.7735,
      "step": 10220
    },
    {
      "epoch": 0.028769816160031048,
      "grad_norm": 42.25,
      "learning_rate": 1.9995462844542858e-05,
      "loss": 7.7311,
      "step": 10230
    },
    {
      "epoch": 0.02879793914747976,
      "grad_norm": 49.5,
      "learning_rate": 1.9995453969717792e-05,
      "loss": 7.9439,
      "step": 10240
    },
    {
      "epoch": 0.02882606213492847,
      "grad_norm": 22.875,
      "learning_rate": 1.999544508622345e-05,
      "loss": 8.3863,
      "step": 10250
    },
    {
      "epoch": 0.02885418512237718,
      "grad_norm": 18.625,
      "learning_rate": 1.9995436194059846e-05,
      "loss": 7.4996,
      "step": 10260
    },
    {
      "epoch": 0.02888230810982589,
      "grad_norm": 29.5,
      "learning_rate": 1.9995427293226983e-05,
      "loss": 7.8813,
      "step": 10270
    },
    {
      "epoch": 0.028910431097274602,
      "grad_norm": 51.0,
      "learning_rate": 1.9995418383724872e-05,
      "loss": 8.077,
      "step": 10280
    },
    {
      "epoch": 0.028938554084723313,
      "grad_norm": 26.75,
      "learning_rate": 1.9995409465553516e-05,
      "loss": 8.3224,
      "step": 10290
    },
    {
      "epoch": 0.02896667707217202,
      "grad_norm": 46.0,
      "learning_rate": 1.999540053871293e-05,
      "loss": 6.9259,
      "step": 10300
    },
    {
      "epoch": 0.028994800059620733,
      "grad_norm": 47.25,
      "learning_rate": 1.9995391603203113e-05,
      "loss": 7.1346,
      "step": 10310
    },
    {
      "epoch": 0.029022923047069445,
      "grad_norm": 52.5,
      "learning_rate": 1.999538265902408e-05,
      "loss": 8.4722,
      "step": 10320
    },
    {
      "epoch": 0.029051046034518156,
      "grad_norm": 24.0,
      "learning_rate": 1.9995373706175838e-05,
      "loss": 7.4183,
      "step": 10330
    },
    {
      "epoch": 0.029079169021966864,
      "grad_norm": 33.0,
      "learning_rate": 1.9995364744658394e-05,
      "loss": 7.4469,
      "step": 10340
    },
    {
      "epoch": 0.029107292009415576,
      "grad_norm": 45.75,
      "learning_rate": 1.999535577447175e-05,
      "loss": 7.8218,
      "step": 10350
    },
    {
      "epoch": 0.029135414996864287,
      "grad_norm": 42.75,
      "learning_rate": 1.9995346795615917e-05,
      "loss": 7.8033,
      "step": 10360
    },
    {
      "epoch": 0.029163537984313,
      "grad_norm": 28.25,
      "learning_rate": 1.999533780809091e-05,
      "loss": 7.5636,
      "step": 10370
    },
    {
      "epoch": 0.029191660971761707,
      "grad_norm": 28.0,
      "learning_rate": 1.9995328811896728e-05,
      "loss": 7.9838,
      "step": 10380
    },
    {
      "epoch": 0.02921978395921042,
      "grad_norm": 31.875,
      "learning_rate": 1.999531980703338e-05,
      "loss": 7.9796,
      "step": 10390
    },
    {
      "epoch": 0.02924790694665913,
      "grad_norm": 47.25,
      "learning_rate": 1.9995310793500877e-05,
      "loss": 8.6816,
      "step": 10400
    },
    {
      "epoch": 0.02927602993410784,
      "grad_norm": 29.25,
      "learning_rate": 1.999530177129923e-05,
      "loss": 8.2196,
      "step": 10410
    },
    {
      "epoch": 0.02930415292155655,
      "grad_norm": 29.375,
      "learning_rate": 1.9995292740428436e-05,
      "loss": 8.2338,
      "step": 10420
    },
    {
      "epoch": 0.02933227590900526,
      "grad_norm": 36.0,
      "learning_rate": 1.999528370088851e-05,
      "loss": 7.6998,
      "step": 10430
    },
    {
      "epoch": 0.029360398896453973,
      "grad_norm": 39.25,
      "learning_rate": 1.9995274652679462e-05,
      "loss": 8.2312,
      "step": 10440
    },
    {
      "epoch": 0.029388521883902684,
      "grad_norm": 24.75,
      "learning_rate": 1.9995265595801294e-05,
      "loss": 8.754,
      "step": 10450
    },
    {
      "epoch": 0.029416644871351393,
      "grad_norm": 25.5,
      "learning_rate": 1.999525653025402e-05,
      "loss": 7.8424,
      "step": 10460
    },
    {
      "epoch": 0.029444767858800104,
      "grad_norm": 28.125,
      "learning_rate": 1.9995247456037643e-05,
      "loss": 8.4582,
      "step": 10470
    },
    {
      "epoch": 0.029472890846248816,
      "grad_norm": 22.375,
      "learning_rate": 1.9995238373152174e-05,
      "loss": 7.4337,
      "step": 10480
    },
    {
      "epoch": 0.029501013833697527,
      "grad_norm": 29.375,
      "learning_rate": 1.9995229281597615e-05,
      "loss": 6.7875,
      "step": 10490
    },
    {
      "epoch": 0.029529136821146235,
      "grad_norm": 20.875,
      "learning_rate": 1.999522018137398e-05,
      "loss": 7.8073,
      "step": 10500
    },
    {
      "epoch": 0.029557259808594947,
      "grad_norm": 35.25,
      "learning_rate": 1.9995211072481275e-05,
      "loss": 8.0483,
      "step": 10510
    },
    {
      "epoch": 0.02958538279604366,
      "grad_norm": 38.25,
      "learning_rate": 1.9995201954919507e-05,
      "loss": 8.0767,
      "step": 10520
    },
    {
      "epoch": 0.02961350578349237,
      "grad_norm": 37.75,
      "learning_rate": 1.9995192828688686e-05,
      "loss": 7.9907,
      "step": 10530
    },
    {
      "epoch": 0.029641628770941078,
      "grad_norm": 24.875,
      "learning_rate": 1.999518369378882e-05,
      "loss": 8.0382,
      "step": 10540
    },
    {
      "epoch": 0.02966975175838979,
      "grad_norm": 27.875,
      "learning_rate": 1.9995174550219914e-05,
      "loss": 8.3722,
      "step": 10550
    },
    {
      "epoch": 0.0296978747458385,
      "grad_norm": 43.75,
      "learning_rate": 1.999516539798198e-05,
      "loss": 8.1127,
      "step": 10560
    },
    {
      "epoch": 0.029725997733287213,
      "grad_norm": 28.25,
      "learning_rate": 1.9995156237075023e-05,
      "loss": 7.9002,
      "step": 10570
    },
    {
      "epoch": 0.02975412072073592,
      "grad_norm": 39.25,
      "learning_rate": 1.999514706749905e-05,
      "loss": 7.143,
      "step": 10580
    },
    {
      "epoch": 0.029782243708184632,
      "grad_norm": 23.625,
      "learning_rate": 1.999513788925407e-05,
      "loss": 7.8294,
      "step": 10590
    },
    {
      "epoch": 0.029810366695633344,
      "grad_norm": 33.5,
      "learning_rate": 1.9995128702340093e-05,
      "loss": 7.8767,
      "step": 10600
    },
    {
      "epoch": 0.029838489683082055,
      "grad_norm": 49.25,
      "learning_rate": 1.9995119506757125e-05,
      "loss": 8.1221,
      "step": 10610
    },
    {
      "epoch": 0.029866612670530764,
      "grad_norm": 28.0,
      "learning_rate": 1.999511030250517e-05,
      "loss": 8.5001,
      "step": 10620
    },
    {
      "epoch": 0.029894735657979475,
      "grad_norm": 33.25,
      "learning_rate": 1.9995101089584245e-05,
      "loss": 8.5017,
      "step": 10630
    },
    {
      "epoch": 0.029922858645428187,
      "grad_norm": 25.625,
      "learning_rate": 1.9995091867994353e-05,
      "loss": 7.3184,
      "step": 10640
    },
    {
      "epoch": 0.029950981632876898,
      "grad_norm": 25.0,
      "learning_rate": 1.9995082637735502e-05,
      "loss": 7.6897,
      "step": 10650
    },
    {
      "epoch": 0.029979104620325606,
      "grad_norm": 23.875,
      "learning_rate": 1.9995073398807703e-05,
      "loss": 8.3613,
      "step": 10660
    },
    {
      "epoch": 0.030007227607774318,
      "grad_norm": 42.0,
      "learning_rate": 1.9995064151210956e-05,
      "loss": 8.2334,
      "step": 10670
    },
    {
      "epoch": 0.03003535059522303,
      "grad_norm": 21.25,
      "learning_rate": 1.999505489494528e-05,
      "loss": 8.3809,
      "step": 10680
    },
    {
      "epoch": 0.03006347358267174,
      "grad_norm": 20.75,
      "learning_rate": 1.9995045630010672e-05,
      "loss": 7.5654,
      "step": 10690
    },
    {
      "epoch": 0.03009159657012045,
      "grad_norm": 26.875,
      "learning_rate": 1.9995036356407147e-05,
      "loss": 8.6393,
      "step": 10700
    },
    {
      "epoch": 0.03011971955756916,
      "grad_norm": 27.625,
      "learning_rate": 1.9995027074134714e-05,
      "loss": 7.8963,
      "step": 10710
    },
    {
      "epoch": 0.030147842545017872,
      "grad_norm": 44.0,
      "learning_rate": 1.9995017783193376e-05,
      "loss": 7.8087,
      "step": 10720
    },
    {
      "epoch": 0.030175965532466584,
      "grad_norm": 31.25,
      "learning_rate": 1.9995008483583144e-05,
      "loss": 7.5284,
      "step": 10730
    },
    {
      "epoch": 0.030204088519915292,
      "grad_norm": 23.125,
      "learning_rate": 1.9994999175304027e-05,
      "loss": 8.9282,
      "step": 10740
    },
    {
      "epoch": 0.030232211507364003,
      "grad_norm": 22.5,
      "learning_rate": 1.999498985835603e-05,
      "loss": 8.3088,
      "step": 10750
    },
    {
      "epoch": 0.030260334494812715,
      "grad_norm": 21.125,
      "learning_rate": 1.9994980532739166e-05,
      "loss": 8.0229,
      "step": 10760
    },
    {
      "epoch": 0.030288457482261427,
      "grad_norm": 20.5,
      "learning_rate": 1.9994971198453438e-05,
      "loss": 7.7891,
      "step": 10770
    },
    {
      "epoch": 0.030316580469710138,
      "grad_norm": 40.0,
      "learning_rate": 1.9994961855498856e-05,
      "loss": 8.3593,
      "step": 10780
    },
    {
      "epoch": 0.030344703457158846,
      "grad_norm": 32.0,
      "learning_rate": 1.9994952503875424e-05,
      "loss": 7.6064,
      "step": 10790
    },
    {
      "epoch": 0.030372826444607558,
      "grad_norm": 21.875,
      "learning_rate": 1.999494314358316e-05,
      "loss": 7.7996,
      "step": 10800
    },
    {
      "epoch": 0.03040094943205627,
      "grad_norm": 21.0,
      "learning_rate": 1.9994933774622063e-05,
      "loss": 8.5965,
      "step": 10810
    },
    {
      "epoch": 0.03042907241950498,
      "grad_norm": 19.375,
      "learning_rate": 1.9994924396992145e-05,
      "loss": 7.965,
      "step": 10820
    },
    {
      "epoch": 0.03045719540695369,
      "grad_norm": 29.0,
      "learning_rate": 1.9994915010693417e-05,
      "loss": 8.6208,
      "step": 10830
    },
    {
      "epoch": 0.0304853183944024,
      "grad_norm": 28.875,
      "learning_rate": 1.999490561572588e-05,
      "loss": 8.1938,
      "step": 10840
    },
    {
      "epoch": 0.030513441381851112,
      "grad_norm": 37.5,
      "learning_rate": 1.999489621208955e-05,
      "loss": 8.6939,
      "step": 10850
    },
    {
      "epoch": 0.030541564369299824,
      "grad_norm": 24.75,
      "learning_rate": 1.9994886799784425e-05,
      "loss": 7.5773,
      "step": 10860
    },
    {
      "epoch": 0.03056968735674853,
      "grad_norm": 46.75,
      "learning_rate": 1.9994877378810523e-05,
      "loss": 7.5652,
      "step": 10870
    },
    {
      "epoch": 0.030597810344197243,
      "grad_norm": 45.25,
      "learning_rate": 1.9994867949167845e-05,
      "loss": 8.1565,
      "step": 10880
    },
    {
      "epoch": 0.030625933331645955,
      "grad_norm": 27.125,
      "learning_rate": 1.9994858510856405e-05,
      "loss": 7.4442,
      "step": 10890
    },
    {
      "epoch": 0.030654056319094666,
      "grad_norm": 38.5,
      "learning_rate": 1.999484906387621e-05,
      "loss": 8.0493,
      "step": 10900
    },
    {
      "epoch": 0.030682179306543374,
      "grad_norm": 28.625,
      "learning_rate": 1.9994839608227265e-05,
      "loss": 8.7882,
      "step": 10910
    },
    {
      "epoch": 0.030710302293992086,
      "grad_norm": 31.625,
      "learning_rate": 1.999483014390958e-05,
      "loss": 7.1223,
      "step": 10920
    },
    {
      "epoch": 0.030738425281440798,
      "grad_norm": 36.75,
      "learning_rate": 1.9994820670923164e-05,
      "loss": 8.1631,
      "step": 10930
    },
    {
      "epoch": 0.03076654826888951,
      "grad_norm": 39.5,
      "learning_rate": 1.9994811189268025e-05,
      "loss": 7.4024,
      "step": 10940
    },
    {
      "epoch": 0.030794671256338217,
      "grad_norm": 35.25,
      "learning_rate": 1.999480169894417e-05,
      "loss": 8.0403,
      "step": 10950
    },
    {
      "epoch": 0.03082279424378693,
      "grad_norm": 22.875,
      "learning_rate": 1.999479219995161e-05,
      "loss": 7.9412,
      "step": 10960
    },
    {
      "epoch": 0.03085091723123564,
      "grad_norm": 27.0,
      "learning_rate": 1.999478269229035e-05,
      "loss": 7.3164,
      "step": 10970
    },
    {
      "epoch": 0.030879040218684352,
      "grad_norm": 36.75,
      "learning_rate": 1.99947731759604e-05,
      "loss": 7.8037,
      "step": 10980
    },
    {
      "epoch": 0.03090716320613306,
      "grad_norm": 23.375,
      "learning_rate": 1.9994763650961767e-05,
      "loss": 7.5078,
      "step": 10990
    },
    {
      "epoch": 0.03093528619358177,
      "grad_norm": 21.75,
      "learning_rate": 1.9994754117294457e-05,
      "loss": 8.8212,
      "step": 11000
    },
    {
      "epoch": 0.030963409181030483,
      "grad_norm": 33.0,
      "learning_rate": 1.9994744574958488e-05,
      "loss": 8.0928,
      "step": 11010
    },
    {
      "epoch": 0.030991532168479195,
      "grad_norm": 32.5,
      "learning_rate": 1.9994735023953854e-05,
      "loss": 8.3206,
      "step": 11020
    },
    {
      "epoch": 0.031019655155927903,
      "grad_norm": 23.5,
      "learning_rate": 1.9994725464280576e-05,
      "loss": 8.1828,
      "step": 11030
    },
    {
      "epoch": 0.031047778143376614,
      "grad_norm": 27.25,
      "learning_rate": 1.9994715895938656e-05,
      "loss": 7.9468,
      "step": 11040
    },
    {
      "epoch": 0.031075901130825326,
      "grad_norm": 29.875,
      "learning_rate": 1.9994706318928105e-05,
      "loss": 7.9164,
      "step": 11050
    },
    {
      "epoch": 0.031104024118274037,
      "grad_norm": 31.125,
      "learning_rate": 1.9994696733248924e-05,
      "loss": 7.7643,
      "step": 11060
    },
    {
      "epoch": 0.031132147105722745,
      "grad_norm": 29.25,
      "learning_rate": 1.9994687138901133e-05,
      "loss": 7.1076,
      "step": 11070
    },
    {
      "epoch": 0.031160270093171457,
      "grad_norm": 36.25,
      "learning_rate": 1.9994677535884732e-05,
      "loss": 8.8965,
      "step": 11080
    },
    {
      "epoch": 0.03118839308062017,
      "grad_norm": 24.5,
      "learning_rate": 1.9994667924199732e-05,
      "loss": 8.5676,
      "step": 11090
    },
    {
      "epoch": 0.03121651606806888,
      "grad_norm": 24.25,
      "learning_rate": 1.999465830384614e-05,
      "loss": 7.8746,
      "step": 11100
    },
    {
      "epoch": 0.031244639055517588,
      "grad_norm": 29.5,
      "learning_rate": 1.9994648674823968e-05,
      "loss": 7.8671,
      "step": 11110
    },
    {
      "epoch": 0.0312727620429663,
      "grad_norm": 72.5,
      "learning_rate": 1.999463903713322e-05,
      "loss": 8.3754,
      "step": 11120
    },
    {
      "epoch": 0.03130088503041501,
      "grad_norm": 25.25,
      "learning_rate": 1.9994629390773906e-05,
      "loss": 8.0884,
      "step": 11130
    },
    {
      "epoch": 0.03132900801786372,
      "grad_norm": 32.5,
      "learning_rate": 1.9994619735746034e-05,
      "loss": 8.2682,
      "step": 11140
    },
    {
      "epoch": 0.031357131005312434,
      "grad_norm": 31.0,
      "learning_rate": 1.9994610072049612e-05,
      "loss": 8.4546,
      "step": 11150
    },
    {
      "epoch": 0.031385253992761146,
      "grad_norm": 25.5,
      "learning_rate": 1.9994600399684653e-05,
      "loss": 7.207,
      "step": 11160
    },
    {
      "epoch": 0.03141337698020985,
      "grad_norm": 31.625,
      "learning_rate": 1.9994590718651158e-05,
      "loss": 8.0186,
      "step": 11170
    },
    {
      "epoch": 0.03144149996765856,
      "grad_norm": 27.25,
      "learning_rate": 1.9994581028949138e-05,
      "loss": 8.6081,
      "step": 11180
    },
    {
      "epoch": 0.031469622955107274,
      "grad_norm": 24.125,
      "learning_rate": 1.9994571330578606e-05,
      "loss": 8.436,
      "step": 11190
    },
    {
      "epoch": 0.031497745942555985,
      "grad_norm": 25.25,
      "learning_rate": 1.9994561623539567e-05,
      "loss": 8.1119,
      "step": 11200
    },
    {
      "epoch": 0.0315258689300047,
      "grad_norm": 40.0,
      "learning_rate": 1.9994551907832026e-05,
      "loss": 8.5074,
      "step": 11210
    },
    {
      "epoch": 0.03155399191745341,
      "grad_norm": 33.5,
      "learning_rate": 1.9994542183455998e-05,
      "loss": 8.0558,
      "step": 11220
    },
    {
      "epoch": 0.03158211490490212,
      "grad_norm": 28.125,
      "learning_rate": 1.9994532450411485e-05,
      "loss": 7.3696,
      "step": 11230
    },
    {
      "epoch": 0.03161023789235083,
      "grad_norm": 26.75,
      "learning_rate": 1.99945227086985e-05,
      "loss": 7.8066,
      "step": 11240
    },
    {
      "epoch": 0.031638360879799536,
      "grad_norm": 25.625,
      "learning_rate": 1.999451295831705e-05,
      "loss": 8.5742,
      "step": 11250
    },
    {
      "epoch": 0.03166648386724825,
      "grad_norm": 23.125,
      "learning_rate": 1.9994503199267145e-05,
      "loss": 8.7722,
      "step": 11260
    },
    {
      "epoch": 0.03169460685469696,
      "grad_norm": 24.875,
      "learning_rate": 1.9994493431548786e-05,
      "loss": 7.8013,
      "step": 11270
    },
    {
      "epoch": 0.03172272984214567,
      "grad_norm": 41.75,
      "learning_rate": 1.9994483655161994e-05,
      "loss": 8.2136,
      "step": 11280
    },
    {
      "epoch": 0.03175085282959438,
      "grad_norm": 21.125,
      "learning_rate": 1.999447387010677e-05,
      "loss": 7.6348,
      "step": 11290
    },
    {
      "epoch": 0.031778975817043094,
      "grad_norm": 24.625,
      "learning_rate": 1.999446407638312e-05,
      "loss": 8.0521,
      "step": 11300
    },
    {
      "epoch": 0.031807098804491805,
      "grad_norm": 28.75,
      "learning_rate": 1.9994454273991056e-05,
      "loss": 7.9637,
      "step": 11310
    },
    {
      "epoch": 0.03183522179194052,
      "grad_norm": 30.875,
      "learning_rate": 1.999444446293059e-05,
      "loss": 7.9769,
      "step": 11320
    },
    {
      "epoch": 0.03186334477938922,
      "grad_norm": 23.625,
      "learning_rate": 1.9994434643201725e-05,
      "loss": 7.6544,
      "step": 11330
    },
    {
      "epoch": 0.03189146776683793,
      "grad_norm": 26.625,
      "learning_rate": 1.999442481480447e-05,
      "loss": 8.424,
      "step": 11340
    },
    {
      "epoch": 0.031919590754286645,
      "grad_norm": 47.0,
      "learning_rate": 1.9994414977738836e-05,
      "loss": 8.4298,
      "step": 11350
    },
    {
      "epoch": 0.031947713741735356,
      "grad_norm": 22.125,
      "learning_rate": 1.999440513200483e-05,
      "loss": 7.9579,
      "step": 11360
    },
    {
      "epoch": 0.03197583672918407,
      "grad_norm": 42.25,
      "learning_rate": 1.9994395277602464e-05,
      "loss": 7.7313,
      "step": 11370
    },
    {
      "epoch": 0.03200395971663278,
      "grad_norm": 46.0,
      "learning_rate": 1.999438541453174e-05,
      "loss": 7.3802,
      "step": 11380
    },
    {
      "epoch": 0.03203208270408149,
      "grad_norm": 20.625,
      "learning_rate": 1.999437554279267e-05,
      "loss": 7.9468,
      "step": 11390
    },
    {
      "epoch": 0.0320602056915302,
      "grad_norm": 25.875,
      "learning_rate": 1.9994365662385264e-05,
      "loss": 7.1312,
      "step": 11400
    },
    {
      "epoch": 0.03208832867897891,
      "grad_norm": 29.125,
      "learning_rate": 1.9994355773309527e-05,
      "loss": 8.3429,
      "step": 11410
    },
    {
      "epoch": 0.03211645166642762,
      "grad_norm": 28.5,
      "learning_rate": 1.999434587556547e-05,
      "loss": 8.2113,
      "step": 11420
    },
    {
      "epoch": 0.03214457465387633,
      "grad_norm": 42.75,
      "learning_rate": 1.9994335969153104e-05,
      "loss": 7.454,
      "step": 11430
    },
    {
      "epoch": 0.03217269764132504,
      "grad_norm": 25.5,
      "learning_rate": 1.9994326054072436e-05,
      "loss": 8.7546,
      "step": 11440
    },
    {
      "epoch": 0.03220082062877375,
      "grad_norm": 23.25,
      "learning_rate": 1.9994316130323473e-05,
      "loss": 7.9789,
      "step": 11450
    },
    {
      "epoch": 0.032228943616222465,
      "grad_norm": 57.0,
      "learning_rate": 1.999430619790622e-05,
      "loss": 8.4451,
      "step": 11460
    },
    {
      "epoch": 0.032257066603671176,
      "grad_norm": 39.25,
      "learning_rate": 1.9994296256820693e-05,
      "loss": 7.3193,
      "step": 11470
    },
    {
      "epoch": 0.03228518959111989,
      "grad_norm": 30.75,
      "learning_rate": 1.9994286307066897e-05,
      "loss": 7.4242,
      "step": 11480
    },
    {
      "epoch": 0.0323133125785686,
      "grad_norm": 21.125,
      "learning_rate": 1.999427634864484e-05,
      "loss": 8.7769,
      "step": 11490
    },
    {
      "epoch": 0.032341435566017304,
      "grad_norm": 38.0,
      "learning_rate": 1.9994266381554532e-05,
      "loss": 8.1196,
      "step": 11500
    },
    {
      "epoch": 0.032369558553466016,
      "grad_norm": 24.875,
      "learning_rate": 1.9994256405795984e-05,
      "loss": 8.0855,
      "step": 11510
    },
    {
      "epoch": 0.03239768154091473,
      "grad_norm": 37.75,
      "learning_rate": 1.9994246421369196e-05,
      "loss": 8.1357,
      "step": 11520
    },
    {
      "epoch": 0.03242580452836344,
      "grad_norm": 29.5,
      "learning_rate": 1.9994236428274187e-05,
      "loss": 7.7256,
      "step": 11530
    },
    {
      "epoch": 0.03245392751581215,
      "grad_norm": 23.875,
      "learning_rate": 1.9994226426510962e-05,
      "loss": 7.7523,
      "step": 11540
    },
    {
      "epoch": 0.03248205050326086,
      "grad_norm": 32.0,
      "learning_rate": 1.9994216416079527e-05,
      "loss": 7.8625,
      "step": 11550
    },
    {
      "epoch": 0.032510173490709574,
      "grad_norm": 34.0,
      "learning_rate": 1.9994206396979892e-05,
      "loss": 7.8557,
      "step": 11560
    },
    {
      "epoch": 0.032538296478158285,
      "grad_norm": 43.75,
      "learning_rate": 1.9994196369212068e-05,
      "loss": 7.6083,
      "step": 11570
    },
    {
      "epoch": 0.03256641946560699,
      "grad_norm": 25.75,
      "learning_rate": 1.999418633277606e-05,
      "loss": 7.8477,
      "step": 11580
    },
    {
      "epoch": 0.0325945424530557,
      "grad_norm": 27.875,
      "learning_rate": 1.999417628767188e-05,
      "loss": 7.5025,
      "step": 11590
    },
    {
      "epoch": 0.03262266544050441,
      "grad_norm": 22.875,
      "learning_rate": 1.9994166233899535e-05,
      "loss": 7.4354,
      "step": 11600
    },
    {
      "epoch": 0.032650788427953124,
      "grad_norm": 42.0,
      "learning_rate": 1.9994156171459037e-05,
      "loss": 7.8523,
      "step": 11610
    },
    {
      "epoch": 0.032678911415401836,
      "grad_norm": 21.0,
      "learning_rate": 1.999414610035039e-05,
      "loss": 8.1717,
      "step": 11620
    },
    {
      "epoch": 0.03270703440285055,
      "grad_norm": 22.5,
      "learning_rate": 1.9994136020573604e-05,
      "loss": 8.0534,
      "step": 11630
    },
    {
      "epoch": 0.03273515739029926,
      "grad_norm": 40.75,
      "learning_rate": 1.9994125932128687e-05,
      "loss": 8.8001,
      "step": 11640
    },
    {
      "epoch": 0.03276328037774797,
      "grad_norm": 30.0,
      "learning_rate": 1.9994115835015654e-05,
      "loss": 7.7579,
      "step": 11650
    },
    {
      "epoch": 0.032791403365196675,
      "grad_norm": 33.25,
      "learning_rate": 1.9994105729234502e-05,
      "loss": 8.0191,
      "step": 11660
    },
    {
      "epoch": 0.03281952635264539,
      "grad_norm": 31.5,
      "learning_rate": 1.9994095614785253e-05,
      "loss": 7.9548,
      "step": 11670
    },
    {
      "epoch": 0.0328476493400941,
      "grad_norm": 42.75,
      "learning_rate": 1.999408549166791e-05,
      "loss": 8.3128,
      "step": 11680
    },
    {
      "epoch": 0.03287577232754281,
      "grad_norm": 32.0,
      "learning_rate": 1.9994075359882473e-05,
      "loss": 8.1475,
      "step": 11690
    },
    {
      "epoch": 0.03290389531499152,
      "grad_norm": 23.0,
      "learning_rate": 1.9994065219428964e-05,
      "loss": 7.2656,
      "step": 11700
    },
    {
      "epoch": 0.03293201830244023,
      "grad_norm": 36.25,
      "learning_rate": 1.9994055070307386e-05,
      "loss": 8.206,
      "step": 11710
    },
    {
      "epoch": 0.032960141289888945,
      "grad_norm": 36.25,
      "learning_rate": 1.999404491251775e-05,
      "loss": 7.915,
      "step": 11720
    },
    {
      "epoch": 0.032988264277337656,
      "grad_norm": 30.375,
      "learning_rate": 1.9994034746060062e-05,
      "loss": 7.8346,
      "step": 11730
    },
    {
      "epoch": 0.03301638726478636,
      "grad_norm": 43.75,
      "learning_rate": 1.9994024570934333e-05,
      "loss": 7.5286,
      "step": 11740
    },
    {
      "epoch": 0.03304451025223507,
      "grad_norm": 21.25,
      "learning_rate": 1.999401438714057e-05,
      "loss": 7.4084,
      "step": 11750
    },
    {
      "epoch": 0.033072633239683784,
      "grad_norm": 26.625,
      "learning_rate": 1.9994004194678786e-05,
      "loss": 7.5272,
      "step": 11760
    },
    {
      "epoch": 0.033100756227132495,
      "grad_norm": 29.625,
      "learning_rate": 1.9993993993548982e-05,
      "loss": 7.8422,
      "step": 11770
    },
    {
      "epoch": 0.03312887921458121,
      "grad_norm": 36.0,
      "learning_rate": 1.9993983783751176e-05,
      "loss": 7.7822,
      "step": 11780
    },
    {
      "epoch": 0.03315700220202992,
      "grad_norm": 45.75,
      "learning_rate": 1.999397356528537e-05,
      "loss": 7.2096,
      "step": 11790
    },
    {
      "epoch": 0.03318512518947863,
      "grad_norm": 24.375,
      "learning_rate": 1.9993963338151572e-05,
      "loss": 7.7872,
      "step": 11800
    },
    {
      "epoch": 0.03321324817692734,
      "grad_norm": 29.125,
      "learning_rate": 1.99939531023498e-05,
      "loss": 8.0382,
      "step": 11810
    },
    {
      "epoch": 0.033241371164376046,
      "grad_norm": 30.25,
      "learning_rate": 1.9993942857880055e-05,
      "loss": 8.1769,
      "step": 11820
    },
    {
      "epoch": 0.03326949415182476,
      "grad_norm": 21.75,
      "learning_rate": 1.9993932604742347e-05,
      "loss": 7.7626,
      "step": 11830
    },
    {
      "epoch": 0.03329761713927347,
      "grad_norm": 31.625,
      "learning_rate": 1.9993922342936686e-05,
      "loss": 8.1302,
      "step": 11840
    },
    {
      "epoch": 0.03332574012672218,
      "grad_norm": 32.5,
      "learning_rate": 1.999391207246308e-05,
      "loss": 7.6275,
      "step": 11850
    },
    {
      "epoch": 0.03335386311417089,
      "grad_norm": 35.0,
      "learning_rate": 1.9993901793321542e-05,
      "loss": 8.7084,
      "step": 11860
    },
    {
      "epoch": 0.033381986101619604,
      "grad_norm": 29.375,
      "learning_rate": 1.9993891505512076e-05,
      "loss": 8.89,
      "step": 11870
    },
    {
      "epoch": 0.033410109089068316,
      "grad_norm": 27.75,
      "learning_rate": 1.999388120903469e-05,
      "loss": 8.1222,
      "step": 11880
    },
    {
      "epoch": 0.03343823207651703,
      "grad_norm": 29.25,
      "learning_rate": 1.9993870903889397e-05,
      "loss": 8.0613,
      "step": 11890
    },
    {
      "epoch": 0.03346635506396573,
      "grad_norm": 28.375,
      "learning_rate": 1.9993860590076205e-05,
      "loss": 8.7048,
      "step": 11900
    },
    {
      "epoch": 0.03349447805141444,
      "grad_norm": 27.0,
      "learning_rate": 1.999385026759512e-05,
      "loss": 8.092,
      "step": 11910
    },
    {
      "epoch": 0.033522601038863155,
      "grad_norm": 28.875,
      "learning_rate": 1.9993839936446157e-05,
      "loss": 6.9908,
      "step": 11920
    },
    {
      "epoch": 0.033550724026311866,
      "grad_norm": 24.625,
      "learning_rate": 1.9993829596629318e-05,
      "loss": 8.0606,
      "step": 11930
    },
    {
      "epoch": 0.03357884701376058,
      "grad_norm": 41.0,
      "learning_rate": 1.9993819248144614e-05,
      "loss": 8.3727,
      "step": 11940
    },
    {
      "epoch": 0.03360697000120929,
      "grad_norm": 38.75,
      "learning_rate": 1.999380889099206e-05,
      "loss": 7.431,
      "step": 11950
    },
    {
      "epoch": 0.033635092988658,
      "grad_norm": 24.0,
      "learning_rate": 1.9993798525171655e-05,
      "loss": 8.2877,
      "step": 11960
    },
    {
      "epoch": 0.03366321597610671,
      "grad_norm": 36.25,
      "learning_rate": 1.9993788150683416e-05,
      "loss": 7.9029,
      "step": 11970
    },
    {
      "epoch": 0.03369133896355542,
      "grad_norm": 30.5,
      "learning_rate": 1.9993777767527347e-05,
      "loss": 8.789,
      "step": 11980
    },
    {
      "epoch": 0.03371946195100413,
      "grad_norm": 32.0,
      "learning_rate": 1.999376737570346e-05,
      "loss": 8.5905,
      "step": 11990
    },
    {
      "epoch": 0.03374758493845284,
      "grad_norm": 31.75,
      "learning_rate": 1.9993756975211764e-05,
      "loss": 7.7265,
      "step": 12000
    },
    {
      "epoch": 0.03377570792590155,
      "grad_norm": 33.25,
      "learning_rate": 1.9993746566052268e-05,
      "loss": 8.0885,
      "step": 12010
    },
    {
      "epoch": 0.03380383091335026,
      "grad_norm": 26.5,
      "learning_rate": 1.999373614822498e-05,
      "loss": 7.5074,
      "step": 12020
    },
    {
      "epoch": 0.033831953900798975,
      "grad_norm": 37.5,
      "learning_rate": 1.9993725721729906e-05,
      "loss": 7.9845,
      "step": 12030
    },
    {
      "epoch": 0.03386007688824769,
      "grad_norm": 37.25,
      "learning_rate": 1.9993715286567058e-05,
      "loss": 8.4465,
      "step": 12040
    },
    {
      "epoch": 0.0338881998756964,
      "grad_norm": 22.875,
      "learning_rate": 1.9993704842736446e-05,
      "loss": 6.9398,
      "step": 12050
    },
    {
      "epoch": 0.0339163228631451,
      "grad_norm": 27.625,
      "learning_rate": 1.9993694390238082e-05,
      "loss": 8.5352,
      "step": 12060
    },
    {
      "epoch": 0.033944445850593814,
      "grad_norm": 26.0,
      "learning_rate": 1.999368392907197e-05,
      "loss": 8.141,
      "step": 12070
    },
    {
      "epoch": 0.033972568838042526,
      "grad_norm": 42.75,
      "learning_rate": 1.9993673459238117e-05,
      "loss": 7.7464,
      "step": 12080
    },
    {
      "epoch": 0.03400069182549124,
      "grad_norm": 31.875,
      "learning_rate": 1.999366298073654e-05,
      "loss": 7.4315,
      "step": 12090
    },
    {
      "epoch": 0.03402881481293995,
      "grad_norm": 26.5,
      "learning_rate": 1.9993652493567244e-05,
      "loss": 8.5784,
      "step": 12100
    },
    {
      "epoch": 0.03405693780038866,
      "grad_norm": 36.75,
      "learning_rate": 1.9993641997730236e-05,
      "loss": 8.0436,
      "step": 12110
    },
    {
      "epoch": 0.03408506078783737,
      "grad_norm": 22.75,
      "learning_rate": 1.9993631493225526e-05,
      "loss": 8.2972,
      "step": 12120
    },
    {
      "epoch": 0.034113183775286084,
      "grad_norm": 26.375,
      "learning_rate": 1.999362098005312e-05,
      "loss": 8.21,
      "step": 12130
    },
    {
      "epoch": 0.034141306762734795,
      "grad_norm": 34.5,
      "learning_rate": 1.999361045821304e-05,
      "loss": 8.1532,
      "step": 12140
    },
    {
      "epoch": 0.0341694297501835,
      "grad_norm": 31.5,
      "learning_rate": 1.999359992770528e-05,
      "loss": 8.0812,
      "step": 12150
    },
    {
      "epoch": 0.03419755273763221,
      "grad_norm": 27.0,
      "learning_rate": 1.9993589388529857e-05,
      "loss": 8.1976,
      "step": 12160
    },
    {
      "epoch": 0.03422567572508092,
      "grad_norm": 30.125,
      "learning_rate": 1.9993578840686777e-05,
      "loss": 8.38,
      "step": 12170
    },
    {
      "epoch": 0.034253798712529634,
      "grad_norm": 23.875,
      "learning_rate": 1.9993568284176054e-05,
      "loss": 7.4026,
      "step": 12180
    },
    {
      "epoch": 0.034281921699978346,
      "grad_norm": 41.25,
      "learning_rate": 1.999355771899769e-05,
      "loss": 7.2106,
      "step": 12190
    },
    {
      "epoch": 0.03431004468742706,
      "grad_norm": 35.0,
      "learning_rate": 1.9993547145151702e-05,
      "loss": 8.3326,
      "step": 12200
    },
    {
      "epoch": 0.03433816767487577,
      "grad_norm": 35.75,
      "learning_rate": 1.999353656263809e-05,
      "loss": 8.4656,
      "step": 12210
    },
    {
      "epoch": 0.03436629066232448,
      "grad_norm": 36.5,
      "learning_rate": 1.9993525971456872e-05,
      "loss": 8.5133,
      "step": 12220
    },
    {
      "epoch": 0.034394413649773185,
      "grad_norm": 32.0,
      "learning_rate": 1.9993515371608054e-05,
      "loss": 8.1739,
      "step": 12230
    },
    {
      "epoch": 0.0344225366372219,
      "grad_norm": 34.75,
      "learning_rate": 1.9993504763091642e-05,
      "loss": 7.9026,
      "step": 12240
    },
    {
      "epoch": 0.03445065962467061,
      "grad_norm": 29.25,
      "learning_rate": 1.999349414590765e-05,
      "loss": 8.2083,
      "step": 12250
    },
    {
      "epoch": 0.03447878261211932,
      "grad_norm": 42.25,
      "learning_rate": 1.9993483520056082e-05,
      "loss": 8.1157,
      "step": 12260
    },
    {
      "epoch": 0.03450690559956803,
      "grad_norm": 33.5,
      "learning_rate": 1.9993472885536952e-05,
      "loss": 7.5783,
      "step": 12270
    },
    {
      "epoch": 0.03453502858701674,
      "grad_norm": 51.25,
      "learning_rate": 1.999346224235027e-05,
      "loss": 7.4398,
      "step": 12280
    },
    {
      "epoch": 0.034563151574465455,
      "grad_norm": 27.375,
      "learning_rate": 1.999345159049604e-05,
      "loss": 7.348,
      "step": 12290
    },
    {
      "epoch": 0.034591274561914166,
      "grad_norm": 22.875,
      "learning_rate": 1.9993440929974275e-05,
      "loss": 8.1754,
      "step": 12300
    },
    {
      "epoch": 0.03461939754936287,
      "grad_norm": 52.25,
      "learning_rate": 1.9993430260784984e-05,
      "loss": 9.4514,
      "step": 12310
    },
    {
      "epoch": 0.03464752053681158,
      "grad_norm": 21.25,
      "learning_rate": 1.9993419582928172e-05,
      "loss": 7.6919,
      "step": 12320
    },
    {
      "epoch": 0.034675643524260294,
      "grad_norm": 24.0,
      "learning_rate": 1.9993408896403858e-05,
      "loss": 7.3727,
      "step": 12330
    },
    {
      "epoch": 0.034703766511709006,
      "grad_norm": 35.5,
      "learning_rate": 1.9993398201212038e-05,
      "loss": 7.8566,
      "step": 12340
    },
    {
      "epoch": 0.03473188949915772,
      "grad_norm": 39.0,
      "learning_rate": 1.999338749735273e-05,
      "loss": 8.3421,
      "step": 12350
    },
    {
      "epoch": 0.03476001248660643,
      "grad_norm": 27.375,
      "learning_rate": 1.9993376784825945e-05,
      "loss": 7.8186,
      "step": 12360
    },
    {
      "epoch": 0.03478813547405514,
      "grad_norm": 44.25,
      "learning_rate": 1.9993366063631688e-05,
      "loss": 8.126,
      "step": 12370
    },
    {
      "epoch": 0.03481625846150385,
      "grad_norm": 26.875,
      "learning_rate": 1.999335533376997e-05,
      "loss": 7.9688,
      "step": 12380
    },
    {
      "epoch": 0.034844381448952556,
      "grad_norm": 30.0,
      "learning_rate": 1.9993344595240797e-05,
      "loss": 8.0047,
      "step": 12390
    },
    {
      "epoch": 0.03487250443640127,
      "grad_norm": 42.25,
      "learning_rate": 1.999333384804418e-05,
      "loss": 7.9628,
      "step": 12400
    },
    {
      "epoch": 0.03490062742384998,
      "grad_norm": 35.75,
      "learning_rate": 1.999332309218013e-05,
      "loss": 8.5207,
      "step": 12410
    },
    {
      "epoch": 0.03492875041129869,
      "grad_norm": 47.75,
      "learning_rate": 1.9993312327648654e-05,
      "loss": 8.3276,
      "step": 12420
    },
    {
      "epoch": 0.0349568733987474,
      "grad_norm": 26.125,
      "learning_rate": 1.9993301554449766e-05,
      "loss": 8.159,
      "step": 12430
    },
    {
      "epoch": 0.034984996386196114,
      "grad_norm": 29.0,
      "learning_rate": 1.999329077258347e-05,
      "loss": 8.2998,
      "step": 12440
    },
    {
      "epoch": 0.035013119373644826,
      "grad_norm": 35.0,
      "learning_rate": 1.999327998204978e-05,
      "loss": 7.2331,
      "step": 12450
    },
    {
      "epoch": 0.03504124236109354,
      "grad_norm": 31.375,
      "learning_rate": 1.99932691828487e-05,
      "loss": 7.7339,
      "step": 12460
    },
    {
      "epoch": 0.03506936534854224,
      "grad_norm": 34.0,
      "learning_rate": 1.999325837498024e-05,
      "loss": 7.8044,
      "step": 12470
    },
    {
      "epoch": 0.03509748833599095,
      "grad_norm": 25.75,
      "learning_rate": 1.9993247558444416e-05,
      "loss": 7.456,
      "step": 12480
    },
    {
      "epoch": 0.035125611323439665,
      "grad_norm": 43.75,
      "learning_rate": 1.9993236733241232e-05,
      "loss": 7.4126,
      "step": 12490
    },
    {
      "epoch": 0.035153734310888377,
      "grad_norm": 34.0,
      "learning_rate": 1.99932258993707e-05,
      "loss": 7.8842,
      "step": 12500
    },
    {
      "epoch": 0.03518185729833709,
      "grad_norm": 24.875,
      "learning_rate": 1.9993215056832826e-05,
      "loss": 8.0112,
      "step": 12510
    },
    {
      "epoch": 0.0352099802857858,
      "grad_norm": 48.0,
      "learning_rate": 1.999320420562762e-05,
      "loss": 8.3875,
      "step": 12520
    },
    {
      "epoch": 0.03523810327323451,
      "grad_norm": 40.5,
      "learning_rate": 1.9993193345755093e-05,
      "loss": 7.612,
      "step": 12530
    },
    {
      "epoch": 0.03526622626068322,
      "grad_norm": 36.5,
      "learning_rate": 1.9993182477215253e-05,
      "loss": 7.7024,
      "step": 12540
    },
    {
      "epoch": 0.03529434924813193,
      "grad_norm": 32.5,
      "learning_rate": 1.9993171600008115e-05,
      "loss": 7.7255,
      "step": 12550
    },
    {
      "epoch": 0.03532247223558064,
      "grad_norm": 23.625,
      "learning_rate": 1.999316071413368e-05,
      "loss": 8.9419,
      "step": 12560
    },
    {
      "epoch": 0.03535059522302935,
      "grad_norm": 41.25,
      "learning_rate": 1.999314981959196e-05,
      "loss": 8.1375,
      "step": 12570
    },
    {
      "epoch": 0.03537871821047806,
      "grad_norm": 27.75,
      "learning_rate": 1.9993138916382968e-05,
      "loss": 8.4635,
      "step": 12580
    },
    {
      "epoch": 0.035406841197926774,
      "grad_norm": 39.25,
      "learning_rate": 1.999312800450671e-05,
      "loss": 7.6638,
      "step": 12590
    },
    {
      "epoch": 0.035434964185375485,
      "grad_norm": 43.0,
      "learning_rate": 1.9993117083963198e-05,
      "loss": 7.3149,
      "step": 12600
    },
    {
      "epoch": 0.0354630871728242,
      "grad_norm": 41.0,
      "learning_rate": 1.9993106154752438e-05,
      "loss": 8.4439,
      "step": 12610
    },
    {
      "epoch": 0.03549121016027291,
      "grad_norm": 22.625,
      "learning_rate": 1.9993095216874444e-05,
      "loss": 7.4968,
      "step": 12620
    },
    {
      "epoch": 0.03551933314772161,
      "grad_norm": 38.25,
      "learning_rate": 1.9993084270329223e-05,
      "loss": 8.0752,
      "step": 12630
    },
    {
      "epoch": 0.035547456135170324,
      "grad_norm": 44.25,
      "learning_rate": 1.9993073315116785e-05,
      "loss": 7.7841,
      "step": 12640
    },
    {
      "epoch": 0.035575579122619036,
      "grad_norm": 38.0,
      "learning_rate": 1.999306235123714e-05,
      "loss": 7.6506,
      "step": 12650
    },
    {
      "epoch": 0.03560370211006775,
      "grad_norm": 28.375,
      "learning_rate": 1.999305137869029e-05,
      "loss": 8.454,
      "step": 12660
    },
    {
      "epoch": 0.03563182509751646,
      "grad_norm": 31.875,
      "learning_rate": 1.9993040397476258e-05,
      "loss": 7.8975,
      "step": 12670
    },
    {
      "epoch": 0.03565994808496517,
      "grad_norm": 25.875,
      "learning_rate": 1.9993029407595045e-05,
      "loss": 7.9324,
      "step": 12680
    },
    {
      "epoch": 0.03568807107241388,
      "grad_norm": 32.5,
      "learning_rate": 1.999301840904666e-05,
      "loss": 8.3967,
      "step": 12690
    },
    {
      "epoch": 0.035716194059862594,
      "grad_norm": 32.0,
      "learning_rate": 1.9993007401831115e-05,
      "loss": 8.6223,
      "step": 12700
    },
    {
      "epoch": 0.0357443170473113,
      "grad_norm": 31.625,
      "learning_rate": 1.999299638594842e-05,
      "loss": 8.4912,
      "step": 12710
    },
    {
      "epoch": 0.03577244003476001,
      "grad_norm": 26.75,
      "learning_rate": 1.9992985361398586e-05,
      "loss": 7.5636,
      "step": 12720
    },
    {
      "epoch": 0.03580056302220872,
      "grad_norm": 34.25,
      "learning_rate": 1.9992974328181616e-05,
      "loss": 7.9519,
      "step": 12730
    },
    {
      "epoch": 0.03582868600965743,
      "grad_norm": 21.125,
      "learning_rate": 1.9992963286297528e-05,
      "loss": 7.8472,
      "step": 12740
    },
    {
      "epoch": 0.035856808997106145,
      "grad_norm": 27.875,
      "learning_rate": 1.9992952235746324e-05,
      "loss": 7.9171,
      "step": 12750
    },
    {
      "epoch": 0.035884931984554856,
      "grad_norm": 34.5,
      "learning_rate": 1.9992941176528017e-05,
      "loss": 8.0738,
      "step": 12760
    },
    {
      "epoch": 0.03591305497200357,
      "grad_norm": 27.125,
      "learning_rate": 1.999293010864262e-05,
      "loss": 8.637,
      "step": 12770
    },
    {
      "epoch": 0.03594117795945228,
      "grad_norm": 27.625,
      "learning_rate": 1.999291903209014e-05,
      "loss": 8.0261,
      "step": 12780
    },
    {
      "epoch": 0.03596930094690099,
      "grad_norm": 29.0,
      "learning_rate": 1.9992907946870585e-05,
      "loss": 7.9682,
      "step": 12790
    },
    {
      "epoch": 0.035997423934349695,
      "grad_norm": 33.25,
      "learning_rate": 1.9992896852983962e-05,
      "loss": 7.8657,
      "step": 12800
    },
    {
      "epoch": 0.03602554692179841,
      "grad_norm": 20.75,
      "learning_rate": 1.9992885750430285e-05,
      "loss": 8.2606,
      "step": 12810
    },
    {
      "epoch": 0.03605366990924712,
      "grad_norm": 28.0,
      "learning_rate": 1.9992874639209563e-05,
      "loss": 8.5209,
      "step": 12820
    },
    {
      "epoch": 0.03608179289669583,
      "grad_norm": 30.5,
      "learning_rate": 1.9992863519321807e-05,
      "loss": 8.5619,
      "step": 12830
    },
    {
      "epoch": 0.03610991588414454,
      "grad_norm": 36.0,
      "learning_rate": 1.9992852390767027e-05,
      "loss": 8.8402,
      "step": 12840
    },
    {
      "epoch": 0.03613803887159325,
      "grad_norm": 24.5,
      "learning_rate": 1.9992841253545227e-05,
      "loss": 7.9641,
      "step": 12850
    },
    {
      "epoch": 0.036166161859041965,
      "grad_norm": 34.75,
      "learning_rate": 1.999283010765642e-05,
      "loss": 7.7746,
      "step": 12860
    },
    {
      "epoch": 0.036194284846490676,
      "grad_norm": 48.0,
      "learning_rate": 1.9992818953100617e-05,
      "loss": 8.4606,
      "step": 12870
    },
    {
      "epoch": 0.03622240783393938,
      "grad_norm": 24.875,
      "learning_rate": 1.999280778987783e-05,
      "loss": 8.8518,
      "step": 12880
    },
    {
      "epoch": 0.03625053082138809,
      "grad_norm": 43.25,
      "learning_rate": 1.999279661798806e-05,
      "loss": 8.6471,
      "step": 12890
    },
    {
      "epoch": 0.036278653808836804,
      "grad_norm": 23.125,
      "learning_rate": 1.9992785437431326e-05,
      "loss": 8.2689,
      "step": 12900
    },
    {
      "epoch": 0.036306776796285516,
      "grad_norm": 43.25,
      "learning_rate": 1.999277424820763e-05,
      "loss": 8.4438,
      "step": 12910
    },
    {
      "epoch": 0.03633489978373423,
      "grad_norm": 23.25,
      "learning_rate": 1.9992763050316988e-05,
      "loss": 8.7991,
      "step": 12920
    },
    {
      "epoch": 0.03636302277118294,
      "grad_norm": 22.875,
      "learning_rate": 1.9992751843759408e-05,
      "loss": 7.9707,
      "step": 12930
    },
    {
      "epoch": 0.03639114575863165,
      "grad_norm": 28.75,
      "learning_rate": 1.99927406285349e-05,
      "loss": 8.4509,
      "step": 12940
    },
    {
      "epoch": 0.03641926874608036,
      "grad_norm": 33.5,
      "learning_rate": 1.999272940464347e-05,
      "loss": 8.2446,
      "step": 12950
    },
    {
      "epoch": 0.036447391733529066,
      "grad_norm": 32.5,
      "learning_rate": 1.999271817208513e-05,
      "loss": 7.7981,
      "step": 12960
    },
    {
      "epoch": 0.03647551472097778,
      "grad_norm": 25.625,
      "learning_rate": 1.9992706930859887e-05,
      "loss": 8.3955,
      "step": 12970
    },
    {
      "epoch": 0.03650363770842649,
      "grad_norm": 30.625,
      "learning_rate": 1.999269568096776e-05,
      "loss": 8.6762,
      "step": 12980
    },
    {
      "epoch": 0.0365317606958752,
      "grad_norm": 24.875,
      "learning_rate": 1.999268442240875e-05,
      "loss": 8.1404,
      "step": 12990
    },
    {
      "epoch": 0.03655988368332391,
      "grad_norm": 38.5,
      "learning_rate": 1.9992673155182872e-05,
      "loss": 7.8047,
      "step": 13000
    },
    {
      "epoch": 0.036588006670772624,
      "grad_norm": 20.375,
      "learning_rate": 1.999266187929013e-05,
      "loss": 7.6434,
      "step": 13010
    },
    {
      "epoch": 0.036616129658221336,
      "grad_norm": 28.0,
      "learning_rate": 1.9992650594730536e-05,
      "loss": 8.3842,
      "step": 13020
    },
    {
      "epoch": 0.03664425264567005,
      "grad_norm": 34.25,
      "learning_rate": 1.9992639301504104e-05,
      "loss": 8.2192,
      "step": 13030
    },
    {
      "epoch": 0.03667237563311875,
      "grad_norm": 34.25,
      "learning_rate": 1.9992627999610838e-05,
      "loss": 8.4654,
      "step": 13040
    },
    {
      "epoch": 0.036700498620567464,
      "grad_norm": 28.625,
      "learning_rate": 1.999261668905075e-05,
      "loss": 7.954,
      "step": 13050
    },
    {
      "epoch": 0.036728621608016175,
      "grad_norm": 24.875,
      "learning_rate": 1.9992605369823854e-05,
      "loss": 7.7056,
      "step": 13060
    },
    {
      "epoch": 0.03675674459546489,
      "grad_norm": 29.75,
      "learning_rate": 1.9992594041930153e-05,
      "loss": 7.5381,
      "step": 13070
    },
    {
      "epoch": 0.0367848675829136,
      "grad_norm": 25.875,
      "learning_rate": 1.999258270536966e-05,
      "loss": 9.2101,
      "step": 13080
    },
    {
      "epoch": 0.03681299057036231,
      "grad_norm": 21.0,
      "learning_rate": 1.9992571360142383e-05,
      "loss": 8.0199,
      "step": 13090
    },
    {
      "epoch": 0.03684111355781102,
      "grad_norm": 21.5,
      "learning_rate": 1.9992560006248335e-05,
      "loss": 7.6696,
      "step": 13100
    },
    {
      "epoch": 0.03686923654525973,
      "grad_norm": 38.25,
      "learning_rate": 1.9992548643687526e-05,
      "loss": 7.5022,
      "step": 13110
    },
    {
      "epoch": 0.03689735953270844,
      "grad_norm": 41.0,
      "learning_rate": 1.9992537272459962e-05,
      "loss": 8.3703,
      "step": 13120
    },
    {
      "epoch": 0.03692548252015715,
      "grad_norm": 36.0,
      "learning_rate": 1.9992525892565658e-05,
      "loss": 7.651,
      "step": 13130
    },
    {
      "epoch": 0.03695360550760586,
      "grad_norm": 43.75,
      "learning_rate": 1.9992514504004618e-05,
      "loss": 7.1515,
      "step": 13140
    },
    {
      "epoch": 0.03698172849505457,
      "grad_norm": 43.5,
      "learning_rate": 1.999250310677685e-05,
      "loss": 7.8423,
      "step": 13150
    },
    {
      "epoch": 0.037009851482503284,
      "grad_norm": 24.0,
      "learning_rate": 1.9992491700882377e-05,
      "loss": 8.0301,
      "step": 13160
    },
    {
      "epoch": 0.037037974469951995,
      "grad_norm": 48.25,
      "learning_rate": 1.9992480286321196e-05,
      "loss": 8.5987,
      "step": 13170
    },
    {
      "epoch": 0.03706609745740071,
      "grad_norm": 49.25,
      "learning_rate": 1.9992468863093323e-05,
      "loss": 9.1781,
      "step": 13180
    },
    {
      "epoch": 0.03709422044484942,
      "grad_norm": 20.875,
      "learning_rate": 1.9992457431198765e-05,
      "loss": 7.9495,
      "step": 13190
    },
    {
      "epoch": 0.03712234343229812,
      "grad_norm": 42.0,
      "learning_rate": 1.9992445990637536e-05,
      "loss": 8.3674,
      "step": 13200
    },
    {
      "epoch": 0.037150466419746835,
      "grad_norm": 31.0,
      "learning_rate": 1.999243454140964e-05,
      "loss": 7.5379,
      "step": 13210
    },
    {
      "epoch": 0.037178589407195546,
      "grad_norm": 22.375,
      "learning_rate": 1.9992423083515092e-05,
      "loss": 7.5037,
      "step": 13220
    },
    {
      "epoch": 0.03720671239464426,
      "grad_norm": 39.75,
      "learning_rate": 1.9992411616953898e-05,
      "loss": 7.6243,
      "step": 13230
    },
    {
      "epoch": 0.03723483538209297,
      "grad_norm": 35.5,
      "learning_rate": 1.9992400141726073e-05,
      "loss": 7.7285,
      "step": 13240
    },
    {
      "epoch": 0.03726295836954168,
      "grad_norm": 29.875,
      "learning_rate": 1.9992388657831624e-05,
      "loss": 8.0422,
      "step": 13250
    },
    {
      "epoch": 0.03729108135699039,
      "grad_norm": 33.75,
      "learning_rate": 1.9992377165270557e-05,
      "loss": 7.9258,
      "step": 13260
    },
    {
      "epoch": 0.037319204344439104,
      "grad_norm": 25.5,
      "learning_rate": 1.9992365664042886e-05,
      "loss": 7.5074,
      "step": 13270
    },
    {
      "epoch": 0.03734732733188781,
      "grad_norm": 28.25,
      "learning_rate": 1.9992354154148625e-05,
      "loss": 7.5044,
      "step": 13280
    },
    {
      "epoch": 0.03737545031933652,
      "grad_norm": 33.25,
      "learning_rate": 1.9992342635587777e-05,
      "loss": 8.3605,
      "step": 13290
    },
    {
      "epoch": 0.03740357330678523,
      "grad_norm": 48.25,
      "learning_rate": 1.9992331108360356e-05,
      "loss": 8.1142,
      "step": 13300
    },
    {
      "epoch": 0.03743169629423394,
      "grad_norm": 31.125,
      "learning_rate": 1.999231957246637e-05,
      "loss": 7.85,
      "step": 13310
    },
    {
      "epoch": 0.037459819281682655,
      "grad_norm": 45.75,
      "learning_rate": 1.9992308027905833e-05,
      "loss": 8.0692,
      "step": 13320
    },
    {
      "epoch": 0.037487942269131366,
      "grad_norm": 42.0,
      "learning_rate": 1.999229647467875e-05,
      "loss": 7.7516,
      "step": 13330
    },
    {
      "epoch": 0.03751606525658008,
      "grad_norm": 23.875,
      "learning_rate": 1.9992284912785128e-05,
      "loss": 7.626,
      "step": 13340
    },
    {
      "epoch": 0.03754418824402879,
      "grad_norm": 23.375,
      "learning_rate": 1.9992273342224985e-05,
      "loss": 8.0699,
      "step": 13350
    },
    {
      "epoch": 0.037572311231477494,
      "grad_norm": 33.0,
      "learning_rate": 1.999226176299833e-05,
      "loss": 8.6411,
      "step": 13360
    },
    {
      "epoch": 0.037600434218926206,
      "grad_norm": 22.625,
      "learning_rate": 1.999225017510517e-05,
      "loss": 7.7081,
      "step": 13370
    },
    {
      "epoch": 0.03762855720637492,
      "grad_norm": 23.5,
      "learning_rate": 1.9992238578545514e-05,
      "loss": 7.9727,
      "step": 13380
    },
    {
      "epoch": 0.03765668019382363,
      "grad_norm": 32.75,
      "learning_rate": 1.999222697331938e-05,
      "loss": 8.0928,
      "step": 13390
    },
    {
      "epoch": 0.03768480318127234,
      "grad_norm": 43.25,
      "learning_rate": 1.9992215359426765e-05,
      "loss": 8.4706,
      "step": 13400
    },
    {
      "epoch": 0.03771292616872105,
      "grad_norm": 26.625,
      "learning_rate": 1.999220373686769e-05,
      "loss": 8.1479,
      "step": 13410
    },
    {
      "epoch": 0.03774104915616976,
      "grad_norm": 33.75,
      "learning_rate": 1.999219210564216e-05,
      "loss": 9.0185,
      "step": 13420
    },
    {
      "epoch": 0.037769172143618475,
      "grad_norm": 30.75,
      "learning_rate": 1.9992180465750187e-05,
      "loss": 8.1988,
      "step": 13430
    },
    {
      "epoch": 0.037797295131067186,
      "grad_norm": 36.0,
      "learning_rate": 1.999216881719178e-05,
      "loss": 7.8366,
      "step": 13440
    },
    {
      "epoch": 0.03782541811851589,
      "grad_norm": 31.25,
      "learning_rate": 1.9992157159966946e-05,
      "loss": 7.7922,
      "step": 13450
    },
    {
      "epoch": 0.0378535411059646,
      "grad_norm": 29.375,
      "learning_rate": 1.99921454940757e-05,
      "loss": 7.4517,
      "step": 13460
    },
    {
      "epoch": 0.037881664093413314,
      "grad_norm": 25.0,
      "learning_rate": 1.9992133819518056e-05,
      "loss": 8.6028,
      "step": 13470
    },
    {
      "epoch": 0.037909787080862026,
      "grad_norm": 27.125,
      "learning_rate": 1.9992122136294016e-05,
      "loss": 7.9583,
      "step": 13480
    },
    {
      "epoch": 0.03793791006831074,
      "grad_norm": 42.25,
      "learning_rate": 1.9992110444403593e-05,
      "loss": 7.9869,
      "step": 13490
    },
    {
      "epoch": 0.03796603305575945,
      "grad_norm": 44.25,
      "learning_rate": 1.9992098743846796e-05,
      "loss": 8.5041,
      "step": 13500
    },
    {
      "epoch": 0.03799415604320816,
      "grad_norm": 37.25,
      "learning_rate": 1.999208703462364e-05,
      "loss": 7.2959,
      "step": 13510
    },
    {
      "epoch": 0.03802227903065687,
      "grad_norm": 45.5,
      "learning_rate": 1.9992075316734126e-05,
      "loss": 7.6549,
      "step": 13520
    },
    {
      "epoch": 0.03805040201810558,
      "grad_norm": 20.0,
      "learning_rate": 1.9992063590178273e-05,
      "loss": 8.3891,
      "step": 13530
    },
    {
      "epoch": 0.03807852500555429,
      "grad_norm": 65.0,
      "learning_rate": 1.999205185495609e-05,
      "loss": 8.8544,
      "step": 13540
    },
    {
      "epoch": 0.038106647993003,
      "grad_norm": 40.5,
      "learning_rate": 1.9992040111067582e-05,
      "loss": 8.2655,
      "step": 13550
    },
    {
      "epoch": 0.03813477098045171,
      "grad_norm": 27.5,
      "learning_rate": 1.9992028358512764e-05,
      "loss": 8.0888,
      "step": 13560
    },
    {
      "epoch": 0.03816289396790042,
      "grad_norm": 63.25,
      "learning_rate": 1.999201659729164e-05,
      "loss": 8.1617,
      "step": 13570
    },
    {
      "epoch": 0.038191016955349134,
      "grad_norm": 47.25,
      "learning_rate": 1.999200482740423e-05,
      "loss": 7.8589,
      "step": 13580
    },
    {
      "epoch": 0.038219139942797846,
      "grad_norm": 23.875,
      "learning_rate": 1.9991993048850537e-05,
      "loss": 8.0865,
      "step": 13590
    },
    {
      "epoch": 0.03824726293024656,
      "grad_norm": 30.125,
      "learning_rate": 1.999198126163057e-05,
      "loss": 7.8325,
      "step": 13600
    },
    {
      "epoch": 0.03827538591769526,
      "grad_norm": 28.0,
      "learning_rate": 1.9991969465744345e-05,
      "loss": 7.9923,
      "step": 13610
    },
    {
      "epoch": 0.038303508905143974,
      "grad_norm": 23.25,
      "learning_rate": 1.999195766119187e-05,
      "loss": 7.0202,
      "step": 13620
    },
    {
      "epoch": 0.038331631892592685,
      "grad_norm": 31.125,
      "learning_rate": 1.9991945847973154e-05,
      "loss": 7.9617,
      "step": 13630
    },
    {
      "epoch": 0.0383597548800414,
      "grad_norm": 34.5,
      "learning_rate": 1.9991934026088205e-05,
      "loss": 7.7364,
      "step": 13640
    },
    {
      "epoch": 0.03838787786749011,
      "grad_norm": 31.125,
      "learning_rate": 1.9991922195537042e-05,
      "loss": 8.0584,
      "step": 13650
    },
    {
      "epoch": 0.03841600085493882,
      "grad_norm": 30.5,
      "learning_rate": 1.9991910356319668e-05,
      "loss": 7.6093,
      "step": 13660
    },
    {
      "epoch": 0.03844412384238753,
      "grad_norm": 22.75,
      "learning_rate": 1.999189850843609e-05,
      "loss": 8.0979,
      "step": 13670
    },
    {
      "epoch": 0.03847224682983624,
      "grad_norm": 26.25,
      "learning_rate": 1.9991886651886327e-05,
      "loss": 7.878,
      "step": 13680
    },
    {
      "epoch": 0.03850036981728495,
      "grad_norm": 102.0,
      "learning_rate": 1.999187478667038e-05,
      "loss": 7.6886,
      "step": 13690
    },
    {
      "epoch": 0.03852849280473366,
      "grad_norm": 43.25,
      "learning_rate": 1.9991862912788274e-05,
      "loss": 7.7273,
      "step": 13700
    },
    {
      "epoch": 0.03855661579218237,
      "grad_norm": 23.25,
      "learning_rate": 1.9991851030240005e-05,
      "loss": 7.7588,
      "step": 13710
    },
    {
      "epoch": 0.03858473877963108,
      "grad_norm": 31.625,
      "learning_rate": 1.9991839139025585e-05,
      "loss": 8.3302,
      "step": 13720
    },
    {
      "epoch": 0.038612861767079794,
      "grad_norm": 40.0,
      "learning_rate": 1.9991827239145032e-05,
      "loss": 7.8507,
      "step": 13730
    },
    {
      "epoch": 0.038640984754528505,
      "grad_norm": 22.75,
      "learning_rate": 1.999181533059835e-05,
      "loss": 7.9962,
      "step": 13740
    },
    {
      "epoch": 0.03866910774197722,
      "grad_norm": 36.5,
      "learning_rate": 1.9991803413385552e-05,
      "loss": 7.6173,
      "step": 13750
    },
    {
      "epoch": 0.03869723072942593,
      "grad_norm": 32.0,
      "learning_rate": 1.9991791487506648e-05,
      "loss": 7.7178,
      "step": 13760
    },
    {
      "epoch": 0.03872535371687463,
      "grad_norm": 31.875,
      "learning_rate": 1.9991779552961644e-05,
      "loss": 8.2425,
      "step": 13770
    },
    {
      "epoch": 0.038753476704323345,
      "grad_norm": 23.375,
      "learning_rate": 1.999176760975056e-05,
      "loss": 7.9868,
      "step": 13780
    },
    {
      "epoch": 0.038781599691772056,
      "grad_norm": 24.5,
      "learning_rate": 1.9991755657873394e-05,
      "loss": 8.6608,
      "step": 13790
    },
    {
      "epoch": 0.03880972267922077,
      "grad_norm": 39.5,
      "learning_rate": 1.999174369733017e-05,
      "loss": 8.0978,
      "step": 13800
    },
    {
      "epoch": 0.03883784566666948,
      "grad_norm": 34.25,
      "learning_rate": 1.9991731728120887e-05,
      "loss": 7.7132,
      "step": 13810
    },
    {
      "epoch": 0.03886596865411819,
      "grad_norm": 38.0,
      "learning_rate": 1.999171975024556e-05,
      "loss": 8.1656,
      "step": 13820
    },
    {
      "epoch": 0.0388940916415669,
      "grad_norm": 33.0,
      "learning_rate": 1.99917077637042e-05,
      "loss": 7.8016,
      "step": 13830
    },
    {
      "epoch": 0.038922214629015614,
      "grad_norm": 30.125,
      "learning_rate": 1.9991695768496815e-05,
      "loss": 8.6725,
      "step": 13840
    },
    {
      "epoch": 0.03895033761646432,
      "grad_norm": 24.25,
      "learning_rate": 1.9991683764623418e-05,
      "loss": 7.4398,
      "step": 13850
    },
    {
      "epoch": 0.03897846060391303,
      "grad_norm": 38.75,
      "learning_rate": 1.9991671752084018e-05,
      "loss": 7.9079,
      "step": 13860
    },
    {
      "epoch": 0.03900658359136174,
      "grad_norm": 23.25,
      "learning_rate": 1.9991659730878626e-05,
      "loss": 7.7781,
      "step": 13870
    },
    {
      "epoch": 0.03903470657881045,
      "grad_norm": 32.75,
      "learning_rate": 1.999164770100725e-05,
      "loss": 7.5926,
      "step": 13880
    },
    {
      "epoch": 0.039062829566259165,
      "grad_norm": 40.0,
      "learning_rate": 1.9991635662469905e-05,
      "loss": 7.6708,
      "step": 13890
    },
    {
      "epoch": 0.039090952553707876,
      "grad_norm": 33.25,
      "learning_rate": 1.99916236152666e-05,
      "loss": 8.1736,
      "step": 13900
    },
    {
      "epoch": 0.03911907554115659,
      "grad_norm": 31.5,
      "learning_rate": 1.999161155939734e-05,
      "loss": 8.3694,
      "step": 13910
    },
    {
      "epoch": 0.0391471985286053,
      "grad_norm": 45.25,
      "learning_rate": 1.9991599494862144e-05,
      "loss": 8.5924,
      "step": 13920
    },
    {
      "epoch": 0.039175321516054004,
      "grad_norm": 32.75,
      "learning_rate": 1.9991587421661015e-05,
      "loss": 7.9896,
      "step": 13930
    },
    {
      "epoch": 0.039203444503502716,
      "grad_norm": 24.875,
      "learning_rate": 1.999157533979397e-05,
      "loss": 8.1262,
      "step": 13940
    },
    {
      "epoch": 0.03923156749095143,
      "grad_norm": 27.875,
      "learning_rate": 1.9991563249261016e-05,
      "loss": 8.1849,
      "step": 13950
    },
    {
      "epoch": 0.03925969047840014,
      "grad_norm": 31.5,
      "learning_rate": 1.9991551150062162e-05,
      "loss": 8.659,
      "step": 13960
    },
    {
      "epoch": 0.03928781346584885,
      "grad_norm": 23.25,
      "learning_rate": 1.999153904219742e-05,
      "loss": 8.0891,
      "step": 13970
    },
    {
      "epoch": 0.03931593645329756,
      "grad_norm": 28.25,
      "learning_rate": 1.99915269256668e-05,
      "loss": 8.1969,
      "step": 13980
    },
    {
      "epoch": 0.039344059440746273,
      "grad_norm": 22.875,
      "learning_rate": 1.9991514800470318e-05,
      "loss": 8.0725,
      "step": 13990
    },
    {
      "epoch": 0.039372182428194985,
      "grad_norm": 26.875,
      "learning_rate": 1.9991502666607976e-05,
      "loss": 7.5807,
      "step": 14000
    },
    {
      "epoch": 0.03940030541564369,
      "grad_norm": 27.25,
      "learning_rate": 1.9991490524079792e-05,
      "loss": 8.0843,
      "step": 14010
    },
    {
      "epoch": 0.0394284284030924,
      "grad_norm": 32.5,
      "learning_rate": 1.9991478372885767e-05,
      "loss": 7.7271,
      "step": 14020
    },
    {
      "epoch": 0.03945655139054111,
      "grad_norm": 34.75,
      "learning_rate": 1.999146621302592e-05,
      "loss": 7.983,
      "step": 14030
    },
    {
      "epoch": 0.039484674377989824,
      "grad_norm": 31.125,
      "learning_rate": 1.999145404450026e-05,
      "loss": 7.5228,
      "step": 14040
    },
    {
      "epoch": 0.039512797365438536,
      "grad_norm": 31.375,
      "learning_rate": 1.9991441867308797e-05,
      "loss": 8.0152,
      "step": 14050
    },
    {
      "epoch": 0.03954092035288725,
      "grad_norm": 26.5,
      "learning_rate": 1.999142968145154e-05,
      "loss": 7.6515,
      "step": 14060
    },
    {
      "epoch": 0.03956904334033596,
      "grad_norm": 48.0,
      "learning_rate": 1.99914174869285e-05,
      "loss": 8.1369,
      "step": 14070
    },
    {
      "epoch": 0.03959716632778467,
      "grad_norm": 22.875,
      "learning_rate": 1.999140528373969e-05,
      "loss": 7.8141,
      "step": 14080
    },
    {
      "epoch": 0.039625289315233375,
      "grad_norm": 38.25,
      "learning_rate": 1.9991393071885115e-05,
      "loss": 8.06,
      "step": 14090
    },
    {
      "epoch": 0.03965341230268209,
      "grad_norm": 26.875,
      "learning_rate": 1.999138085136479e-05,
      "loss": 7.9396,
      "step": 14100
    },
    {
      "epoch": 0.0396815352901308,
      "grad_norm": 27.625,
      "learning_rate": 1.999136862217873e-05,
      "loss": 8.3304,
      "step": 14110
    },
    {
      "epoch": 0.03970965827757951,
      "grad_norm": 24.5,
      "learning_rate": 1.999135638432694e-05,
      "loss": 8.7337,
      "step": 14120
    },
    {
      "epoch": 0.03973778126502822,
      "grad_norm": 50.75,
      "learning_rate": 1.9991344137809428e-05,
      "loss": 7.23,
      "step": 14130
    },
    {
      "epoch": 0.03976590425247693,
      "grad_norm": 27.25,
      "learning_rate": 1.9991331882626208e-05,
      "loss": 7.1619,
      "step": 14140
    },
    {
      "epoch": 0.039794027239925645,
      "grad_norm": 30.0,
      "learning_rate": 1.999131961877729e-05,
      "loss": 7.7033,
      "step": 14150
    },
    {
      "epoch": 0.039822150227374356,
      "grad_norm": 32.5,
      "learning_rate": 1.9991307346262687e-05,
      "loss": 7.6366,
      "step": 14160
    },
    {
      "epoch": 0.03985027321482307,
      "grad_norm": 27.875,
      "learning_rate": 1.999129506508241e-05,
      "loss": 8.2651,
      "step": 14170
    },
    {
      "epoch": 0.03987839620227177,
      "grad_norm": 24.0,
      "learning_rate": 1.999128277523646e-05,
      "loss": 8.1208,
      "step": 14180
    },
    {
      "epoch": 0.039906519189720484,
      "grad_norm": 24.5,
      "learning_rate": 1.999127047672486e-05,
      "loss": 7.7212,
      "step": 14190
    },
    {
      "epoch": 0.039934642177169195,
      "grad_norm": 34.25,
      "learning_rate": 1.9991258169547615e-05,
      "loss": 8.4624,
      "step": 14200
    },
    {
      "epoch": 0.03996276516461791,
      "grad_norm": 28.75,
      "learning_rate": 1.9991245853704737e-05,
      "loss": 8.5372,
      "step": 14210
    },
    {
      "epoch": 0.03999088815206662,
      "grad_norm": 24.125,
      "learning_rate": 1.9991233529196234e-05,
      "loss": 7.9803,
      "step": 14220
    },
    {
      "epoch": 0.04001901113951533,
      "grad_norm": 29.125,
      "learning_rate": 1.999122119602212e-05,
      "loss": 7.6576,
      "step": 14230
    },
    {
      "epoch": 0.04004713412696404,
      "grad_norm": 27.375,
      "learning_rate": 1.9991208854182408e-05,
      "loss": 7.9341,
      "step": 14240
    },
    {
      "epoch": 0.04007525711441275,
      "grad_norm": 23.25,
      "learning_rate": 1.99911965036771e-05,
      "loss": 8.0133,
      "step": 14250
    },
    {
      "epoch": 0.04010338010186146,
      "grad_norm": 32.25,
      "learning_rate": 1.999118414450621e-05,
      "loss": 8.3118,
      "step": 14260
    },
    {
      "epoch": 0.04013150308931017,
      "grad_norm": 22.625,
      "learning_rate": 1.9991171776669758e-05,
      "loss": 7.5477,
      "step": 14270
    },
    {
      "epoch": 0.04015962607675888,
      "grad_norm": 31.625,
      "learning_rate": 1.9991159400167743e-05,
      "loss": 8.2519,
      "step": 14280
    },
    {
      "epoch": 0.04018774906420759,
      "grad_norm": 28.5,
      "learning_rate": 1.9991147015000182e-05,
      "loss": 8.3795,
      "step": 14290
    },
    {
      "epoch": 0.040215872051656304,
      "grad_norm": 28.375,
      "learning_rate": 1.999113462116708e-05,
      "loss": 7.9344,
      "step": 14300
    },
    {
      "epoch": 0.040243995039105016,
      "grad_norm": 24.0,
      "learning_rate": 1.9991122218668456e-05,
      "loss": 7.8801,
      "step": 14310
    },
    {
      "epoch": 0.04027211802655373,
      "grad_norm": 27.375,
      "learning_rate": 1.9991109807504312e-05,
      "loss": 7.9509,
      "step": 14320
    },
    {
      "epoch": 0.04030024101400244,
      "grad_norm": 30.875,
      "learning_rate": 1.999109738767467e-05,
      "loss": 8.0977,
      "step": 14330
    },
    {
      "epoch": 0.04032836400145114,
      "grad_norm": 37.25,
      "learning_rate": 1.9991084959179525e-05,
      "loss": 8.1091,
      "step": 14340
    },
    {
      "epoch": 0.040356486988899855,
      "grad_norm": 34.0,
      "learning_rate": 1.9991072522018903e-05,
      "loss": 7.8797,
      "step": 14350
    },
    {
      "epoch": 0.040384609976348566,
      "grad_norm": 31.5,
      "learning_rate": 1.9991060076192802e-05,
      "loss": 8.8112,
      "step": 14360
    },
    {
      "epoch": 0.04041273296379728,
      "grad_norm": 35.5,
      "learning_rate": 1.9991047621701245e-05,
      "loss": 7.7425,
      "step": 14370
    },
    {
      "epoch": 0.04044085595124599,
      "grad_norm": 30.0,
      "learning_rate": 1.9991035158544237e-05,
      "loss": 7.1846,
      "step": 14380
    },
    {
      "epoch": 0.0404689789386947,
      "grad_norm": 63.5,
      "learning_rate": 1.9991022686721786e-05,
      "loss": 9.0637,
      "step": 14390
    },
    {
      "epoch": 0.04049710192614341,
      "grad_norm": 33.25,
      "learning_rate": 1.9991010206233908e-05,
      "loss": 7.6429,
      "step": 14400
    },
    {
      "epoch": 0.040525224913592124,
      "grad_norm": 34.25,
      "learning_rate": 1.9990997717080608e-05,
      "loss": 7.5113,
      "step": 14410
    },
    {
      "epoch": 0.04055334790104083,
      "grad_norm": 27.125,
      "learning_rate": 1.9990985219261906e-05,
      "loss": 8.1203,
      "step": 14420
    },
    {
      "epoch": 0.04058147088848954,
      "grad_norm": 64.5,
      "learning_rate": 1.9990972712777804e-05,
      "loss": 8.5142,
      "step": 14430
    },
    {
      "epoch": 0.04060959387593825,
      "grad_norm": 41.25,
      "learning_rate": 1.9990960197628314e-05,
      "loss": 8.356,
      "step": 14440
    },
    {
      "epoch": 0.04063771686338696,
      "grad_norm": 26.75,
      "learning_rate": 1.9990947673813446e-05,
      "loss": 7.1781,
      "step": 14450
    },
    {
      "epoch": 0.040665839850835675,
      "grad_norm": 32.5,
      "learning_rate": 1.9990935141333222e-05,
      "loss": 7.6836,
      "step": 14460
    },
    {
      "epoch": 0.04069396283828439,
      "grad_norm": 29.625,
      "learning_rate": 1.999092260018764e-05,
      "loss": 7.677,
      "step": 14470
    },
    {
      "epoch": 0.0407220858257331,
      "grad_norm": 22.25,
      "learning_rate": 1.9990910050376718e-05,
      "loss": 7.6785,
      "step": 14480
    },
    {
      "epoch": 0.04075020881318181,
      "grad_norm": 24.625,
      "learning_rate": 1.999089749190046e-05,
      "loss": 7.7934,
      "step": 14490
    },
    {
      "epoch": 0.040778331800630514,
      "grad_norm": 23.75,
      "learning_rate": 1.9990884924758884e-05,
      "loss": 8.4624,
      "step": 14500
    },
    {
      "epoch": 0.040806454788079226,
      "grad_norm": 33.5,
      "learning_rate": 1.9990872348951996e-05,
      "loss": 8.6198,
      "step": 14510
    },
    {
      "epoch": 0.04083457777552794,
      "grad_norm": 35.25,
      "learning_rate": 1.999085976447981e-05,
      "loss": 8.1483,
      "step": 14520
    },
    {
      "epoch": 0.04086270076297665,
      "grad_norm": 35.25,
      "learning_rate": 1.9990847171342338e-05,
      "loss": 7.164,
      "step": 14530
    },
    {
      "epoch": 0.04089082375042536,
      "grad_norm": 28.125,
      "learning_rate": 1.999083456953959e-05,
      "loss": 7.7325,
      "step": 14540
    },
    {
      "epoch": 0.04091894673787407,
      "grad_norm": 19.875,
      "learning_rate": 1.9990821959071573e-05,
      "loss": 8.1406,
      "step": 14550
    },
    {
      "epoch": 0.040947069725322784,
      "grad_norm": 46.75,
      "learning_rate": 1.99908093399383e-05,
      "loss": 7.9699,
      "step": 14560
    },
    {
      "epoch": 0.040975192712771495,
      "grad_norm": 44.5,
      "learning_rate": 1.9990796712139783e-05,
      "loss": 7.626,
      "step": 14570
    },
    {
      "epoch": 0.0410033157002202,
      "grad_norm": 56.25,
      "learning_rate": 1.9990784075676032e-05,
      "loss": 8.7337,
      "step": 14580
    },
    {
      "epoch": 0.04103143868766891,
      "grad_norm": 44.75,
      "learning_rate": 1.9990771430547063e-05,
      "loss": 8.0525,
      "step": 14590
    },
    {
      "epoch": 0.04105956167511762,
      "grad_norm": 24.125,
      "learning_rate": 1.999075877675288e-05,
      "loss": 7.5899,
      "step": 14600
    },
    {
      "epoch": 0.041087684662566334,
      "grad_norm": 37.75,
      "learning_rate": 1.9990746114293494e-05,
      "loss": 7.6175,
      "step": 14610
    },
    {
      "epoch": 0.041115807650015046,
      "grad_norm": 25.125,
      "learning_rate": 1.999073344316892e-05,
      "loss": 8.2447,
      "step": 14620
    },
    {
      "epoch": 0.04114393063746376,
      "grad_norm": 33.25,
      "learning_rate": 1.999072076337917e-05,
      "loss": 7.923,
      "step": 14630
    },
    {
      "epoch": 0.04117205362491247,
      "grad_norm": 29.125,
      "learning_rate": 1.999070807492425e-05,
      "loss": 8.3361,
      "step": 14640
    },
    {
      "epoch": 0.04120017661236118,
      "grad_norm": 37.25,
      "learning_rate": 1.9990695377804173e-05,
      "loss": 7.8386,
      "step": 14650
    },
    {
      "epoch": 0.041228299599809885,
      "grad_norm": 38.75,
      "learning_rate": 1.999068267201895e-05,
      "loss": 8.1347,
      "step": 14660
    },
    {
      "epoch": 0.0412564225872586,
      "grad_norm": 25.0,
      "learning_rate": 1.9990669957568592e-05,
      "loss": 7.2211,
      "step": 14670
    },
    {
      "epoch": 0.04128454557470731,
      "grad_norm": 22.875,
      "learning_rate": 1.9990657234453116e-05,
      "loss": 7.1225,
      "step": 14680
    },
    {
      "epoch": 0.04131266856215602,
      "grad_norm": 29.125,
      "learning_rate": 1.9990644502672524e-05,
      "loss": 8.1533,
      "step": 14690
    },
    {
      "epoch": 0.04134079154960473,
      "grad_norm": 45.0,
      "learning_rate": 1.999063176222683e-05,
      "loss": 8.1926,
      "step": 14700
    },
    {
      "epoch": 0.04136891453705344,
      "grad_norm": 45.5,
      "learning_rate": 1.9990619013116048e-05,
      "loss": 7.6541,
      "step": 14710
    },
    {
      "epoch": 0.041397037524502155,
      "grad_norm": 45.75,
      "learning_rate": 1.9990606255340184e-05,
      "loss": 7.9876,
      "step": 14720
    },
    {
      "epoch": 0.041425160511950866,
      "grad_norm": 23.5,
      "learning_rate": 1.9990593488899254e-05,
      "loss": 8.3273,
      "step": 14730
    },
    {
      "epoch": 0.04145328349939957,
      "grad_norm": 33.75,
      "learning_rate": 1.9990580713793265e-05,
      "loss": 8.1444,
      "step": 14740
    },
    {
      "epoch": 0.04148140648684828,
      "grad_norm": 40.75,
      "learning_rate": 1.999056793002223e-05,
      "loss": 7.4559,
      "step": 14750
    },
    {
      "epoch": 0.041509529474296994,
      "grad_norm": 30.5,
      "learning_rate": 1.9990555137586162e-05,
      "loss": 8.2001,
      "step": 14760
    },
    {
      "epoch": 0.041537652461745705,
      "grad_norm": 26.5,
      "learning_rate": 1.9990542336485068e-05,
      "loss": 8.6624,
      "step": 14770
    },
    {
      "epoch": 0.04156577544919442,
      "grad_norm": 27.5,
      "learning_rate": 1.999052952671896e-05,
      "loss": 7.7957,
      "step": 14780
    },
    {
      "epoch": 0.04159389843664313,
      "grad_norm": 37.0,
      "learning_rate": 1.9990516708287854e-05,
      "loss": 8.1917,
      "step": 14790
    },
    {
      "epoch": 0.04162202142409184,
      "grad_norm": 44.25,
      "learning_rate": 1.9990503881191756e-05,
      "loss": 7.3718,
      "step": 14800
    },
    {
      "epoch": 0.04165014441154055,
      "grad_norm": 32.5,
      "learning_rate": 1.999049104543068e-05,
      "loss": 7.5645,
      "step": 14810
    },
    {
      "epoch": 0.04167826739898926,
      "grad_norm": 39.0,
      "learning_rate": 1.9990478201004633e-05,
      "loss": 8.3762,
      "step": 14820
    },
    {
      "epoch": 0.04170639038643797,
      "grad_norm": 34.75,
      "learning_rate": 1.9990465347913632e-05,
      "loss": 8.6465,
      "step": 14830
    },
    {
      "epoch": 0.04173451337388668,
      "grad_norm": 26.25,
      "learning_rate": 1.9990452486157684e-05,
      "loss": 7.7829,
      "step": 14840
    },
    {
      "epoch": 0.04176263636133539,
      "grad_norm": 34.5,
      "learning_rate": 1.99904396157368e-05,
      "loss": 7.9722,
      "step": 14850
    },
    {
      "epoch": 0.0417907593487841,
      "grad_norm": 42.25,
      "learning_rate": 1.999042673665099e-05,
      "loss": 7.7614,
      "step": 14860
    },
    {
      "epoch": 0.041818882336232814,
      "grad_norm": 47.0,
      "learning_rate": 1.9990413848900272e-05,
      "loss": 7.9466,
      "step": 14870
    },
    {
      "epoch": 0.041847005323681526,
      "grad_norm": 30.625,
      "learning_rate": 1.999040095248465e-05,
      "loss": 8.0575,
      "step": 14880
    },
    {
      "epoch": 0.04187512831113024,
      "grad_norm": 29.25,
      "learning_rate": 1.999038804740414e-05,
      "loss": 8.1235,
      "step": 14890
    },
    {
      "epoch": 0.04190325129857895,
      "grad_norm": 40.0,
      "learning_rate": 1.999037513365875e-05,
      "loss": 8.1925,
      "step": 14900
    },
    {
      "epoch": 0.04193137428602765,
      "grad_norm": 26.375,
      "learning_rate": 1.999036221124849e-05,
      "loss": 8.5718,
      "step": 14910
    },
    {
      "epoch": 0.041959497273476365,
      "grad_norm": 27.0,
      "learning_rate": 1.9990349280173377e-05,
      "loss": 8.6519,
      "step": 14920
    },
    {
      "epoch": 0.041987620260925076,
      "grad_norm": 23.25,
      "learning_rate": 1.9990336340433417e-05,
      "loss": 7.4102,
      "step": 14930
    },
    {
      "epoch": 0.04201574324837379,
      "grad_norm": 23.25,
      "learning_rate": 1.999032339202862e-05,
      "loss": 7.9522,
      "step": 14940
    },
    {
      "epoch": 0.0420438662358225,
      "grad_norm": 21.625,
      "learning_rate": 1.9990310434959006e-05,
      "loss": 8.0161,
      "step": 14950
    },
    {
      "epoch": 0.04207198922327121,
      "grad_norm": 28.5,
      "learning_rate": 1.9990297469224576e-05,
      "loss": 8.1599,
      "step": 14960
    },
    {
      "epoch": 0.04210011221071992,
      "grad_norm": 52.0,
      "learning_rate": 1.999028449482535e-05,
      "loss": 8.6145,
      "step": 14970
    },
    {
      "epoch": 0.042128235198168634,
      "grad_norm": 41.75,
      "learning_rate": 1.999027151176133e-05,
      "loss": 7.9823,
      "step": 14980
    },
    {
      "epoch": 0.04215635818561734,
      "grad_norm": 29.0,
      "learning_rate": 1.9990258520032536e-05,
      "loss": 7.5883,
      "step": 14990
    },
    {
      "epoch": 0.04218448117306605,
      "grad_norm": 32.25,
      "learning_rate": 1.9990245519638972e-05,
      "loss": 8.4586,
      "step": 15000
    },
    {
      "epoch": 0.04221260416051476,
      "grad_norm": 26.25,
      "learning_rate": 1.9990232510580654e-05,
      "loss": 7.1532,
      "step": 15010
    },
    {
      "epoch": 0.042240727147963474,
      "grad_norm": 28.75,
      "learning_rate": 1.9990219492857595e-05,
      "loss": 8.5273,
      "step": 15020
    },
    {
      "epoch": 0.042268850135412185,
      "grad_norm": 22.5,
      "learning_rate": 1.99902064664698e-05,
      "loss": 8.2674,
      "step": 15030
    },
    {
      "epoch": 0.0422969731228609,
      "grad_norm": 35.25,
      "learning_rate": 1.9990193431417283e-05,
      "loss": 8.1158,
      "step": 15040
    },
    {
      "epoch": 0.04232509611030961,
      "grad_norm": 25.25,
      "learning_rate": 1.9990180387700057e-05,
      "loss": 8.3668,
      "step": 15050
    },
    {
      "epoch": 0.04235321909775832,
      "grad_norm": 46.25,
      "learning_rate": 1.9990167335318135e-05,
      "loss": 7.5332,
      "step": 15060
    },
    {
      "epoch": 0.042381342085207024,
      "grad_norm": 74.0,
      "learning_rate": 1.999015427427152e-05,
      "loss": 8.0822,
      "step": 15070
    },
    {
      "epoch": 0.042409465072655736,
      "grad_norm": 38.0,
      "learning_rate": 1.999014120456023e-05,
      "loss": 7.8682,
      "step": 15080
    },
    {
      "epoch": 0.04243758806010445,
      "grad_norm": 27.875,
      "learning_rate": 1.9990128126184277e-05,
      "loss": 8.1402,
      "step": 15090
    },
    {
      "epoch": 0.04246571104755316,
      "grad_norm": 21.0,
      "learning_rate": 1.999011503914367e-05,
      "loss": 8.1576,
      "step": 15100
    },
    {
      "epoch": 0.04249383403500187,
      "grad_norm": 21.625,
      "learning_rate": 1.9990101943438422e-05,
      "loss": 8.322,
      "step": 15110
    },
    {
      "epoch": 0.04252195702245058,
      "grad_norm": 29.125,
      "learning_rate": 1.999008883906854e-05,
      "loss": 7.13,
      "step": 15120
    },
    {
      "epoch": 0.042550080009899294,
      "grad_norm": 52.0,
      "learning_rate": 1.9990075726034042e-05,
      "loss": 7.9855,
      "step": 15130
    },
    {
      "epoch": 0.042578202997348005,
      "grad_norm": 33.25,
      "learning_rate": 1.9990062604334938e-05,
      "loss": 7.8318,
      "step": 15140
    },
    {
      "epoch": 0.04260632598479671,
      "grad_norm": 51.0,
      "learning_rate": 1.9990049473971232e-05,
      "loss": 7.5613,
      "step": 15150
    },
    {
      "epoch": 0.04263444897224542,
      "grad_norm": 30.375,
      "learning_rate": 1.999003633494294e-05,
      "loss": 8.262,
      "step": 15160
    },
    {
      "epoch": 0.04266257195969413,
      "grad_norm": 39.5,
      "learning_rate": 1.9990023187250077e-05,
      "loss": 8.1489,
      "step": 15170
    },
    {
      "epoch": 0.042690694947142845,
      "grad_norm": 29.125,
      "learning_rate": 1.999001003089265e-05,
      "loss": 8.6367,
      "step": 15180
    },
    {
      "epoch": 0.042718817934591556,
      "grad_norm": 31.125,
      "learning_rate": 1.9989996865870674e-05,
      "loss": 8.2041,
      "step": 15190
    },
    {
      "epoch": 0.04274694092204027,
      "grad_norm": 50.25,
      "learning_rate": 1.9989983692184157e-05,
      "loss": 7.9912,
      "step": 15200
    },
    {
      "epoch": 0.04277506390948898,
      "grad_norm": 23.125,
      "learning_rate": 1.9989970509833117e-05,
      "loss": 8.1719,
      "step": 15210
    },
    {
      "epoch": 0.04280318689693769,
      "grad_norm": 26.625,
      "learning_rate": 1.9989957318817552e-05,
      "loss": 7.8112,
      "step": 15220
    },
    {
      "epoch": 0.042831309884386395,
      "grad_norm": 28.0,
      "learning_rate": 1.9989944119137488e-05,
      "loss": 8.0234,
      "step": 15230
    },
    {
      "epoch": 0.04285943287183511,
      "grad_norm": 20.0,
      "learning_rate": 1.9989930910792927e-05,
      "loss": 7.4964,
      "step": 15240
    },
    {
      "epoch": 0.04288755585928382,
      "grad_norm": 28.25,
      "learning_rate": 1.9989917693783884e-05,
      "loss": 7.9812,
      "step": 15250
    },
    {
      "epoch": 0.04291567884673253,
      "grad_norm": 36.25,
      "learning_rate": 1.9989904468110367e-05,
      "loss": 7.743,
      "step": 15260
    },
    {
      "epoch": 0.04294380183418124,
      "grad_norm": 46.25,
      "learning_rate": 1.9989891233772395e-05,
      "loss": 8.2282,
      "step": 15270
    },
    {
      "epoch": 0.04297192482162995,
      "grad_norm": 31.75,
      "learning_rate": 1.9989877990769973e-05,
      "loss": 7.965,
      "step": 15280
    },
    {
      "epoch": 0.043000047809078665,
      "grad_norm": 39.75,
      "learning_rate": 1.9989864739103116e-05,
      "loss": 7.4655,
      "step": 15290
    },
    {
      "epoch": 0.043028170796527376,
      "grad_norm": 36.5,
      "learning_rate": 1.998985147877183e-05,
      "loss": 8.2348,
      "step": 15300
    },
    {
      "epoch": 0.04305629378397608,
      "grad_norm": 41.25,
      "learning_rate": 1.9989838209776136e-05,
      "loss": 8.841,
      "step": 15310
    },
    {
      "epoch": 0.04308441677142479,
      "grad_norm": 28.75,
      "learning_rate": 1.9989824932116037e-05,
      "loss": 8.0753,
      "step": 15320
    },
    {
      "epoch": 0.043112539758873504,
      "grad_norm": 29.5,
      "learning_rate": 1.9989811645791546e-05,
      "loss": 8.1346,
      "step": 15330
    },
    {
      "epoch": 0.043140662746322216,
      "grad_norm": 52.25,
      "learning_rate": 1.9989798350802677e-05,
      "loss": 8.1031,
      "step": 15340
    },
    {
      "epoch": 0.04316878573377093,
      "grad_norm": 36.0,
      "learning_rate": 1.998978504714944e-05,
      "loss": 7.9402,
      "step": 15350
    },
    {
      "epoch": 0.04319690872121964,
      "grad_norm": 34.75,
      "learning_rate": 1.998977173483185e-05,
      "loss": 7.9225,
      "step": 15360
    },
    {
      "epoch": 0.04322503170866835,
      "grad_norm": 39.5,
      "learning_rate": 1.9989758413849915e-05,
      "loss": 7.7493,
      "step": 15370
    },
    {
      "epoch": 0.04325315469611706,
      "grad_norm": 34.75,
      "learning_rate": 1.9989745084203643e-05,
      "loss": 7.652,
      "step": 15380
    },
    {
      "epoch": 0.043281277683565766,
      "grad_norm": 33.25,
      "learning_rate": 1.9989731745893054e-05,
      "loss": 8.4337,
      "step": 15390
    },
    {
      "epoch": 0.04330940067101448,
      "grad_norm": 25.75,
      "learning_rate": 1.9989718398918154e-05,
      "loss": 7.3164,
      "step": 15400
    },
    {
      "epoch": 0.04333752365846319,
      "grad_norm": 24.0,
      "learning_rate": 1.9989705043278955e-05,
      "loss": 7.8767,
      "step": 15410
    },
    {
      "epoch": 0.0433656466459119,
      "grad_norm": 27.25,
      "learning_rate": 1.998969167897547e-05,
      "loss": 7.9032,
      "step": 15420
    },
    {
      "epoch": 0.04339376963336061,
      "grad_norm": 32.25,
      "learning_rate": 1.998967830600771e-05,
      "loss": 8.4123,
      "step": 15430
    },
    {
      "epoch": 0.043421892620809324,
      "grad_norm": 28.625,
      "learning_rate": 1.9989664924375686e-05,
      "loss": 7.3174,
      "step": 15440
    },
    {
      "epoch": 0.043450015608258036,
      "grad_norm": 23.875,
      "learning_rate": 1.9989651534079412e-05,
      "loss": 8.1008,
      "step": 15450
    },
    {
      "epoch": 0.04347813859570675,
      "grad_norm": 34.75,
      "learning_rate": 1.9989638135118895e-05,
      "loss": 7.6594,
      "step": 15460
    },
    {
      "epoch": 0.04350626158315546,
      "grad_norm": 41.75,
      "learning_rate": 1.9989624727494154e-05,
      "loss": 7.7052,
      "step": 15470
    },
    {
      "epoch": 0.043534384570604164,
      "grad_norm": 30.25,
      "learning_rate": 1.998961131120519e-05,
      "loss": 8.4516,
      "step": 15480
    },
    {
      "epoch": 0.043562507558052875,
      "grad_norm": 36.75,
      "learning_rate": 1.9989597886252027e-05,
      "loss": 8.3436,
      "step": 15490
    },
    {
      "epoch": 0.04359063054550159,
      "grad_norm": 24.125,
      "learning_rate": 1.9989584452634666e-05,
      "loss": 7.9859,
      "step": 15500
    },
    {
      "epoch": 0.0436187535329503,
      "grad_norm": 34.5,
      "learning_rate": 1.9989571010353124e-05,
      "loss": 7.7361,
      "step": 15510
    },
    {
      "epoch": 0.04364687652039901,
      "grad_norm": 24.25,
      "learning_rate": 1.9989557559407414e-05,
      "loss": 7.8217,
      "step": 15520
    },
    {
      "epoch": 0.04367499950784772,
      "grad_norm": 39.5,
      "learning_rate": 1.9989544099797543e-05,
      "loss": 8.1218,
      "step": 15530
    },
    {
      "epoch": 0.04370312249529643,
      "grad_norm": 20.375,
      "learning_rate": 1.9989530631523524e-05,
      "loss": 7.802,
      "step": 15540
    },
    {
      "epoch": 0.043731245482745144,
      "grad_norm": 27.375,
      "learning_rate": 1.9989517154585375e-05,
      "loss": 7.899,
      "step": 15550
    },
    {
      "epoch": 0.04375936847019385,
      "grad_norm": 23.125,
      "learning_rate": 1.9989503668983095e-05,
      "loss": 8.2947,
      "step": 15560
    },
    {
      "epoch": 0.04378749145764256,
      "grad_norm": 35.5,
      "learning_rate": 1.998949017471671e-05,
      "loss": 8.2794,
      "step": 15570
    },
    {
      "epoch": 0.04381561444509127,
      "grad_norm": 32.75,
      "learning_rate": 1.998947667178622e-05,
      "loss": 8.3958,
      "step": 15580
    },
    {
      "epoch": 0.043843737432539984,
      "grad_norm": 23.625,
      "learning_rate": 1.9989463160191645e-05,
      "loss": 7.7927,
      "step": 15590
    },
    {
      "epoch": 0.043871860419988695,
      "grad_norm": 37.75,
      "learning_rate": 1.9989449639932988e-05,
      "loss": 8.208,
      "step": 15600
    },
    {
      "epoch": 0.04389998340743741,
      "grad_norm": 48.75,
      "learning_rate": 1.998943611101027e-05,
      "loss": 8.1944,
      "step": 15610
    },
    {
      "epoch": 0.04392810639488612,
      "grad_norm": 35.75,
      "learning_rate": 1.99894225734235e-05,
      "loss": 8.2754,
      "step": 15620
    },
    {
      "epoch": 0.04395622938233483,
      "grad_norm": 22.875,
      "learning_rate": 1.9989409027172685e-05,
      "loss": 8.0983,
      "step": 15630
    },
    {
      "epoch": 0.043984352369783535,
      "grad_norm": 44.0,
      "learning_rate": 1.998939547225784e-05,
      "loss": 8.6797,
      "step": 15640
    },
    {
      "epoch": 0.044012475357232246,
      "grad_norm": 24.375,
      "learning_rate": 1.998938190867898e-05,
      "loss": 8.6991,
      "step": 15650
    },
    {
      "epoch": 0.04404059834468096,
      "grad_norm": 31.5,
      "learning_rate": 1.998936833643611e-05,
      "loss": 8.1683,
      "step": 15660
    },
    {
      "epoch": 0.04406872133212967,
      "grad_norm": 30.625,
      "learning_rate": 1.998935475552925e-05,
      "loss": 7.5811,
      "step": 15670
    },
    {
      "epoch": 0.04409684431957838,
      "grad_norm": 33.25,
      "learning_rate": 1.9989341165958404e-05,
      "loss": 7.0774,
      "step": 15680
    },
    {
      "epoch": 0.04412496730702709,
      "grad_norm": 29.625,
      "learning_rate": 1.998932756772359e-05,
      "loss": 7.6559,
      "step": 15690
    },
    {
      "epoch": 0.044153090294475804,
      "grad_norm": 51.5,
      "learning_rate": 1.9989313960824814e-05,
      "loss": 8.0974,
      "step": 15700
    },
    {
      "epoch": 0.044181213281924515,
      "grad_norm": 26.125,
      "learning_rate": 1.9989300345262093e-05,
      "loss": 7.4047,
      "step": 15710
    },
    {
      "epoch": 0.04420933626937322,
      "grad_norm": 45.75,
      "learning_rate": 1.9989286721035435e-05,
      "loss": 7.7699,
      "step": 15720
    },
    {
      "epoch": 0.04423745925682193,
      "grad_norm": 28.625,
      "learning_rate": 1.9989273088144855e-05,
      "loss": 8.1135,
      "step": 15730
    },
    {
      "epoch": 0.04426558224427064,
      "grad_norm": 32.25,
      "learning_rate": 1.998925944659036e-05,
      "loss": 6.9867,
      "step": 15740
    },
    {
      "epoch": 0.044293705231719355,
      "grad_norm": 27.5,
      "learning_rate": 1.9989245796371966e-05,
      "loss": 7.4603,
      "step": 15750
    },
    {
      "epoch": 0.044321828219168066,
      "grad_norm": 25.125,
      "learning_rate": 1.9989232137489686e-05,
      "loss": 7.7796,
      "step": 15760
    },
    {
      "epoch": 0.04434995120661678,
      "grad_norm": 33.25,
      "learning_rate": 1.9989218469943526e-05,
      "loss": 7.7801,
      "step": 15770
    },
    {
      "epoch": 0.04437807419406549,
      "grad_norm": 41.75,
      "learning_rate": 1.9989204793733503e-05,
      "loss": 8.3392,
      "step": 15780
    },
    {
      "epoch": 0.0444061971815142,
      "grad_norm": 27.625,
      "learning_rate": 1.9989191108859628e-05,
      "loss": 7.6179,
      "step": 15790
    },
    {
      "epoch": 0.044434320168962906,
      "grad_norm": 25.625,
      "learning_rate": 1.998917741532191e-05,
      "loss": 7.6459,
      "step": 15800
    },
    {
      "epoch": 0.04446244315641162,
      "grad_norm": 24.125,
      "learning_rate": 1.9989163713120366e-05,
      "loss": 8.2462,
      "step": 15810
    },
    {
      "epoch": 0.04449056614386033,
      "grad_norm": 25.125,
      "learning_rate": 1.9989150002255003e-05,
      "loss": 7.9194,
      "step": 15820
    },
    {
      "epoch": 0.04451868913130904,
      "grad_norm": 25.75,
      "learning_rate": 1.9989136282725835e-05,
      "loss": 8.3128,
      "step": 15830
    },
    {
      "epoch": 0.04454681211875775,
      "grad_norm": 25.0,
      "learning_rate": 1.9989122554532875e-05,
      "loss": 8.1845,
      "step": 15840
    },
    {
      "epoch": 0.04457493510620646,
      "grad_norm": 26.625,
      "learning_rate": 1.9989108817676134e-05,
      "loss": 7.6273,
      "step": 15850
    },
    {
      "epoch": 0.044603058093655175,
      "grad_norm": 73.5,
      "learning_rate": 1.998909507215562e-05,
      "loss": 8.0485,
      "step": 15860
    },
    {
      "epoch": 0.044631181081103886,
      "grad_norm": 35.0,
      "learning_rate": 1.9989081317971353e-05,
      "loss": 7.9216,
      "step": 15870
    },
    {
      "epoch": 0.04465930406855259,
      "grad_norm": 26.125,
      "learning_rate": 1.9989067555123338e-05,
      "loss": 7.7466,
      "step": 15880
    },
    {
      "epoch": 0.0446874270560013,
      "grad_norm": 23.0,
      "learning_rate": 1.998905378361159e-05,
      "loss": 7.5943,
      "step": 15890
    },
    {
      "epoch": 0.044715550043450014,
      "grad_norm": 36.25,
      "learning_rate": 1.998904000343612e-05,
      "loss": 7.4588,
      "step": 15900
    },
    {
      "epoch": 0.044743673030898726,
      "grad_norm": 45.0,
      "learning_rate": 1.9989026214596938e-05,
      "loss": 8.4078,
      "step": 15910
    },
    {
      "epoch": 0.04477179601834744,
      "grad_norm": 39.75,
      "learning_rate": 1.9989012417094064e-05,
      "loss": 8.678,
      "step": 15920
    },
    {
      "epoch": 0.04479991900579615,
      "grad_norm": 52.5,
      "learning_rate": 1.99889986109275e-05,
      "loss": 8.3793,
      "step": 15930
    },
    {
      "epoch": 0.04482804199324486,
      "grad_norm": 24.0,
      "learning_rate": 1.9988984796097262e-05,
      "loss": 7.7575,
      "step": 15940
    },
    {
      "epoch": 0.04485616498069357,
      "grad_norm": 36.75,
      "learning_rate": 1.998897097260336e-05,
      "loss": 8.2181,
      "step": 15950
    },
    {
      "epoch": 0.04488428796814228,
      "grad_norm": 27.125,
      "learning_rate": 1.9988957140445815e-05,
      "loss": 7.629,
      "step": 15960
    },
    {
      "epoch": 0.04491241095559099,
      "grad_norm": 43.25,
      "learning_rate": 1.998894329962463e-05,
      "loss": 7.9694,
      "step": 15970
    },
    {
      "epoch": 0.0449405339430397,
      "grad_norm": 37.5,
      "learning_rate": 1.9988929450139814e-05,
      "loss": 8.0665,
      "step": 15980
    },
    {
      "epoch": 0.04496865693048841,
      "grad_norm": 31.875,
      "learning_rate": 1.9988915591991386e-05,
      "loss": 8.6759,
      "step": 15990
    },
    {
      "epoch": 0.04499677991793712,
      "grad_norm": 21.625,
      "learning_rate": 1.998890172517936e-05,
      "loss": 7.3393,
      "step": 16000
    },
    {
      "epoch": 0.045024902905385834,
      "grad_norm": 33.5,
      "learning_rate": 1.9988887849703743e-05,
      "loss": 8.1811,
      "step": 16010
    },
    {
      "epoch": 0.045053025892834546,
      "grad_norm": 18.0,
      "learning_rate": 1.9988873965564543e-05,
      "loss": 8.4981,
      "step": 16020
    },
    {
      "epoch": 0.04508114888028326,
      "grad_norm": 26.125,
      "learning_rate": 1.9988860072761782e-05,
      "loss": 7.6231,
      "step": 16030
    },
    {
      "epoch": 0.04510927186773196,
      "grad_norm": 32.75,
      "learning_rate": 1.9988846171295466e-05,
      "loss": 8.4545,
      "step": 16040
    },
    {
      "epoch": 0.045137394855180674,
      "grad_norm": 45.25,
      "learning_rate": 1.998883226116561e-05,
      "loss": 7.7677,
      "step": 16050
    },
    {
      "epoch": 0.045165517842629385,
      "grad_norm": 24.875,
      "learning_rate": 1.9988818342372225e-05,
      "loss": 8.2254,
      "step": 16060
    },
    {
      "epoch": 0.0451936408300781,
      "grad_norm": 25.5,
      "learning_rate": 1.998880441491532e-05,
      "loss": 8.5046,
      "step": 16070
    },
    {
      "epoch": 0.04522176381752681,
      "grad_norm": 24.75,
      "learning_rate": 1.9988790478794908e-05,
      "loss": 7.6883,
      "step": 16080
    },
    {
      "epoch": 0.04524988680497552,
      "grad_norm": 34.25,
      "learning_rate": 1.998877653401101e-05,
      "loss": 7.7976,
      "step": 16090
    },
    {
      "epoch": 0.04527800979242423,
      "grad_norm": 23.5,
      "learning_rate": 1.9988762580563623e-05,
      "loss": 7.7764,
      "step": 16100
    },
    {
      "epoch": 0.04530613277987294,
      "grad_norm": 37.0,
      "learning_rate": 1.998874861845277e-05,
      "loss": 7.5296,
      "step": 16110
    },
    {
      "epoch": 0.04533425576732165,
      "grad_norm": 26.5,
      "learning_rate": 1.9988734647678458e-05,
      "loss": 8.0913,
      "step": 16120
    },
    {
      "epoch": 0.04536237875477036,
      "grad_norm": 23.5,
      "learning_rate": 1.9988720668240703e-05,
      "loss": 7.766,
      "step": 16130
    },
    {
      "epoch": 0.04539050174221907,
      "grad_norm": 25.875,
      "learning_rate": 1.9988706680139515e-05,
      "loss": 7.9757,
      "step": 16140
    },
    {
      "epoch": 0.04541862472966778,
      "grad_norm": 24.375,
      "learning_rate": 1.9988692683374905e-05,
      "loss": 7.9978,
      "step": 16150
    },
    {
      "epoch": 0.045446747717116494,
      "grad_norm": 34.0,
      "learning_rate": 1.998867867794689e-05,
      "loss": 7.8265,
      "step": 16160
    },
    {
      "epoch": 0.045474870704565205,
      "grad_norm": 24.5,
      "learning_rate": 1.9988664663855478e-05,
      "loss": 7.7763,
      "step": 16170
    },
    {
      "epoch": 0.04550299369201392,
      "grad_norm": 45.25,
      "learning_rate": 1.9988650641100678e-05,
      "loss": 7.5871,
      "step": 16180
    },
    {
      "epoch": 0.04553111667946263,
      "grad_norm": 38.25,
      "learning_rate": 1.9988636609682507e-05,
      "loss": 8.633,
      "step": 16190
    },
    {
      "epoch": 0.04555923966691134,
      "grad_norm": 34.5,
      "learning_rate": 1.9988622569600977e-05,
      "loss": 7.1273,
      "step": 16200
    },
    {
      "epoch": 0.045587362654360045,
      "grad_norm": 25.5,
      "learning_rate": 1.99886085208561e-05,
      "loss": 7.1782,
      "step": 16210
    },
    {
      "epoch": 0.045615485641808756,
      "grad_norm": 28.625,
      "learning_rate": 1.9988594463447888e-05,
      "loss": 8.4275,
      "step": 16220
    },
    {
      "epoch": 0.04564360862925747,
      "grad_norm": 25.0,
      "learning_rate": 1.998858039737635e-05,
      "loss": 7.6835,
      "step": 16230
    },
    {
      "epoch": 0.04567173161670618,
      "grad_norm": 18.5,
      "learning_rate": 1.9988566322641505e-05,
      "loss": 7.8056,
      "step": 16240
    },
    {
      "epoch": 0.04569985460415489,
      "grad_norm": 40.25,
      "learning_rate": 1.9988552239243357e-05,
      "loss": 8.3347,
      "step": 16250
    },
    {
      "epoch": 0.0457279775916036,
      "grad_norm": 30.75,
      "learning_rate": 1.9988538147181926e-05,
      "loss": 7.8491,
      "step": 16260
    },
    {
      "epoch": 0.045756100579052314,
      "grad_norm": 45.75,
      "learning_rate": 1.998852404645722e-05,
      "loss": 8.3845,
      "step": 16270
    },
    {
      "epoch": 0.045784223566501026,
      "grad_norm": 29.125,
      "learning_rate": 1.9988509937069248e-05,
      "loss": 8.0623,
      "step": 16280
    },
    {
      "epoch": 0.04581234655394973,
      "grad_norm": 31.375,
      "learning_rate": 1.9988495819018027e-05,
      "loss": 8.7344,
      "step": 16290
    },
    {
      "epoch": 0.04584046954139844,
      "grad_norm": 28.0,
      "learning_rate": 1.998848169230357e-05,
      "loss": 8.33,
      "step": 16300
    },
    {
      "epoch": 0.04586859252884715,
      "grad_norm": 42.75,
      "learning_rate": 1.9988467556925887e-05,
      "loss": 8.2033,
      "step": 16310
    },
    {
      "epoch": 0.045896715516295865,
      "grad_norm": 30.625,
      "learning_rate": 1.9988453412884992e-05,
      "loss": 7.4832,
      "step": 16320
    },
    {
      "epoch": 0.045924838503744576,
      "grad_norm": 26.375,
      "learning_rate": 1.9988439260180895e-05,
      "loss": 7.3331,
      "step": 16330
    },
    {
      "epoch": 0.04595296149119329,
      "grad_norm": 23.875,
      "learning_rate": 1.9988425098813605e-05,
      "loss": 7.341,
      "step": 16340
    },
    {
      "epoch": 0.045981084478642,
      "grad_norm": 33.0,
      "learning_rate": 1.9988410928783147e-05,
      "loss": 7.932,
      "step": 16350
    },
    {
      "epoch": 0.04600920746609071,
      "grad_norm": 41.25,
      "learning_rate": 1.9988396750089517e-05,
      "loss": 8.0818,
      "step": 16360
    },
    {
      "epoch": 0.046037330453539416,
      "grad_norm": 20.75,
      "learning_rate": 1.998838256273274e-05,
      "loss": 8.1024,
      "step": 16370
    },
    {
      "epoch": 0.04606545344098813,
      "grad_norm": 35.0,
      "learning_rate": 1.998836836671282e-05,
      "loss": 7.487,
      "step": 16380
    },
    {
      "epoch": 0.04609357642843684,
      "grad_norm": 29.0,
      "learning_rate": 1.9988354162029777e-05,
      "loss": 8.4373,
      "step": 16390
    },
    {
      "epoch": 0.04612169941588555,
      "grad_norm": 25.375,
      "learning_rate": 1.9988339948683616e-05,
      "loss": 7.7159,
      "step": 16400
    },
    {
      "epoch": 0.04614982240333426,
      "grad_norm": 43.5,
      "learning_rate": 1.9988325726674354e-05,
      "loss": 7.2631,
      "step": 16410
    },
    {
      "epoch": 0.04617794539078297,
      "grad_norm": 19.375,
      "learning_rate": 1.9988311496002e-05,
      "loss": 8.0049,
      "step": 16420
    },
    {
      "epoch": 0.046206068378231685,
      "grad_norm": 28.125,
      "learning_rate": 1.998829725666657e-05,
      "loss": 8.6466,
      "step": 16430
    },
    {
      "epoch": 0.0462341913656804,
      "grad_norm": 39.25,
      "learning_rate": 1.998828300866807e-05,
      "loss": 8.6418,
      "step": 16440
    },
    {
      "epoch": 0.0462623143531291,
      "grad_norm": 37.0,
      "learning_rate": 1.9988268752006522e-05,
      "loss": 7.5935,
      "step": 16450
    },
    {
      "epoch": 0.04629043734057781,
      "grad_norm": 32.25,
      "learning_rate": 1.9988254486681934e-05,
      "loss": 6.8411,
      "step": 16460
    },
    {
      "epoch": 0.046318560328026524,
      "grad_norm": 29.25,
      "learning_rate": 1.9988240212694316e-05,
      "loss": 7.8838,
      "step": 16470
    },
    {
      "epoch": 0.046346683315475236,
      "grad_norm": 29.75,
      "learning_rate": 1.998822593004368e-05,
      "loss": 7.3165,
      "step": 16480
    },
    {
      "epoch": 0.04637480630292395,
      "grad_norm": 27.25,
      "learning_rate": 1.9988211638730042e-05,
      "loss": 8.1123,
      "step": 16490
    },
    {
      "epoch": 0.04640292929037266,
      "grad_norm": 24.375,
      "learning_rate": 1.9988197338753414e-05,
      "loss": 8.2335,
      "step": 16500
    },
    {
      "epoch": 0.04643105227782137,
      "grad_norm": 39.0,
      "learning_rate": 1.99881830301138e-05,
      "loss": 8.4012,
      "step": 16510
    },
    {
      "epoch": 0.04645917526527008,
      "grad_norm": 30.0,
      "learning_rate": 1.9988168712811228e-05,
      "loss": 7.7348,
      "step": 16520
    },
    {
      "epoch": 0.04648729825271879,
      "grad_norm": 30.0,
      "learning_rate": 1.9988154386845697e-05,
      "loss": 7.0193,
      "step": 16530
    },
    {
      "epoch": 0.0465154212401675,
      "grad_norm": 30.125,
      "learning_rate": 1.9988140052217225e-05,
      "loss": 7.3901,
      "step": 16540
    },
    {
      "epoch": 0.04654354422761621,
      "grad_norm": 28.625,
      "learning_rate": 1.998812570892583e-05,
      "loss": 8.1108,
      "step": 16550
    },
    {
      "epoch": 0.04657166721506492,
      "grad_norm": 33.25,
      "learning_rate": 1.9988111356971508e-05,
      "loss": 8.0769,
      "step": 16560
    },
    {
      "epoch": 0.04659979020251363,
      "grad_norm": 46.0,
      "learning_rate": 1.9988096996354287e-05,
      "loss": 7.6357,
      "step": 16570
    },
    {
      "epoch": 0.046627913189962344,
      "grad_norm": 36.75,
      "learning_rate": 1.9988082627074175e-05,
      "loss": 8.1465,
      "step": 16580
    },
    {
      "epoch": 0.046656036177411056,
      "grad_norm": 36.5,
      "learning_rate": 1.9988068249131182e-05,
      "loss": 6.9119,
      "step": 16590
    },
    {
      "epoch": 0.04668415916485977,
      "grad_norm": 35.5,
      "learning_rate": 1.9988053862525323e-05,
      "loss": 7.6147,
      "step": 16600
    },
    {
      "epoch": 0.04671228215230847,
      "grad_norm": 28.25,
      "learning_rate": 1.9988039467256607e-05,
      "loss": 8.1272,
      "step": 16610
    },
    {
      "epoch": 0.046740405139757184,
      "grad_norm": 42.5,
      "learning_rate": 1.9988025063325055e-05,
      "loss": 8.4363,
      "step": 16620
    },
    {
      "epoch": 0.046768528127205895,
      "grad_norm": 25.125,
      "learning_rate": 1.9988010650730666e-05,
      "loss": 8.2167,
      "step": 16630
    },
    {
      "epoch": 0.04679665111465461,
      "grad_norm": 29.75,
      "learning_rate": 1.998799622947346e-05,
      "loss": 8.1811,
      "step": 16640
    },
    {
      "epoch": 0.04682477410210332,
      "grad_norm": 29.5,
      "learning_rate": 1.9987981799553458e-05,
      "loss": 7.9032,
      "step": 16650
    },
    {
      "epoch": 0.04685289708955203,
      "grad_norm": 44.75,
      "learning_rate": 1.998796736097066e-05,
      "loss": 8.8709,
      "step": 16660
    },
    {
      "epoch": 0.04688102007700074,
      "grad_norm": 33.5,
      "learning_rate": 1.998795291372508e-05,
      "loss": 8.5267,
      "step": 16670
    },
    {
      "epoch": 0.04690914306444945,
      "grad_norm": 21.5,
      "learning_rate": 1.9987938457816737e-05,
      "loss": 7.6331,
      "step": 16680
    },
    {
      "epoch": 0.04693726605189816,
      "grad_norm": 34.5,
      "learning_rate": 1.998792399324564e-05,
      "loss": 8.2361,
      "step": 16690
    },
    {
      "epoch": 0.04696538903934687,
      "grad_norm": 41.75,
      "learning_rate": 1.9987909520011795e-05,
      "loss": 8.6306,
      "step": 16700
    },
    {
      "epoch": 0.04699351202679558,
      "grad_norm": 43.0,
      "learning_rate": 1.9987895038115224e-05,
      "loss": 8.5579,
      "step": 16710
    },
    {
      "epoch": 0.04702163501424429,
      "grad_norm": 21.25,
      "learning_rate": 1.998788054755594e-05,
      "loss": 7.9455,
      "step": 16720
    },
    {
      "epoch": 0.047049758001693004,
      "grad_norm": 33.5,
      "learning_rate": 1.9987866048333945e-05,
      "loss": 8.5222,
      "step": 16730
    },
    {
      "epoch": 0.047077880989141715,
      "grad_norm": 24.625,
      "learning_rate": 1.998785154044926e-05,
      "loss": 7.2171,
      "step": 16740
    },
    {
      "epoch": 0.04710600397659043,
      "grad_norm": 23.5,
      "learning_rate": 1.99878370239019e-05,
      "loss": 8.4293,
      "step": 16750
    },
    {
      "epoch": 0.04713412696403914,
      "grad_norm": 38.75,
      "learning_rate": 1.998782249869187e-05,
      "loss": 8.419,
      "step": 16760
    },
    {
      "epoch": 0.04716224995148784,
      "grad_norm": 25.5,
      "learning_rate": 1.9987807964819192e-05,
      "loss": 7.6227,
      "step": 16770
    },
    {
      "epoch": 0.047190372938936555,
      "grad_norm": 31.0,
      "learning_rate": 1.9987793422283864e-05,
      "loss": 8.2135,
      "step": 16780
    },
    {
      "epoch": 0.047218495926385266,
      "grad_norm": 23.875,
      "learning_rate": 1.9987778871085917e-05,
      "loss": 8.5281,
      "step": 16790
    },
    {
      "epoch": 0.04724661891383398,
      "grad_norm": 37.0,
      "learning_rate": 1.9987764311225346e-05,
      "loss": 8.2783,
      "step": 16800
    },
    {
      "epoch": 0.04727474190128269,
      "grad_norm": 27.75,
      "learning_rate": 1.9987749742702177e-05,
      "loss": 8.7025,
      "step": 16810
    },
    {
      "epoch": 0.0473028648887314,
      "grad_norm": 39.0,
      "learning_rate": 1.9987735165516415e-05,
      "loss": 8.2682,
      "step": 16820
    },
    {
      "epoch": 0.04733098787618011,
      "grad_norm": 29.5,
      "learning_rate": 1.9987720579668075e-05,
      "loss": 7.9711,
      "step": 16830
    },
    {
      "epoch": 0.047359110863628824,
      "grad_norm": 28.625,
      "learning_rate": 1.9987705985157172e-05,
      "loss": 7.2494,
      "step": 16840
    },
    {
      "epoch": 0.047387233851077536,
      "grad_norm": 29.25,
      "learning_rate": 1.998769138198371e-05,
      "loss": 7.2451,
      "step": 16850
    },
    {
      "epoch": 0.04741535683852624,
      "grad_norm": 22.25,
      "learning_rate": 1.9987676770147715e-05,
      "loss": 8.2972,
      "step": 16860
    },
    {
      "epoch": 0.04744347982597495,
      "grad_norm": 25.25,
      "learning_rate": 1.9987662149649188e-05,
      "loss": 8.4163,
      "step": 16870
    },
    {
      "epoch": 0.04747160281342366,
      "grad_norm": 27.125,
      "learning_rate": 1.998764752048815e-05,
      "loss": 7.5057,
      "step": 16880
    },
    {
      "epoch": 0.047499725800872375,
      "grad_norm": 32.25,
      "learning_rate": 1.9987632882664604e-05,
      "loss": 7.6181,
      "step": 16890
    },
    {
      "epoch": 0.047527848788321087,
      "grad_norm": 44.5,
      "learning_rate": 1.9987618236178574e-05,
      "loss": 7.6507,
      "step": 16900
    },
    {
      "epoch": 0.0475559717757698,
      "grad_norm": 32.5,
      "learning_rate": 1.9987603581030066e-05,
      "loss": 8.1165,
      "step": 16910
    },
    {
      "epoch": 0.04758409476321851,
      "grad_norm": 35.0,
      "learning_rate": 1.9987588917219095e-05,
      "loss": 7.9291,
      "step": 16920
    },
    {
      "epoch": 0.04761221775066722,
      "grad_norm": 25.25,
      "learning_rate": 1.998757424474567e-05,
      "loss": 8.4136,
      "step": 16930
    },
    {
      "epoch": 0.047640340738115926,
      "grad_norm": 28.0,
      "learning_rate": 1.998755956360981e-05,
      "loss": 7.6407,
      "step": 16940
    },
    {
      "epoch": 0.04766846372556464,
      "grad_norm": 88.0,
      "learning_rate": 1.9987544873811523e-05,
      "loss": 8.2023,
      "step": 16950
    },
    {
      "epoch": 0.04769658671301335,
      "grad_norm": 40.0,
      "learning_rate": 1.9987530175350823e-05,
      "loss": 7.8967,
      "step": 16960
    },
    {
      "epoch": 0.04772470970046206,
      "grad_norm": 35.25,
      "learning_rate": 1.998751546822772e-05,
      "loss": 7.9824,
      "step": 16970
    },
    {
      "epoch": 0.04775283268791077,
      "grad_norm": 21.75,
      "learning_rate": 1.998750075244223e-05,
      "loss": 7.1745,
      "step": 16980
    },
    {
      "epoch": 0.047780955675359484,
      "grad_norm": 31.75,
      "learning_rate": 1.998748602799437e-05,
      "loss": 8.5975,
      "step": 16990
    },
    {
      "epoch": 0.047809078662808195,
      "grad_norm": 26.75,
      "learning_rate": 1.9987471294884148e-05,
      "loss": 8.3548,
      "step": 17000
    },
    {
      "epoch": 0.04783720165025691,
      "grad_norm": 19.5,
      "learning_rate": 1.998745655311157e-05,
      "loss": 7.9876,
      "step": 17010
    },
    {
      "epoch": 0.04786532463770561,
      "grad_norm": 25.25,
      "learning_rate": 1.998744180267666e-05,
      "loss": 7.9681,
      "step": 17020
    },
    {
      "epoch": 0.04789344762515432,
      "grad_norm": 35.0,
      "learning_rate": 1.9987427043579427e-05,
      "loss": 7.7028,
      "step": 17030
    },
    {
      "epoch": 0.047921570612603034,
      "grad_norm": 27.625,
      "learning_rate": 1.998741227581988e-05,
      "loss": 7.6756,
      "step": 17040
    },
    {
      "epoch": 0.047949693600051746,
      "grad_norm": 21.75,
      "learning_rate": 1.998739749939804e-05,
      "loss": 8.0954,
      "step": 17050
    },
    {
      "epoch": 0.04797781658750046,
      "grad_norm": 41.0,
      "learning_rate": 1.998738271431391e-05,
      "loss": 7.5269,
      "step": 17060
    },
    {
      "epoch": 0.04800593957494917,
      "grad_norm": 26.125,
      "learning_rate": 1.998736792056751e-05,
      "loss": 8.3666,
      "step": 17070
    },
    {
      "epoch": 0.04803406256239788,
      "grad_norm": 25.875,
      "learning_rate": 1.9987353118158852e-05,
      "loss": 7.4585,
      "step": 17080
    },
    {
      "epoch": 0.04806218554984659,
      "grad_norm": 33.0,
      "learning_rate": 1.9987338307087944e-05,
      "loss": 8.1088,
      "step": 17090
    },
    {
      "epoch": 0.0480903085372953,
      "grad_norm": 39.0,
      "learning_rate": 1.9987323487354806e-05,
      "loss": 8.3615,
      "step": 17100
    },
    {
      "epoch": 0.04811843152474401,
      "grad_norm": 50.75,
      "learning_rate": 1.9987308658959442e-05,
      "loss": 7.3314,
      "step": 17110
    },
    {
      "epoch": 0.04814655451219272,
      "grad_norm": 22.625,
      "learning_rate": 1.9987293821901874e-05,
      "loss": 8.1899,
      "step": 17120
    },
    {
      "epoch": 0.04817467749964143,
      "grad_norm": 25.5,
      "learning_rate": 1.9987278976182108e-05,
      "loss": 8.1695,
      "step": 17130
    },
    {
      "epoch": 0.04820280048709014,
      "grad_norm": 37.25,
      "learning_rate": 1.9987264121800162e-05,
      "loss": 7.9669,
      "step": 17140
    },
    {
      "epoch": 0.048230923474538855,
      "grad_norm": 25.125,
      "learning_rate": 1.9987249258756044e-05,
      "loss": 8.2335,
      "step": 17150
    },
    {
      "epoch": 0.048259046461987566,
      "grad_norm": 45.75,
      "learning_rate": 1.9987234387049774e-05,
      "loss": 8.9114,
      "step": 17160
    },
    {
      "epoch": 0.04828716944943628,
      "grad_norm": 25.25,
      "learning_rate": 1.9987219506681356e-05,
      "loss": 7.5578,
      "step": 17170
    },
    {
      "epoch": 0.04831529243688498,
      "grad_norm": 50.0,
      "learning_rate": 1.998720461765081e-05,
      "loss": 8.1219,
      "step": 17180
    },
    {
      "epoch": 0.048343415424333694,
      "grad_norm": 27.875,
      "learning_rate": 1.998718971995814e-05,
      "loss": 6.9166,
      "step": 17190
    },
    {
      "epoch": 0.048371538411782405,
      "grad_norm": 23.625,
      "learning_rate": 1.9987174813603372e-05,
      "loss": 7.5407,
      "step": 17200
    },
    {
      "epoch": 0.04839966139923112,
      "grad_norm": 35.75,
      "learning_rate": 1.9987159898586512e-05,
      "loss": 8.1419,
      "step": 17210
    },
    {
      "epoch": 0.04842778438667983,
      "grad_norm": 37.25,
      "learning_rate": 1.9987144974907567e-05,
      "loss": 7.3759,
      "step": 17220
    },
    {
      "epoch": 0.04845590737412854,
      "grad_norm": 29.125,
      "learning_rate": 1.998713004256656e-05,
      "loss": 7.9778,
      "step": 17230
    },
    {
      "epoch": 0.04848403036157725,
      "grad_norm": 32.75,
      "learning_rate": 1.99871151015635e-05,
      "loss": 7.4636,
      "step": 17240
    },
    {
      "epoch": 0.04851215334902596,
      "grad_norm": 23.625,
      "learning_rate": 1.99871001518984e-05,
      "loss": 7.6179,
      "step": 17250
    },
    {
      "epoch": 0.04854027633647467,
      "grad_norm": 52.5,
      "learning_rate": 1.998708519357127e-05,
      "loss": 7.882,
      "step": 17260
    },
    {
      "epoch": 0.04856839932392338,
      "grad_norm": 27.625,
      "learning_rate": 1.9987070226582128e-05,
      "loss": 7.7648,
      "step": 17270
    },
    {
      "epoch": 0.04859652231137209,
      "grad_norm": 39.75,
      "learning_rate": 1.9987055250930984e-05,
      "loss": 8.3472,
      "step": 17280
    },
    {
      "epoch": 0.0486246452988208,
      "grad_norm": 36.75,
      "learning_rate": 1.998704026661785e-05,
      "loss": 7.5586,
      "step": 17290
    },
    {
      "epoch": 0.048652768286269514,
      "grad_norm": 41.0,
      "learning_rate": 1.9987025273642744e-05,
      "loss": 7.918,
      "step": 17300
    },
    {
      "epoch": 0.048680891273718226,
      "grad_norm": 35.25,
      "learning_rate": 1.9987010272005672e-05,
      "loss": 8.3589,
      "step": 17310
    },
    {
      "epoch": 0.04870901426116694,
      "grad_norm": 28.625,
      "learning_rate": 1.998699526170665e-05,
      "loss": 8.0466,
      "step": 17320
    },
    {
      "epoch": 0.04873713724861565,
      "grad_norm": 25.75,
      "learning_rate": 1.9986980242745696e-05,
      "loss": 7.717,
      "step": 17330
    },
    {
      "epoch": 0.04876526023606435,
      "grad_norm": 23.25,
      "learning_rate": 1.998696521512282e-05,
      "loss": 8.0517,
      "step": 17340
    },
    {
      "epoch": 0.048793383223513065,
      "grad_norm": 38.0,
      "learning_rate": 1.998695017883803e-05,
      "loss": 7.4055,
      "step": 17350
    },
    {
      "epoch": 0.048821506210961776,
      "grad_norm": 45.5,
      "learning_rate": 1.9986935133891344e-05,
      "loss": 8.9037,
      "step": 17360
    },
    {
      "epoch": 0.04884962919841049,
      "grad_norm": 37.75,
      "learning_rate": 1.9986920080282773e-05,
      "loss": 8.8098,
      "step": 17370
    },
    {
      "epoch": 0.0488777521858592,
      "grad_norm": 22.25,
      "learning_rate": 1.998690501801233e-05,
      "loss": 7.6807,
      "step": 17380
    },
    {
      "epoch": 0.04890587517330791,
      "grad_norm": 30.0,
      "learning_rate": 1.998688994708003e-05,
      "loss": 8.4379,
      "step": 17390
    },
    {
      "epoch": 0.04893399816075662,
      "grad_norm": 29.625,
      "learning_rate": 1.9986874867485885e-05,
      "loss": 8.2624,
      "step": 17400
    },
    {
      "epoch": 0.048962121148205334,
      "grad_norm": 24.375,
      "learning_rate": 1.998685977922991e-05,
      "loss": 7.7149,
      "step": 17410
    },
    {
      "epoch": 0.04899024413565404,
      "grad_norm": 22.75,
      "learning_rate": 1.9986844682312114e-05,
      "loss": 8.695,
      "step": 17420
    },
    {
      "epoch": 0.04901836712310275,
      "grad_norm": 47.75,
      "learning_rate": 1.9986829576732512e-05,
      "loss": 8.0961,
      "step": 17430
    },
    {
      "epoch": 0.04904649011055146,
      "grad_norm": 25.625,
      "learning_rate": 1.998681446249112e-05,
      "loss": 8.2192,
      "step": 17440
    },
    {
      "epoch": 0.049074613098000174,
      "grad_norm": 39.0,
      "learning_rate": 1.9986799339587947e-05,
      "loss": 8.5481,
      "step": 17450
    },
    {
      "epoch": 0.049102736085448885,
      "grad_norm": 36.5,
      "learning_rate": 1.9986784208023007e-05,
      "loss": 8.0833,
      "step": 17460
    },
    {
      "epoch": 0.0491308590728976,
      "grad_norm": 23.5,
      "learning_rate": 1.9986769067796312e-05,
      "loss": 8.305,
      "step": 17470
    },
    {
      "epoch": 0.04915898206034631,
      "grad_norm": 25.625,
      "learning_rate": 1.998675391890788e-05,
      "loss": 7.3399,
      "step": 17480
    },
    {
      "epoch": 0.04918710504779502,
      "grad_norm": 30.25,
      "learning_rate": 1.998673876135772e-05,
      "loss": 7.5145,
      "step": 17490
    },
    {
      "epoch": 0.04921522803524373,
      "grad_norm": 32.25,
      "learning_rate": 1.9986723595145846e-05,
      "loss": 7.2597,
      "step": 17500
    },
    {
      "epoch": 0.049243351022692436,
      "grad_norm": 36.75,
      "learning_rate": 1.998670842027227e-05,
      "loss": 8.5303,
      "step": 17510
    },
    {
      "epoch": 0.04927147401014115,
      "grad_norm": 47.75,
      "learning_rate": 1.998669323673701e-05,
      "loss": 7.783,
      "step": 17520
    },
    {
      "epoch": 0.04929959699758986,
      "grad_norm": 27.0,
      "learning_rate": 1.998667804454007e-05,
      "loss": 7.5953,
      "step": 17530
    },
    {
      "epoch": 0.04932771998503857,
      "grad_norm": 35.5,
      "learning_rate": 1.9986662843681474e-05,
      "loss": 8.1571,
      "step": 17540
    },
    {
      "epoch": 0.04935584297248728,
      "grad_norm": 34.5,
      "learning_rate": 1.9986647634161233e-05,
      "loss": 7.6336,
      "step": 17550
    },
    {
      "epoch": 0.049383965959935994,
      "grad_norm": 25.875,
      "learning_rate": 1.998663241597935e-05,
      "loss": 7.7515,
      "step": 17560
    },
    {
      "epoch": 0.049412088947384705,
      "grad_norm": 29.875,
      "learning_rate": 1.998661718913585e-05,
      "loss": 7.2571,
      "step": 17570
    },
    {
      "epoch": 0.04944021193483342,
      "grad_norm": 37.5,
      "learning_rate": 1.998660195363074e-05,
      "loss": 7.9719,
      "step": 17580
    },
    {
      "epoch": 0.04946833492228212,
      "grad_norm": 27.125,
      "learning_rate": 1.9986586709464034e-05,
      "loss": 8.0792,
      "step": 17590
    },
    {
      "epoch": 0.04949645790973083,
      "grad_norm": 21.125,
      "learning_rate": 1.9986571456635743e-05,
      "loss": 7.165,
      "step": 17600
    },
    {
      "epoch": 0.049524580897179545,
      "grad_norm": 31.5,
      "learning_rate": 1.998655619514589e-05,
      "loss": 7.9473,
      "step": 17610
    },
    {
      "epoch": 0.049552703884628256,
      "grad_norm": 20.0,
      "learning_rate": 1.9986540924994478e-05,
      "loss": 8.5033,
      "step": 17620
    },
    {
      "epoch": 0.04958082687207697,
      "grad_norm": 30.625,
      "learning_rate": 1.9986525646181524e-05,
      "loss": 8.4058,
      "step": 17630
    },
    {
      "epoch": 0.04960894985952568,
      "grad_norm": 58.0,
      "learning_rate": 1.9986510358707042e-05,
      "loss": 8.6028,
      "step": 17640
    },
    {
      "epoch": 0.04963707284697439,
      "grad_norm": 31.25,
      "learning_rate": 1.9986495062571044e-05,
      "loss": 8.8688,
      "step": 17650
    },
    {
      "epoch": 0.0496651958344231,
      "grad_norm": 31.75,
      "learning_rate": 1.9986479757773542e-05,
      "loss": 6.848,
      "step": 17660
    },
    {
      "epoch": 0.04969331882187181,
      "grad_norm": 25.125,
      "learning_rate": 1.9986464444314552e-05,
      "loss": 8.0212,
      "step": 17670
    },
    {
      "epoch": 0.04972144180932052,
      "grad_norm": 31.0,
      "learning_rate": 1.9986449122194084e-05,
      "loss": 7.974,
      "step": 17680
    },
    {
      "epoch": 0.04974956479676923,
      "grad_norm": 35.25,
      "learning_rate": 1.9986433791412155e-05,
      "loss": 7.947,
      "step": 17690
    },
    {
      "epoch": 0.04977768778421794,
      "grad_norm": 29.875,
      "learning_rate": 1.9986418451968776e-05,
      "loss": 8.0968,
      "step": 17700
    },
    {
      "epoch": 0.04980581077166665,
      "grad_norm": 45.25,
      "learning_rate": 1.9986403103863963e-05,
      "loss": 8.0338,
      "step": 17710
    },
    {
      "epoch": 0.049833933759115365,
      "grad_norm": 40.5,
      "learning_rate": 1.9986387747097727e-05,
      "loss": 8.4278,
      "step": 17720
    },
    {
      "epoch": 0.049862056746564076,
      "grad_norm": 38.0,
      "learning_rate": 1.998637238167008e-05,
      "loss": 8.6942,
      "step": 17730
    },
    {
      "epoch": 0.04989017973401279,
      "grad_norm": 25.125,
      "learning_rate": 1.9986357007581038e-05,
      "loss": 7.4679,
      "step": 17740
    },
    {
      "epoch": 0.04991830272146149,
      "grad_norm": 32.75,
      "learning_rate": 1.9986341624830616e-05,
      "loss": 8.0009,
      "step": 17750
    },
    {
      "epoch": 0.049946425708910204,
      "grad_norm": 35.75,
      "learning_rate": 1.9986326233418817e-05,
      "loss": 7.3796,
      "step": 17760
    },
    {
      "epoch": 0.049974548696358916,
      "grad_norm": 38.25,
      "learning_rate": 1.9986310833345667e-05,
      "loss": 7.4392,
      "step": 17770
    },
    {
      "epoch": 0.05000267168380763,
      "grad_norm": 38.75,
      "learning_rate": 1.9986295424611174e-05,
      "loss": 8.1885,
      "step": 17780
    },
    {
      "epoch": 0.05003079467125634,
      "grad_norm": 26.75,
      "learning_rate": 1.998628000721535e-05,
      "loss": 7.3953,
      "step": 17790
    },
    {
      "epoch": 0.05005891765870505,
      "grad_norm": 27.125,
      "learning_rate": 1.9986264581158213e-05,
      "loss": 8.018,
      "step": 17800
    },
    {
      "epoch": 0.05008704064615376,
      "grad_norm": 25.375,
      "learning_rate": 1.998624914643977e-05,
      "loss": 8.089,
      "step": 17810
    },
    {
      "epoch": 0.05011516363360247,
      "grad_norm": 26.5,
      "learning_rate": 1.9986233703060042e-05,
      "loss": 8.2092,
      "step": 17820
    },
    {
      "epoch": 0.05014328662105118,
      "grad_norm": 35.0,
      "learning_rate": 1.9986218251019035e-05,
      "loss": 7.3687,
      "step": 17830
    },
    {
      "epoch": 0.05017140960849989,
      "grad_norm": 30.5,
      "learning_rate": 1.998620279031677e-05,
      "loss": 8.5225,
      "step": 17840
    },
    {
      "epoch": 0.0501995325959486,
      "grad_norm": 52.0,
      "learning_rate": 1.9986187320953248e-05,
      "loss": 8.477,
      "step": 17850
    },
    {
      "epoch": 0.05022765558339731,
      "grad_norm": 32.75,
      "learning_rate": 1.99861718429285e-05,
      "loss": 7.926,
      "step": 17860
    },
    {
      "epoch": 0.050255778570846024,
      "grad_norm": 41.0,
      "learning_rate": 1.998615635624252e-05,
      "loss": 7.7154,
      "step": 17870
    },
    {
      "epoch": 0.050283901558294736,
      "grad_norm": 52.75,
      "learning_rate": 1.9986140860895337e-05,
      "loss": 8.6033,
      "step": 17880
    },
    {
      "epoch": 0.05031202454574345,
      "grad_norm": 26.375,
      "learning_rate": 1.9986125356886958e-05,
      "loss": 8.026,
      "step": 17890
    },
    {
      "epoch": 0.05034014753319216,
      "grad_norm": 39.5,
      "learning_rate": 1.99861098442174e-05,
      "loss": 7.4468,
      "step": 17900
    },
    {
      "epoch": 0.05036827052064086,
      "grad_norm": 36.5,
      "learning_rate": 1.9986094322886668e-05,
      "loss": 7.8802,
      "step": 17910
    },
    {
      "epoch": 0.050396393508089575,
      "grad_norm": 24.375,
      "learning_rate": 1.9986078792894784e-05,
      "loss": 8.1702,
      "step": 17920
    },
    {
      "epoch": 0.05042451649553829,
      "grad_norm": 49.75,
      "learning_rate": 1.9986063254241757e-05,
      "loss": 8.2462,
      "step": 17930
    },
    {
      "epoch": 0.050452639482987,
      "grad_norm": 52.75,
      "learning_rate": 1.9986047706927606e-05,
      "loss": 7.8778,
      "step": 17940
    },
    {
      "epoch": 0.05048076247043571,
      "grad_norm": 46.5,
      "learning_rate": 1.998603215095234e-05,
      "loss": 7.7282,
      "step": 17950
    },
    {
      "epoch": 0.05050888545788442,
      "grad_norm": 21.125,
      "learning_rate": 1.9986016586315967e-05,
      "loss": 8.4388,
      "step": 17960
    },
    {
      "epoch": 0.05053700844533313,
      "grad_norm": 41.75,
      "learning_rate": 1.998600101301851e-05,
      "loss": 8.1653,
      "step": 17970
    },
    {
      "epoch": 0.050565131432781844,
      "grad_norm": 29.5,
      "learning_rate": 1.9985985431059983e-05,
      "loss": 8.4912,
      "step": 17980
    },
    {
      "epoch": 0.05059325442023055,
      "grad_norm": 32.5,
      "learning_rate": 1.998596984044039e-05,
      "loss": 7.657,
      "step": 17990
    },
    {
      "epoch": 0.05062137740767926,
      "grad_norm": 39.0,
      "learning_rate": 1.9985954241159752e-05,
      "loss": 7.8324,
      "step": 18000
    },
    {
      "epoch": 0.05064950039512797,
      "grad_norm": 25.625,
      "learning_rate": 1.9985938633218082e-05,
      "loss": 8.4936,
      "step": 18010
    },
    {
      "epoch": 0.050677623382576684,
      "grad_norm": 38.75,
      "learning_rate": 1.9985923016615392e-05,
      "loss": 8.2165,
      "step": 18020
    },
    {
      "epoch": 0.050705746370025395,
      "grad_norm": 22.375,
      "learning_rate": 1.9985907391351692e-05,
      "loss": 7.7616,
      "step": 18030
    },
    {
      "epoch": 0.05073386935747411,
      "grad_norm": 22.625,
      "learning_rate": 1.9985891757427005e-05,
      "loss": 7.5339,
      "step": 18040
    },
    {
      "epoch": 0.05076199234492282,
      "grad_norm": 21.0,
      "learning_rate": 1.9985876114841332e-05,
      "loss": 7.8015,
      "step": 18050
    },
    {
      "epoch": 0.05079011533237153,
      "grad_norm": 35.25,
      "learning_rate": 1.99858604635947e-05,
      "loss": 7.9375,
      "step": 18060
    },
    {
      "epoch": 0.050818238319820234,
      "grad_norm": 31.25,
      "learning_rate": 1.9985844803687115e-05,
      "loss": 7.9773,
      "step": 18070
    },
    {
      "epoch": 0.050846361307268946,
      "grad_norm": 38.0,
      "learning_rate": 1.9985829135118588e-05,
      "loss": 7.6728,
      "step": 18080
    },
    {
      "epoch": 0.05087448429471766,
      "grad_norm": 23.375,
      "learning_rate": 1.9985813457889135e-05,
      "loss": 7.7526,
      "step": 18090
    },
    {
      "epoch": 0.05090260728216637,
      "grad_norm": 26.0,
      "learning_rate": 1.9985797771998774e-05,
      "loss": 7.8189,
      "step": 18100
    },
    {
      "epoch": 0.05093073026961508,
      "grad_norm": 31.875,
      "learning_rate": 1.9985782077447514e-05,
      "loss": 7.5563,
      "step": 18110
    },
    {
      "epoch": 0.05095885325706379,
      "grad_norm": 23.75,
      "learning_rate": 1.9985766374235373e-05,
      "loss": 7.5335,
      "step": 18120
    },
    {
      "epoch": 0.050986976244512504,
      "grad_norm": 26.875,
      "learning_rate": 1.9985750662362357e-05,
      "loss": 8.925,
      "step": 18130
    },
    {
      "epoch": 0.051015099231961215,
      "grad_norm": 20.375,
      "learning_rate": 1.9985734941828484e-05,
      "loss": 8.1326,
      "step": 18140
    },
    {
      "epoch": 0.05104322221940993,
      "grad_norm": 31.75,
      "learning_rate": 1.9985719212633774e-05,
      "loss": 8.5344,
      "step": 18150
    },
    {
      "epoch": 0.05107134520685863,
      "grad_norm": 30.5,
      "learning_rate": 1.998570347477823e-05,
      "loss": 7.3642,
      "step": 18160
    },
    {
      "epoch": 0.05109946819430734,
      "grad_norm": 26.625,
      "learning_rate": 1.9985687728261867e-05,
      "loss": 8.3164,
      "step": 18170
    },
    {
      "epoch": 0.051127591181756055,
      "grad_norm": 19.75,
      "learning_rate": 1.998567197308471e-05,
      "loss": 8.3561,
      "step": 18180
    },
    {
      "epoch": 0.051155714169204766,
      "grad_norm": 46.0,
      "learning_rate": 1.9985656209246757e-05,
      "loss": 7.9917,
      "step": 18190
    },
    {
      "epoch": 0.05118383715665348,
      "grad_norm": 33.75,
      "learning_rate": 1.998564043674803e-05,
      "loss": 7.7524,
      "step": 18200
    },
    {
      "epoch": 0.05121196014410219,
      "grad_norm": 34.0,
      "learning_rate": 1.9985624655588545e-05,
      "loss": 8.1255,
      "step": 18210
    },
    {
      "epoch": 0.0512400831315509,
      "grad_norm": 28.75,
      "learning_rate": 1.9985608865768312e-05,
      "loss": 8.3894,
      "step": 18220
    },
    {
      "epoch": 0.05126820611899961,
      "grad_norm": 26.25,
      "learning_rate": 1.9985593067287343e-05,
      "loss": 7.3203,
      "step": 18230
    },
    {
      "epoch": 0.05129632910644832,
      "grad_norm": 25.5,
      "learning_rate": 1.9985577260145656e-05,
      "loss": 8.2529,
      "step": 18240
    },
    {
      "epoch": 0.05132445209389703,
      "grad_norm": 32.5,
      "learning_rate": 1.9985561444343263e-05,
      "loss": 8.1071,
      "step": 18250
    },
    {
      "epoch": 0.05135257508134574,
      "grad_norm": 28.375,
      "learning_rate": 1.9985545619880175e-05,
      "loss": 8.1236,
      "step": 18260
    },
    {
      "epoch": 0.05138069806879445,
      "grad_norm": 24.125,
      "learning_rate": 1.998552978675641e-05,
      "loss": 7.6847,
      "step": 18270
    },
    {
      "epoch": 0.05140882105624316,
      "grad_norm": 23.25,
      "learning_rate": 1.9985513944971978e-05,
      "loss": 8.397,
      "step": 18280
    },
    {
      "epoch": 0.051436944043691875,
      "grad_norm": 22.625,
      "learning_rate": 1.9985498094526893e-05,
      "loss": 8.4665,
      "step": 18290
    },
    {
      "epoch": 0.051465067031140586,
      "grad_norm": 34.75,
      "learning_rate": 1.998548223542117e-05,
      "loss": 7.8867,
      "step": 18300
    },
    {
      "epoch": 0.0514931900185893,
      "grad_norm": 36.25,
      "learning_rate": 1.9985466367654828e-05,
      "loss": 7.9034,
      "step": 18310
    },
    {
      "epoch": 0.051521313006038,
      "grad_norm": 23.5,
      "learning_rate": 1.9985450491227874e-05,
      "loss": 8.4502,
      "step": 18320
    },
    {
      "epoch": 0.051549435993486714,
      "grad_norm": 24.0,
      "learning_rate": 1.998543460614032e-05,
      "loss": 7.1839,
      "step": 18330
    },
    {
      "epoch": 0.051577558980935426,
      "grad_norm": 25.375,
      "learning_rate": 1.9985418712392187e-05,
      "loss": 8.578,
      "step": 18340
    },
    {
      "epoch": 0.05160568196838414,
      "grad_norm": 31.375,
      "learning_rate": 1.9985402809983484e-05,
      "loss": 7.6102,
      "step": 18350
    },
    {
      "epoch": 0.05163380495583285,
      "grad_norm": 22.75,
      "learning_rate": 1.9985386898914225e-05,
      "loss": 8.1065,
      "step": 18360
    },
    {
      "epoch": 0.05166192794328156,
      "grad_norm": 34.0,
      "learning_rate": 1.9985370979184427e-05,
      "loss": 7.7248,
      "step": 18370
    },
    {
      "epoch": 0.05169005093073027,
      "grad_norm": 26.125,
      "learning_rate": 1.99853550507941e-05,
      "loss": 7.8967,
      "step": 18380
    },
    {
      "epoch": 0.05171817391817898,
      "grad_norm": 41.25,
      "learning_rate": 1.9985339113743257e-05,
      "loss": 8.4286,
      "step": 18390
    },
    {
      "epoch": 0.05174629690562769,
      "grad_norm": 28.0,
      "learning_rate": 1.9985323168031915e-05,
      "loss": 8.4294,
      "step": 18400
    },
    {
      "epoch": 0.0517744198930764,
      "grad_norm": 35.75,
      "learning_rate": 1.998530721366009e-05,
      "loss": 8.3683,
      "step": 18410
    },
    {
      "epoch": 0.05180254288052511,
      "grad_norm": 28.875,
      "learning_rate": 1.998529125062779e-05,
      "loss": 8.508,
      "step": 18420
    },
    {
      "epoch": 0.05183066586797382,
      "grad_norm": 44.5,
      "learning_rate": 1.9985275278935032e-05,
      "loss": 8.2264,
      "step": 18430
    },
    {
      "epoch": 0.051858788855422534,
      "grad_norm": 33.5,
      "learning_rate": 1.998525929858183e-05,
      "loss": 8.018,
      "step": 18440
    },
    {
      "epoch": 0.051886911842871246,
      "grad_norm": 23.75,
      "learning_rate": 1.9985243309568198e-05,
      "loss": 7.7806,
      "step": 18450
    },
    {
      "epoch": 0.05191503483031996,
      "grad_norm": 40.5,
      "learning_rate": 1.998522731189415e-05,
      "loss": 7.8555,
      "step": 18460
    },
    {
      "epoch": 0.05194315781776867,
      "grad_norm": 29.875,
      "learning_rate": 1.9985211305559695e-05,
      "loss": 8.0372,
      "step": 18470
    },
    {
      "epoch": 0.051971280805217374,
      "grad_norm": 33.0,
      "learning_rate": 1.9985195290564853e-05,
      "loss": 7.2746,
      "step": 18480
    },
    {
      "epoch": 0.051999403792666085,
      "grad_norm": 35.5,
      "learning_rate": 1.998517926690964e-05,
      "loss": 7.6454,
      "step": 18490
    },
    {
      "epoch": 0.0520275267801148,
      "grad_norm": 33.5,
      "learning_rate": 1.998516323459406e-05,
      "loss": 8.0477,
      "step": 18500
    },
    {
      "epoch": 0.05205564976756351,
      "grad_norm": 22.0,
      "learning_rate": 1.9985147193618137e-05,
      "loss": 7.5663,
      "step": 18510
    },
    {
      "epoch": 0.05208377275501222,
      "grad_norm": 22.375,
      "learning_rate": 1.9985131143981877e-05,
      "loss": 7.5779,
      "step": 18520
    },
    {
      "epoch": 0.05211189574246093,
      "grad_norm": 22.25,
      "learning_rate": 1.9985115085685298e-05,
      "loss": 7.7189,
      "step": 18530
    },
    {
      "epoch": 0.05214001872990964,
      "grad_norm": 27.875,
      "learning_rate": 1.9985099018728414e-05,
      "loss": 7.9264,
      "step": 18540
    },
    {
      "epoch": 0.052168141717358354,
      "grad_norm": 24.25,
      "learning_rate": 1.9985082943111237e-05,
      "loss": 8.363,
      "step": 18550
    },
    {
      "epoch": 0.05219626470480706,
      "grad_norm": 29.625,
      "learning_rate": 1.9985066858833786e-05,
      "loss": 8.0014,
      "step": 18560
    },
    {
      "epoch": 0.05222438769225577,
      "grad_norm": 28.375,
      "learning_rate": 1.998505076589607e-05,
      "loss": 8.6115,
      "step": 18570
    },
    {
      "epoch": 0.05225251067970448,
      "grad_norm": 27.375,
      "learning_rate": 1.9985034664298104e-05,
      "loss": 7.4145,
      "step": 18580
    },
    {
      "epoch": 0.052280633667153194,
      "grad_norm": 34.75,
      "learning_rate": 1.9985018554039902e-05,
      "loss": 8.1568,
      "step": 18590
    },
    {
      "epoch": 0.052308756654601905,
      "grad_norm": 28.0,
      "learning_rate": 1.9985002435121474e-05,
      "loss": 7.9962,
      "step": 18600
    },
    {
      "epoch": 0.05233687964205062,
      "grad_norm": 27.0,
      "learning_rate": 1.9984986307542844e-05,
      "loss": 7.7467,
      "step": 18610
    },
    {
      "epoch": 0.05236500262949933,
      "grad_norm": 32.0,
      "learning_rate": 1.998497017130402e-05,
      "loss": 8.1312,
      "step": 18620
    },
    {
      "epoch": 0.05239312561694804,
      "grad_norm": 26.0,
      "learning_rate": 1.9984954026405012e-05,
      "loss": 9.0233,
      "step": 18630
    },
    {
      "epoch": 0.052421248604396745,
      "grad_norm": 37.5,
      "learning_rate": 1.998493787284584e-05,
      "loss": 7.7665,
      "step": 18640
    },
    {
      "epoch": 0.052449371591845456,
      "grad_norm": 25.25,
      "learning_rate": 1.9984921710626514e-05,
      "loss": 7.9155,
      "step": 18650
    },
    {
      "epoch": 0.05247749457929417,
      "grad_norm": 30.0,
      "learning_rate": 1.9984905539747052e-05,
      "loss": 8.1381,
      "step": 18660
    },
    {
      "epoch": 0.05250561756674288,
      "grad_norm": 30.0,
      "learning_rate": 1.9984889360207468e-05,
      "loss": 7.8947,
      "step": 18670
    },
    {
      "epoch": 0.05253374055419159,
      "grad_norm": 29.0,
      "learning_rate": 1.9984873172007772e-05,
      "loss": 7.7746,
      "step": 18680
    },
    {
      "epoch": 0.0525618635416403,
      "grad_norm": 24.875,
      "learning_rate": 1.998485697514798e-05,
      "loss": 8.0488,
      "step": 18690
    },
    {
      "epoch": 0.052589986529089014,
      "grad_norm": 29.375,
      "learning_rate": 1.998484076962811e-05,
      "loss": 7.9374,
      "step": 18700
    },
    {
      "epoch": 0.052618109516537726,
      "grad_norm": 46.0,
      "learning_rate": 1.998482455544817e-05,
      "loss": 8.4672,
      "step": 18710
    },
    {
      "epoch": 0.05264623250398643,
      "grad_norm": 27.875,
      "learning_rate": 1.9984808332608175e-05,
      "loss": 7.9785,
      "step": 18720
    },
    {
      "epoch": 0.05267435549143514,
      "grad_norm": 28.875,
      "learning_rate": 1.998479210110814e-05,
      "loss": 7.398,
      "step": 18730
    },
    {
      "epoch": 0.05270247847888385,
      "grad_norm": 53.25,
      "learning_rate": 1.9984775860948083e-05,
      "loss": 7.5566,
      "step": 18740
    },
    {
      "epoch": 0.052730601466332565,
      "grad_norm": 33.25,
      "learning_rate": 1.998475961212801e-05,
      "loss": 8.4809,
      "step": 18750
    },
    {
      "epoch": 0.052758724453781276,
      "grad_norm": 33.5,
      "learning_rate": 1.9984743354647944e-05,
      "loss": 7.761,
      "step": 18760
    },
    {
      "epoch": 0.05278684744122999,
      "grad_norm": 26.625,
      "learning_rate": 1.9984727088507895e-05,
      "loss": 8.0479,
      "step": 18770
    },
    {
      "epoch": 0.0528149704286787,
      "grad_norm": 20.125,
      "learning_rate": 1.9984710813707872e-05,
      "loss": 8.3646,
      "step": 18780
    },
    {
      "epoch": 0.05284309341612741,
      "grad_norm": 48.0,
      "learning_rate": 1.99846945302479e-05,
      "loss": 8.3399,
      "step": 18790
    },
    {
      "epoch": 0.052871216403576116,
      "grad_norm": 20.75,
      "learning_rate": 1.9984678238127984e-05,
      "loss": 7.2663,
      "step": 18800
    },
    {
      "epoch": 0.05289933939102483,
      "grad_norm": 21.5,
      "learning_rate": 1.998466193734814e-05,
      "loss": 8.3039,
      "step": 18810
    },
    {
      "epoch": 0.05292746237847354,
      "grad_norm": 38.75,
      "learning_rate": 1.9984645627908385e-05,
      "loss": 7.9138,
      "step": 18820
    },
    {
      "epoch": 0.05295558536592225,
      "grad_norm": 29.0,
      "learning_rate": 1.998462930980873e-05,
      "loss": 8.6257,
      "step": 18830
    },
    {
      "epoch": 0.05298370835337096,
      "grad_norm": 29.625,
      "learning_rate": 1.9984612983049194e-05,
      "loss": 7.2328,
      "step": 18840
    },
    {
      "epoch": 0.05301183134081967,
      "grad_norm": 28.5,
      "learning_rate": 1.9984596647629785e-05,
      "loss": 8.441,
      "step": 18850
    },
    {
      "epoch": 0.053039954328268385,
      "grad_norm": 30.5,
      "learning_rate": 1.9984580303550523e-05,
      "loss": 7.9144,
      "step": 18860
    },
    {
      "epoch": 0.053068077315717097,
      "grad_norm": 43.75,
      "learning_rate": 1.9984563950811418e-05,
      "loss": 7.9972,
      "step": 18870
    },
    {
      "epoch": 0.05309620030316581,
      "grad_norm": 35.0,
      "learning_rate": 1.9984547589412484e-05,
      "loss": 7.6836,
      "step": 18880
    },
    {
      "epoch": 0.05312432329061451,
      "grad_norm": 22.75,
      "learning_rate": 1.9984531219353737e-05,
      "loss": 7.9876,
      "step": 18890
    },
    {
      "epoch": 0.053152446278063224,
      "grad_norm": 36.25,
      "learning_rate": 1.998451484063519e-05,
      "loss": 8.9144,
      "step": 18900
    },
    {
      "epoch": 0.053180569265511936,
      "grad_norm": 32.75,
      "learning_rate": 1.998449845325686e-05,
      "loss": 8.1803,
      "step": 18910
    },
    {
      "epoch": 0.05320869225296065,
      "grad_norm": 35.0,
      "learning_rate": 1.998448205721876e-05,
      "loss": 7.7282,
      "step": 18920
    },
    {
      "epoch": 0.05323681524040936,
      "grad_norm": 26.875,
      "learning_rate": 1.99844656525209e-05,
      "loss": 7.6988,
      "step": 18930
    },
    {
      "epoch": 0.05326493822785807,
      "grad_norm": 27.0,
      "learning_rate": 1.99844492391633e-05,
      "loss": 8.6295,
      "step": 18940
    },
    {
      "epoch": 0.05329306121530678,
      "grad_norm": 25.75,
      "learning_rate": 1.998443281714597e-05,
      "loss": 7.3244,
      "step": 18950
    },
    {
      "epoch": 0.053321184202755494,
      "grad_norm": 22.0,
      "learning_rate": 1.998441638646893e-05,
      "loss": 7.9429,
      "step": 18960
    },
    {
      "epoch": 0.0533493071902042,
      "grad_norm": 26.625,
      "learning_rate": 1.998439994713219e-05,
      "loss": 8.728,
      "step": 18970
    },
    {
      "epoch": 0.05337743017765291,
      "grad_norm": 25.125,
      "learning_rate": 1.998438349913576e-05,
      "loss": 8.4232,
      "step": 18980
    },
    {
      "epoch": 0.05340555316510162,
      "grad_norm": 28.625,
      "learning_rate": 1.9984367042479662e-05,
      "loss": 8.0412,
      "step": 18990
    },
    {
      "epoch": 0.05343367615255033,
      "grad_norm": 39.5,
      "learning_rate": 1.9984350577163908e-05,
      "loss": 8.3863,
      "step": 19000
    },
    {
      "epoch": 0.053461799139999044,
      "grad_norm": 29.875,
      "learning_rate": 1.998433410318851e-05,
      "loss": 7.8856,
      "step": 19010
    },
    {
      "epoch": 0.053489922127447756,
      "grad_norm": 39.5,
      "learning_rate": 1.9984317620553485e-05,
      "loss": 7.4815,
      "step": 19020
    },
    {
      "epoch": 0.05351804511489647,
      "grad_norm": 23.25,
      "learning_rate": 1.9984301129258844e-05,
      "loss": 8.1617,
      "step": 19030
    },
    {
      "epoch": 0.05354616810234518,
      "grad_norm": 26.0,
      "learning_rate": 1.9984284629304606e-05,
      "loss": 8.1392,
      "step": 19040
    },
    {
      "epoch": 0.053574291089793884,
      "grad_norm": 31.375,
      "learning_rate": 1.998426812069078e-05,
      "loss": 8.3843,
      "step": 19050
    },
    {
      "epoch": 0.053602414077242595,
      "grad_norm": 30.375,
      "learning_rate": 1.9984251603417385e-05,
      "loss": 7.8697,
      "step": 19060
    },
    {
      "epoch": 0.05363053706469131,
      "grad_norm": 26.5,
      "learning_rate": 1.9984235077484433e-05,
      "loss": 8.6182,
      "step": 19070
    },
    {
      "epoch": 0.05365866005214002,
      "grad_norm": 25.5,
      "learning_rate": 1.9984218542891937e-05,
      "loss": 7.8591,
      "step": 19080
    },
    {
      "epoch": 0.05368678303958873,
      "grad_norm": 32.75,
      "learning_rate": 1.9984201999639916e-05,
      "loss": 8.221,
      "step": 19090
    },
    {
      "epoch": 0.05371490602703744,
      "grad_norm": 34.0,
      "learning_rate": 1.9984185447728382e-05,
      "loss": 7.8907,
      "step": 19100
    },
    {
      "epoch": 0.05374302901448615,
      "grad_norm": 23.625,
      "learning_rate": 1.9984168887157345e-05,
      "loss": 8.3338,
      "step": 19110
    },
    {
      "epoch": 0.053771152001934865,
      "grad_norm": 23.5,
      "learning_rate": 1.9984152317926827e-05,
      "loss": 8.3301,
      "step": 19120
    },
    {
      "epoch": 0.05379927498938357,
      "grad_norm": 36.0,
      "learning_rate": 1.9984135740036833e-05,
      "loss": 7.7672,
      "step": 19130
    },
    {
      "epoch": 0.05382739797683228,
      "grad_norm": 27.0,
      "learning_rate": 1.9984119153487387e-05,
      "loss": 8.6214,
      "step": 19140
    },
    {
      "epoch": 0.05385552096428099,
      "grad_norm": 31.75,
      "learning_rate": 1.9984102558278497e-05,
      "loss": 7.5795,
      "step": 19150
    },
    {
      "epoch": 0.053883643951729704,
      "grad_norm": 29.0,
      "learning_rate": 1.998408595441018e-05,
      "loss": 7.3124,
      "step": 19160
    },
    {
      "epoch": 0.053911766939178415,
      "grad_norm": 26.25,
      "learning_rate": 1.9984069341882453e-05,
      "loss": 8.3848,
      "step": 19170
    },
    {
      "epoch": 0.05393988992662713,
      "grad_norm": 28.5,
      "learning_rate": 1.9984052720695323e-05,
      "loss": 7.653,
      "step": 19180
    },
    {
      "epoch": 0.05396801291407584,
      "grad_norm": 46.5,
      "learning_rate": 1.998403609084881e-05,
      "loss": 7.4101,
      "step": 19190
    },
    {
      "epoch": 0.05399613590152455,
      "grad_norm": 38.0,
      "learning_rate": 1.9984019452342932e-05,
      "loss": 7.9043,
      "step": 19200
    },
    {
      "epoch": 0.054024258888973255,
      "grad_norm": 42.0,
      "learning_rate": 1.9984002805177694e-05,
      "loss": 8.5598,
      "step": 19210
    },
    {
      "epoch": 0.054052381876421966,
      "grad_norm": 34.75,
      "learning_rate": 1.9983986149353114e-05,
      "loss": 8.9097,
      "step": 19220
    },
    {
      "epoch": 0.05408050486387068,
      "grad_norm": 23.125,
      "learning_rate": 1.998396948486921e-05,
      "loss": 8.4042,
      "step": 19230
    },
    {
      "epoch": 0.05410862785131939,
      "grad_norm": 27.0,
      "learning_rate": 1.9983952811725994e-05,
      "loss": 7.7505,
      "step": 19240
    },
    {
      "epoch": 0.0541367508387681,
      "grad_norm": 39.25,
      "learning_rate": 1.998393612992348e-05,
      "loss": 7.9688,
      "step": 19250
    },
    {
      "epoch": 0.05416487382621681,
      "grad_norm": 31.25,
      "learning_rate": 1.9983919439461685e-05,
      "loss": 7.7305,
      "step": 19260
    },
    {
      "epoch": 0.054192996813665524,
      "grad_norm": 30.5,
      "learning_rate": 1.9983902740340615e-05,
      "loss": 7.654,
      "step": 19270
    },
    {
      "epoch": 0.054221119801114236,
      "grad_norm": 34.75,
      "learning_rate": 1.9983886032560294e-05,
      "loss": 7.4599,
      "step": 19280
    },
    {
      "epoch": 0.05424924278856294,
      "grad_norm": 21.875,
      "learning_rate": 1.9983869316120738e-05,
      "loss": 8.2612,
      "step": 19290
    },
    {
      "epoch": 0.05427736577601165,
      "grad_norm": 32.0,
      "learning_rate": 1.9983852591021954e-05,
      "loss": 8.106,
      "step": 19300
    },
    {
      "epoch": 0.05430548876346036,
      "grad_norm": 30.875,
      "learning_rate": 1.998383585726396e-05,
      "loss": 8.6365,
      "step": 19310
    },
    {
      "epoch": 0.054333611750909075,
      "grad_norm": 24.875,
      "learning_rate": 1.9983819114846767e-05,
      "loss": 8.0632,
      "step": 19320
    },
    {
      "epoch": 0.054361734738357786,
      "grad_norm": 24.25,
      "learning_rate": 1.9983802363770394e-05,
      "loss": 7.5131,
      "step": 19330
    },
    {
      "epoch": 0.0543898577258065,
      "grad_norm": 32.75,
      "learning_rate": 1.9983785604034854e-05,
      "loss": 7.3817,
      "step": 19340
    },
    {
      "epoch": 0.05441798071325521,
      "grad_norm": 20.5,
      "learning_rate": 1.9983768835640163e-05,
      "loss": 7.6893,
      "step": 19350
    },
    {
      "epoch": 0.05444610370070392,
      "grad_norm": 41.25,
      "learning_rate": 1.9983752058586334e-05,
      "loss": 8.0993,
      "step": 19360
    },
    {
      "epoch": 0.054474226688152626,
      "grad_norm": 48.0,
      "learning_rate": 1.998373527287338e-05,
      "loss": 7.7485,
      "step": 19370
    },
    {
      "epoch": 0.05450234967560134,
      "grad_norm": 33.25,
      "learning_rate": 1.9983718478501318e-05,
      "loss": 7.7104,
      "step": 19380
    },
    {
      "epoch": 0.05453047266305005,
      "grad_norm": 43.5,
      "learning_rate": 1.998370167547016e-05,
      "loss": 8.0659,
      "step": 19390
    },
    {
      "epoch": 0.05455859565049876,
      "grad_norm": 24.25,
      "learning_rate": 1.9983684863779923e-05,
      "loss": 8.0713,
      "step": 19400
    },
    {
      "epoch": 0.05458671863794747,
      "grad_norm": 55.75,
      "learning_rate": 1.998366804343062e-05,
      "loss": 7.9499,
      "step": 19410
    },
    {
      "epoch": 0.054614841625396184,
      "grad_norm": 41.75,
      "learning_rate": 1.998365121442227e-05,
      "loss": 8.6007,
      "step": 19420
    },
    {
      "epoch": 0.054642964612844895,
      "grad_norm": 37.5,
      "learning_rate": 1.998363437675488e-05,
      "loss": 8.2319,
      "step": 19430
    },
    {
      "epoch": 0.05467108760029361,
      "grad_norm": 28.25,
      "learning_rate": 1.998361753042847e-05,
      "loss": 7.6701,
      "step": 19440
    },
    {
      "epoch": 0.05469921058774231,
      "grad_norm": 35.5,
      "learning_rate": 1.9983600675443055e-05,
      "loss": 8.1674,
      "step": 19450
    },
    {
      "epoch": 0.05472733357519102,
      "grad_norm": 24.375,
      "learning_rate": 1.9983583811798646e-05,
      "loss": 7.8227,
      "step": 19460
    },
    {
      "epoch": 0.054755456562639734,
      "grad_norm": 25.375,
      "learning_rate": 1.9983566939495264e-05,
      "loss": 8.0151,
      "step": 19470
    },
    {
      "epoch": 0.054783579550088446,
      "grad_norm": 27.5,
      "learning_rate": 1.9983550058532914e-05,
      "loss": 8.8682,
      "step": 19480
    },
    {
      "epoch": 0.05481170253753716,
      "grad_norm": 25.0,
      "learning_rate": 1.9983533168911617e-05,
      "loss": 7.8411,
      "step": 19490
    },
    {
      "epoch": 0.05483982552498587,
      "grad_norm": 47.0,
      "learning_rate": 1.9983516270631384e-05,
      "loss": 8.2282,
      "step": 19500
    },
    {
      "epoch": 0.05486794851243458,
      "grad_norm": 23.5,
      "learning_rate": 1.9983499363692237e-05,
      "loss": 7.628,
      "step": 19510
    },
    {
      "epoch": 0.05489607149988329,
      "grad_norm": 30.25,
      "learning_rate": 1.998348244809418e-05,
      "loss": 8.7042,
      "step": 19520
    },
    {
      "epoch": 0.054924194487332004,
      "grad_norm": 19.125,
      "learning_rate": 1.998346552383724e-05,
      "loss": 8.2588,
      "step": 19530
    },
    {
      "epoch": 0.05495231747478071,
      "grad_norm": 28.625,
      "learning_rate": 1.9983448590921423e-05,
      "loss": 8.3071,
      "step": 19540
    },
    {
      "epoch": 0.05498044046222942,
      "grad_norm": 34.75,
      "learning_rate": 1.9983431649346744e-05,
      "loss": 8.3151,
      "step": 19550
    },
    {
      "epoch": 0.05500856344967813,
      "grad_norm": 32.75,
      "learning_rate": 1.998341469911322e-05,
      "loss": 8.528,
      "step": 19560
    },
    {
      "epoch": 0.05503668643712684,
      "grad_norm": 27.5,
      "learning_rate": 1.9983397740220867e-05,
      "loss": 7.2263,
      "step": 19570
    },
    {
      "epoch": 0.055064809424575555,
      "grad_norm": 26.875,
      "learning_rate": 1.9983380772669697e-05,
      "loss": 7.0595,
      "step": 19580
    },
    {
      "epoch": 0.055092932412024266,
      "grad_norm": 32.75,
      "learning_rate": 1.998336379645972e-05,
      "loss": 7.4332,
      "step": 19590
    },
    {
      "epoch": 0.05512105539947298,
      "grad_norm": 29.875,
      "learning_rate": 1.9983346811590966e-05,
      "loss": 7.8423,
      "step": 19600
    },
    {
      "epoch": 0.05514917838692169,
      "grad_norm": 22.625,
      "learning_rate": 1.9983329818063436e-05,
      "loss": 7.1624,
      "step": 19610
    },
    {
      "epoch": 0.055177301374370394,
      "grad_norm": 28.625,
      "learning_rate": 1.9983312815877146e-05,
      "loss": 7.8154,
      "step": 19620
    },
    {
      "epoch": 0.055205424361819105,
      "grad_norm": 21.25,
      "learning_rate": 1.998329580503212e-05,
      "loss": 7.1226,
      "step": 19630
    },
    {
      "epoch": 0.05523354734926782,
      "grad_norm": 24.875,
      "learning_rate": 1.998327878552836e-05,
      "loss": 8.4382,
      "step": 19640
    },
    {
      "epoch": 0.05526167033671653,
      "grad_norm": 29.625,
      "learning_rate": 1.998326175736589e-05,
      "loss": 7.3483,
      "step": 19650
    },
    {
      "epoch": 0.05528979332416524,
      "grad_norm": 49.75,
      "learning_rate": 1.998324472054472e-05,
      "loss": 7.8986,
      "step": 19660
    },
    {
      "epoch": 0.05531791631161395,
      "grad_norm": 29.125,
      "learning_rate": 1.998322767506487e-05,
      "loss": 8.07,
      "step": 19670
    },
    {
      "epoch": 0.05534603929906266,
      "grad_norm": 31.625,
      "learning_rate": 1.9983210620926344e-05,
      "loss": 7.5803,
      "step": 19680
    },
    {
      "epoch": 0.055374162286511375,
      "grad_norm": 21.5,
      "learning_rate": 1.9983193558129172e-05,
      "loss": 7.7817,
      "step": 19690
    },
    {
      "epoch": 0.05540228527396008,
      "grad_norm": 33.25,
      "learning_rate": 1.998317648667336e-05,
      "loss": 7.8277,
      "step": 19700
    },
    {
      "epoch": 0.05543040826140879,
      "grad_norm": 29.25,
      "learning_rate": 1.998315940655892e-05,
      "loss": 8.0376,
      "step": 19710
    },
    {
      "epoch": 0.0554585312488575,
      "grad_norm": 31.75,
      "learning_rate": 1.9983142317785874e-05,
      "loss": 7.7938,
      "step": 19720
    },
    {
      "epoch": 0.055486654236306214,
      "grad_norm": 26.0,
      "learning_rate": 1.9983125220354232e-05,
      "loss": 8.2105,
      "step": 19730
    },
    {
      "epoch": 0.055514777223754926,
      "grad_norm": 41.0,
      "learning_rate": 1.998310811426401e-05,
      "loss": 8.4287,
      "step": 19740
    },
    {
      "epoch": 0.05554290021120364,
      "grad_norm": 31.5,
      "learning_rate": 1.9983090999515223e-05,
      "loss": 8.45,
      "step": 19750
    },
    {
      "epoch": 0.05557102319865235,
      "grad_norm": 46.25,
      "learning_rate": 1.9983073876107886e-05,
      "loss": 7.348,
      "step": 19760
    },
    {
      "epoch": 0.05559914618610106,
      "grad_norm": 39.75,
      "learning_rate": 1.9983056744042016e-05,
      "loss": 8.0936,
      "step": 19770
    },
    {
      "epoch": 0.055627269173549765,
      "grad_norm": 30.75,
      "learning_rate": 1.9983039603317626e-05,
      "loss": 8.2962,
      "step": 19780
    },
    {
      "epoch": 0.055655392160998476,
      "grad_norm": 22.0,
      "learning_rate": 1.9983022453934726e-05,
      "loss": 7.9339,
      "step": 19790
    },
    {
      "epoch": 0.05568351514844719,
      "grad_norm": 19.75,
      "learning_rate": 1.9983005295893337e-05,
      "loss": 7.7592,
      "step": 19800
    },
    {
      "epoch": 0.0557116381358959,
      "grad_norm": 28.0,
      "learning_rate": 1.9982988129193475e-05,
      "loss": 8.0519,
      "step": 19810
    },
    {
      "epoch": 0.05573976112334461,
      "grad_norm": 32.5,
      "learning_rate": 1.998297095383515e-05,
      "loss": 7.3724,
      "step": 19820
    },
    {
      "epoch": 0.05576788411079332,
      "grad_norm": 39.0,
      "learning_rate": 1.9982953769818382e-05,
      "loss": 7.8578,
      "step": 19830
    },
    {
      "epoch": 0.055796007098242034,
      "grad_norm": 28.375,
      "learning_rate": 1.9982936577143178e-05,
      "loss": 7.5796,
      "step": 19840
    },
    {
      "epoch": 0.055824130085690746,
      "grad_norm": 37.0,
      "learning_rate": 1.998291937580956e-05,
      "loss": 8.6568,
      "step": 19850
    },
    {
      "epoch": 0.05585225307313945,
      "grad_norm": 35.0,
      "learning_rate": 1.998290216581754e-05,
      "loss": 7.3343,
      "step": 19860
    },
    {
      "epoch": 0.05588037606058816,
      "grad_norm": 39.75,
      "learning_rate": 1.9982884947167137e-05,
      "loss": 6.9955,
      "step": 19870
    },
    {
      "epoch": 0.055908499048036873,
      "grad_norm": 58.75,
      "learning_rate": 1.998286771985836e-05,
      "loss": 7.7882,
      "step": 19880
    },
    {
      "epoch": 0.055936622035485585,
      "grad_norm": 26.125,
      "learning_rate": 1.998285048389123e-05,
      "loss": 8.1404,
      "step": 19890
    },
    {
      "epoch": 0.0559647450229343,
      "grad_norm": 36.5,
      "learning_rate": 1.9982833239265752e-05,
      "loss": 7.1027,
      "step": 19900
    },
    {
      "epoch": 0.05599286801038301,
      "grad_norm": 28.625,
      "learning_rate": 1.9982815985981952e-05,
      "loss": 7.8337,
      "step": 19910
    },
    {
      "epoch": 0.05602099099783172,
      "grad_norm": 26.625,
      "learning_rate": 1.998279872403984e-05,
      "loss": 7.9801,
      "step": 19920
    },
    {
      "epoch": 0.05604911398528043,
      "grad_norm": 35.0,
      "learning_rate": 1.998278145343943e-05,
      "loss": 7.5082,
      "step": 19930
    },
    {
      "epoch": 0.056077236972729136,
      "grad_norm": 31.25,
      "learning_rate": 1.998276417418074e-05,
      "loss": 7.9007,
      "step": 19940
    },
    {
      "epoch": 0.05610535996017785,
      "grad_norm": 38.0,
      "learning_rate": 1.9982746886263783e-05,
      "loss": 7.2864,
      "step": 19950
    },
    {
      "epoch": 0.05613348294762656,
      "grad_norm": 33.0,
      "learning_rate": 1.9982729589688573e-05,
      "loss": 7.6332,
      "step": 19960
    },
    {
      "epoch": 0.05616160593507527,
      "grad_norm": 32.0,
      "learning_rate": 1.9982712284455128e-05,
      "loss": 8.0916,
      "step": 19970
    },
    {
      "epoch": 0.05618972892252398,
      "grad_norm": 37.0,
      "learning_rate": 1.998269497056346e-05,
      "loss": 7.149,
      "step": 19980
    },
    {
      "epoch": 0.056217851909972694,
      "grad_norm": 27.125,
      "learning_rate": 1.9982677648013586e-05,
      "loss": 7.4871,
      "step": 19990
    },
    {
      "epoch": 0.056245974897421405,
      "grad_norm": 27.125,
      "learning_rate": 1.9982660316805523e-05,
      "loss": 7.9153,
      "step": 20000
    },
    {
      "epoch": 0.05627409788487012,
      "grad_norm": 41.5,
      "learning_rate": 1.9982642976939282e-05,
      "loss": 7.8897,
      "step": 20010
    },
    {
      "epoch": 0.05630222087231882,
      "grad_norm": 31.25,
      "learning_rate": 1.9982625628414878e-05,
      "loss": 7.7555,
      "step": 20020
    },
    {
      "epoch": 0.05633034385976753,
      "grad_norm": 50.5,
      "learning_rate": 1.998260827123233e-05,
      "loss": 8.5057,
      "step": 20030
    },
    {
      "epoch": 0.056358466847216244,
      "grad_norm": 23.5,
      "learning_rate": 1.9982590905391647e-05,
      "loss": 7.9091,
      "step": 20040
    },
    {
      "epoch": 0.056386589834664956,
      "grad_norm": 31.5,
      "learning_rate": 1.9982573530892852e-05,
      "loss": 7.8127,
      "step": 20050
    },
    {
      "epoch": 0.05641471282211367,
      "grad_norm": 28.75,
      "learning_rate": 1.9982556147735953e-05,
      "loss": 8.5595,
      "step": 20060
    },
    {
      "epoch": 0.05644283580956238,
      "grad_norm": 33.75,
      "learning_rate": 1.998253875592097e-05,
      "loss": 7.6451,
      "step": 20070
    },
    {
      "epoch": 0.05647095879701109,
      "grad_norm": 22.375,
      "learning_rate": 1.998252135544791e-05,
      "loss": 7.6805,
      "step": 20080
    },
    {
      "epoch": 0.0564990817844598,
      "grad_norm": 28.375,
      "learning_rate": 1.99825039463168e-05,
      "loss": 7.5887,
      "step": 20090
    },
    {
      "epoch": 0.05652720477190851,
      "grad_norm": 27.125,
      "learning_rate": 1.9982486528527646e-05,
      "loss": 8.1485,
      "step": 20100
    },
    {
      "epoch": 0.05655532775935722,
      "grad_norm": 27.75,
      "learning_rate": 1.998246910208047e-05,
      "loss": 7.5987,
      "step": 20110
    },
    {
      "epoch": 0.05658345074680593,
      "grad_norm": 44.25,
      "learning_rate": 1.998245166697528e-05,
      "loss": 8.5167,
      "step": 20120
    },
    {
      "epoch": 0.05661157373425464,
      "grad_norm": 37.0,
      "learning_rate": 1.9982434223212098e-05,
      "loss": 7.832,
      "step": 20130
    },
    {
      "epoch": 0.05663969672170335,
      "grad_norm": 25.375,
      "learning_rate": 1.998241677079093e-05,
      "loss": 8.0407,
      "step": 20140
    },
    {
      "epoch": 0.056667819709152065,
      "grad_norm": 34.5,
      "learning_rate": 1.99823993097118e-05,
      "loss": 7.4102,
      "step": 20150
    },
    {
      "epoch": 0.056695942696600776,
      "grad_norm": 27.75,
      "learning_rate": 1.998238183997472e-05,
      "loss": 7.7821,
      "step": 20160
    },
    {
      "epoch": 0.05672406568404949,
      "grad_norm": 37.0,
      "learning_rate": 1.998236436157971e-05,
      "loss": 8.1257,
      "step": 20170
    },
    {
      "epoch": 0.0567521886714982,
      "grad_norm": 46.5,
      "learning_rate": 1.9982346874526772e-05,
      "loss": 8.6723,
      "step": 20180
    },
    {
      "epoch": 0.056780311658946904,
      "grad_norm": 30.375,
      "learning_rate": 1.9982329378815932e-05,
      "loss": 8.0866,
      "step": 20190
    },
    {
      "epoch": 0.056808434646395616,
      "grad_norm": 25.25,
      "learning_rate": 1.9982311874447203e-05,
      "loss": 7.0809,
      "step": 20200
    },
    {
      "epoch": 0.05683655763384433,
      "grad_norm": 31.125,
      "learning_rate": 1.9982294361420597e-05,
      "loss": 8.0938,
      "step": 20210
    },
    {
      "epoch": 0.05686468062129304,
      "grad_norm": 30.375,
      "learning_rate": 1.9982276839736136e-05,
      "loss": 7.746,
      "step": 20220
    },
    {
      "epoch": 0.05689280360874175,
      "grad_norm": 42.25,
      "learning_rate": 1.998225930939383e-05,
      "loss": 7.6016,
      "step": 20230
    },
    {
      "epoch": 0.05692092659619046,
      "grad_norm": 50.75,
      "learning_rate": 1.9982241770393696e-05,
      "loss": 8.4288,
      "step": 20240
    },
    {
      "epoch": 0.05694904958363917,
      "grad_norm": 35.5,
      "learning_rate": 1.9982224222735745e-05,
      "loss": 7.7718,
      "step": 20250
    },
    {
      "epoch": 0.056977172571087885,
      "grad_norm": 25.0,
      "learning_rate": 1.998220666642e-05,
      "loss": 8.005,
      "step": 20260
    },
    {
      "epoch": 0.05700529555853659,
      "grad_norm": 28.375,
      "learning_rate": 1.9982189101446466e-05,
      "loss": 7.8985,
      "step": 20270
    },
    {
      "epoch": 0.0570334185459853,
      "grad_norm": 23.25,
      "learning_rate": 1.9982171527815168e-05,
      "loss": 8.2806,
      "step": 20280
    },
    {
      "epoch": 0.05706154153343401,
      "grad_norm": 24.25,
      "learning_rate": 1.998215394552612e-05,
      "loss": 7.9291,
      "step": 20290
    },
    {
      "epoch": 0.057089664520882724,
      "grad_norm": 42.5,
      "learning_rate": 1.998213635457933e-05,
      "loss": 7.793,
      "step": 20300
    },
    {
      "epoch": 0.057117787508331436,
      "grad_norm": 25.375,
      "learning_rate": 1.998211875497482e-05,
      "loss": 7.5269,
      "step": 20310
    },
    {
      "epoch": 0.05714591049578015,
      "grad_norm": 26.375,
      "learning_rate": 1.9982101146712604e-05,
      "loss": 7.5538,
      "step": 20320
    },
    {
      "epoch": 0.05717403348322886,
      "grad_norm": 26.25,
      "learning_rate": 1.9982083529792694e-05,
      "loss": 7.853,
      "step": 20330
    },
    {
      "epoch": 0.05720215647067757,
      "grad_norm": 25.5,
      "learning_rate": 1.9982065904215112e-05,
      "loss": 7.0636,
      "step": 20340
    },
    {
      "epoch": 0.057230279458126275,
      "grad_norm": 28.375,
      "learning_rate": 1.9982048269979867e-05,
      "loss": 7.5472,
      "step": 20350
    },
    {
      "epoch": 0.05725840244557499,
      "grad_norm": 22.75,
      "learning_rate": 1.9982030627086976e-05,
      "loss": 7.5599,
      "step": 20360
    },
    {
      "epoch": 0.0572865254330237,
      "grad_norm": 20.75,
      "learning_rate": 1.9982012975536454e-05,
      "loss": 7.6208,
      "step": 20370
    },
    {
      "epoch": 0.05731464842047241,
      "grad_norm": 24.625,
      "learning_rate": 1.998199531532832e-05,
      "loss": 7.8392,
      "step": 20380
    },
    {
      "epoch": 0.05734277140792112,
      "grad_norm": 24.0,
      "learning_rate": 1.9981977646462582e-05,
      "loss": 8.524,
      "step": 20390
    },
    {
      "epoch": 0.05737089439536983,
      "grad_norm": 24.5,
      "learning_rate": 1.998195996893926e-05,
      "loss": 7.6165,
      "step": 20400
    },
    {
      "epoch": 0.057399017382818544,
      "grad_norm": 32.0,
      "learning_rate": 1.9981942282758372e-05,
      "loss": 7.5146,
      "step": 20410
    },
    {
      "epoch": 0.057427140370267256,
      "grad_norm": 27.875,
      "learning_rate": 1.998192458791993e-05,
      "loss": 7.4979,
      "step": 20420
    },
    {
      "epoch": 0.05745526335771596,
      "grad_norm": 35.5,
      "learning_rate": 1.9981906884423945e-05,
      "loss": 8.1473,
      "step": 20430
    },
    {
      "epoch": 0.05748338634516467,
      "grad_norm": 45.5,
      "learning_rate": 1.998188917227044e-05,
      "loss": 8.1192,
      "step": 20440
    },
    {
      "epoch": 0.057511509332613384,
      "grad_norm": 24.375,
      "learning_rate": 1.9981871451459426e-05,
      "loss": 8.0211,
      "step": 20450
    },
    {
      "epoch": 0.057539632320062095,
      "grad_norm": 37.75,
      "learning_rate": 1.9981853721990922e-05,
      "loss": 7.8262,
      "step": 20460
    },
    {
      "epoch": 0.05756775530751081,
      "grad_norm": 24.75,
      "learning_rate": 1.9981835983864938e-05,
      "loss": 7.8506,
      "step": 20470
    },
    {
      "epoch": 0.05759587829495952,
      "grad_norm": 22.875,
      "learning_rate": 1.9981818237081495e-05,
      "loss": 7.6987,
      "step": 20480
    },
    {
      "epoch": 0.05762400128240823,
      "grad_norm": 21.875,
      "learning_rate": 1.998180048164061e-05,
      "loss": 8.0511,
      "step": 20490
    },
    {
      "epoch": 0.05765212426985694,
      "grad_norm": 24.375,
      "learning_rate": 1.9981782717542287e-05,
      "loss": 7.8948,
      "step": 20500
    },
    {
      "epoch": 0.057680247257305646,
      "grad_norm": 38.75,
      "learning_rate": 1.998176494478655e-05,
      "loss": 7.9494,
      "step": 20510
    },
    {
      "epoch": 0.05770837024475436,
      "grad_norm": 45.5,
      "learning_rate": 1.9981747163373415e-05,
      "loss": 7.9632,
      "step": 20520
    },
    {
      "epoch": 0.05773649323220307,
      "grad_norm": 42.25,
      "learning_rate": 1.9981729373302895e-05,
      "loss": 7.441,
      "step": 20530
    },
    {
      "epoch": 0.05776461621965178,
      "grad_norm": 35.0,
      "learning_rate": 1.9981711574575007e-05,
      "loss": 8.0552,
      "step": 20540
    },
    {
      "epoch": 0.05779273920710049,
      "grad_norm": 22.125,
      "learning_rate": 1.998169376718976e-05,
      "loss": 7.7305,
      "step": 20550
    },
    {
      "epoch": 0.057820862194549204,
      "grad_norm": 23.75,
      "learning_rate": 1.998167595114718e-05,
      "loss": 7.6503,
      "step": 20560
    },
    {
      "epoch": 0.057848985181997915,
      "grad_norm": 29.875,
      "learning_rate": 1.9981658126447278e-05,
      "loss": 8.7395,
      "step": 20570
    },
    {
      "epoch": 0.05787710816944663,
      "grad_norm": 33.0,
      "learning_rate": 1.9981640293090064e-05,
      "loss": 7.163,
      "step": 20580
    },
    {
      "epoch": 0.05790523115689533,
      "grad_norm": 28.125,
      "learning_rate": 1.9981622451075563e-05,
      "loss": 8.8052,
      "step": 20590
    },
    {
      "epoch": 0.05793335414434404,
      "grad_norm": 30.5,
      "learning_rate": 1.9981604600403783e-05,
      "loss": 8.1743,
      "step": 20600
    },
    {
      "epoch": 0.057961477131792755,
      "grad_norm": 22.0,
      "learning_rate": 1.9981586741074747e-05,
      "loss": 7.8118,
      "step": 20610
    },
    {
      "epoch": 0.057989600119241466,
      "grad_norm": 27.25,
      "learning_rate": 1.998156887308846e-05,
      "loss": 8.459,
      "step": 20620
    },
    {
      "epoch": 0.05801772310669018,
      "grad_norm": 32.25,
      "learning_rate": 1.9981550996444943e-05,
      "loss": 7.7447,
      "step": 20630
    },
    {
      "epoch": 0.05804584609413889,
      "grad_norm": 29.125,
      "learning_rate": 1.9981533111144213e-05,
      "loss": 7.8542,
      "step": 20640
    },
    {
      "epoch": 0.0580739690815876,
      "grad_norm": 29.25,
      "learning_rate": 1.9981515217186284e-05,
      "loss": 7.7968,
      "step": 20650
    },
    {
      "epoch": 0.05810209206903631,
      "grad_norm": 29.875,
      "learning_rate": 1.998149731457117e-05,
      "loss": 8.4871,
      "step": 20660
    },
    {
      "epoch": 0.05813021505648502,
      "grad_norm": 22.25,
      "learning_rate": 1.9981479403298892e-05,
      "loss": 6.876,
      "step": 20670
    },
    {
      "epoch": 0.05815833804393373,
      "grad_norm": 29.5,
      "learning_rate": 1.998146148336946e-05,
      "loss": 8.569,
      "step": 20680
    },
    {
      "epoch": 0.05818646103138244,
      "grad_norm": 27.875,
      "learning_rate": 1.998144355478289e-05,
      "loss": 8.5098,
      "step": 20690
    },
    {
      "epoch": 0.05821458401883115,
      "grad_norm": 28.625,
      "learning_rate": 1.9981425617539202e-05,
      "loss": 8.0638,
      "step": 20700
    },
    {
      "epoch": 0.05824270700627986,
      "grad_norm": 53.5,
      "learning_rate": 1.9981407671638405e-05,
      "loss": 7.7695,
      "step": 20710
    },
    {
      "epoch": 0.058270829993728575,
      "grad_norm": 31.5,
      "learning_rate": 1.998138971708052e-05,
      "loss": 7.7284,
      "step": 20720
    },
    {
      "epoch": 0.058298952981177286,
      "grad_norm": 34.25,
      "learning_rate": 1.998137175386556e-05,
      "loss": 8.0025,
      "step": 20730
    },
    {
      "epoch": 0.058327075968626,
      "grad_norm": 25.0,
      "learning_rate": 1.998135378199354e-05,
      "loss": 8.1154,
      "step": 20740
    },
    {
      "epoch": 0.0583551989560747,
      "grad_norm": 33.25,
      "learning_rate": 1.9981335801464476e-05,
      "loss": 7.317,
      "step": 20750
    },
    {
      "epoch": 0.058383321943523414,
      "grad_norm": 41.5,
      "learning_rate": 1.9981317812278387e-05,
      "loss": 8.2104,
      "step": 20760
    },
    {
      "epoch": 0.058411444930972126,
      "grad_norm": 28.25,
      "learning_rate": 1.9981299814435286e-05,
      "loss": 8.4275,
      "step": 20770
    },
    {
      "epoch": 0.05843956791842084,
      "grad_norm": 34.75,
      "learning_rate": 1.9981281807935183e-05,
      "loss": 8.0986,
      "step": 20780
    },
    {
      "epoch": 0.05846769090586955,
      "grad_norm": 57.25,
      "learning_rate": 1.998126379277811e-05,
      "loss": 7.777,
      "step": 20790
    },
    {
      "epoch": 0.05849581389331826,
      "grad_norm": 27.0,
      "learning_rate": 1.9981245768964063e-05,
      "loss": 7.597,
      "step": 20800
    },
    {
      "epoch": 0.05852393688076697,
      "grad_norm": 30.125,
      "learning_rate": 1.9981227736493066e-05,
      "loss": 7.7218,
      "step": 20810
    },
    {
      "epoch": 0.05855205986821568,
      "grad_norm": 27.875,
      "learning_rate": 1.9981209695365134e-05,
      "loss": 7.4805,
      "step": 20820
    },
    {
      "epoch": 0.058580182855664395,
      "grad_norm": 24.75,
      "learning_rate": 1.9981191645580292e-05,
      "loss": 7.2145,
      "step": 20830
    },
    {
      "epoch": 0.0586083058431131,
      "grad_norm": 26.5,
      "learning_rate": 1.998117358713854e-05,
      "loss": 7.1732,
      "step": 20840
    },
    {
      "epoch": 0.05863642883056181,
      "grad_norm": 22.875,
      "learning_rate": 1.9981155520039904e-05,
      "loss": 8.2664,
      "step": 20850
    },
    {
      "epoch": 0.05866455181801052,
      "grad_norm": 32.75,
      "learning_rate": 1.9981137444284395e-05,
      "loss": 7.7409,
      "step": 20860
    },
    {
      "epoch": 0.058692674805459234,
      "grad_norm": 36.0,
      "learning_rate": 1.9981119359872034e-05,
      "loss": 8.3165,
      "step": 20870
    },
    {
      "epoch": 0.058720797792907946,
      "grad_norm": 27.5,
      "learning_rate": 1.9981101266802826e-05,
      "loss": 8.2613,
      "step": 20880
    },
    {
      "epoch": 0.05874892078035666,
      "grad_norm": 29.5,
      "learning_rate": 1.99810831650768e-05,
      "loss": 8.018,
      "step": 20890
    },
    {
      "epoch": 0.05877704376780537,
      "grad_norm": 43.5,
      "learning_rate": 1.9981065054693963e-05,
      "loss": 7.8357,
      "step": 20900
    },
    {
      "epoch": 0.05880516675525408,
      "grad_norm": 49.5,
      "learning_rate": 1.9981046935654335e-05,
      "loss": 7.8559,
      "step": 20910
    },
    {
      "epoch": 0.058833289742702785,
      "grad_norm": 30.0,
      "learning_rate": 1.9981028807957925e-05,
      "loss": 8.2599,
      "step": 20920
    },
    {
      "epoch": 0.0588614127301515,
      "grad_norm": 24.375,
      "learning_rate": 1.9981010671604758e-05,
      "loss": 7.7116,
      "step": 20930
    },
    {
      "epoch": 0.05888953571760021,
      "grad_norm": 25.125,
      "learning_rate": 1.9980992526594846e-05,
      "loss": 8.4724,
      "step": 20940
    },
    {
      "epoch": 0.05891765870504892,
      "grad_norm": 48.5,
      "learning_rate": 1.99809743729282e-05,
      "loss": 7.5649,
      "step": 20950
    },
    {
      "epoch": 0.05894578169249763,
      "grad_norm": 21.75,
      "learning_rate": 1.998095621060484e-05,
      "loss": 7.1761,
      "step": 20960
    },
    {
      "epoch": 0.05897390467994634,
      "grad_norm": 25.25,
      "learning_rate": 1.9980938039624785e-05,
      "loss": 8.121,
      "step": 20970
    },
    {
      "epoch": 0.059002027667395054,
      "grad_norm": 21.75,
      "learning_rate": 1.9980919859988044e-05,
      "loss": 7.9191,
      "step": 20980
    },
    {
      "epoch": 0.059030150654843766,
      "grad_norm": 32.75,
      "learning_rate": 1.9980901671694637e-05,
      "loss": 8.073,
      "step": 20990
    },
    {
      "epoch": 0.05905827364229247,
      "grad_norm": 24.875,
      "learning_rate": 1.998088347474458e-05,
      "loss": 8.4625,
      "step": 21000
    },
    {
      "epoch": 0.05908639662974118,
      "grad_norm": 37.25,
      "learning_rate": 1.9980865269137886e-05,
      "loss": 8.4095,
      "step": 21010
    },
    {
      "epoch": 0.059114519617189894,
      "grad_norm": 27.875,
      "learning_rate": 1.9980847054874574e-05,
      "loss": 8.7108,
      "step": 21020
    },
    {
      "epoch": 0.059142642604638605,
      "grad_norm": 27.0,
      "learning_rate": 1.9980828831954656e-05,
      "loss": 7.7098,
      "step": 21030
    },
    {
      "epoch": 0.05917076559208732,
      "grad_norm": 29.5,
      "learning_rate": 1.998081060037815e-05,
      "loss": 8.2652,
      "step": 21040
    },
    {
      "epoch": 0.05919888857953603,
      "grad_norm": 39.5,
      "learning_rate": 1.9980792360145075e-05,
      "loss": 8.1287,
      "step": 21050
    },
    {
      "epoch": 0.05922701156698474,
      "grad_norm": 34.75,
      "learning_rate": 1.9980774111255442e-05,
      "loss": 8.1299,
      "step": 21060
    },
    {
      "epoch": 0.05925513455443345,
      "grad_norm": 42.75,
      "learning_rate": 1.9980755853709265e-05,
      "loss": 7.7294,
      "step": 21070
    },
    {
      "epoch": 0.059283257541882156,
      "grad_norm": 24.375,
      "learning_rate": 1.998073758750657e-05,
      "loss": 7.7457,
      "step": 21080
    },
    {
      "epoch": 0.05931138052933087,
      "grad_norm": 53.75,
      "learning_rate": 1.998071931264736e-05,
      "loss": 7.8513,
      "step": 21090
    },
    {
      "epoch": 0.05933950351677958,
      "grad_norm": 25.375,
      "learning_rate": 1.9980701029131658e-05,
      "loss": 7.5523,
      "step": 21100
    },
    {
      "epoch": 0.05936762650422829,
      "grad_norm": 36.0,
      "learning_rate": 1.998068273695948e-05,
      "loss": 7.7128,
      "step": 21110
    },
    {
      "epoch": 0.059395749491677,
      "grad_norm": 26.75,
      "learning_rate": 1.9980664436130842e-05,
      "loss": 7.7281,
      "step": 21120
    },
    {
      "epoch": 0.059423872479125714,
      "grad_norm": 37.5,
      "learning_rate": 1.998064612664576e-05,
      "loss": 8.2988,
      "step": 21130
    },
    {
      "epoch": 0.059451995466574425,
      "grad_norm": 23.25,
      "learning_rate": 1.9980627808504244e-05,
      "loss": 8.0079,
      "step": 21140
    },
    {
      "epoch": 0.05948011845402314,
      "grad_norm": 32.25,
      "learning_rate": 1.9980609481706317e-05,
      "loss": 7.5455,
      "step": 21150
    },
    {
      "epoch": 0.05950824144147184,
      "grad_norm": 27.125,
      "learning_rate": 1.9980591146251994e-05,
      "loss": 8.1719,
      "step": 21160
    },
    {
      "epoch": 0.05953636442892055,
      "grad_norm": 42.25,
      "learning_rate": 1.998057280214128e-05,
      "loss": 8.4557,
      "step": 21170
    },
    {
      "epoch": 0.059564487416369265,
      "grad_norm": 26.625,
      "learning_rate": 1.998055444937421e-05,
      "loss": 8.1354,
      "step": 21180
    },
    {
      "epoch": 0.059592610403817976,
      "grad_norm": 26.75,
      "learning_rate": 1.9980536087950787e-05,
      "loss": 7.8391,
      "step": 21190
    },
    {
      "epoch": 0.05962073339126669,
      "grad_norm": 43.25,
      "learning_rate": 1.998051771787103e-05,
      "loss": 8.4365,
      "step": 21200
    },
    {
      "epoch": 0.0596488563787154,
      "grad_norm": 30.25,
      "learning_rate": 1.9980499339134953e-05,
      "loss": 7.9535,
      "step": 21210
    },
    {
      "epoch": 0.05967697936616411,
      "grad_norm": 32.75,
      "learning_rate": 1.9980480951742576e-05,
      "loss": 7.9947,
      "step": 21220
    },
    {
      "epoch": 0.05970510235361282,
      "grad_norm": 22.25,
      "learning_rate": 1.998046255569391e-05,
      "loss": 7.6822,
      "step": 21230
    },
    {
      "epoch": 0.05973322534106153,
      "grad_norm": 21.25,
      "learning_rate": 1.998044415098898e-05,
      "loss": 7.849,
      "step": 21240
    },
    {
      "epoch": 0.05976134832851024,
      "grad_norm": 28.375,
      "learning_rate": 1.9980425737627788e-05,
      "loss": 7.8298,
      "step": 21250
    },
    {
      "epoch": 0.05978947131595895,
      "grad_norm": 26.0,
      "learning_rate": 1.998040731561036e-05,
      "loss": 8.5821,
      "step": 21260
    },
    {
      "epoch": 0.05981759430340766,
      "grad_norm": 51.75,
      "learning_rate": 1.998038888493671e-05,
      "loss": 7.5051,
      "step": 21270
    },
    {
      "epoch": 0.05984571729085637,
      "grad_norm": 24.0,
      "learning_rate": 1.9980370445606853e-05,
      "loss": 7.2337,
      "step": 21280
    },
    {
      "epoch": 0.059873840278305085,
      "grad_norm": 28.125,
      "learning_rate": 1.9980351997620804e-05,
      "loss": 7.8206,
      "step": 21290
    },
    {
      "epoch": 0.059901963265753796,
      "grad_norm": 36.5,
      "learning_rate": 1.9980333540978587e-05,
      "loss": 8.2115,
      "step": 21300
    },
    {
      "epoch": 0.05993008625320251,
      "grad_norm": 39.0,
      "learning_rate": 1.998031507568021e-05,
      "loss": 8.2074,
      "step": 21310
    },
    {
      "epoch": 0.05995820924065121,
      "grad_norm": 26.0,
      "learning_rate": 1.9980296601725685e-05,
      "loss": 7.7596,
      "step": 21320
    },
    {
      "epoch": 0.059986332228099924,
      "grad_norm": 24.0,
      "learning_rate": 1.9980278119115034e-05,
      "loss": 7.5902,
      "step": 21330
    },
    {
      "epoch": 0.060014455215548636,
      "grad_norm": 23.125,
      "learning_rate": 1.9980259627848278e-05,
      "loss": 7.5393,
      "step": 21340
    },
    {
      "epoch": 0.06004257820299735,
      "grad_norm": 33.5,
      "learning_rate": 1.9980241127925425e-05,
      "loss": 8.0611,
      "step": 21350
    },
    {
      "epoch": 0.06007070119044606,
      "grad_norm": 32.0,
      "learning_rate": 1.9980222619346492e-05,
      "loss": 8.417,
      "step": 21360
    },
    {
      "epoch": 0.06009882417789477,
      "grad_norm": 24.5,
      "learning_rate": 1.99802041021115e-05,
      "loss": 7.9667,
      "step": 21370
    },
    {
      "epoch": 0.06012694716534348,
      "grad_norm": 53.25,
      "learning_rate": 1.9980185576220455e-05,
      "loss": 7.9464,
      "step": 21380
    },
    {
      "epoch": 0.060155070152792194,
      "grad_norm": 31.25,
      "learning_rate": 1.9980167041673386e-05,
      "loss": 7.2817,
      "step": 21390
    },
    {
      "epoch": 0.0601831931402409,
      "grad_norm": 30.875,
      "learning_rate": 1.99801484984703e-05,
      "loss": 7.3288,
      "step": 21400
    },
    {
      "epoch": 0.06021131612768961,
      "grad_norm": 30.625,
      "learning_rate": 1.9980129946611223e-05,
      "loss": 7.5807,
      "step": 21410
    },
    {
      "epoch": 0.06023943911513832,
      "grad_norm": 31.75,
      "learning_rate": 1.9980111386096156e-05,
      "loss": 7.8464,
      "step": 21420
    },
    {
      "epoch": 0.06026756210258703,
      "grad_norm": 45.5,
      "learning_rate": 1.9980092816925127e-05,
      "loss": 7.4515,
      "step": 21430
    },
    {
      "epoch": 0.060295685090035744,
      "grad_norm": 29.375,
      "learning_rate": 1.9980074239098147e-05,
      "loss": 8.4607,
      "step": 21440
    },
    {
      "epoch": 0.060323808077484456,
      "grad_norm": 17.875,
      "learning_rate": 1.9980055652615236e-05,
      "loss": 8.4047,
      "step": 21450
    },
    {
      "epoch": 0.06035193106493317,
      "grad_norm": 29.375,
      "learning_rate": 1.9980037057476406e-05,
      "loss": 7.3221,
      "step": 21460
    },
    {
      "epoch": 0.06038005405238188,
      "grad_norm": 30.875,
      "learning_rate": 1.9980018453681673e-05,
      "loss": 7.9673,
      "step": 21470
    },
    {
      "epoch": 0.060408177039830584,
      "grad_norm": 23.625,
      "learning_rate": 1.9979999841231055e-05,
      "loss": 8.3618,
      "step": 21480
    },
    {
      "epoch": 0.060436300027279295,
      "grad_norm": 32.25,
      "learning_rate": 1.997998122012457e-05,
      "loss": 7.7866,
      "step": 21490
    },
    {
      "epoch": 0.06046442301472801,
      "grad_norm": 33.75,
      "learning_rate": 1.9979962590362233e-05,
      "loss": 8.4761,
      "step": 21500
    },
    {
      "epoch": 0.06049254600217672,
      "grad_norm": 30.0,
      "learning_rate": 1.9979943951944055e-05,
      "loss": 8.1006,
      "step": 21510
    },
    {
      "epoch": 0.06052066898962543,
      "grad_norm": 31.125,
      "learning_rate": 1.997992530487006e-05,
      "loss": 8.0258,
      "step": 21520
    },
    {
      "epoch": 0.06054879197707414,
      "grad_norm": 27.5,
      "learning_rate": 1.9979906649140258e-05,
      "loss": 7.5496,
      "step": 21530
    },
    {
      "epoch": 0.06057691496452285,
      "grad_norm": 27.125,
      "learning_rate": 1.9979887984754673e-05,
      "loss": 8.1998,
      "step": 21540
    },
    {
      "epoch": 0.060605037951971565,
      "grad_norm": 29.375,
      "learning_rate": 1.997986931171331e-05,
      "loss": 8.098,
      "step": 21550
    },
    {
      "epoch": 0.060633160939420276,
      "grad_norm": 21.75,
      "learning_rate": 1.9979850630016192e-05,
      "loss": 8.0011,
      "step": 21560
    },
    {
      "epoch": 0.06066128392686898,
      "grad_norm": 27.25,
      "learning_rate": 1.9979831939663338e-05,
      "loss": 7.7268,
      "step": 21570
    },
    {
      "epoch": 0.06068940691431769,
      "grad_norm": 28.75,
      "learning_rate": 1.9979813240654756e-05,
      "loss": 7.9227,
      "step": 21580
    },
    {
      "epoch": 0.060717529901766404,
      "grad_norm": 32.25,
      "learning_rate": 1.9979794532990473e-05,
      "loss": 8.003,
      "step": 21590
    },
    {
      "epoch": 0.060745652889215115,
      "grad_norm": 32.25,
      "learning_rate": 1.9979775816670495e-05,
      "loss": 7.8461,
      "step": 21600
    },
    {
      "epoch": 0.06077377587666383,
      "grad_norm": 48.0,
      "learning_rate": 1.997975709169484e-05,
      "loss": 8.3639,
      "step": 21610
    },
    {
      "epoch": 0.06080189886411254,
      "grad_norm": 24.375,
      "learning_rate": 1.9979738358063532e-05,
      "loss": 6.7699,
      "step": 21620
    },
    {
      "epoch": 0.06083002185156125,
      "grad_norm": 33.25,
      "learning_rate": 1.9979719615776578e-05,
      "loss": 8.0705,
      "step": 21630
    },
    {
      "epoch": 0.06085814483900996,
      "grad_norm": 77.0,
      "learning_rate": 1.9979700864833996e-05,
      "loss": 7.6375,
      "step": 21640
    },
    {
      "epoch": 0.060886267826458666,
      "grad_norm": 23.875,
      "learning_rate": 1.997968210523581e-05,
      "loss": 8.3532,
      "step": 21650
    },
    {
      "epoch": 0.06091439081390738,
      "grad_norm": 41.75,
      "learning_rate": 1.997966333698203e-05,
      "loss": 8.5647,
      "step": 21660
    },
    {
      "epoch": 0.06094251380135609,
      "grad_norm": 51.0,
      "learning_rate": 1.9979644560072667e-05,
      "loss": 7.767,
      "step": 21670
    },
    {
      "epoch": 0.0609706367888048,
      "grad_norm": 45.5,
      "learning_rate": 1.9979625774507746e-05,
      "loss": 8.4055,
      "step": 21680
    },
    {
      "epoch": 0.06099875977625351,
      "grad_norm": 22.875,
      "learning_rate": 1.9979606980287284e-05,
      "loss": 8.0404,
      "step": 21690
    },
    {
      "epoch": 0.061026882763702224,
      "grad_norm": 61.25,
      "learning_rate": 1.997958817741129e-05,
      "loss": 8.2913,
      "step": 21700
    },
    {
      "epoch": 0.061055005751150936,
      "grad_norm": 40.25,
      "learning_rate": 1.9979569365879784e-05,
      "loss": 7.6914,
      "step": 21710
    },
    {
      "epoch": 0.06108312873859965,
      "grad_norm": 29.0,
      "learning_rate": 1.9979550545692785e-05,
      "loss": 8.1403,
      "step": 21720
    },
    {
      "epoch": 0.06111125172604835,
      "grad_norm": 31.25,
      "learning_rate": 1.9979531716850304e-05,
      "loss": 7.5098,
      "step": 21730
    },
    {
      "epoch": 0.06113937471349706,
      "grad_norm": 27.25,
      "learning_rate": 1.997951287935236e-05,
      "loss": 7.9102,
      "step": 21740
    },
    {
      "epoch": 0.061167497700945775,
      "grad_norm": 23.0,
      "learning_rate": 1.997949403319897e-05,
      "loss": 7.7421,
      "step": 21750
    },
    {
      "epoch": 0.061195620688394486,
      "grad_norm": 25.125,
      "learning_rate": 1.9979475178390152e-05,
      "loss": 7.9892,
      "step": 21760
    },
    {
      "epoch": 0.0612237436758432,
      "grad_norm": 28.75,
      "learning_rate": 1.997945631492592e-05,
      "loss": 8.8751,
      "step": 21770
    },
    {
      "epoch": 0.06125186666329191,
      "grad_norm": 28.875,
      "learning_rate": 1.9979437442806285e-05,
      "loss": 8.0082,
      "step": 21780
    },
    {
      "epoch": 0.06127998965074062,
      "grad_norm": 41.0,
      "learning_rate": 1.9979418562031273e-05,
      "loss": 7.0084,
      "step": 21790
    },
    {
      "epoch": 0.06130811263818933,
      "grad_norm": 23.125,
      "learning_rate": 1.9979399672600896e-05,
      "loss": 7.8341,
      "step": 21800
    },
    {
      "epoch": 0.06133623562563804,
      "grad_norm": 25.5,
      "learning_rate": 1.997938077451517e-05,
      "loss": 7.8657,
      "step": 21810
    },
    {
      "epoch": 0.06136435861308675,
      "grad_norm": 26.5,
      "learning_rate": 1.9979361867774108e-05,
      "loss": 6.9899,
      "step": 21820
    },
    {
      "epoch": 0.06139248160053546,
      "grad_norm": 22.125,
      "learning_rate": 1.9979342952377735e-05,
      "loss": 7.7231,
      "step": 21830
    },
    {
      "epoch": 0.06142060458798417,
      "grad_norm": 40.0,
      "learning_rate": 1.9979324028326062e-05,
      "loss": 8.3969,
      "step": 21840
    },
    {
      "epoch": 0.061448727575432883,
      "grad_norm": 27.0,
      "learning_rate": 1.9979305095619107e-05,
      "loss": 7.9266,
      "step": 21850
    },
    {
      "epoch": 0.061476850562881595,
      "grad_norm": 21.375,
      "learning_rate": 1.9979286154256882e-05,
      "loss": 8.3608,
      "step": 21860
    },
    {
      "epoch": 0.06150497355033031,
      "grad_norm": 82.5,
      "learning_rate": 1.997926720423941e-05,
      "loss": 7.8816,
      "step": 21870
    },
    {
      "epoch": 0.06153309653777902,
      "grad_norm": 25.625,
      "learning_rate": 1.9979248245566704e-05,
      "loss": 7.2015,
      "step": 21880
    },
    {
      "epoch": 0.06156121952522772,
      "grad_norm": 42.0,
      "learning_rate": 1.997922927823878e-05,
      "loss": 8.785,
      "step": 21890
    },
    {
      "epoch": 0.061589342512676434,
      "grad_norm": 17.0,
      "learning_rate": 1.9979210302255654e-05,
      "loss": 7.5792,
      "step": 21900
    },
    {
      "epoch": 0.061617465500125146,
      "grad_norm": 40.25,
      "learning_rate": 1.9979191317617347e-05,
      "loss": 8.1866,
      "step": 21910
    },
    {
      "epoch": 0.06164558848757386,
      "grad_norm": 37.0,
      "learning_rate": 1.997917232432387e-05,
      "loss": 7.8444,
      "step": 21920
    },
    {
      "epoch": 0.06167371147502257,
      "grad_norm": 33.25,
      "learning_rate": 1.9979153322375246e-05,
      "loss": 8.1389,
      "step": 21930
    },
    {
      "epoch": 0.06170183446247128,
      "grad_norm": 33.25,
      "learning_rate": 1.9979134311771482e-05,
      "loss": 7.9676,
      "step": 21940
    },
    {
      "epoch": 0.06172995744991999,
      "grad_norm": 28.0,
      "learning_rate": 1.99791152925126e-05,
      "loss": 7.6079,
      "step": 21950
    },
    {
      "epoch": 0.061758080437368704,
      "grad_norm": 51.75,
      "learning_rate": 1.9979096264598618e-05,
      "loss": 8.2742,
      "step": 21960
    },
    {
      "epoch": 0.06178620342481741,
      "grad_norm": 34.75,
      "learning_rate": 1.997907722802955e-05,
      "loss": 8.3118,
      "step": 21970
    },
    {
      "epoch": 0.06181432641226612,
      "grad_norm": 24.25,
      "learning_rate": 1.9979058182805416e-05,
      "loss": 7.8057,
      "step": 21980
    },
    {
      "epoch": 0.06184244939971483,
      "grad_norm": 25.25,
      "learning_rate": 1.9979039128926224e-05,
      "loss": 7.8729,
      "step": 21990
    },
    {
      "epoch": 0.06187057238716354,
      "grad_norm": 31.25,
      "learning_rate": 1.9979020066392e-05,
      "loss": 7.5631,
      "step": 22000
    },
    {
      "epoch": 0.061898695374612255,
      "grad_norm": 33.25,
      "learning_rate": 1.9979000995202757e-05,
      "loss": 8.5476,
      "step": 22010
    },
    {
      "epoch": 0.061926818362060966,
      "grad_norm": 29.25,
      "learning_rate": 1.9978981915358512e-05,
      "loss": 8.0555,
      "step": 22020
    },
    {
      "epoch": 0.06195494134950968,
      "grad_norm": 50.5,
      "learning_rate": 1.9978962826859278e-05,
      "loss": 7.8041,
      "step": 22030
    },
    {
      "epoch": 0.06198306433695839,
      "grad_norm": 31.625,
      "learning_rate": 1.9978943729705074e-05,
      "loss": 8.1923,
      "step": 22040
    },
    {
      "epoch": 0.062011187324407094,
      "grad_norm": 43.0,
      "learning_rate": 1.997892462389592e-05,
      "loss": 8.066,
      "step": 22050
    },
    {
      "epoch": 0.062039310311855805,
      "grad_norm": 42.0,
      "learning_rate": 1.9978905509431828e-05,
      "loss": 8.7097,
      "step": 22060
    },
    {
      "epoch": 0.06206743329930452,
      "grad_norm": 40.0,
      "learning_rate": 1.9978886386312816e-05,
      "loss": 8.1241,
      "step": 22070
    },
    {
      "epoch": 0.06209555628675323,
      "grad_norm": 28.25,
      "learning_rate": 1.99788672545389e-05,
      "loss": 6.6821,
      "step": 22080
    },
    {
      "epoch": 0.06212367927420194,
      "grad_norm": 38.25,
      "learning_rate": 1.9978848114110097e-05,
      "loss": 7.7368,
      "step": 22090
    },
    {
      "epoch": 0.06215180226165065,
      "grad_norm": 25.625,
      "learning_rate": 1.9978828965026428e-05,
      "loss": 8.2003,
      "step": 22100
    },
    {
      "epoch": 0.06217992524909936,
      "grad_norm": 21.125,
      "learning_rate": 1.99788098072879e-05,
      "loss": 7.9211,
      "step": 22110
    },
    {
      "epoch": 0.062208048236548075,
      "grad_norm": 27.75,
      "learning_rate": 1.997879064089454e-05,
      "loss": 8.2872,
      "step": 22120
    },
    {
      "epoch": 0.06223617122399678,
      "grad_norm": 25.375,
      "learning_rate": 1.9978771465846356e-05,
      "loss": 7.5588,
      "step": 22130
    },
    {
      "epoch": 0.06226429421144549,
      "grad_norm": 40.5,
      "learning_rate": 1.997875228214337e-05,
      "loss": 8.0747,
      "step": 22140
    },
    {
      "epoch": 0.0622924171988942,
      "grad_norm": 41.75,
      "learning_rate": 1.9978733089785597e-05,
      "loss": 7.539,
      "step": 22150
    },
    {
      "epoch": 0.062320540186342914,
      "grad_norm": 34.75,
      "learning_rate": 1.997871388877305e-05,
      "loss": 8.34,
      "step": 22160
    },
    {
      "epoch": 0.062348663173791626,
      "grad_norm": 31.625,
      "learning_rate": 1.9978694679105754e-05,
      "loss": 6.9892,
      "step": 22170
    },
    {
      "epoch": 0.06237678616124034,
      "grad_norm": 27.5,
      "learning_rate": 1.997867546078372e-05,
      "loss": 8.4835,
      "step": 22180
    },
    {
      "epoch": 0.06240490914868905,
      "grad_norm": 31.875,
      "learning_rate": 1.9978656233806967e-05,
      "loss": 7.5068,
      "step": 22190
    },
    {
      "epoch": 0.06243303213613776,
      "grad_norm": 21.875,
      "learning_rate": 1.9978636998175508e-05,
      "loss": 8.2025,
      "step": 22200
    },
    {
      "epoch": 0.06246115512358647,
      "grad_norm": 36.0,
      "learning_rate": 1.997861775388936e-05,
      "loss": 7.1691,
      "step": 22210
    },
    {
      "epoch": 0.062489278111035176,
      "grad_norm": 42.25,
      "learning_rate": 1.9978598500948546e-05,
      "loss": 7.2822,
      "step": 22220
    },
    {
      "epoch": 0.06251740109848389,
      "grad_norm": 25.125,
      "learning_rate": 1.9978579239353077e-05,
      "loss": 7.6716,
      "step": 22230
    },
    {
      "epoch": 0.0625455240859326,
      "grad_norm": 33.5,
      "learning_rate": 1.997855996910297e-05,
      "loss": 7.9005,
      "step": 22240
    },
    {
      "epoch": 0.06257364707338131,
      "grad_norm": 25.375,
      "learning_rate": 1.9978540690198246e-05,
      "loss": 7.7899,
      "step": 22250
    },
    {
      "epoch": 0.06260177006083002,
      "grad_norm": 20.625,
      "learning_rate": 1.9978521402638917e-05,
      "loss": 8.4835,
      "step": 22260
    },
    {
      "epoch": 0.06262989304827873,
      "grad_norm": 23.625,
      "learning_rate": 1.9978502106424998e-05,
      "loss": 8.2887,
      "step": 22270
    },
    {
      "epoch": 0.06265801603572745,
      "grad_norm": 28.25,
      "learning_rate": 1.997848280155651e-05,
      "loss": 6.7868,
      "step": 22280
    },
    {
      "epoch": 0.06268613902317616,
      "grad_norm": 24.875,
      "learning_rate": 1.9978463488033473e-05,
      "loss": 7.5978,
      "step": 22290
    },
    {
      "epoch": 0.06271426201062487,
      "grad_norm": 26.625,
      "learning_rate": 1.9978444165855894e-05,
      "loss": 8.4458,
      "step": 22300
    },
    {
      "epoch": 0.06274238499807358,
      "grad_norm": 23.5,
      "learning_rate": 1.99784248350238e-05,
      "loss": 8.8546,
      "step": 22310
    },
    {
      "epoch": 0.06277050798552229,
      "grad_norm": 27.875,
      "learning_rate": 1.99784054955372e-05,
      "loss": 8.3741,
      "step": 22320
    },
    {
      "epoch": 0.062798630972971,
      "grad_norm": 22.375,
      "learning_rate": 1.9978386147396116e-05,
      "loss": 8.1057,
      "step": 22330
    },
    {
      "epoch": 0.0628267539604197,
      "grad_norm": 26.125,
      "learning_rate": 1.997836679060056e-05,
      "loss": 8.0162,
      "step": 22340
    },
    {
      "epoch": 0.06285487694786841,
      "grad_norm": 34.0,
      "learning_rate": 1.9978347425150553e-05,
      "loss": 8.1395,
      "step": 22350
    },
    {
      "epoch": 0.06288299993531712,
      "grad_norm": 39.0,
      "learning_rate": 1.997832805104611e-05,
      "loss": 8.2695,
      "step": 22360
    },
    {
      "epoch": 0.06291112292276584,
      "grad_norm": 112.0,
      "learning_rate": 1.9978308668287247e-05,
      "loss": 7.8599,
      "step": 22370
    },
    {
      "epoch": 0.06293924591021455,
      "grad_norm": 28.625,
      "learning_rate": 1.9978289276873984e-05,
      "loss": 8.5067,
      "step": 22380
    },
    {
      "epoch": 0.06296736889766326,
      "grad_norm": 31.375,
      "learning_rate": 1.9978269876806335e-05,
      "loss": 8.0129,
      "step": 22390
    },
    {
      "epoch": 0.06299549188511197,
      "grad_norm": 23.25,
      "learning_rate": 1.9978250468084316e-05,
      "loss": 7.2293,
      "step": 22400
    },
    {
      "epoch": 0.06302361487256068,
      "grad_norm": 37.75,
      "learning_rate": 1.9978231050707945e-05,
      "loss": 7.6755,
      "step": 22410
    },
    {
      "epoch": 0.0630517378600094,
      "grad_norm": 25.5,
      "learning_rate": 1.997821162467724e-05,
      "loss": 7.6361,
      "step": 22420
    },
    {
      "epoch": 0.0630798608474581,
      "grad_norm": 27.375,
      "learning_rate": 1.9978192189992217e-05,
      "loss": 7.3591,
      "step": 22430
    },
    {
      "epoch": 0.06310798383490682,
      "grad_norm": 36.5,
      "learning_rate": 1.9978172746652893e-05,
      "loss": 7.3798,
      "step": 22440
    },
    {
      "epoch": 0.06313610682235553,
      "grad_norm": 22.75,
      "learning_rate": 1.9978153294659282e-05,
      "loss": 7.3457,
      "step": 22450
    },
    {
      "epoch": 0.06316422980980424,
      "grad_norm": 38.0,
      "learning_rate": 1.9978133834011406e-05,
      "loss": 7.782,
      "step": 22460
    },
    {
      "epoch": 0.06319235279725295,
      "grad_norm": 24.5,
      "learning_rate": 1.997811436470928e-05,
      "loss": 7.7166,
      "step": 22470
    },
    {
      "epoch": 0.06322047578470166,
      "grad_norm": 44.75,
      "learning_rate": 1.997809488675292e-05,
      "loss": 8.1827,
      "step": 22480
    },
    {
      "epoch": 0.06324859877215037,
      "grad_norm": 44.25,
      "learning_rate": 1.997807540014234e-05,
      "loss": 7.8697,
      "step": 22490
    },
    {
      "epoch": 0.06327672175959907,
      "grad_norm": 39.0,
      "learning_rate": 1.9978055904877564e-05,
      "loss": 7.6617,
      "step": 22500
    },
    {
      "epoch": 0.06330484474704778,
      "grad_norm": 29.5,
      "learning_rate": 1.9978036400958603e-05,
      "loss": 8.3092,
      "step": 22510
    },
    {
      "epoch": 0.0633329677344965,
      "grad_norm": 44.0,
      "learning_rate": 1.9978016888385476e-05,
      "loss": 7.6671,
      "step": 22520
    },
    {
      "epoch": 0.0633610907219452,
      "grad_norm": 32.25,
      "learning_rate": 1.99779973671582e-05,
      "loss": 7.9161,
      "step": 22530
    },
    {
      "epoch": 0.06338921370939392,
      "grad_norm": 32.0,
      "learning_rate": 1.9977977837276793e-05,
      "loss": 8.2919,
      "step": 22540
    },
    {
      "epoch": 0.06341733669684263,
      "grad_norm": 22.875,
      "learning_rate": 1.997795829874127e-05,
      "loss": 8.2044,
      "step": 22550
    },
    {
      "epoch": 0.06344545968429134,
      "grad_norm": 23.5,
      "learning_rate": 1.9977938751551647e-05,
      "loss": 9.1357,
      "step": 22560
    },
    {
      "epoch": 0.06347358267174005,
      "grad_norm": 24.375,
      "learning_rate": 1.9977919195707948e-05,
      "loss": 8.6587,
      "step": 22570
    },
    {
      "epoch": 0.06350170565918876,
      "grad_norm": 33.25,
      "learning_rate": 1.9977899631210177e-05,
      "loss": 7.6374,
      "step": 22580
    },
    {
      "epoch": 0.06352982864663748,
      "grad_norm": 25.875,
      "learning_rate": 1.997788005805836e-05,
      "loss": 7.7163,
      "step": 22590
    },
    {
      "epoch": 0.06355795163408619,
      "grad_norm": 35.25,
      "learning_rate": 1.997786047625252e-05,
      "loss": 8.3111,
      "step": 22600
    },
    {
      "epoch": 0.0635860746215349,
      "grad_norm": 35.75,
      "learning_rate": 1.997784088579266e-05,
      "loss": 7.9682,
      "step": 22610
    },
    {
      "epoch": 0.06361419760898361,
      "grad_norm": 31.375,
      "learning_rate": 1.9977821286678804e-05,
      "loss": 8.1675,
      "step": 22620
    },
    {
      "epoch": 0.06364232059643232,
      "grad_norm": 23.0,
      "learning_rate": 1.9977801678910968e-05,
      "loss": 8.3013,
      "step": 22630
    },
    {
      "epoch": 0.06367044358388103,
      "grad_norm": 43.5,
      "learning_rate": 1.9977782062489172e-05,
      "loss": 8.2498,
      "step": 22640
    },
    {
      "epoch": 0.06369856657132975,
      "grad_norm": 24.75,
      "learning_rate": 1.9977762437413428e-05,
      "loss": 7.3001,
      "step": 22650
    },
    {
      "epoch": 0.06372668955877844,
      "grad_norm": 25.125,
      "learning_rate": 1.997774280368376e-05,
      "loss": 7.5571,
      "step": 22660
    },
    {
      "epoch": 0.06375481254622715,
      "grad_norm": 56.25,
      "learning_rate": 1.9977723161300176e-05,
      "loss": 7.8752,
      "step": 22670
    },
    {
      "epoch": 0.06378293553367587,
      "grad_norm": 26.5,
      "learning_rate": 1.99777035102627e-05,
      "loss": 7.0414,
      "step": 22680
    },
    {
      "epoch": 0.06381105852112458,
      "grad_norm": 33.25,
      "learning_rate": 1.9977683850571343e-05,
      "loss": 7.8153,
      "step": 22690
    },
    {
      "epoch": 0.06383918150857329,
      "grad_norm": 41.0,
      "learning_rate": 1.997766418222613e-05,
      "loss": 7.9327,
      "step": 22700
    },
    {
      "epoch": 0.063867304496022,
      "grad_norm": 24.5,
      "learning_rate": 1.997764450522707e-05,
      "loss": 7.7052,
      "step": 22710
    },
    {
      "epoch": 0.06389542748347071,
      "grad_norm": 37.25,
      "learning_rate": 1.9977624819574188e-05,
      "loss": 7.5229,
      "step": 22720
    },
    {
      "epoch": 0.06392355047091942,
      "grad_norm": 34.25,
      "learning_rate": 1.9977605125267493e-05,
      "loss": 7.8829,
      "step": 22730
    },
    {
      "epoch": 0.06395167345836814,
      "grad_norm": 25.625,
      "learning_rate": 1.9977585422307008e-05,
      "loss": 8.1108,
      "step": 22740
    },
    {
      "epoch": 0.06397979644581685,
      "grad_norm": 24.75,
      "learning_rate": 1.9977565710692746e-05,
      "loss": 7.2474,
      "step": 22750
    },
    {
      "epoch": 0.06400791943326556,
      "grad_norm": 32.0,
      "learning_rate": 1.997754599042473e-05,
      "loss": 7.7898,
      "step": 22760
    },
    {
      "epoch": 0.06403604242071427,
      "grad_norm": 26.125,
      "learning_rate": 1.9977526261502968e-05,
      "loss": 7.7363,
      "step": 22770
    },
    {
      "epoch": 0.06406416540816298,
      "grad_norm": 29.5,
      "learning_rate": 1.997750652392749e-05,
      "loss": 7.9805,
      "step": 22780
    },
    {
      "epoch": 0.0640922883956117,
      "grad_norm": 37.0,
      "learning_rate": 1.9977486777698298e-05,
      "loss": 8.2608,
      "step": 22790
    },
    {
      "epoch": 0.0641204113830604,
      "grad_norm": 57.25,
      "learning_rate": 1.9977467022815418e-05,
      "loss": 8.3327,
      "step": 22800
    },
    {
      "epoch": 0.06414853437050912,
      "grad_norm": 26.125,
      "learning_rate": 1.9977447259278867e-05,
      "loss": 8.0268,
      "step": 22810
    },
    {
      "epoch": 0.06417665735795781,
      "grad_norm": 26.375,
      "learning_rate": 1.9977427487088665e-05,
      "loss": 7.559,
      "step": 22820
    },
    {
      "epoch": 0.06420478034540653,
      "grad_norm": 35.75,
      "learning_rate": 1.997740770624482e-05,
      "loss": 8.1212,
      "step": 22830
    },
    {
      "epoch": 0.06423290333285524,
      "grad_norm": 36.0,
      "learning_rate": 1.9977387916747354e-05,
      "loss": 7.5903,
      "step": 22840
    },
    {
      "epoch": 0.06426102632030395,
      "grad_norm": 25.75,
      "learning_rate": 1.9977368118596287e-05,
      "loss": 8.2833,
      "step": 22850
    },
    {
      "epoch": 0.06428914930775266,
      "grad_norm": 29.375,
      "learning_rate": 1.9977348311791632e-05,
      "loss": 8.1548,
      "step": 22860
    },
    {
      "epoch": 0.06431727229520137,
      "grad_norm": 27.25,
      "learning_rate": 1.997732849633341e-05,
      "loss": 8.2613,
      "step": 22870
    },
    {
      "epoch": 0.06434539528265008,
      "grad_norm": 25.25,
      "learning_rate": 1.9977308672221628e-05,
      "loss": 8.0939,
      "step": 22880
    },
    {
      "epoch": 0.0643735182700988,
      "grad_norm": 45.25,
      "learning_rate": 1.997728883945632e-05,
      "loss": 7.6367,
      "step": 22890
    },
    {
      "epoch": 0.0644016412575475,
      "grad_norm": 28.5,
      "learning_rate": 1.997726899803749e-05,
      "loss": 7.9615,
      "step": 22900
    },
    {
      "epoch": 0.06442976424499622,
      "grad_norm": 97.0,
      "learning_rate": 1.997724914796516e-05,
      "loss": 8.1091,
      "step": 22910
    },
    {
      "epoch": 0.06445788723244493,
      "grad_norm": 37.0,
      "learning_rate": 1.9977229289239348e-05,
      "loss": 7.8084,
      "step": 22920
    },
    {
      "epoch": 0.06448601021989364,
      "grad_norm": 24.875,
      "learning_rate": 1.9977209421860067e-05,
      "loss": 8.6558,
      "step": 22930
    },
    {
      "epoch": 0.06451413320734235,
      "grad_norm": 28.875,
      "learning_rate": 1.9977189545827342e-05,
      "loss": 8.0869,
      "step": 22940
    },
    {
      "epoch": 0.06454225619479106,
      "grad_norm": 28.75,
      "learning_rate": 1.9977169661141183e-05,
      "loss": 8.2552,
      "step": 22950
    },
    {
      "epoch": 0.06457037918223978,
      "grad_norm": 26.125,
      "learning_rate": 1.997714976780161e-05,
      "loss": 7.7331,
      "step": 22960
    },
    {
      "epoch": 0.06459850216968849,
      "grad_norm": 35.5,
      "learning_rate": 1.9977129865808637e-05,
      "loss": 7.1371,
      "step": 22970
    },
    {
      "epoch": 0.0646266251571372,
      "grad_norm": 30.0,
      "learning_rate": 1.9977109955162287e-05,
      "loss": 8.1656,
      "step": 22980
    },
    {
      "epoch": 0.0646547481445859,
      "grad_norm": 33.5,
      "learning_rate": 1.9977090035862577e-05,
      "loss": 7.9018,
      "step": 22990
    },
    {
      "epoch": 0.06468287113203461,
      "grad_norm": 21.0,
      "learning_rate": 1.9977070107909518e-05,
      "loss": 8.1548,
      "step": 23000
    },
    {
      "epoch": 0.06471099411948332,
      "grad_norm": 25.0,
      "learning_rate": 1.9977050171303132e-05,
      "loss": 7.8347,
      "step": 23010
    },
    {
      "epoch": 0.06473911710693203,
      "grad_norm": 58.5,
      "learning_rate": 1.9977030226043435e-05,
      "loss": 8.1935,
      "step": 23020
    },
    {
      "epoch": 0.06476724009438074,
      "grad_norm": 24.5,
      "learning_rate": 1.9977010272130445e-05,
      "loss": 8.4827,
      "step": 23030
    },
    {
      "epoch": 0.06479536308182945,
      "grad_norm": 26.875,
      "learning_rate": 1.9976990309564177e-05,
      "loss": 7.1383,
      "step": 23040
    },
    {
      "epoch": 0.06482348606927817,
      "grad_norm": 44.5,
      "learning_rate": 1.997697033834465e-05,
      "loss": 8.0067,
      "step": 23050
    },
    {
      "epoch": 0.06485160905672688,
      "grad_norm": 36.75,
      "learning_rate": 1.9976950358471884e-05,
      "loss": 7.9116,
      "step": 23060
    },
    {
      "epoch": 0.06487973204417559,
      "grad_norm": 23.375,
      "learning_rate": 1.9976930369945892e-05,
      "loss": 7.728,
      "step": 23070
    },
    {
      "epoch": 0.0649078550316243,
      "grad_norm": 31.375,
      "learning_rate": 1.9976910372766697e-05,
      "loss": 7.7307,
      "step": 23080
    },
    {
      "epoch": 0.06493597801907301,
      "grad_norm": 25.75,
      "learning_rate": 1.9976890366934308e-05,
      "loss": 7.9723,
      "step": 23090
    },
    {
      "epoch": 0.06496410100652172,
      "grad_norm": 31.0,
      "learning_rate": 1.997687035244875e-05,
      "loss": 8.0428,
      "step": 23100
    },
    {
      "epoch": 0.06499222399397044,
      "grad_norm": 30.75,
      "learning_rate": 1.9976850329310037e-05,
      "loss": 8.3067,
      "step": 23110
    },
    {
      "epoch": 0.06502034698141915,
      "grad_norm": 38.25,
      "learning_rate": 1.9976830297518185e-05,
      "loss": 7.6053,
      "step": 23120
    },
    {
      "epoch": 0.06504846996886786,
      "grad_norm": 33.75,
      "learning_rate": 1.9976810257073214e-05,
      "loss": 7.0956,
      "step": 23130
    },
    {
      "epoch": 0.06507659295631657,
      "grad_norm": 22.625,
      "learning_rate": 1.9976790207975138e-05,
      "loss": 8.2208,
      "step": 23140
    },
    {
      "epoch": 0.06510471594376527,
      "grad_norm": 27.25,
      "learning_rate": 1.997677015022398e-05,
      "loss": 7.7637,
      "step": 23150
    },
    {
      "epoch": 0.06513283893121398,
      "grad_norm": 32.75,
      "learning_rate": 1.9976750083819753e-05,
      "loss": 8.0825,
      "step": 23160
    },
    {
      "epoch": 0.06516096191866269,
      "grad_norm": 29.75,
      "learning_rate": 1.9976730008762475e-05,
      "loss": 7.8594,
      "step": 23170
    },
    {
      "epoch": 0.0651890849061114,
      "grad_norm": 29.5,
      "learning_rate": 1.9976709925052162e-05,
      "loss": 7.7548,
      "step": 23180
    },
    {
      "epoch": 0.06521720789356011,
      "grad_norm": 22.625,
      "learning_rate": 1.997668983268884e-05,
      "loss": 8.0494,
      "step": 23190
    },
    {
      "epoch": 0.06524533088100883,
      "grad_norm": 46.5,
      "learning_rate": 1.997666973167251e-05,
      "loss": 8.0951,
      "step": 23200
    },
    {
      "epoch": 0.06527345386845754,
      "grad_norm": 30.875,
      "learning_rate": 1.9976649622003207e-05,
      "loss": 7.6749,
      "step": 23210
    },
    {
      "epoch": 0.06530157685590625,
      "grad_norm": 28.25,
      "learning_rate": 1.9976629503680936e-05,
      "loss": 7.7256,
      "step": 23220
    },
    {
      "epoch": 0.06532969984335496,
      "grad_norm": 28.5,
      "learning_rate": 1.9976609376705725e-05,
      "loss": 8.0738,
      "step": 23230
    },
    {
      "epoch": 0.06535782283080367,
      "grad_norm": 41.75,
      "learning_rate": 1.9976589241077582e-05,
      "loss": 7.3333,
      "step": 23240
    },
    {
      "epoch": 0.06538594581825238,
      "grad_norm": 55.25,
      "learning_rate": 1.9976569096796526e-05,
      "loss": 7.4189,
      "step": 23250
    },
    {
      "epoch": 0.0654140688057011,
      "grad_norm": 33.0,
      "learning_rate": 1.997654894386258e-05,
      "loss": 8.2886,
      "step": 23260
    },
    {
      "epoch": 0.0654421917931498,
      "grad_norm": 26.75,
      "learning_rate": 1.9976528782275758e-05,
      "loss": 7.8644,
      "step": 23270
    },
    {
      "epoch": 0.06547031478059852,
      "grad_norm": 22.625,
      "learning_rate": 1.9976508612036076e-05,
      "loss": 7.9872,
      "step": 23280
    },
    {
      "epoch": 0.06549843776804723,
      "grad_norm": 32.5,
      "learning_rate": 1.9976488433143553e-05,
      "loss": 7.7647,
      "step": 23290
    },
    {
      "epoch": 0.06552656075549594,
      "grad_norm": 27.125,
      "learning_rate": 1.9976468245598206e-05,
      "loss": 7.6811,
      "step": 23300
    },
    {
      "epoch": 0.06555468374294464,
      "grad_norm": 38.75,
      "learning_rate": 1.9976448049400058e-05,
      "loss": 7.9701,
      "step": 23310
    },
    {
      "epoch": 0.06558280673039335,
      "grad_norm": 29.25,
      "learning_rate": 1.9976427844549112e-05,
      "loss": 8.1048,
      "step": 23320
    },
    {
      "epoch": 0.06561092971784206,
      "grad_norm": 25.75,
      "learning_rate": 1.9976407631045404e-05,
      "loss": 7.9037,
      "step": 23330
    },
    {
      "epoch": 0.06563905270529077,
      "grad_norm": 27.625,
      "learning_rate": 1.9976387408888938e-05,
      "loss": 7.1071,
      "step": 23340
    },
    {
      "epoch": 0.06566717569273949,
      "grad_norm": 22.0,
      "learning_rate": 1.9976367178079736e-05,
      "loss": 7.7824,
      "step": 23350
    },
    {
      "epoch": 0.0656952986801882,
      "grad_norm": 32.75,
      "learning_rate": 1.9976346938617815e-05,
      "loss": 8.4946,
      "step": 23360
    },
    {
      "epoch": 0.06572342166763691,
      "grad_norm": 22.875,
      "learning_rate": 1.9976326690503195e-05,
      "loss": 8.2183,
      "step": 23370
    },
    {
      "epoch": 0.06575154465508562,
      "grad_norm": 44.5,
      "learning_rate": 1.9976306433735896e-05,
      "loss": 8.5571,
      "step": 23380
    },
    {
      "epoch": 0.06577966764253433,
      "grad_norm": 25.875,
      "learning_rate": 1.9976286168315925e-05,
      "loss": 7.7022,
      "step": 23390
    },
    {
      "epoch": 0.06580779062998304,
      "grad_norm": 54.0,
      "learning_rate": 1.9976265894243306e-05,
      "loss": 7.8134,
      "step": 23400
    },
    {
      "epoch": 0.06583591361743175,
      "grad_norm": 34.5,
      "learning_rate": 1.9976245611518062e-05,
      "loss": 8.1709,
      "step": 23410
    },
    {
      "epoch": 0.06586403660488047,
      "grad_norm": 24.625,
      "learning_rate": 1.99762253201402e-05,
      "loss": 8.1647,
      "step": 23420
    },
    {
      "epoch": 0.06589215959232918,
      "grad_norm": 24.125,
      "learning_rate": 1.9976205020109745e-05,
      "loss": 8.4993,
      "step": 23430
    },
    {
      "epoch": 0.06592028257977789,
      "grad_norm": 20.75,
      "learning_rate": 1.9976184711426713e-05,
      "loss": 6.8959,
      "step": 23440
    },
    {
      "epoch": 0.0659484055672266,
      "grad_norm": 22.5,
      "learning_rate": 1.9976164394091117e-05,
      "loss": 8.0051,
      "step": 23450
    },
    {
      "epoch": 0.06597652855467531,
      "grad_norm": 25.5,
      "learning_rate": 1.997614406810298e-05,
      "loss": 7.8721,
      "step": 23460
    },
    {
      "epoch": 0.06600465154212401,
      "grad_norm": 38.25,
      "learning_rate": 1.997612373346232e-05,
      "loss": 8.0255,
      "step": 23470
    },
    {
      "epoch": 0.06603277452957272,
      "grad_norm": 52.25,
      "learning_rate": 1.9976103390169154e-05,
      "loss": 8.7649,
      "step": 23480
    },
    {
      "epoch": 0.06606089751702143,
      "grad_norm": 28.75,
      "learning_rate": 1.9976083038223497e-05,
      "loss": 7.916,
      "step": 23490
    },
    {
      "epoch": 0.06608902050447014,
      "grad_norm": 27.0,
      "learning_rate": 1.9976062677625365e-05,
      "loss": 8.2649,
      "step": 23500
    },
    {
      "epoch": 0.06611714349191886,
      "grad_norm": 29.125,
      "learning_rate": 1.9976042308374784e-05,
      "loss": 8.5381,
      "step": 23510
    },
    {
      "epoch": 0.06614526647936757,
      "grad_norm": 23.0,
      "learning_rate": 1.997602193047176e-05,
      "loss": 8.2512,
      "step": 23520
    },
    {
      "epoch": 0.06617338946681628,
      "grad_norm": 36.0,
      "learning_rate": 1.9976001543916325e-05,
      "loss": 8.253,
      "step": 23530
    },
    {
      "epoch": 0.06620151245426499,
      "grad_norm": 33.0,
      "learning_rate": 1.9975981148708483e-05,
      "loss": 9.0981,
      "step": 23540
    },
    {
      "epoch": 0.0662296354417137,
      "grad_norm": 27.125,
      "learning_rate": 1.997596074484826e-05,
      "loss": 7.0276,
      "step": 23550
    },
    {
      "epoch": 0.06625775842916241,
      "grad_norm": 29.0,
      "learning_rate": 1.997594033233567e-05,
      "loss": 7.4713,
      "step": 23560
    },
    {
      "epoch": 0.06628588141661113,
      "grad_norm": 27.25,
      "learning_rate": 1.9975919911170734e-05,
      "loss": 7.8457,
      "step": 23570
    },
    {
      "epoch": 0.06631400440405984,
      "grad_norm": 24.375,
      "learning_rate": 1.9975899481353466e-05,
      "loss": 7.6421,
      "step": 23580
    },
    {
      "epoch": 0.06634212739150855,
      "grad_norm": 27.125,
      "learning_rate": 1.9975879042883884e-05,
      "loss": 8.4034,
      "step": 23590
    },
    {
      "epoch": 0.06637025037895726,
      "grad_norm": 43.0,
      "learning_rate": 1.9975858595762004e-05,
      "loss": 8.0975,
      "step": 23600
    },
    {
      "epoch": 0.06639837336640597,
      "grad_norm": 29.75,
      "learning_rate": 1.9975838139987854e-05,
      "loss": 8.0296,
      "step": 23610
    },
    {
      "epoch": 0.06642649635385468,
      "grad_norm": 35.75,
      "learning_rate": 1.9975817675561443e-05,
      "loss": 8.2287,
      "step": 23620
    },
    {
      "epoch": 0.0664546193413034,
      "grad_norm": 26.0,
      "learning_rate": 1.9975797202482787e-05,
      "loss": 8.0322,
      "step": 23630
    },
    {
      "epoch": 0.06648274232875209,
      "grad_norm": 20.25,
      "learning_rate": 1.997577672075191e-05,
      "loss": 7.7976,
      "step": 23640
    },
    {
      "epoch": 0.0665108653162008,
      "grad_norm": 52.0,
      "learning_rate": 1.9975756230368827e-05,
      "loss": 8.0099,
      "step": 23650
    },
    {
      "epoch": 0.06653898830364952,
      "grad_norm": 38.25,
      "learning_rate": 1.997573573133355e-05,
      "loss": 8.3235,
      "step": 23660
    },
    {
      "epoch": 0.06656711129109823,
      "grad_norm": 39.25,
      "learning_rate": 1.9975715223646107e-05,
      "loss": 8.5622,
      "step": 23670
    },
    {
      "epoch": 0.06659523427854694,
      "grad_norm": 37.75,
      "learning_rate": 1.997569470730651e-05,
      "loss": 7.6335,
      "step": 23680
    },
    {
      "epoch": 0.06662335726599565,
      "grad_norm": 20.875,
      "learning_rate": 1.997567418231478e-05,
      "loss": 7.213,
      "step": 23690
    },
    {
      "epoch": 0.06665148025344436,
      "grad_norm": 27.625,
      "learning_rate": 1.997565364867093e-05,
      "loss": 7.1579,
      "step": 23700
    },
    {
      "epoch": 0.06667960324089307,
      "grad_norm": 34.75,
      "learning_rate": 1.997563310637498e-05,
      "loss": 8.0902,
      "step": 23710
    },
    {
      "epoch": 0.06670772622834178,
      "grad_norm": 29.5,
      "learning_rate": 1.9975612555426953e-05,
      "loss": 8.5122,
      "step": 23720
    },
    {
      "epoch": 0.0667358492157905,
      "grad_norm": 25.0,
      "learning_rate": 1.997559199582686e-05,
      "loss": 8.1568,
      "step": 23730
    },
    {
      "epoch": 0.06676397220323921,
      "grad_norm": 29.5,
      "learning_rate": 1.9975571427574717e-05,
      "loss": 7.5884,
      "step": 23740
    },
    {
      "epoch": 0.06679209519068792,
      "grad_norm": 26.875,
      "learning_rate": 1.997555085067055e-05,
      "loss": 8.1614,
      "step": 23750
    },
    {
      "epoch": 0.06682021817813663,
      "grad_norm": 25.25,
      "learning_rate": 1.997553026511437e-05,
      "loss": 7.8148,
      "step": 23760
    },
    {
      "epoch": 0.06684834116558534,
      "grad_norm": 37.75,
      "learning_rate": 1.9975509670906202e-05,
      "loss": 8.2944,
      "step": 23770
    },
    {
      "epoch": 0.06687646415303405,
      "grad_norm": 22.75,
      "learning_rate": 1.9975489068046055e-05,
      "loss": 7.892,
      "step": 23780
    },
    {
      "epoch": 0.06690458714048277,
      "grad_norm": 28.75,
      "learning_rate": 1.997546845653395e-05,
      "loss": 8.2363,
      "step": 23790
    },
    {
      "epoch": 0.06693271012793146,
      "grad_norm": 40.25,
      "learning_rate": 1.997544783636991e-05,
      "loss": 8.4571,
      "step": 23800
    },
    {
      "epoch": 0.06696083311538018,
      "grad_norm": 26.5,
      "learning_rate": 1.997542720755395e-05,
      "loss": 7.9693,
      "step": 23810
    },
    {
      "epoch": 0.06698895610282889,
      "grad_norm": 30.5,
      "learning_rate": 1.9975406570086084e-05,
      "loss": 8.017,
      "step": 23820
    },
    {
      "epoch": 0.0670170790902776,
      "grad_norm": 27.5,
      "learning_rate": 1.9975385923966335e-05,
      "loss": 7.9965,
      "step": 23830
    },
    {
      "epoch": 0.06704520207772631,
      "grad_norm": 30.625,
      "learning_rate": 1.9975365269194714e-05,
      "loss": 7.7165,
      "step": 23840
    },
    {
      "epoch": 0.06707332506517502,
      "grad_norm": 23.875,
      "learning_rate": 1.997534460577125e-05,
      "loss": 7.7689,
      "step": 23850
    },
    {
      "epoch": 0.06710144805262373,
      "grad_norm": 28.125,
      "learning_rate": 1.9975323933695952e-05,
      "loss": 7.9831,
      "step": 23860
    },
    {
      "epoch": 0.06712957104007244,
      "grad_norm": 24.125,
      "learning_rate": 1.997530325296884e-05,
      "loss": 7.7716,
      "step": 23870
    },
    {
      "epoch": 0.06715769402752116,
      "grad_norm": 26.25,
      "learning_rate": 1.9975282563589932e-05,
      "loss": 8.2295,
      "step": 23880
    },
    {
      "epoch": 0.06718581701496987,
      "grad_norm": 32.5,
      "learning_rate": 1.9975261865559247e-05,
      "loss": 7.994,
      "step": 23890
    },
    {
      "epoch": 0.06721394000241858,
      "grad_norm": 42.0,
      "learning_rate": 1.9975241158876805e-05,
      "loss": 8.3927,
      "step": 23900
    },
    {
      "epoch": 0.06724206298986729,
      "grad_norm": 35.0,
      "learning_rate": 1.9975220443542616e-05,
      "loss": 7.6007,
      "step": 23910
    },
    {
      "epoch": 0.067270185977316,
      "grad_norm": 27.0,
      "learning_rate": 1.9975199719556707e-05,
      "loss": 8.5861,
      "step": 23920
    },
    {
      "epoch": 0.06729830896476471,
      "grad_norm": 31.0,
      "learning_rate": 1.9975178986919088e-05,
      "loss": 8.6574,
      "step": 23930
    },
    {
      "epoch": 0.06732643195221343,
      "grad_norm": 36.0,
      "learning_rate": 1.9975158245629783e-05,
      "loss": 8.2572,
      "step": 23940
    },
    {
      "epoch": 0.06735455493966214,
      "grad_norm": 34.0,
      "learning_rate": 1.9975137495688813e-05,
      "loss": 8.3509,
      "step": 23950
    },
    {
      "epoch": 0.06738267792711083,
      "grad_norm": 22.875,
      "learning_rate": 1.9975116737096187e-05,
      "loss": 8.2402,
      "step": 23960
    },
    {
      "epoch": 0.06741080091455955,
      "grad_norm": 27.125,
      "learning_rate": 1.9975095969851926e-05,
      "loss": 8.2014,
      "step": 23970
    },
    {
      "epoch": 0.06743892390200826,
      "grad_norm": 26.75,
      "learning_rate": 1.9975075193956053e-05,
      "loss": 7.5507,
      "step": 23980
    },
    {
      "epoch": 0.06746704688945697,
      "grad_norm": 44.0,
      "learning_rate": 1.997505440940858e-05,
      "loss": 8.6654,
      "step": 23990
    },
    {
      "epoch": 0.06749516987690568,
      "grad_norm": 53.75,
      "learning_rate": 1.9975033616209528e-05,
      "loss": 7.841,
      "step": 24000
    },
    {
      "epoch": 0.06752329286435439,
      "grad_norm": 28.25,
      "learning_rate": 1.9975012814358913e-05,
      "loss": 7.6185,
      "step": 24010
    },
    {
      "epoch": 0.0675514158518031,
      "grad_norm": 29.125,
      "learning_rate": 1.9974992003856757e-05,
      "loss": 8.4279,
      "step": 24020
    },
    {
      "epoch": 0.06757953883925182,
      "grad_norm": 23.25,
      "learning_rate": 1.997497118470307e-05,
      "loss": 8.2732,
      "step": 24030
    },
    {
      "epoch": 0.06760766182670053,
      "grad_norm": 32.0,
      "learning_rate": 1.997495035689788e-05,
      "loss": 7.9184,
      "step": 24040
    },
    {
      "epoch": 0.06763578481414924,
      "grad_norm": 50.75,
      "learning_rate": 1.99749295204412e-05,
      "loss": 8.6721,
      "step": 24050
    },
    {
      "epoch": 0.06766390780159795,
      "grad_norm": 32.25,
      "learning_rate": 1.997490867533305e-05,
      "loss": 7.8413,
      "step": 24060
    },
    {
      "epoch": 0.06769203078904666,
      "grad_norm": 25.375,
      "learning_rate": 1.997488782157344e-05,
      "loss": 7.4932,
      "step": 24070
    },
    {
      "epoch": 0.06772015377649537,
      "grad_norm": 30.75,
      "learning_rate": 1.99748669591624e-05,
      "loss": 7.2798,
      "step": 24080
    },
    {
      "epoch": 0.06774827676394408,
      "grad_norm": 32.75,
      "learning_rate": 1.9974846088099944e-05,
      "loss": 7.3896,
      "step": 24090
    },
    {
      "epoch": 0.0677763997513928,
      "grad_norm": 34.25,
      "learning_rate": 1.9974825208386086e-05,
      "loss": 7.7981,
      "step": 24100
    },
    {
      "epoch": 0.06780452273884151,
      "grad_norm": 49.75,
      "learning_rate": 1.997480432002085e-05,
      "loss": 8.5522,
      "step": 24110
    },
    {
      "epoch": 0.0678326457262902,
      "grad_norm": 31.75,
      "learning_rate": 1.997478342300425e-05,
      "loss": 8.1082,
      "step": 24120
    },
    {
      "epoch": 0.06786076871373892,
      "grad_norm": 31.625,
      "learning_rate": 1.9974762517336303e-05,
      "loss": 7.6424,
      "step": 24130
    },
    {
      "epoch": 0.06788889170118763,
      "grad_norm": 20.25,
      "learning_rate": 1.9974741603017032e-05,
      "loss": 8.1578,
      "step": 24140
    },
    {
      "epoch": 0.06791701468863634,
      "grad_norm": 35.75,
      "learning_rate": 1.997472068004645e-05,
      "loss": 7.946,
      "step": 24150
    },
    {
      "epoch": 0.06794513767608505,
      "grad_norm": 33.75,
      "learning_rate": 1.997469974842458e-05,
      "loss": 7.9592,
      "step": 24160
    },
    {
      "epoch": 0.06797326066353376,
      "grad_norm": 29.0,
      "learning_rate": 1.9974678808151438e-05,
      "loss": 7.9576,
      "step": 24170
    },
    {
      "epoch": 0.06800138365098247,
      "grad_norm": 33.75,
      "learning_rate": 1.997465785922704e-05,
      "loss": 7.9868,
      "step": 24180
    },
    {
      "epoch": 0.06802950663843119,
      "grad_norm": 26.0,
      "learning_rate": 1.997463690165141e-05,
      "loss": 7.6569,
      "step": 24190
    },
    {
      "epoch": 0.0680576296258799,
      "grad_norm": 27.5,
      "learning_rate": 1.9974615935424557e-05,
      "loss": 8.1518,
      "step": 24200
    },
    {
      "epoch": 0.06808575261332861,
      "grad_norm": 37.0,
      "learning_rate": 1.997459496054651e-05,
      "loss": 8.8376,
      "step": 24210
    },
    {
      "epoch": 0.06811387560077732,
      "grad_norm": 30.375,
      "learning_rate": 1.9974573977017278e-05,
      "loss": 8.1753,
      "step": 24220
    },
    {
      "epoch": 0.06814199858822603,
      "grad_norm": 26.25,
      "learning_rate": 1.9974552984836882e-05,
      "loss": 7.8068,
      "step": 24230
    },
    {
      "epoch": 0.06817012157567474,
      "grad_norm": 51.75,
      "learning_rate": 1.9974531984005346e-05,
      "loss": 7.5078,
      "step": 24240
    },
    {
      "epoch": 0.06819824456312346,
      "grad_norm": 34.25,
      "learning_rate": 1.9974510974522682e-05,
      "loss": 7.6289,
      "step": 24250
    },
    {
      "epoch": 0.06822636755057217,
      "grad_norm": 20.5,
      "learning_rate": 1.9974489956388905e-05,
      "loss": 7.7869,
      "step": 24260
    },
    {
      "epoch": 0.06825449053802088,
      "grad_norm": 27.75,
      "learning_rate": 1.9974468929604042e-05,
      "loss": 7.4938,
      "step": 24270
    },
    {
      "epoch": 0.06828261352546959,
      "grad_norm": 23.125,
      "learning_rate": 1.9974447894168106e-05,
      "loss": 7.9507,
      "step": 24280
    },
    {
      "epoch": 0.06831073651291829,
      "grad_norm": 27.125,
      "learning_rate": 1.9974426850081118e-05,
      "loss": 7.9369,
      "step": 24290
    },
    {
      "epoch": 0.068338859500367,
      "grad_norm": 21.75,
      "learning_rate": 1.997440579734309e-05,
      "loss": 7.1179,
      "step": 24300
    },
    {
      "epoch": 0.06836698248781571,
      "grad_norm": 22.25,
      "learning_rate": 1.9974384735954045e-05,
      "loss": 7.6934,
      "step": 24310
    },
    {
      "epoch": 0.06839510547526442,
      "grad_norm": 52.75,
      "learning_rate": 1.9974363665914005e-05,
      "loss": 6.7528,
      "step": 24320
    },
    {
      "epoch": 0.06842322846271313,
      "grad_norm": 29.625,
      "learning_rate": 1.9974342587222986e-05,
      "loss": 7.7815,
      "step": 24330
    },
    {
      "epoch": 0.06845135145016185,
      "grad_norm": 25.875,
      "learning_rate": 1.9974321499881e-05,
      "loss": 8.4692,
      "step": 24340
    },
    {
      "epoch": 0.06847947443761056,
      "grad_norm": 32.5,
      "learning_rate": 1.997430040388807e-05,
      "loss": 8.1893,
      "step": 24350
    },
    {
      "epoch": 0.06850759742505927,
      "grad_norm": 34.5,
      "learning_rate": 1.9974279299244216e-05,
      "loss": 7.8831,
      "step": 24360
    },
    {
      "epoch": 0.06853572041250798,
      "grad_norm": 23.125,
      "learning_rate": 1.9974258185949456e-05,
      "loss": 7.4445,
      "step": 24370
    },
    {
      "epoch": 0.06856384339995669,
      "grad_norm": 25.0,
      "learning_rate": 1.9974237064003806e-05,
      "loss": 8.2006,
      "step": 24380
    },
    {
      "epoch": 0.0685919663874054,
      "grad_norm": 26.25,
      "learning_rate": 1.9974215933407282e-05,
      "loss": 8.0715,
      "step": 24390
    },
    {
      "epoch": 0.06862008937485412,
      "grad_norm": 26.0,
      "learning_rate": 1.997419479415991e-05,
      "loss": 8.259,
      "step": 24400
    },
    {
      "epoch": 0.06864821236230283,
      "grad_norm": 42.5,
      "learning_rate": 1.9974173646261702e-05,
      "loss": 8.6407,
      "step": 24410
    },
    {
      "epoch": 0.06867633534975154,
      "grad_norm": 22.625,
      "learning_rate": 1.9974152489712676e-05,
      "loss": 7.4938,
      "step": 24420
    },
    {
      "epoch": 0.06870445833720025,
      "grad_norm": 35.5,
      "learning_rate": 1.9974131324512854e-05,
      "loss": 8.3035,
      "step": 24430
    },
    {
      "epoch": 0.06873258132464896,
      "grad_norm": 24.75,
      "learning_rate": 1.9974110150662255e-05,
      "loss": 8.2383,
      "step": 24440
    },
    {
      "epoch": 0.06876070431209766,
      "grad_norm": 26.875,
      "learning_rate": 1.997408896816089e-05,
      "loss": 8.6597,
      "step": 24450
    },
    {
      "epoch": 0.06878882729954637,
      "grad_norm": 21.75,
      "learning_rate": 1.9974067777008786e-05,
      "loss": 7.4632,
      "step": 24460
    },
    {
      "epoch": 0.06881695028699508,
      "grad_norm": 24.25,
      "learning_rate": 1.997404657720596e-05,
      "loss": 8.5503,
      "step": 24470
    },
    {
      "epoch": 0.0688450732744438,
      "grad_norm": 25.5,
      "learning_rate": 1.9974025368752426e-05,
      "loss": 7.443,
      "step": 24480
    },
    {
      "epoch": 0.0688731962618925,
      "grad_norm": 31.375,
      "learning_rate": 1.9974004151648208e-05,
      "loss": 7.4162,
      "step": 24490
    },
    {
      "epoch": 0.06890131924934122,
      "grad_norm": 26.0,
      "learning_rate": 1.9973982925893317e-05,
      "loss": 7.9443,
      "step": 24500
    },
    {
      "epoch": 0.06892944223678993,
      "grad_norm": 31.5,
      "learning_rate": 1.9973961691487776e-05,
      "loss": 7.5871,
      "step": 24510
    },
    {
      "epoch": 0.06895756522423864,
      "grad_norm": 26.625,
      "learning_rate": 1.9973940448431604e-05,
      "loss": 8.2074,
      "step": 24520
    },
    {
      "epoch": 0.06898568821168735,
      "grad_norm": 25.625,
      "learning_rate": 1.997391919672482e-05,
      "loss": 8.0772,
      "step": 24530
    },
    {
      "epoch": 0.06901381119913606,
      "grad_norm": 29.25,
      "learning_rate": 1.9973897936367436e-05,
      "loss": 8.0236,
      "step": 24540
    },
    {
      "epoch": 0.06904193418658477,
      "grad_norm": 23.375,
      "learning_rate": 1.997387666735948e-05,
      "loss": 7.7054,
      "step": 24550
    },
    {
      "epoch": 0.06907005717403349,
      "grad_norm": 33.25,
      "learning_rate": 1.9973855389700966e-05,
      "loss": 7.3754,
      "step": 24560
    },
    {
      "epoch": 0.0690981801614822,
      "grad_norm": 69.5,
      "learning_rate": 1.997383410339191e-05,
      "loss": 7.9906,
      "step": 24570
    },
    {
      "epoch": 0.06912630314893091,
      "grad_norm": 35.75,
      "learning_rate": 1.9973812808432332e-05,
      "loss": 8.6023,
      "step": 24580
    },
    {
      "epoch": 0.06915442613637962,
      "grad_norm": 64.5,
      "learning_rate": 1.9973791504822252e-05,
      "loss": 8.0812,
      "step": 24590
    },
    {
      "epoch": 0.06918254912382833,
      "grad_norm": 29.75,
      "learning_rate": 1.9973770192561688e-05,
      "loss": 8.6807,
      "step": 24600
    },
    {
      "epoch": 0.06921067211127703,
      "grad_norm": 40.75,
      "learning_rate": 1.997374887165066e-05,
      "loss": 8.3598,
      "step": 24610
    },
    {
      "epoch": 0.06923879509872574,
      "grad_norm": 32.75,
      "learning_rate": 1.997372754208918e-05,
      "loss": 7.3308,
      "step": 24620
    },
    {
      "epoch": 0.06926691808617445,
      "grad_norm": 23.625,
      "learning_rate": 1.997370620387727e-05,
      "loss": 7.3374,
      "step": 24630
    },
    {
      "epoch": 0.06929504107362316,
      "grad_norm": 55.25,
      "learning_rate": 1.9973684857014957e-05,
      "loss": 8.8595,
      "step": 24640
    },
    {
      "epoch": 0.06932316406107188,
      "grad_norm": 35.75,
      "learning_rate": 1.997366350150225e-05,
      "loss": 7.8599,
      "step": 24650
    },
    {
      "epoch": 0.06935128704852059,
      "grad_norm": 42.25,
      "learning_rate": 1.9973642137339165e-05,
      "loss": 8.3071,
      "step": 24660
    },
    {
      "epoch": 0.0693794100359693,
      "grad_norm": 55.25,
      "learning_rate": 1.9973620764525727e-05,
      "loss": 7.1628,
      "step": 24670
    },
    {
      "epoch": 0.06940753302341801,
      "grad_norm": 24.125,
      "learning_rate": 1.9973599383061955e-05,
      "loss": 7.7399,
      "step": 24680
    },
    {
      "epoch": 0.06943565601086672,
      "grad_norm": 24.625,
      "learning_rate": 1.9973577992947862e-05,
      "loss": 8.583,
      "step": 24690
    },
    {
      "epoch": 0.06946377899831543,
      "grad_norm": 20.0,
      "learning_rate": 1.9973556594183473e-05,
      "loss": 7.7831,
      "step": 24700
    },
    {
      "epoch": 0.06949190198576415,
      "grad_norm": 58.0,
      "learning_rate": 1.9973535186768803e-05,
      "loss": 7.9116,
      "step": 24710
    },
    {
      "epoch": 0.06952002497321286,
      "grad_norm": 39.75,
      "learning_rate": 1.9973513770703867e-05,
      "loss": 8.5319,
      "step": 24720
    },
    {
      "epoch": 0.06954814796066157,
      "grad_norm": 28.25,
      "learning_rate": 1.9973492345988692e-05,
      "loss": 7.0798,
      "step": 24730
    },
    {
      "epoch": 0.06957627094811028,
      "grad_norm": 22.0,
      "learning_rate": 1.9973470912623288e-05,
      "loss": 7.8794,
      "step": 24740
    },
    {
      "epoch": 0.06960439393555899,
      "grad_norm": 30.25,
      "learning_rate": 1.9973449470607682e-05,
      "loss": 8.0186,
      "step": 24750
    },
    {
      "epoch": 0.0696325169230077,
      "grad_norm": 32.75,
      "learning_rate": 1.9973428019941887e-05,
      "loss": 7.9679,
      "step": 24760
    },
    {
      "epoch": 0.0696606399104564,
      "grad_norm": 30.125,
      "learning_rate": 1.9973406560625924e-05,
      "loss": 8.7424,
      "step": 24770
    },
    {
      "epoch": 0.06968876289790511,
      "grad_norm": 47.0,
      "learning_rate": 1.9973385092659807e-05,
      "loss": 8.1056,
      "step": 24780
    },
    {
      "epoch": 0.06971688588535382,
      "grad_norm": 25.0,
      "learning_rate": 1.997336361604356e-05,
      "loss": 8.0202,
      "step": 24790
    },
    {
      "epoch": 0.06974500887280254,
      "grad_norm": 27.125,
      "learning_rate": 1.99733421307772e-05,
      "loss": 7.5639,
      "step": 24800
    },
    {
      "epoch": 0.06977313186025125,
      "grad_norm": 45.0,
      "learning_rate": 1.9973320636860744e-05,
      "loss": 7.6013,
      "step": 24810
    },
    {
      "epoch": 0.06980125484769996,
      "grad_norm": 37.75,
      "learning_rate": 1.9973299134294212e-05,
      "loss": 8.8776,
      "step": 24820
    },
    {
      "epoch": 0.06982937783514867,
      "grad_norm": 33.75,
      "learning_rate": 1.9973277623077623e-05,
      "loss": 7.1823,
      "step": 24830
    },
    {
      "epoch": 0.06985750082259738,
      "grad_norm": 24.75,
      "learning_rate": 1.9973256103210995e-05,
      "loss": 7.9224,
      "step": 24840
    },
    {
      "epoch": 0.0698856238100461,
      "grad_norm": 23.0,
      "learning_rate": 1.997323457469435e-05,
      "loss": 8.2273,
      "step": 24850
    },
    {
      "epoch": 0.0699137467974948,
      "grad_norm": 36.0,
      "learning_rate": 1.9973213037527703e-05,
      "loss": 6.5648,
      "step": 24860
    },
    {
      "epoch": 0.06994186978494352,
      "grad_norm": 28.75,
      "learning_rate": 1.9973191491711068e-05,
      "loss": 8.4503,
      "step": 24870
    },
    {
      "epoch": 0.06996999277239223,
      "grad_norm": 42.5,
      "learning_rate": 1.9973169937244476e-05,
      "loss": 8.266,
      "step": 24880
    },
    {
      "epoch": 0.06999811575984094,
      "grad_norm": 40.75,
      "learning_rate": 1.9973148374127932e-05,
      "loss": 7.3302,
      "step": 24890
    },
    {
      "epoch": 0.07002623874728965,
      "grad_norm": 32.5,
      "learning_rate": 1.9973126802361468e-05,
      "loss": 7.5562,
      "step": 24900
    },
    {
      "epoch": 0.07005436173473836,
      "grad_norm": 33.75,
      "learning_rate": 1.997310522194509e-05,
      "loss": 7.698,
      "step": 24910
    },
    {
      "epoch": 0.07008248472218707,
      "grad_norm": 28.875,
      "learning_rate": 1.9973083632878827e-05,
      "loss": 7.8756,
      "step": 24920
    },
    {
      "epoch": 0.07011060770963579,
      "grad_norm": 35.25,
      "learning_rate": 1.9973062035162692e-05,
      "loss": 8.3029,
      "step": 24930
    },
    {
      "epoch": 0.07013873069708448,
      "grad_norm": 36.0,
      "learning_rate": 1.9973040428796704e-05,
      "loss": 8.2363,
      "step": 24940
    },
    {
      "epoch": 0.0701668536845332,
      "grad_norm": 33.75,
      "learning_rate": 1.9973018813780886e-05,
      "loss": 7.9192,
      "step": 24950
    },
    {
      "epoch": 0.0701949766719819,
      "grad_norm": 23.75,
      "learning_rate": 1.997299719011525e-05,
      "loss": 7.7031,
      "step": 24960
    },
    {
      "epoch": 0.07022309965943062,
      "grad_norm": 56.75,
      "learning_rate": 1.9972975557799824e-05,
      "loss": 8.0672,
      "step": 24970
    },
    {
      "epoch": 0.07025122264687933,
      "grad_norm": 24.375,
      "learning_rate": 1.9972953916834616e-05,
      "loss": 7.9685,
      "step": 24980
    },
    {
      "epoch": 0.07027934563432804,
      "grad_norm": 33.75,
      "learning_rate": 1.9972932267219654e-05,
      "loss": 7.5474,
      "step": 24990
    },
    {
      "epoch": 0.07030746862177675,
      "grad_norm": 42.25,
      "learning_rate": 1.997291060895495e-05,
      "loss": 7.2416,
      "step": 25000
    },
    {
      "epoch": 0.07033559160922546,
      "grad_norm": 28.375,
      "learning_rate": 1.9972888942040527e-05,
      "loss": 8.4164,
      "step": 25010
    },
    {
      "epoch": 0.07036371459667418,
      "grad_norm": 40.5,
      "learning_rate": 1.9972867266476402e-05,
      "loss": 7.7495,
      "step": 25020
    },
    {
      "epoch": 0.07039183758412289,
      "grad_norm": 32.75,
      "learning_rate": 1.9972845582262594e-05,
      "loss": 8.6208,
      "step": 25030
    },
    {
      "epoch": 0.0704199605715716,
      "grad_norm": 42.5,
      "learning_rate": 1.9972823889399122e-05,
      "loss": 7.58,
      "step": 25040
    },
    {
      "epoch": 0.07044808355902031,
      "grad_norm": 23.625,
      "learning_rate": 1.9972802187886007e-05,
      "loss": 7.3101,
      "step": 25050
    },
    {
      "epoch": 0.07047620654646902,
      "grad_norm": 37.75,
      "learning_rate": 1.9972780477723263e-05,
      "loss": 8.8442,
      "step": 25060
    },
    {
      "epoch": 0.07050432953391773,
      "grad_norm": 26.25,
      "learning_rate": 1.997275875891091e-05,
      "loss": 7.3642,
      "step": 25070
    },
    {
      "epoch": 0.07053245252136645,
      "grad_norm": 37.25,
      "learning_rate": 1.9972737031448972e-05,
      "loss": 8.5764,
      "step": 25080
    },
    {
      "epoch": 0.07056057550881516,
      "grad_norm": 26.375,
      "learning_rate": 1.9972715295337465e-05,
      "loss": 8.2793,
      "step": 25090
    },
    {
      "epoch": 0.07058869849626385,
      "grad_norm": 39.5,
      "learning_rate": 1.9972693550576403e-05,
      "loss": 8.0837,
      "step": 25100
    },
    {
      "epoch": 0.07061682148371257,
      "grad_norm": 28.625,
      "learning_rate": 1.997267179716581e-05,
      "loss": 8.0924,
      "step": 25110
    },
    {
      "epoch": 0.07064494447116128,
      "grad_norm": 36.25,
      "learning_rate": 1.9972650035105703e-05,
      "loss": 8.1165,
      "step": 25120
    },
    {
      "epoch": 0.07067306745860999,
      "grad_norm": 22.125,
      "learning_rate": 1.9972628264396102e-05,
      "loss": 8.1356,
      "step": 25130
    },
    {
      "epoch": 0.0707011904460587,
      "grad_norm": 29.625,
      "learning_rate": 1.9972606485037028e-05,
      "loss": 8.1969,
      "step": 25140
    },
    {
      "epoch": 0.07072931343350741,
      "grad_norm": 41.75,
      "learning_rate": 1.9972584697028497e-05,
      "loss": 8.1606,
      "step": 25150
    },
    {
      "epoch": 0.07075743642095612,
      "grad_norm": 21.0,
      "learning_rate": 1.9972562900370526e-05,
      "loss": 8.3056,
      "step": 25160
    },
    {
      "epoch": 0.07078555940840484,
      "grad_norm": 45.5,
      "learning_rate": 1.997254109506314e-05,
      "loss": 8.3785,
      "step": 25170
    },
    {
      "epoch": 0.07081368239585355,
      "grad_norm": 29.125,
      "learning_rate": 1.997251928110635e-05,
      "loss": 7.9498,
      "step": 25180
    },
    {
      "epoch": 0.07084180538330226,
      "grad_norm": 28.375,
      "learning_rate": 1.997249745850018e-05,
      "loss": 7.8461,
      "step": 25190
    },
    {
      "epoch": 0.07086992837075097,
      "grad_norm": 31.125,
      "learning_rate": 1.997247562724465e-05,
      "loss": 7.0328,
      "step": 25200
    },
    {
      "epoch": 0.07089805135819968,
      "grad_norm": 42.0,
      "learning_rate": 1.9972453787339775e-05,
      "loss": 7.8984,
      "step": 25210
    },
    {
      "epoch": 0.0709261743456484,
      "grad_norm": 28.125,
      "learning_rate": 1.9972431938785576e-05,
      "loss": 7.9376,
      "step": 25220
    },
    {
      "epoch": 0.0709542973330971,
      "grad_norm": 36.5,
      "learning_rate": 1.997241008158207e-05,
      "loss": 8.0867,
      "step": 25230
    },
    {
      "epoch": 0.07098242032054582,
      "grad_norm": 21.625,
      "learning_rate": 1.9972388215729282e-05,
      "loss": 8.2967,
      "step": 25240
    },
    {
      "epoch": 0.07101054330799453,
      "grad_norm": 24.875,
      "learning_rate": 1.9972366341227222e-05,
      "loss": 7.565,
      "step": 25250
    },
    {
      "epoch": 0.07103866629544323,
      "grad_norm": 29.25,
      "learning_rate": 1.9972344458075915e-05,
      "loss": 8.3284,
      "step": 25260
    },
    {
      "epoch": 0.07106678928289194,
      "grad_norm": 28.75,
      "learning_rate": 1.997232256627538e-05,
      "loss": 7.8378,
      "step": 25270
    },
    {
      "epoch": 0.07109491227034065,
      "grad_norm": 37.25,
      "learning_rate": 1.9972300665825636e-05,
      "loss": 8.1903,
      "step": 25280
    },
    {
      "epoch": 0.07112303525778936,
      "grad_norm": 32.5,
      "learning_rate": 1.99722787567267e-05,
      "loss": 7.521,
      "step": 25290
    },
    {
      "epoch": 0.07115115824523807,
      "grad_norm": 51.5,
      "learning_rate": 1.997225683897859e-05,
      "loss": 7.5105,
      "step": 25300
    },
    {
      "epoch": 0.07117928123268678,
      "grad_norm": 50.0,
      "learning_rate": 1.997223491258133e-05,
      "loss": 8.3483,
      "step": 25310
    },
    {
      "epoch": 0.0712074042201355,
      "grad_norm": 24.875,
      "learning_rate": 1.9972212977534932e-05,
      "loss": 7.9997,
      "step": 25320
    },
    {
      "epoch": 0.0712355272075842,
      "grad_norm": 32.5,
      "learning_rate": 1.9972191033839417e-05,
      "loss": 8.0054,
      "step": 25330
    },
    {
      "epoch": 0.07126365019503292,
      "grad_norm": 39.0,
      "learning_rate": 1.997216908149481e-05,
      "loss": 8.231,
      "step": 25340
    },
    {
      "epoch": 0.07129177318248163,
      "grad_norm": 41.25,
      "learning_rate": 1.9972147120501123e-05,
      "loss": 8.2103,
      "step": 25350
    },
    {
      "epoch": 0.07131989616993034,
      "grad_norm": 25.25,
      "learning_rate": 1.997212515085838e-05,
      "loss": 8.1515,
      "step": 25360
    },
    {
      "epoch": 0.07134801915737905,
      "grad_norm": 32.75,
      "learning_rate": 1.9972103172566598e-05,
      "loss": 8.2716,
      "step": 25370
    },
    {
      "epoch": 0.07137614214482776,
      "grad_norm": 26.25,
      "learning_rate": 1.9972081185625793e-05,
      "loss": 7.4486,
      "step": 25380
    },
    {
      "epoch": 0.07140426513227648,
      "grad_norm": 24.75,
      "learning_rate": 1.9972059190035988e-05,
      "loss": 8.3785,
      "step": 25390
    },
    {
      "epoch": 0.07143238811972519,
      "grad_norm": 31.375,
      "learning_rate": 1.99720371857972e-05,
      "loss": 7.3092,
      "step": 25400
    },
    {
      "epoch": 0.0714605111071739,
      "grad_norm": 33.5,
      "learning_rate": 1.9972015172909452e-05,
      "loss": 8.2852,
      "step": 25410
    },
    {
      "epoch": 0.0714886340946226,
      "grad_norm": 49.25,
      "learning_rate": 1.9971993151372762e-05,
      "loss": 7.9641,
      "step": 25420
    },
    {
      "epoch": 0.07151675708207131,
      "grad_norm": 28.75,
      "learning_rate": 1.9971971121187144e-05,
      "loss": 8.8815,
      "step": 25430
    },
    {
      "epoch": 0.07154488006952002,
      "grad_norm": 25.75,
      "learning_rate": 1.997194908235262e-05,
      "loss": 7.9302,
      "step": 25440
    },
    {
      "epoch": 0.07157300305696873,
      "grad_norm": 27.5,
      "learning_rate": 1.997192703486921e-05,
      "loss": 7.878,
      "step": 25450
    },
    {
      "epoch": 0.07160112604441744,
      "grad_norm": 30.75,
      "learning_rate": 1.9971904978736933e-05,
      "loss": 8.4253,
      "step": 25460
    },
    {
      "epoch": 0.07162924903186615,
      "grad_norm": 38.25,
      "learning_rate": 1.997188291395581e-05,
      "loss": 8.1017,
      "step": 25470
    },
    {
      "epoch": 0.07165737201931487,
      "grad_norm": 32.5,
      "learning_rate": 1.9971860840525855e-05,
      "loss": 7.901,
      "step": 25480
    },
    {
      "epoch": 0.07168549500676358,
      "grad_norm": 49.25,
      "learning_rate": 1.9971838758447092e-05,
      "loss": 8.2286,
      "step": 25490
    },
    {
      "epoch": 0.07171361799421229,
      "grad_norm": 28.375,
      "learning_rate": 1.9971816667719536e-05,
      "loss": 8.9143,
      "step": 25500
    },
    {
      "epoch": 0.071741740981661,
      "grad_norm": 30.375,
      "learning_rate": 1.997179456834321e-05,
      "loss": 7.7938,
      "step": 25510
    },
    {
      "epoch": 0.07176986396910971,
      "grad_norm": 23.25,
      "learning_rate": 1.997177246031813e-05,
      "loss": 7.9066,
      "step": 25520
    },
    {
      "epoch": 0.07179798695655842,
      "grad_norm": 25.75,
      "learning_rate": 1.9971750343644317e-05,
      "loss": 8.3859,
      "step": 25530
    },
    {
      "epoch": 0.07182610994400714,
      "grad_norm": 27.125,
      "learning_rate": 1.9971728218321794e-05,
      "loss": 8.1234,
      "step": 25540
    },
    {
      "epoch": 0.07185423293145585,
      "grad_norm": 30.75,
      "learning_rate": 1.9971706084350574e-05,
      "loss": 8.2773,
      "step": 25550
    },
    {
      "epoch": 0.07188235591890456,
      "grad_norm": 36.75,
      "learning_rate": 1.9971683941730676e-05,
      "loss": 7.9438,
      "step": 25560
    },
    {
      "epoch": 0.07191047890635327,
      "grad_norm": 31.0,
      "learning_rate": 1.997166179046212e-05,
      "loss": 7.5717,
      "step": 25570
    },
    {
      "epoch": 0.07193860189380198,
      "grad_norm": 48.5,
      "learning_rate": 1.9971639630544933e-05,
      "loss": 8.0687,
      "step": 25580
    },
    {
      "epoch": 0.07196672488125068,
      "grad_norm": 43.0,
      "learning_rate": 1.997161746197913e-05,
      "loss": 8.1877,
      "step": 25590
    },
    {
      "epoch": 0.07199484786869939,
      "grad_norm": 34.25,
      "learning_rate": 1.997159528476472e-05,
      "loss": 7.9821,
      "step": 25600
    },
    {
      "epoch": 0.0720229708561481,
      "grad_norm": 22.125,
      "learning_rate": 1.9971573098901732e-05,
      "loss": 7.8202,
      "step": 25610
    },
    {
      "epoch": 0.07205109384359681,
      "grad_norm": 36.5,
      "learning_rate": 1.9971550904390183e-05,
      "loss": 8.0715,
      "step": 25620
    },
    {
      "epoch": 0.07207921683104553,
      "grad_norm": 23.125,
      "learning_rate": 1.9971528701230096e-05,
      "loss": 7.7052,
      "step": 25630
    },
    {
      "epoch": 0.07210733981849424,
      "grad_norm": 26.75,
      "learning_rate": 1.9971506489421485e-05,
      "loss": 7.7468,
      "step": 25640
    },
    {
      "epoch": 0.07213546280594295,
      "grad_norm": 29.75,
      "learning_rate": 1.9971484268964375e-05,
      "loss": 8.5877,
      "step": 25650
    },
    {
      "epoch": 0.07216358579339166,
      "grad_norm": 21.625,
      "learning_rate": 1.997146203985878e-05,
      "loss": 7.6238,
      "step": 25660
    },
    {
      "epoch": 0.07219170878084037,
      "grad_norm": 38.25,
      "learning_rate": 1.997143980210472e-05,
      "loss": 7.8328,
      "step": 25670
    },
    {
      "epoch": 0.07221983176828908,
      "grad_norm": 29.625,
      "learning_rate": 1.9971417555702216e-05,
      "loss": 7.8446,
      "step": 25680
    },
    {
      "epoch": 0.0722479547557378,
      "grad_norm": 28.75,
      "learning_rate": 1.9971395300651285e-05,
      "loss": 7.0579,
      "step": 25690
    },
    {
      "epoch": 0.0722760777431865,
      "grad_norm": 27.875,
      "learning_rate": 1.9971373036951947e-05,
      "loss": 7.6562,
      "step": 25700
    },
    {
      "epoch": 0.07230420073063522,
      "grad_norm": 24.375,
      "learning_rate": 1.9971350764604227e-05,
      "loss": 8.8901,
      "step": 25710
    },
    {
      "epoch": 0.07233232371808393,
      "grad_norm": 40.75,
      "learning_rate": 1.9971328483608138e-05,
      "loss": 7.5222,
      "step": 25720
    },
    {
      "epoch": 0.07236044670553264,
      "grad_norm": 34.25,
      "learning_rate": 1.9971306193963697e-05,
      "loss": 8.1287,
      "step": 25730
    },
    {
      "epoch": 0.07238856969298135,
      "grad_norm": 30.375,
      "learning_rate": 1.997128389567093e-05,
      "loss": 8.1927,
      "step": 25740
    },
    {
      "epoch": 0.07241669268043005,
      "grad_norm": 40.75,
      "learning_rate": 1.9971261588729853e-05,
      "loss": 8.3004,
      "step": 25750
    },
    {
      "epoch": 0.07244481566787876,
      "grad_norm": 23.375,
      "learning_rate": 1.9971239273140485e-05,
      "loss": 7.6001,
      "step": 25760
    },
    {
      "epoch": 0.07247293865532747,
      "grad_norm": 38.5,
      "learning_rate": 1.997121694890285e-05,
      "loss": 7.216,
      "step": 25770
    },
    {
      "epoch": 0.07250106164277619,
      "grad_norm": 24.625,
      "learning_rate": 1.997119461601696e-05,
      "loss": 7.666,
      "step": 25780
    },
    {
      "epoch": 0.0725291846302249,
      "grad_norm": 28.0,
      "learning_rate": 1.997117227448284e-05,
      "loss": 8.5669,
      "step": 25790
    },
    {
      "epoch": 0.07255730761767361,
      "grad_norm": 39.0,
      "learning_rate": 1.9971149924300506e-05,
      "loss": 7.8265,
      "step": 25800
    },
    {
      "epoch": 0.07258543060512232,
      "grad_norm": 47.75,
      "learning_rate": 1.997112756546998e-05,
      "loss": 8.1448,
      "step": 25810
    },
    {
      "epoch": 0.07261355359257103,
      "grad_norm": 35.0,
      "learning_rate": 1.9971105197991278e-05,
      "loss": 8.3019,
      "step": 25820
    },
    {
      "epoch": 0.07264167658001974,
      "grad_norm": 78.0,
      "learning_rate": 1.9971082821864424e-05,
      "loss": 7.1803,
      "step": 25830
    },
    {
      "epoch": 0.07266979956746845,
      "grad_norm": 25.125,
      "learning_rate": 1.997106043708943e-05,
      "loss": 7.9368,
      "step": 25840
    },
    {
      "epoch": 0.07269792255491717,
      "grad_norm": 36.5,
      "learning_rate": 1.9971038043666323e-05,
      "loss": 8.7768,
      "step": 25850
    },
    {
      "epoch": 0.07272604554236588,
      "grad_norm": 31.375,
      "learning_rate": 1.997101564159512e-05,
      "loss": 8.032,
      "step": 25860
    },
    {
      "epoch": 0.07275416852981459,
      "grad_norm": 42.0,
      "learning_rate": 1.9970993230875845e-05,
      "loss": 7.9507,
      "step": 25870
    },
    {
      "epoch": 0.0727822915172633,
      "grad_norm": 29.5,
      "learning_rate": 1.9970970811508504e-05,
      "loss": 7.6736,
      "step": 25880
    },
    {
      "epoch": 0.07281041450471201,
      "grad_norm": 45.75,
      "learning_rate": 1.997094838349313e-05,
      "loss": 8.3579,
      "step": 25890
    },
    {
      "epoch": 0.07283853749216072,
      "grad_norm": 20.75,
      "learning_rate": 1.997092594682974e-05,
      "loss": 8.0961,
      "step": 25900
    },
    {
      "epoch": 0.07286666047960942,
      "grad_norm": 32.0,
      "learning_rate": 1.9970903501518342e-05,
      "loss": 7.8379,
      "step": 25910
    },
    {
      "epoch": 0.07289478346705813,
      "grad_norm": 30.625,
      "learning_rate": 1.9970881047558972e-05,
      "loss": 8.0047,
      "step": 25920
    },
    {
      "epoch": 0.07292290645450684,
      "grad_norm": 40.25,
      "learning_rate": 1.9970858584951643e-05,
      "loss": 7.9365,
      "step": 25930
    },
    {
      "epoch": 0.07295102944195556,
      "grad_norm": 23.5,
      "learning_rate": 1.997083611369637e-05,
      "loss": 8.2069,
      "step": 25940
    },
    {
      "epoch": 0.07297915242940427,
      "grad_norm": 23.625,
      "learning_rate": 1.9970813633793176e-05,
      "loss": 7.8206,
      "step": 25950
    },
    {
      "epoch": 0.07300727541685298,
      "grad_norm": 28.375,
      "learning_rate": 1.997079114524208e-05,
      "loss": 8.1541,
      "step": 25960
    },
    {
      "epoch": 0.07303539840430169,
      "grad_norm": 22.625,
      "learning_rate": 1.9970768648043103e-05,
      "loss": 8.0545,
      "step": 25970
    },
    {
      "epoch": 0.0730635213917504,
      "grad_norm": 41.5,
      "learning_rate": 1.9970746142196263e-05,
      "loss": 7.9122,
      "step": 25980
    },
    {
      "epoch": 0.07309164437919911,
      "grad_norm": 33.0,
      "learning_rate": 1.9970723627701577e-05,
      "loss": 7.4905,
      "step": 25990
    },
    {
      "epoch": 0.07311976736664783,
      "grad_norm": 34.25,
      "learning_rate": 1.9970701104559073e-05,
      "loss": 8.0898,
      "step": 26000
    },
    {
      "epoch": 0.07314789035409654,
      "grad_norm": 22.75,
      "learning_rate": 1.9970678572768763e-05,
      "loss": 7.8744,
      "step": 26010
    },
    {
      "epoch": 0.07317601334154525,
      "grad_norm": 30.125,
      "learning_rate": 1.9970656032330663e-05,
      "loss": 7.903,
      "step": 26020
    },
    {
      "epoch": 0.07320413632899396,
      "grad_norm": 26.375,
      "learning_rate": 1.9970633483244807e-05,
      "loss": 7.7607,
      "step": 26030
    },
    {
      "epoch": 0.07323225931644267,
      "grad_norm": 35.5,
      "learning_rate": 1.99706109255112e-05,
      "loss": 7.996,
      "step": 26040
    },
    {
      "epoch": 0.07326038230389138,
      "grad_norm": 30.25,
      "learning_rate": 1.997058835912987e-05,
      "loss": 7.836,
      "step": 26050
    },
    {
      "epoch": 0.0732885052913401,
      "grad_norm": 25.875,
      "learning_rate": 1.997056578410083e-05,
      "loss": 8.619,
      "step": 26060
    },
    {
      "epoch": 0.07331662827878879,
      "grad_norm": 33.75,
      "learning_rate": 1.9970543200424105e-05,
      "loss": 7.37,
      "step": 26070
    },
    {
      "epoch": 0.0733447512662375,
      "grad_norm": 24.625,
      "learning_rate": 1.9970520608099714e-05,
      "loss": 7.3459,
      "step": 26080
    },
    {
      "epoch": 0.07337287425368622,
      "grad_norm": 37.25,
      "learning_rate": 1.9970498007127673e-05,
      "loss": 7.7747,
      "step": 26090
    },
    {
      "epoch": 0.07340099724113493,
      "grad_norm": 44.0,
      "learning_rate": 1.9970475397508008e-05,
      "loss": 8.2631,
      "step": 26100
    },
    {
      "epoch": 0.07342912022858364,
      "grad_norm": 40.5,
      "learning_rate": 1.9970452779240733e-05,
      "loss": 7.8677,
      "step": 26110
    },
    {
      "epoch": 0.07345724321603235,
      "grad_norm": 34.75,
      "learning_rate": 1.9970430152325872e-05,
      "loss": 7.909,
      "step": 26120
    },
    {
      "epoch": 0.07348536620348106,
      "grad_norm": 32.75,
      "learning_rate": 1.9970407516763438e-05,
      "loss": 8.0213,
      "step": 26130
    },
    {
      "epoch": 0.07351348919092977,
      "grad_norm": 24.0,
      "learning_rate": 1.9970384872553454e-05,
      "loss": 8.4118,
      "step": 26140
    },
    {
      "epoch": 0.07354161217837848,
      "grad_norm": 30.625,
      "learning_rate": 1.9970362219695946e-05,
      "loss": 7.9723,
      "step": 26150
    },
    {
      "epoch": 0.0735697351658272,
      "grad_norm": 25.25,
      "learning_rate": 1.9970339558190922e-05,
      "loss": 8.612,
      "step": 26160
    },
    {
      "epoch": 0.07359785815327591,
      "grad_norm": 21.0,
      "learning_rate": 1.997031688803841e-05,
      "loss": 7.9131,
      "step": 26170
    },
    {
      "epoch": 0.07362598114072462,
      "grad_norm": 25.5,
      "learning_rate": 1.9970294209238427e-05,
      "loss": 8.1793,
      "step": 26180
    },
    {
      "epoch": 0.07365410412817333,
      "grad_norm": 48.0,
      "learning_rate": 1.9970271521790993e-05,
      "loss": 8.0626,
      "step": 26190
    },
    {
      "epoch": 0.07368222711562204,
      "grad_norm": 21.75,
      "learning_rate": 1.9970248825696125e-05,
      "loss": 7.7248,
      "step": 26200
    },
    {
      "epoch": 0.07371035010307075,
      "grad_norm": 23.5,
      "learning_rate": 1.9970226120953847e-05,
      "loss": 8.4138,
      "step": 26210
    },
    {
      "epoch": 0.07373847309051947,
      "grad_norm": 27.75,
      "learning_rate": 1.997020340756418e-05,
      "loss": 8.2009,
      "step": 26220
    },
    {
      "epoch": 0.07376659607796818,
      "grad_norm": 40.75,
      "learning_rate": 1.9970180685527135e-05,
      "loss": 8.5662,
      "step": 26230
    },
    {
      "epoch": 0.07379471906541687,
      "grad_norm": 62.25,
      "learning_rate": 1.997015795484274e-05,
      "loss": 9.0719,
      "step": 26240
    },
    {
      "epoch": 0.07382284205286559,
      "grad_norm": 26.625,
      "learning_rate": 1.9970135215511016e-05,
      "loss": 7.6147,
      "step": 26250
    },
    {
      "epoch": 0.0738509650403143,
      "grad_norm": 29.625,
      "learning_rate": 1.997011246753197e-05,
      "loss": 7.541,
      "step": 26260
    },
    {
      "epoch": 0.07387908802776301,
      "grad_norm": 35.0,
      "learning_rate": 1.997008971090564e-05,
      "loss": 7.3945,
      "step": 26270
    },
    {
      "epoch": 0.07390721101521172,
      "grad_norm": 26.375,
      "learning_rate": 1.997006694563203e-05,
      "loss": 7.8403,
      "step": 26280
    },
    {
      "epoch": 0.07393533400266043,
      "grad_norm": 45.0,
      "learning_rate": 1.997004417171117e-05,
      "loss": 8.0331,
      "step": 26290
    },
    {
      "epoch": 0.07396345699010914,
      "grad_norm": 25.5,
      "learning_rate": 1.997002138914307e-05,
      "loss": 7.7275,
      "step": 26300
    },
    {
      "epoch": 0.07399157997755786,
      "grad_norm": 29.0,
      "learning_rate": 1.9969998597927762e-05,
      "loss": 8.2184,
      "step": 26310
    },
    {
      "epoch": 0.07401970296500657,
      "grad_norm": 65.5,
      "learning_rate": 1.9969975798065256e-05,
      "loss": 8.6619,
      "step": 26320
    },
    {
      "epoch": 0.07404782595245528,
      "grad_norm": 28.625,
      "learning_rate": 1.9969952989555574e-05,
      "loss": 7.2831,
      "step": 26330
    },
    {
      "epoch": 0.07407594893990399,
      "grad_norm": 25.875,
      "learning_rate": 1.996993017239874e-05,
      "loss": 8.0372,
      "step": 26340
    },
    {
      "epoch": 0.0741040719273527,
      "grad_norm": 33.0,
      "learning_rate": 1.9969907346594768e-05,
      "loss": 8.4383,
      "step": 26350
    },
    {
      "epoch": 0.07413219491480141,
      "grad_norm": 29.625,
      "learning_rate": 1.9969884512143682e-05,
      "loss": 7.9763,
      "step": 26360
    },
    {
      "epoch": 0.07416031790225013,
      "grad_norm": 24.125,
      "learning_rate": 1.99698616690455e-05,
      "loss": 7.6821,
      "step": 26370
    },
    {
      "epoch": 0.07418844088969884,
      "grad_norm": 30.5,
      "learning_rate": 1.996983881730024e-05,
      "loss": 7.7474,
      "step": 26380
    },
    {
      "epoch": 0.07421656387714755,
      "grad_norm": 24.5,
      "learning_rate": 1.9969815956907927e-05,
      "loss": 7.6051,
      "step": 26390
    },
    {
      "epoch": 0.07424468686459625,
      "grad_norm": 46.25,
      "learning_rate": 1.9969793087868576e-05,
      "loss": 8.1298,
      "step": 26400
    },
    {
      "epoch": 0.07427280985204496,
      "grad_norm": 20.625,
      "learning_rate": 1.9969770210182208e-05,
      "loss": 7.1447,
      "step": 26410
    },
    {
      "epoch": 0.07430093283949367,
      "grad_norm": 22.375,
      "learning_rate": 1.9969747323848842e-05,
      "loss": 7.8355,
      "step": 26420
    },
    {
      "epoch": 0.07432905582694238,
      "grad_norm": 39.25,
      "learning_rate": 1.9969724428868504e-05,
      "loss": 8.0793,
      "step": 26430
    },
    {
      "epoch": 0.07435717881439109,
      "grad_norm": 22.875,
      "learning_rate": 1.9969701525241206e-05,
      "loss": 8.8741,
      "step": 26440
    },
    {
      "epoch": 0.0743853018018398,
      "grad_norm": 58.5,
      "learning_rate": 1.996967861296697e-05,
      "loss": 7.9702,
      "step": 26450
    },
    {
      "epoch": 0.07441342478928852,
      "grad_norm": 34.25,
      "learning_rate": 1.9969655692045816e-05,
      "loss": 7.8223,
      "step": 26460
    },
    {
      "epoch": 0.07444154777673723,
      "grad_norm": 38.25,
      "learning_rate": 1.996963276247777e-05,
      "loss": 8.3986,
      "step": 26470
    },
    {
      "epoch": 0.07446967076418594,
      "grad_norm": 26.375,
      "learning_rate": 1.9969609824262842e-05,
      "loss": 7.8544,
      "step": 26480
    },
    {
      "epoch": 0.07449779375163465,
      "grad_norm": 26.5,
      "learning_rate": 1.9969586877401058e-05,
      "loss": 7.868,
      "step": 26490
    },
    {
      "epoch": 0.07452591673908336,
      "grad_norm": 25.125,
      "learning_rate": 1.9969563921892437e-05,
      "loss": 7.4325,
      "step": 26500
    },
    {
      "epoch": 0.07455403972653207,
      "grad_norm": 23.25,
      "learning_rate": 1.9969540957736998e-05,
      "loss": 8.7173,
      "step": 26510
    },
    {
      "epoch": 0.07458216271398078,
      "grad_norm": 29.0,
      "learning_rate": 1.996951798493476e-05,
      "loss": 8.479,
      "step": 26520
    },
    {
      "epoch": 0.0746102857014295,
      "grad_norm": 29.625,
      "learning_rate": 1.9969495003485743e-05,
      "loss": 8.4633,
      "step": 26530
    },
    {
      "epoch": 0.07463840868887821,
      "grad_norm": 23.375,
      "learning_rate": 1.996947201338997e-05,
      "loss": 8.0694,
      "step": 26540
    },
    {
      "epoch": 0.07466653167632692,
      "grad_norm": 27.625,
      "learning_rate": 1.9969449014647456e-05,
      "loss": 8.166,
      "step": 26550
    },
    {
      "epoch": 0.07469465466377562,
      "grad_norm": 32.5,
      "learning_rate": 1.996942600725823e-05,
      "loss": 8.0356,
      "step": 26560
    },
    {
      "epoch": 0.07472277765122433,
      "grad_norm": 28.375,
      "learning_rate": 1.9969402991222303e-05,
      "loss": 8.3624,
      "step": 26570
    },
    {
      "epoch": 0.07475090063867304,
      "grad_norm": 25.625,
      "learning_rate": 1.9969379966539696e-05,
      "loss": 7.8257,
      "step": 26580
    },
    {
      "epoch": 0.07477902362612175,
      "grad_norm": 23.125,
      "learning_rate": 1.996935693321043e-05,
      "loss": 7.9279,
      "step": 26590
    },
    {
      "epoch": 0.07480714661357046,
      "grad_norm": 42.75,
      "learning_rate": 1.996933389123453e-05,
      "loss": 7.7669,
      "step": 26600
    },
    {
      "epoch": 0.07483526960101917,
      "grad_norm": 23.25,
      "learning_rate": 1.996931084061201e-05,
      "loss": 8.0348,
      "step": 26610
    },
    {
      "epoch": 0.07486339258846789,
      "grad_norm": 50.25,
      "learning_rate": 1.9969287781342892e-05,
      "loss": 8.3616,
      "step": 26620
    },
    {
      "epoch": 0.0748915155759166,
      "grad_norm": 24.0,
      "learning_rate": 1.9969264713427195e-05,
      "loss": 8.2894,
      "step": 26630
    },
    {
      "epoch": 0.07491963856336531,
      "grad_norm": 23.125,
      "learning_rate": 1.996924163686494e-05,
      "loss": 7.7881,
      "step": 26640
    },
    {
      "epoch": 0.07494776155081402,
      "grad_norm": 29.125,
      "learning_rate": 1.9969218551656148e-05,
      "loss": 8.1339,
      "step": 26650
    },
    {
      "epoch": 0.07497588453826273,
      "grad_norm": 25.625,
      "learning_rate": 1.9969195457800835e-05,
      "loss": 7.4664,
      "step": 26660
    },
    {
      "epoch": 0.07500400752571144,
      "grad_norm": 27.25,
      "learning_rate": 1.9969172355299027e-05,
      "loss": 8.2754,
      "step": 26670
    },
    {
      "epoch": 0.07503213051316016,
      "grad_norm": 44.25,
      "learning_rate": 1.9969149244150737e-05,
      "loss": 7.5049,
      "step": 26680
    },
    {
      "epoch": 0.07506025350060887,
      "grad_norm": 26.875,
      "learning_rate": 1.9969126124355993e-05,
      "loss": 9.2107,
      "step": 26690
    },
    {
      "epoch": 0.07508837648805758,
      "grad_norm": 30.625,
      "learning_rate": 1.9969102995914807e-05,
      "loss": 7.4617,
      "step": 26700
    },
    {
      "epoch": 0.07511649947550629,
      "grad_norm": 24.25,
      "learning_rate": 1.9969079858827208e-05,
      "loss": 7.648,
      "step": 26710
    },
    {
      "epoch": 0.07514462246295499,
      "grad_norm": 30.125,
      "learning_rate": 1.9969056713093207e-05,
      "loss": 8.0947,
      "step": 26720
    },
    {
      "epoch": 0.0751727454504037,
      "grad_norm": 22.0,
      "learning_rate": 1.996903355871283e-05,
      "loss": 7.5861,
      "step": 26730
    },
    {
      "epoch": 0.07520086843785241,
      "grad_norm": 21.875,
      "learning_rate": 1.9969010395686094e-05,
      "loss": 8.7944,
      "step": 26740
    },
    {
      "epoch": 0.07522899142530112,
      "grad_norm": 30.875,
      "learning_rate": 1.9968987224013024e-05,
      "loss": 8.1003,
      "step": 26750
    },
    {
      "epoch": 0.07525711441274983,
      "grad_norm": 41.5,
      "learning_rate": 1.9968964043693632e-05,
      "loss": 8.1872,
      "step": 26760
    },
    {
      "epoch": 0.07528523740019855,
      "grad_norm": 26.25,
      "learning_rate": 1.9968940854727944e-05,
      "loss": 8.4318,
      "step": 26770
    },
    {
      "epoch": 0.07531336038764726,
      "grad_norm": 29.0,
      "learning_rate": 1.996891765711598e-05,
      "loss": 7.7735,
      "step": 26780
    },
    {
      "epoch": 0.07534148337509597,
      "grad_norm": 28.0,
      "learning_rate": 1.9968894450857757e-05,
      "loss": 7.9705,
      "step": 26790
    },
    {
      "epoch": 0.07536960636254468,
      "grad_norm": 23.375,
      "learning_rate": 1.9968871235953298e-05,
      "loss": 7.4879,
      "step": 26800
    },
    {
      "epoch": 0.07539772934999339,
      "grad_norm": 33.0,
      "learning_rate": 1.996884801240262e-05,
      "loss": 8.0119,
      "step": 26810
    },
    {
      "epoch": 0.0754258523374421,
      "grad_norm": 28.0,
      "learning_rate": 1.9968824780205748e-05,
      "loss": 7.8967,
      "step": 26820
    },
    {
      "epoch": 0.07545397532489082,
      "grad_norm": 24.125,
      "learning_rate": 1.99688015393627e-05,
      "loss": 8.532,
      "step": 26830
    },
    {
      "epoch": 0.07548209831233953,
      "grad_norm": 34.25,
      "learning_rate": 1.9968778289873493e-05,
      "loss": 7.9858,
      "step": 26840
    },
    {
      "epoch": 0.07551022129978824,
      "grad_norm": 22.5,
      "learning_rate": 1.996875503173815e-05,
      "loss": 7.7781,
      "step": 26850
    },
    {
      "epoch": 0.07553834428723695,
      "grad_norm": 49.25,
      "learning_rate": 1.9968731764956694e-05,
      "loss": 8.0445,
      "step": 26860
    },
    {
      "epoch": 0.07556646727468566,
      "grad_norm": 42.5,
      "learning_rate": 1.9968708489529135e-05,
      "loss": 8.1704,
      "step": 26870
    },
    {
      "epoch": 0.07559459026213437,
      "grad_norm": 33.0,
      "learning_rate": 1.9968685205455505e-05,
      "loss": 7.879,
      "step": 26880
    },
    {
      "epoch": 0.07562271324958307,
      "grad_norm": 23.75,
      "learning_rate": 1.996866191273582e-05,
      "loss": 7.7216,
      "step": 26890
    },
    {
      "epoch": 0.07565083623703178,
      "grad_norm": 33.5,
      "learning_rate": 1.9968638611370097e-05,
      "loss": 7.7943,
      "step": 26900
    },
    {
      "epoch": 0.0756789592244805,
      "grad_norm": 43.25,
      "learning_rate": 1.9968615301358363e-05,
      "loss": 7.9838,
      "step": 26910
    },
    {
      "epoch": 0.0757070822119292,
      "grad_norm": 25.0,
      "learning_rate": 1.9968591982700628e-05,
      "loss": 8.035,
      "step": 26920
    },
    {
      "epoch": 0.07573520519937792,
      "grad_norm": 47.0,
      "learning_rate": 1.9968568655396923e-05,
      "loss": 8.625,
      "step": 26930
    },
    {
      "epoch": 0.07576332818682663,
      "grad_norm": 33.75,
      "learning_rate": 1.996854531944726e-05,
      "loss": 7.7114,
      "step": 26940
    },
    {
      "epoch": 0.07579145117427534,
      "grad_norm": 39.75,
      "learning_rate": 1.9968521974851667e-05,
      "loss": 7.8385,
      "step": 26950
    },
    {
      "epoch": 0.07581957416172405,
      "grad_norm": 30.25,
      "learning_rate": 1.9968498621610156e-05,
      "loss": 8.5148,
      "step": 26960
    },
    {
      "epoch": 0.07584769714917276,
      "grad_norm": 32.25,
      "learning_rate": 1.9968475259722753e-05,
      "loss": 8.4401,
      "step": 26970
    },
    {
      "epoch": 0.07587582013662147,
      "grad_norm": 21.5,
      "learning_rate": 1.996845188918948e-05,
      "loss": 8.1494,
      "step": 26980
    },
    {
      "epoch": 0.07590394312407019,
      "grad_norm": 22.625,
      "learning_rate": 1.996842851001035e-05,
      "loss": 8.5612,
      "step": 26990
    },
    {
      "epoch": 0.0759320661115189,
      "grad_norm": 34.5,
      "learning_rate": 1.9968405122185387e-05,
      "loss": 7.6477,
      "step": 27000
    },
    {
      "epoch": 0.07596018909896761,
      "grad_norm": 34.0,
      "learning_rate": 1.9968381725714612e-05,
      "loss": 8.4032,
      "step": 27010
    },
    {
      "epoch": 0.07598831208641632,
      "grad_norm": 26.125,
      "learning_rate": 1.9968358320598042e-05,
      "loss": 7.5209,
      "step": 27020
    },
    {
      "epoch": 0.07601643507386503,
      "grad_norm": 27.375,
      "learning_rate": 1.9968334906835706e-05,
      "loss": 7.9156,
      "step": 27030
    },
    {
      "epoch": 0.07604455806131374,
      "grad_norm": 37.75,
      "learning_rate": 1.9968311484427612e-05,
      "loss": 8.0194,
      "step": 27040
    },
    {
      "epoch": 0.07607268104876244,
      "grad_norm": 38.5,
      "learning_rate": 1.996828805337379e-05,
      "loss": 7.854,
      "step": 27050
    },
    {
      "epoch": 0.07610080403621115,
      "grad_norm": 24.375,
      "learning_rate": 1.9968264613674254e-05,
      "loss": 7.9612,
      "step": 27060
    },
    {
      "epoch": 0.07612892702365986,
      "grad_norm": 21.75,
      "learning_rate": 1.996824116532903e-05,
      "loss": 8.1747,
      "step": 27070
    },
    {
      "epoch": 0.07615705001110858,
      "grad_norm": 30.375,
      "learning_rate": 1.9968217708338136e-05,
      "loss": 8.6858,
      "step": 27080
    },
    {
      "epoch": 0.07618517299855729,
      "grad_norm": 43.5,
      "learning_rate": 1.9968194242701587e-05,
      "loss": 8.5937,
      "step": 27090
    },
    {
      "epoch": 0.076213295986006,
      "grad_norm": 27.0,
      "learning_rate": 1.9968170768419414e-05,
      "loss": 7.9049,
      "step": 27100
    },
    {
      "epoch": 0.07624141897345471,
      "grad_norm": 26.125,
      "learning_rate": 1.996814728549163e-05,
      "loss": 8.4836,
      "step": 27110
    },
    {
      "epoch": 0.07626954196090342,
      "grad_norm": 25.75,
      "learning_rate": 1.9968123793918255e-05,
      "loss": 8.8337,
      "step": 27120
    },
    {
      "epoch": 0.07629766494835213,
      "grad_norm": 28.25,
      "learning_rate": 1.996810029369931e-05,
      "loss": 8.6502,
      "step": 27130
    },
    {
      "epoch": 0.07632578793580085,
      "grad_norm": 34.5,
      "learning_rate": 1.996807678483482e-05,
      "loss": 7.7829,
      "step": 27140
    },
    {
      "epoch": 0.07635391092324956,
      "grad_norm": 30.625,
      "learning_rate": 1.9968053267324802e-05,
      "loss": 8.8064,
      "step": 27150
    },
    {
      "epoch": 0.07638203391069827,
      "grad_norm": 48.25,
      "learning_rate": 1.9968029741169277e-05,
      "loss": 7.8698,
      "step": 27160
    },
    {
      "epoch": 0.07641015689814698,
      "grad_norm": 23.625,
      "learning_rate": 1.9968006206368264e-05,
      "loss": 7.7861,
      "step": 27170
    },
    {
      "epoch": 0.07643827988559569,
      "grad_norm": 29.625,
      "learning_rate": 1.9967982662921785e-05,
      "loss": 7.2716,
      "step": 27180
    },
    {
      "epoch": 0.0764664028730444,
      "grad_norm": 24.375,
      "learning_rate": 1.9967959110829857e-05,
      "loss": 8.5829,
      "step": 27190
    },
    {
      "epoch": 0.07649452586049311,
      "grad_norm": 23.25,
      "learning_rate": 1.9967935550092505e-05,
      "loss": 8.052,
      "step": 27200
    },
    {
      "epoch": 0.07652264884794181,
      "grad_norm": 34.0,
      "learning_rate": 1.9967911980709748e-05,
      "loss": 7.4781,
      "step": 27210
    },
    {
      "epoch": 0.07655077183539052,
      "grad_norm": 25.0,
      "learning_rate": 1.9967888402681604e-05,
      "loss": 7.6999,
      "step": 27220
    },
    {
      "epoch": 0.07657889482283924,
      "grad_norm": 26.25,
      "learning_rate": 1.99678648160081e-05,
      "loss": 7.7534,
      "step": 27230
    },
    {
      "epoch": 0.07660701781028795,
      "grad_norm": 32.25,
      "learning_rate": 1.9967841220689248e-05,
      "loss": 8.7501,
      "step": 27240
    },
    {
      "epoch": 0.07663514079773666,
      "grad_norm": 23.375,
      "learning_rate": 1.9967817616725074e-05,
      "loss": 8.1692,
      "step": 27250
    },
    {
      "epoch": 0.07666326378518537,
      "grad_norm": 29.125,
      "learning_rate": 1.9967794004115593e-05,
      "loss": 7.9667,
      "step": 27260
    },
    {
      "epoch": 0.07669138677263408,
      "grad_norm": 23.625,
      "learning_rate": 1.9967770382860834e-05,
      "loss": 7.9357,
      "step": 27270
    },
    {
      "epoch": 0.0767195097600828,
      "grad_norm": 22.0,
      "learning_rate": 1.9967746752960812e-05,
      "loss": 7.7933,
      "step": 27280
    },
    {
      "epoch": 0.0767476327475315,
      "grad_norm": 42.25,
      "learning_rate": 1.996772311441555e-05,
      "loss": 7.8295,
      "step": 27290
    },
    {
      "epoch": 0.07677575573498022,
      "grad_norm": 24.125,
      "learning_rate": 1.9967699467225064e-05,
      "loss": 8.3429,
      "step": 27300
    },
    {
      "epoch": 0.07680387872242893,
      "grad_norm": 28.75,
      "learning_rate": 1.9967675811389378e-05,
      "loss": 7.742,
      "step": 27310
    },
    {
      "epoch": 0.07683200170987764,
      "grad_norm": 27.625,
      "learning_rate": 1.996765214690851e-05,
      "loss": 7.7158,
      "step": 27320
    },
    {
      "epoch": 0.07686012469732635,
      "grad_norm": 27.75,
      "learning_rate": 1.9967628473782487e-05,
      "loss": 8.1685,
      "step": 27330
    },
    {
      "epoch": 0.07688824768477506,
      "grad_norm": 19.25,
      "learning_rate": 1.996760479201132e-05,
      "loss": 7.8431,
      "step": 27340
    },
    {
      "epoch": 0.07691637067222377,
      "grad_norm": 41.5,
      "learning_rate": 1.9967581101595036e-05,
      "loss": 7.8701,
      "step": 27350
    },
    {
      "epoch": 0.07694449365967249,
      "grad_norm": 45.25,
      "learning_rate": 1.9967557402533656e-05,
      "loss": 7.083,
      "step": 27360
    },
    {
      "epoch": 0.07697261664712118,
      "grad_norm": 23.375,
      "learning_rate": 1.9967533694827197e-05,
      "loss": 7.8217,
      "step": 27370
    },
    {
      "epoch": 0.0770007396345699,
      "grad_norm": 32.0,
      "learning_rate": 1.996750997847568e-05,
      "loss": 8.0724,
      "step": 27380
    },
    {
      "epoch": 0.0770288626220186,
      "grad_norm": 38.0,
      "learning_rate": 1.996748625347913e-05,
      "loss": 8.8073,
      "step": 27390
    },
    {
      "epoch": 0.07705698560946732,
      "grad_norm": 28.25,
      "learning_rate": 1.996746251983756e-05,
      "loss": 8.43,
      "step": 27400
    },
    {
      "epoch": 0.07708510859691603,
      "grad_norm": 22.0,
      "learning_rate": 1.9967438777551e-05,
      "loss": 8.5713,
      "step": 27410
    },
    {
      "epoch": 0.07711323158436474,
      "grad_norm": 29.375,
      "learning_rate": 1.996741502661946e-05,
      "loss": 7.3314,
      "step": 27420
    },
    {
      "epoch": 0.07714135457181345,
      "grad_norm": 26.5,
      "learning_rate": 1.996739126704297e-05,
      "loss": 7.9049,
      "step": 27430
    },
    {
      "epoch": 0.07716947755926216,
      "grad_norm": 18.25,
      "learning_rate": 1.9967367498821546e-05,
      "loss": 7.3143,
      "step": 27440
    },
    {
      "epoch": 0.07719760054671088,
      "grad_norm": 22.75,
      "learning_rate": 1.996734372195521e-05,
      "loss": 7.9868,
      "step": 27450
    },
    {
      "epoch": 0.07722572353415959,
      "grad_norm": 25.0,
      "learning_rate": 1.996731993644398e-05,
      "loss": 8.8749,
      "step": 27460
    },
    {
      "epoch": 0.0772538465216083,
      "grad_norm": 43.5,
      "learning_rate": 1.9967296142287878e-05,
      "loss": 7.805,
      "step": 27470
    },
    {
      "epoch": 0.07728196950905701,
      "grad_norm": 32.0,
      "learning_rate": 1.9967272339486924e-05,
      "loss": 7.9415,
      "step": 27480
    },
    {
      "epoch": 0.07731009249650572,
      "grad_norm": 27.75,
      "learning_rate": 1.9967248528041144e-05,
      "loss": 8.5043,
      "step": 27490
    },
    {
      "epoch": 0.07733821548395443,
      "grad_norm": 47.25,
      "learning_rate": 1.996722470795055e-05,
      "loss": 7.3558,
      "step": 27500
    },
    {
      "epoch": 0.07736633847140315,
      "grad_norm": 26.75,
      "learning_rate": 1.9967200879215172e-05,
      "loss": 8.6099,
      "step": 27510
    },
    {
      "epoch": 0.07739446145885186,
      "grad_norm": 28.5,
      "learning_rate": 1.9967177041835023e-05,
      "loss": 7.7975,
      "step": 27520
    },
    {
      "epoch": 0.07742258444630055,
      "grad_norm": 52.25,
      "learning_rate": 1.9967153195810127e-05,
      "loss": 7.9012,
      "step": 27530
    },
    {
      "epoch": 0.07745070743374927,
      "grad_norm": 25.75,
      "learning_rate": 1.9967129341140504e-05,
      "loss": 8.0525,
      "step": 27540
    },
    {
      "epoch": 0.07747883042119798,
      "grad_norm": 26.625,
      "learning_rate": 1.9967105477826174e-05,
      "loss": 8.437,
      "step": 27550
    },
    {
      "epoch": 0.07750695340864669,
      "grad_norm": 46.5,
      "learning_rate": 1.9967081605867162e-05,
      "loss": 8.1046,
      "step": 27560
    },
    {
      "epoch": 0.0775350763960954,
      "grad_norm": 26.5,
      "learning_rate": 1.996705772526348e-05,
      "loss": 7.3295,
      "step": 27570
    },
    {
      "epoch": 0.07756319938354411,
      "grad_norm": 37.25,
      "learning_rate": 1.9967033836015157e-05,
      "loss": 7.8688,
      "step": 27580
    },
    {
      "epoch": 0.07759132237099282,
      "grad_norm": 18.625,
      "learning_rate": 1.996700993812221e-05,
      "loss": 7.8513,
      "step": 27590
    },
    {
      "epoch": 0.07761944535844154,
      "grad_norm": 35.0,
      "learning_rate": 1.9966986031584665e-05,
      "loss": 8.4104,
      "step": 27600
    },
    {
      "epoch": 0.07764756834589025,
      "grad_norm": 61.5,
      "learning_rate": 1.996696211640253e-05,
      "loss": 7.727,
      "step": 27610
    },
    {
      "epoch": 0.07767569133333896,
      "grad_norm": 23.625,
      "learning_rate": 1.9966938192575838e-05,
      "loss": 8.2852,
      "step": 27620
    },
    {
      "epoch": 0.07770381432078767,
      "grad_norm": 19.75,
      "learning_rate": 1.9966914260104606e-05,
      "loss": 7.4344,
      "step": 27630
    },
    {
      "epoch": 0.07773193730823638,
      "grad_norm": 27.75,
      "learning_rate": 1.9966890318988853e-05,
      "loss": 7.8142,
      "step": 27640
    },
    {
      "epoch": 0.0777600602956851,
      "grad_norm": 44.75,
      "learning_rate": 1.9966866369228605e-05,
      "loss": 8.4584,
      "step": 27650
    },
    {
      "epoch": 0.0777881832831338,
      "grad_norm": 26.5,
      "learning_rate": 1.9966842410823873e-05,
      "loss": 8.0484,
      "step": 27660
    },
    {
      "epoch": 0.07781630627058252,
      "grad_norm": 26.75,
      "learning_rate": 1.9966818443774685e-05,
      "loss": 7.6098,
      "step": 27670
    },
    {
      "epoch": 0.07784442925803123,
      "grad_norm": 28.125,
      "learning_rate": 1.9966794468081063e-05,
      "loss": 8.4103,
      "step": 27680
    },
    {
      "epoch": 0.07787255224547994,
      "grad_norm": 28.25,
      "learning_rate": 1.9966770483743025e-05,
      "loss": 7.9141,
      "step": 27690
    },
    {
      "epoch": 0.07790067523292864,
      "grad_norm": 23.625,
      "learning_rate": 1.9966746490760592e-05,
      "loss": 8.0464,
      "step": 27700
    },
    {
      "epoch": 0.07792879822037735,
      "grad_norm": 37.75,
      "learning_rate": 1.996672248913378e-05,
      "loss": 7.9216,
      "step": 27710
    },
    {
      "epoch": 0.07795692120782606,
      "grad_norm": 44.25,
      "learning_rate": 1.9966698478862624e-05,
      "loss": 7.2609,
      "step": 27720
    },
    {
      "epoch": 0.07798504419527477,
      "grad_norm": 32.25,
      "learning_rate": 1.996667445994713e-05,
      "loss": 8.0394,
      "step": 27730
    },
    {
      "epoch": 0.07801316718272348,
      "grad_norm": 29.625,
      "learning_rate": 1.9966650432387324e-05,
      "loss": 8.3917,
      "step": 27740
    },
    {
      "epoch": 0.0780412901701722,
      "grad_norm": 28.0,
      "learning_rate": 1.9966626396183226e-05,
      "loss": 8.2311,
      "step": 27750
    },
    {
      "epoch": 0.0780694131576209,
      "grad_norm": 24.625,
      "learning_rate": 1.9966602351334862e-05,
      "loss": 7.7782,
      "step": 27760
    },
    {
      "epoch": 0.07809753614506962,
      "grad_norm": 23.5,
      "learning_rate": 1.996657829784225e-05,
      "loss": 7.8627,
      "step": 27770
    },
    {
      "epoch": 0.07812565913251833,
      "grad_norm": 21.625,
      "learning_rate": 1.9966554235705404e-05,
      "loss": 7.7049,
      "step": 27780
    },
    {
      "epoch": 0.07815378211996704,
      "grad_norm": 45.0,
      "learning_rate": 1.9966530164924353e-05,
      "loss": 7.8896,
      "step": 27790
    },
    {
      "epoch": 0.07818190510741575,
      "grad_norm": 57.25,
      "learning_rate": 1.9966506085499115e-05,
      "loss": 7.9455,
      "step": 27800
    },
    {
      "epoch": 0.07821002809486446,
      "grad_norm": 24.375,
      "learning_rate": 1.9966481997429713e-05,
      "loss": 8.2275,
      "step": 27810
    },
    {
      "epoch": 0.07823815108231318,
      "grad_norm": 43.5,
      "learning_rate": 1.9966457900716166e-05,
      "loss": 7.3201,
      "step": 27820
    },
    {
      "epoch": 0.07826627406976189,
      "grad_norm": 24.375,
      "learning_rate": 1.9966433795358492e-05,
      "loss": 7.492,
      "step": 27830
    },
    {
      "epoch": 0.0782943970572106,
      "grad_norm": 31.125,
      "learning_rate": 1.996640968135672e-05,
      "loss": 7.8319,
      "step": 27840
    },
    {
      "epoch": 0.07832252004465931,
      "grad_norm": 24.25,
      "learning_rate": 1.9966385558710865e-05,
      "loss": 7.8667,
      "step": 27850
    },
    {
      "epoch": 0.07835064303210801,
      "grad_norm": 30.5,
      "learning_rate": 1.996636142742095e-05,
      "loss": 7.6707,
      "step": 27860
    },
    {
      "epoch": 0.07837876601955672,
      "grad_norm": 36.5,
      "learning_rate": 1.996633728748699e-05,
      "loss": 7.3885,
      "step": 27870
    },
    {
      "epoch": 0.07840688900700543,
      "grad_norm": 25.5,
      "learning_rate": 1.9966313138909013e-05,
      "loss": 6.9396,
      "step": 27880
    },
    {
      "epoch": 0.07843501199445414,
      "grad_norm": 36.25,
      "learning_rate": 1.9966288981687038e-05,
      "loss": 8.0507,
      "step": 27890
    },
    {
      "epoch": 0.07846313498190285,
      "grad_norm": 24.625,
      "learning_rate": 1.9966264815821085e-05,
      "loss": 8.5828,
      "step": 27900
    },
    {
      "epoch": 0.07849125796935157,
      "grad_norm": 35.25,
      "learning_rate": 1.996624064131118e-05,
      "loss": 7.747,
      "step": 27910
    },
    {
      "epoch": 0.07851938095680028,
      "grad_norm": 30.0,
      "learning_rate": 1.9966216458157335e-05,
      "loss": 7.4919,
      "step": 27920
    },
    {
      "epoch": 0.07854750394424899,
      "grad_norm": 31.0,
      "learning_rate": 1.9966192266359577e-05,
      "loss": 7.862,
      "step": 27930
    },
    {
      "epoch": 0.0785756269316977,
      "grad_norm": 30.875,
      "learning_rate": 1.9966168065917925e-05,
      "loss": 8.4959,
      "step": 27940
    },
    {
      "epoch": 0.07860374991914641,
      "grad_norm": 22.625,
      "learning_rate": 1.99661438568324e-05,
      "loss": 8.0447,
      "step": 27950
    },
    {
      "epoch": 0.07863187290659512,
      "grad_norm": 30.75,
      "learning_rate": 1.9966119639103025e-05,
      "loss": 7.9502,
      "step": 27960
    },
    {
      "epoch": 0.07865999589404384,
      "grad_norm": 28.125,
      "learning_rate": 1.996609541272982e-05,
      "loss": 7.9825,
      "step": 27970
    },
    {
      "epoch": 0.07868811888149255,
      "grad_norm": 35.0,
      "learning_rate": 1.9966071177712804e-05,
      "loss": 7.2857,
      "step": 27980
    },
    {
      "epoch": 0.07871624186894126,
      "grad_norm": 28.25,
      "learning_rate": 1.9966046934052e-05,
      "loss": 7.3708,
      "step": 27990
    },
    {
      "epoch": 0.07874436485638997,
      "grad_norm": 23.0,
      "learning_rate": 1.996602268174743e-05,
      "loss": 7.7806,
      "step": 28000
    },
    {
      "epoch": 0.07877248784383868,
      "grad_norm": 23.375,
      "learning_rate": 1.996599842079911e-05,
      "loss": 8.0325,
      "step": 28010
    },
    {
      "epoch": 0.07880061083128738,
      "grad_norm": 29.75,
      "learning_rate": 1.9965974151207067e-05,
      "loss": 8.085,
      "step": 28020
    },
    {
      "epoch": 0.07882873381873609,
      "grad_norm": 37.0,
      "learning_rate": 1.996594987297132e-05,
      "loss": 8.2522,
      "step": 28030
    },
    {
      "epoch": 0.0788568568061848,
      "grad_norm": 54.5,
      "learning_rate": 1.996592558609189e-05,
      "loss": 8.2149,
      "step": 28040
    },
    {
      "epoch": 0.07888497979363351,
      "grad_norm": 23.5,
      "learning_rate": 1.99659012905688e-05,
      "loss": 7.6538,
      "step": 28050
    },
    {
      "epoch": 0.07891310278108223,
      "grad_norm": 47.75,
      "learning_rate": 1.9965876986402067e-05,
      "loss": 7.736,
      "step": 28060
    },
    {
      "epoch": 0.07894122576853094,
      "grad_norm": 41.5,
      "learning_rate": 1.9965852673591713e-05,
      "loss": 8.3534,
      "step": 28070
    },
    {
      "epoch": 0.07896934875597965,
      "grad_norm": 25.875,
      "learning_rate": 1.996582835213776e-05,
      "loss": 7.958,
      "step": 28080
    },
    {
      "epoch": 0.07899747174342836,
      "grad_norm": 20.875,
      "learning_rate": 1.9965804022040233e-05,
      "loss": 8.0366,
      "step": 28090
    },
    {
      "epoch": 0.07902559473087707,
      "grad_norm": 37.0,
      "learning_rate": 1.9965779683299145e-05,
      "loss": 7.2943,
      "step": 28100
    },
    {
      "epoch": 0.07905371771832578,
      "grad_norm": 21.125,
      "learning_rate": 1.9965755335914522e-05,
      "loss": 7.8049,
      "step": 28110
    },
    {
      "epoch": 0.0790818407057745,
      "grad_norm": 24.625,
      "learning_rate": 1.9965730979886385e-05,
      "loss": 7.2822,
      "step": 28120
    },
    {
      "epoch": 0.0791099636932232,
      "grad_norm": 25.5,
      "learning_rate": 1.996570661521476e-05,
      "loss": 8.0912,
      "step": 28130
    },
    {
      "epoch": 0.07913808668067192,
      "grad_norm": 35.75,
      "learning_rate": 1.9965682241899652e-05,
      "loss": 7.5906,
      "step": 28140
    },
    {
      "epoch": 0.07916620966812063,
      "grad_norm": 29.625,
      "learning_rate": 1.9965657859941103e-05,
      "loss": 7.7365,
      "step": 28150
    },
    {
      "epoch": 0.07919433265556934,
      "grad_norm": 31.625,
      "learning_rate": 1.9965633469339117e-05,
      "loss": 8.2284,
      "step": 28160
    },
    {
      "epoch": 0.07922245564301805,
      "grad_norm": 27.875,
      "learning_rate": 1.9965609070093727e-05,
      "loss": 8.0788,
      "step": 28170
    },
    {
      "epoch": 0.07925057863046675,
      "grad_norm": 45.75,
      "learning_rate": 1.9965584662204948e-05,
      "loss": 8.6868,
      "step": 28180
    },
    {
      "epoch": 0.07927870161791546,
      "grad_norm": 39.25,
      "learning_rate": 1.99655602456728e-05,
      "loss": 8.6296,
      "step": 28190
    },
    {
      "epoch": 0.07930682460536417,
      "grad_norm": 38.5,
      "learning_rate": 1.996553582049731e-05,
      "loss": 7.9807,
      "step": 28200
    },
    {
      "epoch": 0.07933494759281288,
      "grad_norm": 20.875,
      "learning_rate": 1.9965511386678496e-05,
      "loss": 7.6621,
      "step": 28210
    },
    {
      "epoch": 0.0793630705802616,
      "grad_norm": 39.25,
      "learning_rate": 1.9965486944216377e-05,
      "loss": 8.2259,
      "step": 28220
    },
    {
      "epoch": 0.07939119356771031,
      "grad_norm": 33.5,
      "learning_rate": 1.9965462493110977e-05,
      "loss": 8.6496,
      "step": 28230
    },
    {
      "epoch": 0.07941931655515902,
      "grad_norm": 35.5,
      "learning_rate": 1.9965438033362318e-05,
      "loss": 7.7785,
      "step": 28240
    },
    {
      "epoch": 0.07944743954260773,
      "grad_norm": 75.5,
      "learning_rate": 1.9965413564970416e-05,
      "loss": 7.6092,
      "step": 28250
    },
    {
      "epoch": 0.07947556253005644,
      "grad_norm": 22.875,
      "learning_rate": 1.99653890879353e-05,
      "loss": 8.2531,
      "step": 28260
    },
    {
      "epoch": 0.07950368551750515,
      "grad_norm": 19.375,
      "learning_rate": 1.9965364602256985e-05,
      "loss": 8.5099,
      "step": 28270
    },
    {
      "epoch": 0.07953180850495387,
      "grad_norm": 33.75,
      "learning_rate": 1.9965340107935494e-05,
      "loss": 7.5987,
      "step": 28280
    },
    {
      "epoch": 0.07955993149240258,
      "grad_norm": 21.375,
      "learning_rate": 1.996531560497085e-05,
      "loss": 8.6437,
      "step": 28290
    },
    {
      "epoch": 0.07958805447985129,
      "grad_norm": 34.75,
      "learning_rate": 1.9965291093363074e-05,
      "loss": 7.5748,
      "step": 28300
    },
    {
      "epoch": 0.0796161774673,
      "grad_norm": 45.5,
      "learning_rate": 1.9965266573112183e-05,
      "loss": 7.4957,
      "step": 28310
    },
    {
      "epoch": 0.07964430045474871,
      "grad_norm": 30.625,
      "learning_rate": 1.9965242044218204e-05,
      "loss": 8.4343,
      "step": 28320
    },
    {
      "epoch": 0.07967242344219742,
      "grad_norm": 34.5,
      "learning_rate": 1.9965217506681156e-05,
      "loss": 8.1894,
      "step": 28330
    },
    {
      "epoch": 0.07970054642964614,
      "grad_norm": 29.375,
      "learning_rate": 1.996519296050106e-05,
      "loss": 6.7445,
      "step": 28340
    },
    {
      "epoch": 0.07972866941709483,
      "grad_norm": 44.5,
      "learning_rate": 1.9965168405677935e-05,
      "loss": 7.5108,
      "step": 28350
    },
    {
      "epoch": 0.07975679240454354,
      "grad_norm": 25.0,
      "learning_rate": 1.9965143842211806e-05,
      "loss": 8.5677,
      "step": 28360
    },
    {
      "epoch": 0.07978491539199226,
      "grad_norm": 30.5,
      "learning_rate": 1.9965119270102695e-05,
      "loss": 7.7441,
      "step": 28370
    },
    {
      "epoch": 0.07981303837944097,
      "grad_norm": 30.625,
      "learning_rate": 1.996509468935062e-05,
      "loss": 8.0103,
      "step": 28380
    },
    {
      "epoch": 0.07984116136688968,
      "grad_norm": 31.0,
      "learning_rate": 1.9965070099955598e-05,
      "loss": 8.1222,
      "step": 28390
    },
    {
      "epoch": 0.07986928435433839,
      "grad_norm": 24.375,
      "learning_rate": 1.9965045501917663e-05,
      "loss": 8.7416,
      "step": 28400
    },
    {
      "epoch": 0.0798974073417871,
      "grad_norm": 47.0,
      "learning_rate": 1.9965020895236824e-05,
      "loss": 7.5119,
      "step": 28410
    },
    {
      "epoch": 0.07992553032923581,
      "grad_norm": 25.375,
      "learning_rate": 1.9964996279913112e-05,
      "loss": 7.1913,
      "step": 28420
    },
    {
      "epoch": 0.07995365331668453,
      "grad_norm": 33.0,
      "learning_rate": 1.9964971655946543e-05,
      "loss": 8.3894,
      "step": 28430
    },
    {
      "epoch": 0.07998177630413324,
      "grad_norm": 22.25,
      "learning_rate": 1.9964947023337138e-05,
      "loss": 7.8122,
      "step": 28440
    },
    {
      "epoch": 0.08000989929158195,
      "grad_norm": 26.5,
      "learning_rate": 1.996492238208492e-05,
      "loss": 8.3363,
      "step": 28450
    },
    {
      "epoch": 0.08003802227903066,
      "grad_norm": 42.0,
      "learning_rate": 1.996489773218991e-05,
      "loss": 7.2653,
      "step": 28460
    },
    {
      "epoch": 0.08006614526647937,
      "grad_norm": 28.0,
      "learning_rate": 1.996487307365213e-05,
      "loss": 6.5333,
      "step": 28470
    },
    {
      "epoch": 0.08009426825392808,
      "grad_norm": 50.75,
      "learning_rate": 1.9964848406471596e-05,
      "loss": 7.2924,
      "step": 28480
    },
    {
      "epoch": 0.0801223912413768,
      "grad_norm": 27.0,
      "learning_rate": 1.9964823730648342e-05,
      "loss": 8.316,
      "step": 28490
    },
    {
      "epoch": 0.0801505142288255,
      "grad_norm": 28.375,
      "learning_rate": 1.9964799046182377e-05,
      "loss": 7.7852,
      "step": 28500
    },
    {
      "epoch": 0.0801786372162742,
      "grad_norm": 26.0,
      "learning_rate": 1.9964774353073725e-05,
      "loss": 7.6473,
      "step": 28510
    },
    {
      "epoch": 0.08020676020372292,
      "grad_norm": 41.25,
      "learning_rate": 1.9964749651322413e-05,
      "loss": 7.9712,
      "step": 28520
    },
    {
      "epoch": 0.08023488319117163,
      "grad_norm": 48.0,
      "learning_rate": 1.9964724940928458e-05,
      "loss": 8.1065,
      "step": 28530
    },
    {
      "epoch": 0.08026300617862034,
      "grad_norm": 36.5,
      "learning_rate": 1.9964700221891884e-05,
      "loss": 7.3466,
      "step": 28540
    },
    {
      "epoch": 0.08029112916606905,
      "grad_norm": 29.25,
      "learning_rate": 1.996467549421271e-05,
      "loss": 7.9753,
      "step": 28550
    },
    {
      "epoch": 0.08031925215351776,
      "grad_norm": 27.875,
      "learning_rate": 1.9964650757890955e-05,
      "loss": 7.5777,
      "step": 28560
    },
    {
      "epoch": 0.08034737514096647,
      "grad_norm": 35.25,
      "learning_rate": 1.9964626012926644e-05,
      "loss": 7.9163,
      "step": 28570
    },
    {
      "epoch": 0.08037549812841518,
      "grad_norm": 23.375,
      "learning_rate": 1.99646012593198e-05,
      "loss": 7.9393,
      "step": 28580
    },
    {
      "epoch": 0.0804036211158639,
      "grad_norm": 24.625,
      "learning_rate": 1.9964576497070443e-05,
      "loss": 8.0188,
      "step": 28590
    },
    {
      "epoch": 0.08043174410331261,
      "grad_norm": 36.5,
      "learning_rate": 1.9964551726178592e-05,
      "loss": 8.0241,
      "step": 28600
    },
    {
      "epoch": 0.08045986709076132,
      "grad_norm": 23.875,
      "learning_rate": 1.9964526946644272e-05,
      "loss": 7.9075,
      "step": 28610
    },
    {
      "epoch": 0.08048799007821003,
      "grad_norm": 37.25,
      "learning_rate": 1.99645021584675e-05,
      "loss": 8.0868,
      "step": 28620
    },
    {
      "epoch": 0.08051611306565874,
      "grad_norm": 39.5,
      "learning_rate": 1.9964477361648305e-05,
      "loss": 7.472,
      "step": 28630
    },
    {
      "epoch": 0.08054423605310745,
      "grad_norm": 22.25,
      "learning_rate": 1.9964452556186702e-05,
      "loss": 8.1021,
      "step": 28640
    },
    {
      "epoch": 0.08057235904055617,
      "grad_norm": 49.5,
      "learning_rate": 1.9964427742082714e-05,
      "loss": 7.7658,
      "step": 28650
    },
    {
      "epoch": 0.08060048202800488,
      "grad_norm": 24.625,
      "learning_rate": 1.9964402919336364e-05,
      "loss": 8.4861,
      "step": 28660
    },
    {
      "epoch": 0.08062860501545357,
      "grad_norm": 33.0,
      "learning_rate": 1.9964378087947673e-05,
      "loss": 8.0058,
      "step": 28670
    },
    {
      "epoch": 0.08065672800290229,
      "grad_norm": 25.125,
      "learning_rate": 1.9964353247916657e-05,
      "loss": 7.9268,
      "step": 28680
    },
    {
      "epoch": 0.080684850990351,
      "grad_norm": 35.75,
      "learning_rate": 1.9964328399243346e-05,
      "loss": 7.7485,
      "step": 28690
    },
    {
      "epoch": 0.08071297397779971,
      "grad_norm": 37.75,
      "learning_rate": 1.996430354192776e-05,
      "loss": 7.5114,
      "step": 28700
    },
    {
      "epoch": 0.08074109696524842,
      "grad_norm": 26.125,
      "learning_rate": 1.9964278675969918e-05,
      "loss": 7.8527,
      "step": 28710
    },
    {
      "epoch": 0.08076921995269713,
      "grad_norm": 27.625,
      "learning_rate": 1.996425380136984e-05,
      "loss": 7.7686,
      "step": 28720
    },
    {
      "epoch": 0.08079734294014584,
      "grad_norm": 26.125,
      "learning_rate": 1.9964228918127554e-05,
      "loss": 7.6673,
      "step": 28730
    },
    {
      "epoch": 0.08082546592759456,
      "grad_norm": 26.125,
      "learning_rate": 1.9964204026243075e-05,
      "loss": 7.7632,
      "step": 28740
    },
    {
      "epoch": 0.08085358891504327,
      "grad_norm": 27.125,
      "learning_rate": 1.9964179125716424e-05,
      "loss": 7.526,
      "step": 28750
    },
    {
      "epoch": 0.08088171190249198,
      "grad_norm": 34.0,
      "learning_rate": 1.996415421654763e-05,
      "loss": 9.0559,
      "step": 28760
    },
    {
      "epoch": 0.08090983488994069,
      "grad_norm": 27.5,
      "learning_rate": 1.996412929873671e-05,
      "loss": 7.6445,
      "step": 28770
    },
    {
      "epoch": 0.0809379578773894,
      "grad_norm": 32.5,
      "learning_rate": 1.9964104372283684e-05,
      "loss": 7.7676,
      "step": 28780
    },
    {
      "epoch": 0.08096608086483811,
      "grad_norm": 36.0,
      "learning_rate": 1.9964079437188575e-05,
      "loss": 7.4814,
      "step": 28790
    },
    {
      "epoch": 0.08099420385228683,
      "grad_norm": 32.0,
      "learning_rate": 1.9964054493451405e-05,
      "loss": 8.0271,
      "step": 28800
    },
    {
      "epoch": 0.08102232683973554,
      "grad_norm": 26.875,
      "learning_rate": 1.99640295410722e-05,
      "loss": 7.6748,
      "step": 28810
    },
    {
      "epoch": 0.08105044982718425,
      "grad_norm": 34.5,
      "learning_rate": 1.996400458005097e-05,
      "loss": 8.0978,
      "step": 28820
    },
    {
      "epoch": 0.08107857281463295,
      "grad_norm": 30.25,
      "learning_rate": 1.996397961038775e-05,
      "loss": 7.7824,
      "step": 28830
    },
    {
      "epoch": 0.08110669580208166,
      "grad_norm": 31.875,
      "learning_rate": 1.9963954632082553e-05,
      "loss": 8.7066,
      "step": 28840
    },
    {
      "epoch": 0.08113481878953037,
      "grad_norm": 30.125,
      "learning_rate": 1.9963929645135404e-05,
      "loss": 7.3685,
      "step": 28850
    },
    {
      "epoch": 0.08116294177697908,
      "grad_norm": 27.5,
      "learning_rate": 1.9963904649546326e-05,
      "loss": 7.9533,
      "step": 28860
    },
    {
      "epoch": 0.08119106476442779,
      "grad_norm": 47.75,
      "learning_rate": 1.9963879645315337e-05,
      "loss": 7.5284,
      "step": 28870
    },
    {
      "epoch": 0.0812191877518765,
      "grad_norm": 29.625,
      "learning_rate": 1.9963854632442457e-05,
      "loss": 7.7151,
      "step": 28880
    },
    {
      "epoch": 0.08124731073932522,
      "grad_norm": 21.75,
      "learning_rate": 1.9963829610927717e-05,
      "loss": 8.0307,
      "step": 28890
    },
    {
      "epoch": 0.08127543372677393,
      "grad_norm": 25.125,
      "learning_rate": 1.996380458077113e-05,
      "loss": 8.2713,
      "step": 28900
    },
    {
      "epoch": 0.08130355671422264,
      "grad_norm": 25.875,
      "learning_rate": 1.996377954197272e-05,
      "loss": 7.9945,
      "step": 28910
    },
    {
      "epoch": 0.08133167970167135,
      "grad_norm": 24.25,
      "learning_rate": 1.9963754494532506e-05,
      "loss": 7.4128,
      "step": 28920
    },
    {
      "epoch": 0.08135980268912006,
      "grad_norm": 31.375,
      "learning_rate": 1.996372943845052e-05,
      "loss": 7.9222,
      "step": 28930
    },
    {
      "epoch": 0.08138792567656877,
      "grad_norm": 29.625,
      "learning_rate": 1.9963704373726773e-05,
      "loss": 8.6171,
      "step": 28940
    },
    {
      "epoch": 0.08141604866401748,
      "grad_norm": 26.625,
      "learning_rate": 1.996367930036129e-05,
      "loss": 8.2733,
      "step": 28950
    },
    {
      "epoch": 0.0814441716514662,
      "grad_norm": 32.5,
      "learning_rate": 1.9963654218354093e-05,
      "loss": 7.605,
      "step": 28960
    },
    {
      "epoch": 0.08147229463891491,
      "grad_norm": 28.875,
      "learning_rate": 1.9963629127705205e-05,
      "loss": 8.4248,
      "step": 28970
    },
    {
      "epoch": 0.08150041762636362,
      "grad_norm": 36.5,
      "learning_rate": 1.9963604028414646e-05,
      "loss": 7.6258,
      "step": 28980
    },
    {
      "epoch": 0.08152854061381233,
      "grad_norm": 38.75,
      "learning_rate": 1.996357892048244e-05,
      "loss": 7.9703,
      "step": 28990
    },
    {
      "epoch": 0.08155666360126103,
      "grad_norm": 22.0,
      "learning_rate": 1.9963553803908607e-05,
      "loss": 7.5612,
      "step": 29000
    },
    {
      "epoch": 0.08158478658870974,
      "grad_norm": 24.625,
      "learning_rate": 1.9963528678693168e-05,
      "loss": 7.705,
      "step": 29010
    },
    {
      "epoch": 0.08161290957615845,
      "grad_norm": 47.5,
      "learning_rate": 1.9963503544836143e-05,
      "loss": 8.8081,
      "step": 29020
    },
    {
      "epoch": 0.08164103256360716,
      "grad_norm": 37.25,
      "learning_rate": 1.9963478402337563e-05,
      "loss": 8.5169,
      "step": 29030
    },
    {
      "epoch": 0.08166915555105587,
      "grad_norm": 31.0,
      "learning_rate": 1.9963453251197437e-05,
      "loss": 8.1981,
      "step": 29040
    },
    {
      "epoch": 0.08169727853850459,
      "grad_norm": 30.375,
      "learning_rate": 1.9963428091415796e-05,
      "loss": 7.5985,
      "step": 29050
    },
    {
      "epoch": 0.0817254015259533,
      "grad_norm": 23.875,
      "learning_rate": 1.9963402922992658e-05,
      "loss": 7.4507,
      "step": 29060
    },
    {
      "epoch": 0.08175352451340201,
      "grad_norm": 27.25,
      "learning_rate": 1.996337774592805e-05,
      "loss": 7.7355,
      "step": 29070
    },
    {
      "epoch": 0.08178164750085072,
      "grad_norm": 24.375,
      "learning_rate": 1.9963352560221985e-05,
      "loss": 8.3257,
      "step": 29080
    },
    {
      "epoch": 0.08180977048829943,
      "grad_norm": 35.5,
      "learning_rate": 1.996332736587449e-05,
      "loss": 8.1861,
      "step": 29090
    },
    {
      "epoch": 0.08183789347574814,
      "grad_norm": 26.125,
      "learning_rate": 1.9963302162885586e-05,
      "loss": 7.3656,
      "step": 29100
    },
    {
      "epoch": 0.08186601646319686,
      "grad_norm": 43.5,
      "learning_rate": 1.9963276951255296e-05,
      "loss": 8.152,
      "step": 29110
    },
    {
      "epoch": 0.08189413945064557,
      "grad_norm": 30.75,
      "learning_rate": 1.9963251730983645e-05,
      "loss": 7.5156,
      "step": 29120
    },
    {
      "epoch": 0.08192226243809428,
      "grad_norm": 22.625,
      "learning_rate": 1.9963226502070645e-05,
      "loss": 8.0142,
      "step": 29130
    },
    {
      "epoch": 0.08195038542554299,
      "grad_norm": 25.625,
      "learning_rate": 1.9963201264516326e-05,
      "loss": 7.6814,
      "step": 29140
    },
    {
      "epoch": 0.0819785084129917,
      "grad_norm": 24.875,
      "learning_rate": 1.996317601832071e-05,
      "loss": 7.5147,
      "step": 29150
    },
    {
      "epoch": 0.0820066314004404,
      "grad_norm": 48.0,
      "learning_rate": 1.9963150763483815e-05,
      "loss": 8.0413,
      "step": 29160
    },
    {
      "epoch": 0.08203475438788911,
      "grad_norm": 42.5,
      "learning_rate": 1.9963125500005663e-05,
      "loss": 8.6127,
      "step": 29170
    },
    {
      "epoch": 0.08206287737533782,
      "grad_norm": 21.25,
      "learning_rate": 1.9963100227886278e-05,
      "loss": 8.3984,
      "step": 29180
    },
    {
      "epoch": 0.08209100036278653,
      "grad_norm": 31.0,
      "learning_rate": 1.9963074947125683e-05,
      "loss": 7.271,
      "step": 29190
    },
    {
      "epoch": 0.08211912335023525,
      "grad_norm": 21.125,
      "learning_rate": 1.9963049657723895e-05,
      "loss": 7.7053,
      "step": 29200
    },
    {
      "epoch": 0.08214724633768396,
      "grad_norm": 33.75,
      "learning_rate": 1.996302435968094e-05,
      "loss": 8.1535,
      "step": 29210
    },
    {
      "epoch": 0.08217536932513267,
      "grad_norm": 51.0,
      "learning_rate": 1.9962999052996845e-05,
      "loss": 7.4601,
      "step": 29220
    },
    {
      "epoch": 0.08220349231258138,
      "grad_norm": 26.25,
      "learning_rate": 1.996297373767162e-05,
      "loss": 8.2985,
      "step": 29230
    },
    {
      "epoch": 0.08223161530003009,
      "grad_norm": 42.25,
      "learning_rate": 1.9962948413705297e-05,
      "loss": 8.3043,
      "step": 29240
    },
    {
      "epoch": 0.0822597382874788,
      "grad_norm": 32.25,
      "learning_rate": 1.996292308109789e-05,
      "loss": 7.7354,
      "step": 29250
    },
    {
      "epoch": 0.08228786127492752,
      "grad_norm": 32.25,
      "learning_rate": 1.996289773984943e-05,
      "loss": 7.2548,
      "step": 29260
    },
    {
      "epoch": 0.08231598426237623,
      "grad_norm": 25.5,
      "learning_rate": 1.996287238995993e-05,
      "loss": 7.4271,
      "step": 29270
    },
    {
      "epoch": 0.08234410724982494,
      "grad_norm": 24.5,
      "learning_rate": 1.996284703142942e-05,
      "loss": 8.5957,
      "step": 29280
    },
    {
      "epoch": 0.08237223023727365,
      "grad_norm": 42.75,
      "learning_rate": 1.9962821664257913e-05,
      "loss": 7.7794,
      "step": 29290
    },
    {
      "epoch": 0.08240035322472236,
      "grad_norm": 24.625,
      "learning_rate": 1.9962796288445436e-05,
      "loss": 8.5384,
      "step": 29300
    },
    {
      "epoch": 0.08242847621217107,
      "grad_norm": 24.375,
      "learning_rate": 1.9962770903992013e-05,
      "loss": 7.4033,
      "step": 29310
    },
    {
      "epoch": 0.08245659919961977,
      "grad_norm": 25.75,
      "learning_rate": 1.9962745510897664e-05,
      "loss": 7.5881,
      "step": 29320
    },
    {
      "epoch": 0.08248472218706848,
      "grad_norm": 44.5,
      "learning_rate": 1.9962720109162414e-05,
      "loss": 8.0466,
      "step": 29330
    },
    {
      "epoch": 0.0825128451745172,
      "grad_norm": 27.5,
      "learning_rate": 1.9962694698786276e-05,
      "loss": 7.9642,
      "step": 29340
    },
    {
      "epoch": 0.0825409681619659,
      "grad_norm": 29.5,
      "learning_rate": 1.9962669279769283e-05,
      "loss": 7.4193,
      "step": 29350
    },
    {
      "epoch": 0.08256909114941462,
      "grad_norm": 58.0,
      "learning_rate": 1.996264385211145e-05,
      "loss": 7.3756,
      "step": 29360
    },
    {
      "epoch": 0.08259721413686333,
      "grad_norm": 33.0,
      "learning_rate": 1.99626184158128e-05,
      "loss": 7.6708,
      "step": 29370
    },
    {
      "epoch": 0.08262533712431204,
      "grad_norm": 39.5,
      "learning_rate": 1.996259297087336e-05,
      "loss": 7.467,
      "step": 29380
    },
    {
      "epoch": 0.08265346011176075,
      "grad_norm": 21.125,
      "learning_rate": 1.9962567517293145e-05,
      "loss": 7.5773,
      "step": 29390
    },
    {
      "epoch": 0.08268158309920946,
      "grad_norm": 31.125,
      "learning_rate": 1.996254205507218e-05,
      "loss": 7.7751,
      "step": 29400
    },
    {
      "epoch": 0.08270970608665817,
      "grad_norm": 48.0,
      "learning_rate": 1.996251658421049e-05,
      "loss": 8.136,
      "step": 29410
    },
    {
      "epoch": 0.08273782907410689,
      "grad_norm": 23.625,
      "learning_rate": 1.9962491104708095e-05,
      "loss": 7.918,
      "step": 29420
    },
    {
      "epoch": 0.0827659520615556,
      "grad_norm": 24.5,
      "learning_rate": 1.9962465616565015e-05,
      "loss": 7.8234,
      "step": 29430
    },
    {
      "epoch": 0.08279407504900431,
      "grad_norm": 26.0,
      "learning_rate": 1.9962440119781274e-05,
      "loss": 7.5626,
      "step": 29440
    },
    {
      "epoch": 0.08282219803645302,
      "grad_norm": 39.0,
      "learning_rate": 1.9962414614356892e-05,
      "loss": 7.8826,
      "step": 29450
    },
    {
      "epoch": 0.08285032102390173,
      "grad_norm": 42.75,
      "learning_rate": 1.9962389100291896e-05,
      "loss": 8.0542,
      "step": 29460
    },
    {
      "epoch": 0.08287844401135044,
      "grad_norm": 26.625,
      "learning_rate": 1.9962363577586304e-05,
      "loss": 7.6912,
      "step": 29470
    },
    {
      "epoch": 0.08290656699879914,
      "grad_norm": 30.875,
      "learning_rate": 1.996233804624014e-05,
      "loss": 7.6208,
      "step": 29480
    },
    {
      "epoch": 0.08293468998624785,
      "grad_norm": 26.875,
      "learning_rate": 1.9962312506253425e-05,
      "loss": 7.9748,
      "step": 29490
    },
    {
      "epoch": 0.08296281297369656,
      "grad_norm": 45.5,
      "learning_rate": 1.9962286957626182e-05,
      "loss": 7.7688,
      "step": 29500
    },
    {
      "epoch": 0.08299093596114528,
      "grad_norm": 45.0,
      "learning_rate": 1.996226140035843e-05,
      "loss": 7.5988,
      "step": 29510
    },
    {
      "epoch": 0.08301905894859399,
      "grad_norm": 29.0,
      "learning_rate": 1.9962235834450197e-05,
      "loss": 7.6292,
      "step": 29520
    },
    {
      "epoch": 0.0830471819360427,
      "grad_norm": 30.125,
      "learning_rate": 1.9962210259901503e-05,
      "loss": 8.3314,
      "step": 29530
    },
    {
      "epoch": 0.08307530492349141,
      "grad_norm": 27.25,
      "learning_rate": 1.9962184676712365e-05,
      "loss": 7.9886,
      "step": 29540
    },
    {
      "epoch": 0.08310342791094012,
      "grad_norm": 24.5,
      "learning_rate": 1.9962159084882816e-05,
      "loss": 8.5218,
      "step": 29550
    },
    {
      "epoch": 0.08313155089838883,
      "grad_norm": 24.5,
      "learning_rate": 1.9962133484412864e-05,
      "loss": 8.0313,
      "step": 29560
    },
    {
      "epoch": 0.08315967388583755,
      "grad_norm": 23.375,
      "learning_rate": 1.9962107875302542e-05,
      "loss": 7.5289,
      "step": 29570
    },
    {
      "epoch": 0.08318779687328626,
      "grad_norm": 52.75,
      "learning_rate": 1.9962082257551868e-05,
      "loss": 8.0753,
      "step": 29580
    },
    {
      "epoch": 0.08321591986073497,
      "grad_norm": 27.125,
      "learning_rate": 1.9962056631160867e-05,
      "loss": 7.7786,
      "step": 29590
    },
    {
      "epoch": 0.08324404284818368,
      "grad_norm": 28.625,
      "learning_rate": 1.9962030996129562e-05,
      "loss": 8.6019,
      "step": 29600
    },
    {
      "epoch": 0.08327216583563239,
      "grad_norm": 22.25,
      "learning_rate": 1.9962005352457968e-05,
      "loss": 8.1079,
      "step": 29610
    },
    {
      "epoch": 0.0833002888230811,
      "grad_norm": 31.75,
      "learning_rate": 1.9961979700146113e-05,
      "loss": 7.9685,
      "step": 29620
    },
    {
      "epoch": 0.08332841181052981,
      "grad_norm": 30.5,
      "learning_rate": 1.9961954039194024e-05,
      "loss": 7.8832,
      "step": 29630
    },
    {
      "epoch": 0.08335653479797853,
      "grad_norm": 73.5,
      "learning_rate": 1.996192836960171e-05,
      "loss": 8.3904,
      "step": 29640
    },
    {
      "epoch": 0.08338465778542722,
      "grad_norm": 35.0,
      "learning_rate": 1.9961902691369205e-05,
      "loss": 8.2105,
      "step": 29650
    },
    {
      "epoch": 0.08341278077287594,
      "grad_norm": 38.0,
      "learning_rate": 1.9961877004496525e-05,
      "loss": 7.7654,
      "step": 29660
    },
    {
      "epoch": 0.08344090376032465,
      "grad_norm": 33.75,
      "learning_rate": 1.9961851308983693e-05,
      "loss": 7.9462,
      "step": 29670
    },
    {
      "epoch": 0.08346902674777336,
      "grad_norm": 25.25,
      "learning_rate": 1.9961825604830735e-05,
      "loss": 7.859,
      "step": 29680
    },
    {
      "epoch": 0.08349714973522207,
      "grad_norm": 42.75,
      "learning_rate": 1.996179989203767e-05,
      "loss": 8.1228,
      "step": 29690
    },
    {
      "epoch": 0.08352527272267078,
      "grad_norm": 22.125,
      "learning_rate": 1.9961774170604522e-05,
      "loss": 7.3687,
      "step": 29700
    },
    {
      "epoch": 0.0835533957101195,
      "grad_norm": 31.25,
      "learning_rate": 1.9961748440531313e-05,
      "loss": 8.2964,
      "step": 29710
    },
    {
      "epoch": 0.0835815186975682,
      "grad_norm": 41.0,
      "learning_rate": 1.9961722701818062e-05,
      "loss": 7.9904,
      "step": 29720
    },
    {
      "epoch": 0.08360964168501692,
      "grad_norm": 69.0,
      "learning_rate": 1.9961696954464797e-05,
      "loss": 7.3973,
      "step": 29730
    },
    {
      "epoch": 0.08363776467246563,
      "grad_norm": 30.75,
      "learning_rate": 1.9961671198471535e-05,
      "loss": 7.7189,
      "step": 29740
    },
    {
      "epoch": 0.08366588765991434,
      "grad_norm": 35.0,
      "learning_rate": 1.9961645433838304e-05,
      "loss": 7.9342,
      "step": 29750
    },
    {
      "epoch": 0.08369401064736305,
      "grad_norm": 57.5,
      "learning_rate": 1.9961619660565123e-05,
      "loss": 8.1702,
      "step": 29760
    },
    {
      "epoch": 0.08372213363481176,
      "grad_norm": 28.75,
      "learning_rate": 1.996159387865201e-05,
      "loss": 8.2349,
      "step": 29770
    },
    {
      "epoch": 0.08375025662226047,
      "grad_norm": 33.0,
      "learning_rate": 1.9961568088098993e-05,
      "loss": 7.8651,
      "step": 29780
    },
    {
      "epoch": 0.08377837960970919,
      "grad_norm": 20.0,
      "learning_rate": 1.9961542288906096e-05,
      "loss": 7.7593,
      "step": 29790
    },
    {
      "epoch": 0.0838065025971579,
      "grad_norm": 30.125,
      "learning_rate": 1.9961516481073337e-05,
      "loss": 7.8211,
      "step": 29800
    },
    {
      "epoch": 0.0838346255846066,
      "grad_norm": 27.625,
      "learning_rate": 1.996149066460074e-05,
      "loss": 8.2812,
      "step": 29810
    },
    {
      "epoch": 0.0838627485720553,
      "grad_norm": 30.5,
      "learning_rate": 1.996146483948833e-05,
      "loss": 8.1712,
      "step": 29820
    },
    {
      "epoch": 0.08389087155950402,
      "grad_norm": 26.625,
      "learning_rate": 1.9961439005736122e-05,
      "loss": 8.0673,
      "step": 29830
    },
    {
      "epoch": 0.08391899454695273,
      "grad_norm": 33.5,
      "learning_rate": 1.9961413163344148e-05,
      "loss": 8.7466,
      "step": 29840
    },
    {
      "epoch": 0.08394711753440144,
      "grad_norm": 27.75,
      "learning_rate": 1.996138731231242e-05,
      "loss": 7.9358,
      "step": 29850
    },
    {
      "epoch": 0.08397524052185015,
      "grad_norm": 29.375,
      "learning_rate": 1.996136145264097e-05,
      "loss": 7.8069,
      "step": 29860
    },
    {
      "epoch": 0.08400336350929886,
      "grad_norm": 32.5,
      "learning_rate": 1.9961335584329816e-05,
      "loss": 7.3317,
      "step": 29870
    },
    {
      "epoch": 0.08403148649674758,
      "grad_norm": 26.125,
      "learning_rate": 1.9961309707378982e-05,
      "loss": 7.6092,
      "step": 29880
    },
    {
      "epoch": 0.08405960948419629,
      "grad_norm": 29.25,
      "learning_rate": 1.9961283821788487e-05,
      "loss": 7.823,
      "step": 29890
    },
    {
      "epoch": 0.084087732471645,
      "grad_norm": 32.5,
      "learning_rate": 1.996125792755836e-05,
      "loss": 8.023,
      "step": 29900
    },
    {
      "epoch": 0.08411585545909371,
      "grad_norm": 20.5,
      "learning_rate": 1.9961232024688614e-05,
      "loss": 8.2337,
      "step": 29910
    },
    {
      "epoch": 0.08414397844654242,
      "grad_norm": 27.0,
      "learning_rate": 1.996120611317928e-05,
      "loss": 8.0252,
      "step": 29920
    },
    {
      "epoch": 0.08417210143399113,
      "grad_norm": 23.5,
      "learning_rate": 1.9961180193030375e-05,
      "loss": 8.1497,
      "step": 29930
    },
    {
      "epoch": 0.08420022442143985,
      "grad_norm": 25.0,
      "learning_rate": 1.9961154264241925e-05,
      "loss": 8.1874,
      "step": 29940
    },
    {
      "epoch": 0.08422834740888856,
      "grad_norm": 24.125,
      "learning_rate": 1.9961128326813952e-05,
      "loss": 7.5061,
      "step": 29950
    },
    {
      "epoch": 0.08425647039633727,
      "grad_norm": 25.0,
      "learning_rate": 1.9961102380746477e-05,
      "loss": 8.493,
      "step": 29960
    },
    {
      "epoch": 0.08428459338378597,
      "grad_norm": 38.75,
      "learning_rate": 1.9961076426039523e-05,
      "loss": 7.8739,
      "step": 29970
    },
    {
      "epoch": 0.08431271637123468,
      "grad_norm": 28.5,
      "learning_rate": 1.9961050462693114e-05,
      "loss": 8.1151,
      "step": 29980
    },
    {
      "epoch": 0.08434083935868339,
      "grad_norm": 29.625,
      "learning_rate": 1.9961024490707267e-05,
      "loss": 7.6801,
      "step": 29990
    },
    {
      "epoch": 0.0843689623461321,
      "grad_norm": 32.0,
      "learning_rate": 1.996099851008201e-05,
      "loss": 7.9921,
      "step": 30000
    },
    {
      "epoch": 0.08439708533358081,
      "grad_norm": 34.25,
      "learning_rate": 1.9960972520817365e-05,
      "loss": 9.0981,
      "step": 30010
    },
    {
      "epoch": 0.08442520832102952,
      "grad_norm": 25.75,
      "learning_rate": 1.9960946522913357e-05,
      "loss": 8.2323,
      "step": 30020
    },
    {
      "epoch": 0.08445333130847824,
      "grad_norm": 28.375,
      "learning_rate": 1.9960920516370003e-05,
      "loss": 8.2221,
      "step": 30030
    },
    {
      "epoch": 0.08448145429592695,
      "grad_norm": 32.0,
      "learning_rate": 1.9960894501187326e-05,
      "loss": 8.0671,
      "step": 30040
    },
    {
      "epoch": 0.08450957728337566,
      "grad_norm": 21.875,
      "learning_rate": 1.996086847736535e-05,
      "loss": 7.8533,
      "step": 30050
    },
    {
      "epoch": 0.08453770027082437,
      "grad_norm": 36.75,
      "learning_rate": 1.99608424449041e-05,
      "loss": 7.1154,
      "step": 30060
    },
    {
      "epoch": 0.08456582325827308,
      "grad_norm": 50.0,
      "learning_rate": 1.99608164038036e-05,
      "loss": 7.6813,
      "step": 30070
    },
    {
      "epoch": 0.0845939462457218,
      "grad_norm": 28.125,
      "learning_rate": 1.9960790354063865e-05,
      "loss": 8.1946,
      "step": 30080
    },
    {
      "epoch": 0.0846220692331705,
      "grad_norm": 22.625,
      "learning_rate": 1.996076429568492e-05,
      "loss": 7.7372,
      "step": 30090
    },
    {
      "epoch": 0.08465019222061922,
      "grad_norm": 25.875,
      "learning_rate": 1.996073822866679e-05,
      "loss": 8.0794,
      "step": 30100
    },
    {
      "epoch": 0.08467831520806793,
      "grad_norm": 34.25,
      "learning_rate": 1.99607121530095e-05,
      "loss": 8.4461,
      "step": 30110
    },
    {
      "epoch": 0.08470643819551664,
      "grad_norm": 29.25,
      "learning_rate": 1.996068606871307e-05,
      "loss": 7.1612,
      "step": 30120
    },
    {
      "epoch": 0.08473456118296534,
      "grad_norm": 34.25,
      "learning_rate": 1.9960659975777518e-05,
      "loss": 8.0237,
      "step": 30130
    },
    {
      "epoch": 0.08476268417041405,
      "grad_norm": 39.5,
      "learning_rate": 1.9960633874202874e-05,
      "loss": 8.2848,
      "step": 30140
    },
    {
      "epoch": 0.08479080715786276,
      "grad_norm": 27.375,
      "learning_rate": 1.996060776398916e-05,
      "loss": 7.913,
      "step": 30150
    },
    {
      "epoch": 0.08481893014531147,
      "grad_norm": 26.75,
      "learning_rate": 1.996058164513639e-05,
      "loss": 8.4027,
      "step": 30160
    },
    {
      "epoch": 0.08484705313276018,
      "grad_norm": 84.0,
      "learning_rate": 1.99605555176446e-05,
      "loss": 7.9528,
      "step": 30170
    },
    {
      "epoch": 0.0848751761202089,
      "grad_norm": 33.0,
      "learning_rate": 1.99605293815138e-05,
      "loss": 7.6112,
      "step": 30180
    },
    {
      "epoch": 0.0849032991076576,
      "grad_norm": 26.125,
      "learning_rate": 1.996050323674402e-05,
      "loss": 6.8627,
      "step": 30190
    },
    {
      "epoch": 0.08493142209510632,
      "grad_norm": 37.0,
      "learning_rate": 1.996047708333528e-05,
      "loss": 7.7391,
      "step": 30200
    },
    {
      "epoch": 0.08495954508255503,
      "grad_norm": 27.875,
      "learning_rate": 1.9960450921287602e-05,
      "loss": 7.6925,
      "step": 30210
    },
    {
      "epoch": 0.08498766807000374,
      "grad_norm": 39.25,
      "learning_rate": 1.9960424750601013e-05,
      "loss": 8.7664,
      "step": 30220
    },
    {
      "epoch": 0.08501579105745245,
      "grad_norm": 37.25,
      "learning_rate": 1.996039857127553e-05,
      "loss": 8.1097,
      "step": 30230
    },
    {
      "epoch": 0.08504391404490116,
      "grad_norm": 30.125,
      "learning_rate": 1.996037238331118e-05,
      "loss": 8.1872,
      "step": 30240
    },
    {
      "epoch": 0.08507203703234988,
      "grad_norm": 22.75,
      "learning_rate": 1.9960346186707985e-05,
      "loss": 7.65,
      "step": 30250
    },
    {
      "epoch": 0.08510016001979859,
      "grad_norm": 26.25,
      "learning_rate": 1.9960319981465968e-05,
      "loss": 8.1697,
      "step": 30260
    },
    {
      "epoch": 0.0851282830072473,
      "grad_norm": 29.375,
      "learning_rate": 1.996029376758515e-05,
      "loss": 7.4226,
      "step": 30270
    },
    {
      "epoch": 0.08515640599469601,
      "grad_norm": 25.0,
      "learning_rate": 1.996026754506555e-05,
      "loss": 7.6463,
      "step": 30280
    },
    {
      "epoch": 0.08518452898214472,
      "grad_norm": 24.75,
      "learning_rate": 1.99602413139072e-05,
      "loss": 8.238,
      "step": 30290
    },
    {
      "epoch": 0.08521265196959342,
      "grad_norm": 37.0,
      "learning_rate": 1.9960215074110118e-05,
      "loss": 8.188,
      "step": 30300
    },
    {
      "epoch": 0.08524077495704213,
      "grad_norm": 35.75,
      "learning_rate": 1.9960188825674327e-05,
      "loss": 7.9658,
      "step": 30310
    },
    {
      "epoch": 0.08526889794449084,
      "grad_norm": 30.875,
      "learning_rate": 1.9960162568599846e-05,
      "loss": 8.2808,
      "step": 30320
    },
    {
      "epoch": 0.08529702093193955,
      "grad_norm": 27.25,
      "learning_rate": 1.9960136302886704e-05,
      "loss": 7.4087,
      "step": 30330
    },
    {
      "epoch": 0.08532514391938827,
      "grad_norm": 25.125,
      "learning_rate": 1.9960110028534923e-05,
      "loss": 7.9866,
      "step": 30340
    },
    {
      "epoch": 0.08535326690683698,
      "grad_norm": 21.75,
      "learning_rate": 1.996008374554452e-05,
      "loss": 8.1708,
      "step": 30350
    },
    {
      "epoch": 0.08538138989428569,
      "grad_norm": 41.5,
      "learning_rate": 1.9960057453915525e-05,
      "loss": 7.8881,
      "step": 30360
    },
    {
      "epoch": 0.0854095128817344,
      "grad_norm": 30.875,
      "learning_rate": 1.9960031153647955e-05,
      "loss": 7.8825,
      "step": 30370
    },
    {
      "epoch": 0.08543763586918311,
      "grad_norm": 30.875,
      "learning_rate": 1.9960004844741835e-05,
      "loss": 8.5317,
      "step": 30380
    },
    {
      "epoch": 0.08546575885663182,
      "grad_norm": 52.5,
      "learning_rate": 1.995997852719719e-05,
      "loss": 7.0384,
      "step": 30390
    },
    {
      "epoch": 0.08549388184408054,
      "grad_norm": 30.0,
      "learning_rate": 1.9959952201014038e-05,
      "loss": 7.9712,
      "step": 30400
    },
    {
      "epoch": 0.08552200483152925,
      "grad_norm": 22.125,
      "learning_rate": 1.9959925866192406e-05,
      "loss": 8.1405,
      "step": 30410
    },
    {
      "epoch": 0.08555012781897796,
      "grad_norm": 23.875,
      "learning_rate": 1.9959899522732318e-05,
      "loss": 7.697,
      "step": 30420
    },
    {
      "epoch": 0.08557825080642667,
      "grad_norm": 33.0,
      "learning_rate": 1.9959873170633792e-05,
      "loss": 7.9128,
      "step": 30430
    },
    {
      "epoch": 0.08560637379387538,
      "grad_norm": 25.25,
      "learning_rate": 1.9959846809896852e-05,
      "loss": 7.2761,
      "step": 30440
    },
    {
      "epoch": 0.0856344967813241,
      "grad_norm": 50.0,
      "learning_rate": 1.9959820440521524e-05,
      "loss": 8.0125,
      "step": 30450
    },
    {
      "epoch": 0.08566261976877279,
      "grad_norm": 22.875,
      "learning_rate": 1.995979406250783e-05,
      "loss": 7.8731,
      "step": 30460
    },
    {
      "epoch": 0.0856907427562215,
      "grad_norm": 31.25,
      "learning_rate": 1.995976767585579e-05,
      "loss": 7.641,
      "step": 30470
    },
    {
      "epoch": 0.08571886574367021,
      "grad_norm": 24.75,
      "learning_rate": 1.995974128056543e-05,
      "loss": 8.1909,
      "step": 30480
    },
    {
      "epoch": 0.08574698873111893,
      "grad_norm": 43.0,
      "learning_rate": 1.9959714876636772e-05,
      "loss": 7.9417,
      "step": 30490
    },
    {
      "epoch": 0.08577511171856764,
      "grad_norm": 28.75,
      "learning_rate": 1.9959688464069838e-05,
      "loss": 7.8041,
      "step": 30500
    },
    {
      "epoch": 0.08580323470601635,
      "grad_norm": 86.5,
      "learning_rate": 1.995966204286465e-05,
      "loss": 7.9301,
      "step": 30510
    },
    {
      "epoch": 0.08583135769346506,
      "grad_norm": 42.75,
      "learning_rate": 1.9959635613021234e-05,
      "loss": 7.9757,
      "step": 30520
    },
    {
      "epoch": 0.08585948068091377,
      "grad_norm": 37.75,
      "learning_rate": 1.995960917453961e-05,
      "loss": 7.7735,
      "step": 30530
    },
    {
      "epoch": 0.08588760366836248,
      "grad_norm": 26.375,
      "learning_rate": 1.9959582727419805e-05,
      "loss": 7.8387,
      "step": 30540
    },
    {
      "epoch": 0.0859157266558112,
      "grad_norm": 20.5,
      "learning_rate": 1.9959556271661834e-05,
      "loss": 7.6276,
      "step": 30550
    },
    {
      "epoch": 0.0859438496432599,
      "grad_norm": 27.5,
      "learning_rate": 1.995952980726573e-05,
      "loss": 8.2386,
      "step": 30560
    },
    {
      "epoch": 0.08597197263070862,
      "grad_norm": 25.5,
      "learning_rate": 1.9959503334231512e-05,
      "loss": 7.6792,
      "step": 30570
    },
    {
      "epoch": 0.08600009561815733,
      "grad_norm": 35.75,
      "learning_rate": 1.99594768525592e-05,
      "loss": 8.2926,
      "step": 30580
    },
    {
      "epoch": 0.08602821860560604,
      "grad_norm": 31.375,
      "learning_rate": 1.9959450362248816e-05,
      "loss": 7.9275,
      "step": 30590
    },
    {
      "epoch": 0.08605634159305475,
      "grad_norm": 34.75,
      "learning_rate": 1.9959423863300388e-05,
      "loss": 7.8441,
      "step": 30600
    },
    {
      "epoch": 0.08608446458050346,
      "grad_norm": 28.125,
      "learning_rate": 1.995939735571394e-05,
      "loss": 7.2813,
      "step": 30610
    },
    {
      "epoch": 0.08611258756795216,
      "grad_norm": 31.125,
      "learning_rate": 1.995937083948949e-05,
      "loss": 7.5703,
      "step": 30620
    },
    {
      "epoch": 0.08614071055540087,
      "grad_norm": 21.125,
      "learning_rate": 1.9959344314627062e-05,
      "loss": 8.4307,
      "step": 30630
    },
    {
      "epoch": 0.08616883354284958,
      "grad_norm": 28.625,
      "learning_rate": 1.995931778112668e-05,
      "loss": 7.6608,
      "step": 30640
    },
    {
      "epoch": 0.0861969565302983,
      "grad_norm": 27.75,
      "learning_rate": 1.9959291238988367e-05,
      "loss": 8.3147,
      "step": 30650
    },
    {
      "epoch": 0.08622507951774701,
      "grad_norm": 26.75,
      "learning_rate": 1.9959264688212144e-05,
      "loss": 7.1333,
      "step": 30660
    },
    {
      "epoch": 0.08625320250519572,
      "grad_norm": 37.75,
      "learning_rate": 1.995923812879804e-05,
      "loss": 7.8794,
      "step": 30670
    },
    {
      "epoch": 0.08628132549264443,
      "grad_norm": 34.25,
      "learning_rate": 1.995921156074607e-05,
      "loss": 8.2676,
      "step": 30680
    },
    {
      "epoch": 0.08630944848009314,
      "grad_norm": 34.25,
      "learning_rate": 1.9959184984056264e-05,
      "loss": 7.2158,
      "step": 30690
    },
    {
      "epoch": 0.08633757146754185,
      "grad_norm": 29.25,
      "learning_rate": 1.9959158398728643e-05,
      "loss": 7.5831,
      "step": 30700
    },
    {
      "epoch": 0.08636569445499057,
      "grad_norm": 29.75,
      "learning_rate": 1.9959131804763226e-05,
      "loss": 8.2412,
      "step": 30710
    },
    {
      "epoch": 0.08639381744243928,
      "grad_norm": 37.75,
      "learning_rate": 1.9959105202160046e-05,
      "loss": 8.1694,
      "step": 30720
    },
    {
      "epoch": 0.08642194042988799,
      "grad_norm": 33.25,
      "learning_rate": 1.9959078590919112e-05,
      "loss": 8.2763,
      "step": 30730
    },
    {
      "epoch": 0.0864500634173367,
      "grad_norm": 45.75,
      "learning_rate": 1.9959051971040457e-05,
      "loss": 7.9991,
      "step": 30740
    },
    {
      "epoch": 0.08647818640478541,
      "grad_norm": 29.125,
      "learning_rate": 1.99590253425241e-05,
      "loss": 7.7628,
      "step": 30750
    },
    {
      "epoch": 0.08650630939223412,
      "grad_norm": 36.0,
      "learning_rate": 1.9958998705370066e-05,
      "loss": 7.9506,
      "step": 30760
    },
    {
      "epoch": 0.08653443237968284,
      "grad_norm": 29.5,
      "learning_rate": 1.995897205957838e-05,
      "loss": 7.8155,
      "step": 30770
    },
    {
      "epoch": 0.08656255536713153,
      "grad_norm": 33.5,
      "learning_rate": 1.995894540514906e-05,
      "loss": 8.0788,
      "step": 30780
    },
    {
      "epoch": 0.08659067835458024,
      "grad_norm": 30.0,
      "learning_rate": 1.9958918742082135e-05,
      "loss": 8.0447,
      "step": 30790
    },
    {
      "epoch": 0.08661880134202896,
      "grad_norm": 25.125,
      "learning_rate": 1.9958892070377624e-05,
      "loss": 7.5548,
      "step": 30800
    },
    {
      "epoch": 0.08664692432947767,
      "grad_norm": 40.25,
      "learning_rate": 1.9958865390035552e-05,
      "loss": 8.4124,
      "step": 30810
    },
    {
      "epoch": 0.08667504731692638,
      "grad_norm": 53.75,
      "learning_rate": 1.995883870105594e-05,
      "loss": 8.1266,
      "step": 30820
    },
    {
      "epoch": 0.08670317030437509,
      "grad_norm": 36.25,
      "learning_rate": 1.9958812003438814e-05,
      "loss": 8.8181,
      "step": 30830
    },
    {
      "epoch": 0.0867312932918238,
      "grad_norm": 24.25,
      "learning_rate": 1.9958785297184195e-05,
      "loss": 7.6482,
      "step": 30840
    },
    {
      "epoch": 0.08675941627927251,
      "grad_norm": 35.0,
      "learning_rate": 1.9958758582292103e-05,
      "loss": 8.2511,
      "step": 30850
    },
    {
      "epoch": 0.08678753926672123,
      "grad_norm": 27.5,
      "learning_rate": 1.995873185876257e-05,
      "loss": 7.4525,
      "step": 30860
    },
    {
      "epoch": 0.08681566225416994,
      "grad_norm": 34.5,
      "learning_rate": 1.9958705126595613e-05,
      "loss": 7.4543,
      "step": 30870
    },
    {
      "epoch": 0.08684378524161865,
      "grad_norm": 49.25,
      "learning_rate": 1.9958678385791255e-05,
      "loss": 8.2882,
      "step": 30880
    },
    {
      "epoch": 0.08687190822906736,
      "grad_norm": 49.0,
      "learning_rate": 1.995865163634952e-05,
      "loss": 7.9163,
      "step": 30890
    },
    {
      "epoch": 0.08690003121651607,
      "grad_norm": 23.0,
      "learning_rate": 1.9958624878270433e-05,
      "loss": 7.9205,
      "step": 30900
    },
    {
      "epoch": 0.08692815420396478,
      "grad_norm": 34.0,
      "learning_rate": 1.9958598111554016e-05,
      "loss": 7.1666,
      "step": 30910
    },
    {
      "epoch": 0.0869562771914135,
      "grad_norm": 35.0,
      "learning_rate": 1.995857133620029e-05,
      "loss": 8.0085,
      "step": 30920
    },
    {
      "epoch": 0.0869844001788622,
      "grad_norm": 37.5,
      "learning_rate": 1.9958544552209282e-05,
      "loss": 7.6574,
      "step": 30930
    },
    {
      "epoch": 0.08701252316631092,
      "grad_norm": 43.0,
      "learning_rate": 1.9958517759581016e-05,
      "loss": 8.1662,
      "step": 30940
    },
    {
      "epoch": 0.08704064615375962,
      "grad_norm": 34.25,
      "learning_rate": 1.995849095831551e-05,
      "loss": 7.8803,
      "step": 30950
    },
    {
      "epoch": 0.08706876914120833,
      "grad_norm": 27.25,
      "learning_rate": 1.995846414841279e-05,
      "loss": 7.7706,
      "step": 30960
    },
    {
      "epoch": 0.08709689212865704,
      "grad_norm": 30.0,
      "learning_rate": 1.995843732987288e-05,
      "loss": 8.1569,
      "step": 30970
    },
    {
      "epoch": 0.08712501511610575,
      "grad_norm": 26.875,
      "learning_rate": 1.9958410502695805e-05,
      "loss": 7.9677,
      "step": 30980
    },
    {
      "epoch": 0.08715313810355446,
      "grad_norm": 30.125,
      "learning_rate": 1.995838366688158e-05,
      "loss": 7.6945,
      "step": 30990
    },
    {
      "epoch": 0.08718126109100317,
      "grad_norm": 33.25,
      "learning_rate": 1.995835682243024e-05,
      "loss": 7.3983,
      "step": 31000
    },
    {
      "epoch": 0.08720938407845188,
      "grad_norm": 27.125,
      "learning_rate": 1.9958329969341802e-05,
      "loss": 7.7148,
      "step": 31010
    },
    {
      "epoch": 0.0872375070659006,
      "grad_norm": 62.25,
      "learning_rate": 1.9958303107616285e-05,
      "loss": 8.2018,
      "step": 31020
    },
    {
      "epoch": 0.08726563005334931,
      "grad_norm": 27.625,
      "learning_rate": 1.995827623725372e-05,
      "loss": 6.7307,
      "step": 31030
    },
    {
      "epoch": 0.08729375304079802,
      "grad_norm": 21.5,
      "learning_rate": 1.995824935825413e-05,
      "loss": 8.2217,
      "step": 31040
    },
    {
      "epoch": 0.08732187602824673,
      "grad_norm": 23.625,
      "learning_rate": 1.995822247061753e-05,
      "loss": 7.5232,
      "step": 31050
    },
    {
      "epoch": 0.08734999901569544,
      "grad_norm": 29.375,
      "learning_rate": 1.9958195574343952e-05,
      "loss": 8.2332,
      "step": 31060
    },
    {
      "epoch": 0.08737812200314415,
      "grad_norm": 33.0,
      "learning_rate": 1.995816866943342e-05,
      "loss": 7.8385,
      "step": 31070
    },
    {
      "epoch": 0.08740624499059287,
      "grad_norm": 27.125,
      "learning_rate": 1.9958141755885948e-05,
      "loss": 8.4511,
      "step": 31080
    },
    {
      "epoch": 0.08743436797804158,
      "grad_norm": 22.75,
      "learning_rate": 1.9958114833701566e-05,
      "loss": 7.6975,
      "step": 31090
    },
    {
      "epoch": 0.08746249096549029,
      "grad_norm": 28.0,
      "learning_rate": 1.9958087902880295e-05,
      "loss": 8.4121,
      "step": 31100
    },
    {
      "epoch": 0.08749061395293899,
      "grad_norm": 32.25,
      "learning_rate": 1.9958060963422164e-05,
      "loss": 7.855,
      "step": 31110
    },
    {
      "epoch": 0.0875187369403877,
      "grad_norm": 23.625,
      "learning_rate": 1.995803401532719e-05,
      "loss": 7.2676,
      "step": 31120
    },
    {
      "epoch": 0.08754685992783641,
      "grad_norm": 48.0,
      "learning_rate": 1.9958007058595398e-05,
      "loss": 7.7548,
      "step": 31130
    },
    {
      "epoch": 0.08757498291528512,
      "grad_norm": 22.875,
      "learning_rate": 1.995798009322681e-05,
      "loss": 7.9743,
      "step": 31140
    },
    {
      "epoch": 0.08760310590273383,
      "grad_norm": 35.75,
      "learning_rate": 1.9957953119221455e-05,
      "loss": 8.5904,
      "step": 31150
    },
    {
      "epoch": 0.08763122889018254,
      "grad_norm": 26.625,
      "learning_rate": 1.9957926136579352e-05,
      "loss": 7.7661,
      "step": 31160
    },
    {
      "epoch": 0.08765935187763126,
      "grad_norm": 36.75,
      "learning_rate": 1.995789914530052e-05,
      "loss": 7.4335,
      "step": 31170
    },
    {
      "epoch": 0.08768747486507997,
      "grad_norm": 34.0,
      "learning_rate": 1.9957872145384993e-05,
      "loss": 7.355,
      "step": 31180
    },
    {
      "epoch": 0.08771559785252868,
      "grad_norm": 33.5,
      "learning_rate": 1.9957845136832786e-05,
      "loss": 8.0948,
      "step": 31190
    },
    {
      "epoch": 0.08774372083997739,
      "grad_norm": 42.0,
      "learning_rate": 1.995781811964393e-05,
      "loss": 8.0791,
      "step": 31200
    },
    {
      "epoch": 0.0877718438274261,
      "grad_norm": 37.0,
      "learning_rate": 1.9957791093818436e-05,
      "loss": 8.0311,
      "step": 31210
    },
    {
      "epoch": 0.08779996681487481,
      "grad_norm": 44.0,
      "learning_rate": 1.995776405935634e-05,
      "loss": 7.581,
      "step": 31220
    },
    {
      "epoch": 0.08782808980232353,
      "grad_norm": 35.75,
      "learning_rate": 1.995773701625766e-05,
      "loss": 7.9411,
      "step": 31230
    },
    {
      "epoch": 0.08785621278977224,
      "grad_norm": 42.25,
      "learning_rate": 1.995770996452242e-05,
      "loss": 8.5506,
      "step": 31240
    },
    {
      "epoch": 0.08788433577722095,
      "grad_norm": 28.5,
      "learning_rate": 1.995768290415064e-05,
      "loss": 8.3206,
      "step": 31250
    },
    {
      "epoch": 0.08791245876466966,
      "grad_norm": 29.0,
      "learning_rate": 1.9957655835142354e-05,
      "loss": 7.8897,
      "step": 31260
    },
    {
      "epoch": 0.08794058175211836,
      "grad_norm": 32.5,
      "learning_rate": 1.9957628757497574e-05,
      "loss": 7.663,
      "step": 31270
    },
    {
      "epoch": 0.08796870473956707,
      "grad_norm": 28.375,
      "learning_rate": 1.9957601671216327e-05,
      "loss": 7.0347,
      "step": 31280
    },
    {
      "epoch": 0.08799682772701578,
      "grad_norm": 29.0,
      "learning_rate": 1.9957574576298638e-05,
      "loss": 7.8589,
      "step": 31290
    },
    {
      "epoch": 0.08802495071446449,
      "grad_norm": 28.75,
      "learning_rate": 1.995754747274453e-05,
      "loss": 7.4941,
      "step": 31300
    },
    {
      "epoch": 0.0880530737019132,
      "grad_norm": 28.0,
      "learning_rate": 1.995752036055403e-05,
      "loss": 7.9891,
      "step": 31310
    },
    {
      "epoch": 0.08808119668936192,
      "grad_norm": 111.5,
      "learning_rate": 1.995749323972715e-05,
      "loss": 8.1565,
      "step": 31320
    },
    {
      "epoch": 0.08810931967681063,
      "grad_norm": 39.0,
      "learning_rate": 1.9957466110263927e-05,
      "loss": 8.0063,
      "step": 31330
    },
    {
      "epoch": 0.08813744266425934,
      "grad_norm": 46.25,
      "learning_rate": 1.995743897216438e-05,
      "loss": 8.3311,
      "step": 31340
    },
    {
      "epoch": 0.08816556565170805,
      "grad_norm": 28.875,
      "learning_rate": 1.9957411825428528e-05,
      "loss": 8.1747,
      "step": 31350
    },
    {
      "epoch": 0.08819368863915676,
      "grad_norm": 23.5,
      "learning_rate": 1.99573846700564e-05,
      "loss": 7.8104,
      "step": 31360
    },
    {
      "epoch": 0.08822181162660547,
      "grad_norm": 33.25,
      "learning_rate": 1.995735750604802e-05,
      "loss": 8.1157,
      "step": 31370
    },
    {
      "epoch": 0.08824993461405418,
      "grad_norm": 29.5,
      "learning_rate": 1.9957330333403405e-05,
      "loss": 8.3593,
      "step": 31380
    },
    {
      "epoch": 0.0882780576015029,
      "grad_norm": 28.875,
      "learning_rate": 1.995730315212258e-05,
      "loss": 7.8612,
      "step": 31390
    },
    {
      "epoch": 0.08830618058895161,
      "grad_norm": 29.5,
      "learning_rate": 1.995727596220558e-05,
      "loss": 7.9469,
      "step": 31400
    },
    {
      "epoch": 0.08833430357640032,
      "grad_norm": 49.25,
      "learning_rate": 1.9957248763652414e-05,
      "loss": 8.0123,
      "step": 31410
    },
    {
      "epoch": 0.08836242656384903,
      "grad_norm": 30.875,
      "learning_rate": 1.9957221556463113e-05,
      "loss": 7.4203,
      "step": 31420
    },
    {
      "epoch": 0.08839054955129773,
      "grad_norm": 27.25,
      "learning_rate": 1.9957194340637698e-05,
      "loss": 7.8473,
      "step": 31430
    },
    {
      "epoch": 0.08841867253874644,
      "grad_norm": 34.5,
      "learning_rate": 1.9957167116176195e-05,
      "loss": 8.434,
      "step": 31440
    },
    {
      "epoch": 0.08844679552619515,
      "grad_norm": 25.375,
      "learning_rate": 1.9957139883078627e-05,
      "loss": 8.1961,
      "step": 31450
    },
    {
      "epoch": 0.08847491851364386,
      "grad_norm": 53.0,
      "learning_rate": 1.9957112641345015e-05,
      "loss": 7.9697,
      "step": 31460
    },
    {
      "epoch": 0.08850304150109257,
      "grad_norm": 29.5,
      "learning_rate": 1.9957085390975387e-05,
      "loss": 7.2683,
      "step": 31470
    },
    {
      "epoch": 0.08853116448854129,
      "grad_norm": 22.0,
      "learning_rate": 1.9957058131969764e-05,
      "loss": 7.7129,
      "step": 31480
    },
    {
      "epoch": 0.08855928747599,
      "grad_norm": 70.0,
      "learning_rate": 1.9957030864328165e-05,
      "loss": 8.0553,
      "step": 31490
    },
    {
      "epoch": 0.08858741046343871,
      "grad_norm": 43.0,
      "learning_rate": 1.995700358805062e-05,
      "loss": 8.5803,
      "step": 31500
    },
    {
      "epoch": 0.08861553345088742,
      "grad_norm": 31.5,
      "learning_rate": 1.9956976303137154e-05,
      "loss": 7.7549,
      "step": 31510
    },
    {
      "epoch": 0.08864365643833613,
      "grad_norm": 29.375,
      "learning_rate": 1.9956949009587785e-05,
      "loss": 8.4476,
      "step": 31520
    },
    {
      "epoch": 0.08867177942578484,
      "grad_norm": 43.75,
      "learning_rate": 1.9956921707402543e-05,
      "loss": 7.8869,
      "step": 31530
    },
    {
      "epoch": 0.08869990241323356,
      "grad_norm": 32.0,
      "learning_rate": 1.9956894396581448e-05,
      "loss": 7.6557,
      "step": 31540
    },
    {
      "epoch": 0.08872802540068227,
      "grad_norm": 28.375,
      "learning_rate": 1.995686707712452e-05,
      "loss": 8.0399,
      "step": 31550
    },
    {
      "epoch": 0.08875614838813098,
      "grad_norm": 36.25,
      "learning_rate": 1.995683974903179e-05,
      "loss": 7.2351,
      "step": 31560
    },
    {
      "epoch": 0.08878427137557969,
      "grad_norm": 33.5,
      "learning_rate": 1.9956812412303277e-05,
      "loss": 7.7686,
      "step": 31570
    },
    {
      "epoch": 0.0888123943630284,
      "grad_norm": 48.25,
      "learning_rate": 1.9956785066939005e-05,
      "loss": 7.9967,
      "step": 31580
    },
    {
      "epoch": 0.08884051735047711,
      "grad_norm": 28.625,
      "learning_rate": 1.9956757712939e-05,
      "loss": 7.9237,
      "step": 31590
    },
    {
      "epoch": 0.08886864033792581,
      "grad_norm": 26.875,
      "learning_rate": 1.9956730350303283e-05,
      "loss": 7.753,
      "step": 31600
    },
    {
      "epoch": 0.08889676332537452,
      "grad_norm": 24.75,
      "learning_rate": 1.9956702979031882e-05,
      "loss": 7.8594,
      "step": 31610
    },
    {
      "epoch": 0.08892488631282323,
      "grad_norm": 23.875,
      "learning_rate": 1.9956675599124815e-05,
      "loss": 7.3744,
      "step": 31620
    },
    {
      "epoch": 0.08895300930027195,
      "grad_norm": 36.0,
      "learning_rate": 1.995664821058211e-05,
      "loss": 8.1098,
      "step": 31630
    },
    {
      "epoch": 0.08898113228772066,
      "grad_norm": 46.0,
      "learning_rate": 1.9956620813403787e-05,
      "loss": 7.8637,
      "step": 31640
    },
    {
      "epoch": 0.08900925527516937,
      "grad_norm": 36.75,
      "learning_rate": 1.995659340758987e-05,
      "loss": 7.9329,
      "step": 31650
    },
    {
      "epoch": 0.08903737826261808,
      "grad_norm": 25.125,
      "learning_rate": 1.9956565993140393e-05,
      "loss": 8.1023,
      "step": 31660
    },
    {
      "epoch": 0.08906550125006679,
      "grad_norm": 24.75,
      "learning_rate": 1.9956538570055367e-05,
      "loss": 7.3029,
      "step": 31670
    },
    {
      "epoch": 0.0890936242375155,
      "grad_norm": 31.25,
      "learning_rate": 1.995651113833482e-05,
      "loss": 7.6285,
      "step": 31680
    },
    {
      "epoch": 0.08912174722496422,
      "grad_norm": 50.25,
      "learning_rate": 1.995648369797878e-05,
      "loss": 6.8398,
      "step": 31690
    },
    {
      "epoch": 0.08914987021241293,
      "grad_norm": 43.75,
      "learning_rate": 1.995645624898726e-05,
      "loss": 7.7186,
      "step": 31700
    },
    {
      "epoch": 0.08917799319986164,
      "grad_norm": 33.75,
      "learning_rate": 1.9956428791360295e-05,
      "loss": 8.1996,
      "step": 31710
    },
    {
      "epoch": 0.08920611618731035,
      "grad_norm": 45.25,
      "learning_rate": 1.9956401325097907e-05,
      "loss": 7.9158,
      "step": 31720
    },
    {
      "epoch": 0.08923423917475906,
      "grad_norm": 34.0,
      "learning_rate": 1.9956373850200116e-05,
      "loss": 8.6522,
      "step": 31730
    },
    {
      "epoch": 0.08926236216220777,
      "grad_norm": 40.0,
      "learning_rate": 1.9956346366666946e-05,
      "loss": 7.9574,
      "step": 31740
    },
    {
      "epoch": 0.08929048514965648,
      "grad_norm": 42.5,
      "learning_rate": 1.995631887449842e-05,
      "loss": 8.4916,
      "step": 31750
    },
    {
      "epoch": 0.08931860813710518,
      "grad_norm": 31.75,
      "learning_rate": 1.9956291373694568e-05,
      "loss": 8.3419,
      "step": 31760
    },
    {
      "epoch": 0.0893467311245539,
      "grad_norm": 27.75,
      "learning_rate": 1.9956263864255407e-05,
      "loss": 8.3432,
      "step": 31770
    },
    {
      "epoch": 0.0893748541120026,
      "grad_norm": 26.375,
      "learning_rate": 1.9956236346180965e-05,
      "loss": 7.7255,
      "step": 31780
    },
    {
      "epoch": 0.08940297709945132,
      "grad_norm": 38.75,
      "learning_rate": 1.9956208819471267e-05,
      "loss": 8.3694,
      "step": 31790
    },
    {
      "epoch": 0.08943110008690003,
      "grad_norm": 29.5,
      "learning_rate": 1.995618128412633e-05,
      "loss": 7.3745,
      "step": 31800
    },
    {
      "epoch": 0.08945922307434874,
      "grad_norm": 36.0,
      "learning_rate": 1.9956153740146183e-05,
      "loss": 8.7561,
      "step": 31810
    },
    {
      "epoch": 0.08948734606179745,
      "grad_norm": 43.25,
      "learning_rate": 1.9956126187530853e-05,
      "loss": 7.9559,
      "step": 31820
    },
    {
      "epoch": 0.08951546904924616,
      "grad_norm": 35.0,
      "learning_rate": 1.9956098626280357e-05,
      "loss": 7.3742,
      "step": 31830
    },
    {
      "epoch": 0.08954359203669487,
      "grad_norm": 34.0,
      "learning_rate": 1.9956071056394723e-05,
      "loss": 7.3911,
      "step": 31840
    },
    {
      "epoch": 0.08957171502414359,
      "grad_norm": 36.75,
      "learning_rate": 1.9956043477873974e-05,
      "loss": 7.6747,
      "step": 31850
    },
    {
      "epoch": 0.0895998380115923,
      "grad_norm": 35.5,
      "learning_rate": 1.9956015890718134e-05,
      "loss": 8.5908,
      "step": 31860
    },
    {
      "epoch": 0.08962796099904101,
      "grad_norm": 31.25,
      "learning_rate": 1.9955988294927223e-05,
      "loss": 7.9108,
      "step": 31870
    },
    {
      "epoch": 0.08965608398648972,
      "grad_norm": 23.625,
      "learning_rate": 1.9955960690501276e-05,
      "loss": 7.5974,
      "step": 31880
    },
    {
      "epoch": 0.08968420697393843,
      "grad_norm": 35.25,
      "learning_rate": 1.9955933077440302e-05,
      "loss": 7.6679,
      "step": 31890
    },
    {
      "epoch": 0.08971232996138714,
      "grad_norm": 29.625,
      "learning_rate": 1.995590545574434e-05,
      "loss": 8.1325,
      "step": 31900
    },
    {
      "epoch": 0.08974045294883586,
      "grad_norm": 23.875,
      "learning_rate": 1.99558778254134e-05,
      "loss": 7.6805,
      "step": 31910
    },
    {
      "epoch": 0.08976857593628455,
      "grad_norm": 28.875,
      "learning_rate": 1.9955850186447513e-05,
      "loss": 8.2192,
      "step": 31920
    },
    {
      "epoch": 0.08979669892373326,
      "grad_norm": 32.75,
      "learning_rate": 1.9955822538846708e-05,
      "loss": 7.4486,
      "step": 31930
    },
    {
      "epoch": 0.08982482191118198,
      "grad_norm": 41.0,
      "learning_rate": 1.9955794882611e-05,
      "loss": 7.97,
      "step": 31940
    },
    {
      "epoch": 0.08985294489863069,
      "grad_norm": 30.375,
      "learning_rate": 1.9955767217740416e-05,
      "loss": 8.2425,
      "step": 31950
    },
    {
      "epoch": 0.0898810678860794,
      "grad_norm": 29.625,
      "learning_rate": 1.995573954423498e-05,
      "loss": 8.5579,
      "step": 31960
    },
    {
      "epoch": 0.08990919087352811,
      "grad_norm": 27.5,
      "learning_rate": 1.995571186209472e-05,
      "loss": 8.605,
      "step": 31970
    },
    {
      "epoch": 0.08993731386097682,
      "grad_norm": 26.875,
      "learning_rate": 1.9955684171319653e-05,
      "loss": 7.4941,
      "step": 31980
    },
    {
      "epoch": 0.08996543684842553,
      "grad_norm": 29.0,
      "learning_rate": 1.9955656471909806e-05,
      "loss": 7.0686,
      "step": 31990
    },
    {
      "epoch": 0.08999355983587425,
      "grad_norm": 27.25,
      "learning_rate": 1.9955628763865204e-05,
      "loss": 8.023,
      "step": 32000
    },
    {
      "epoch": 0.09002168282332296,
      "grad_norm": 30.375,
      "learning_rate": 1.9955601047185873e-05,
      "loss": 7.8517,
      "step": 32010
    },
    {
      "epoch": 0.09004980581077167,
      "grad_norm": 32.5,
      "learning_rate": 1.9955573321871834e-05,
      "loss": 7.2573,
      "step": 32020
    },
    {
      "epoch": 0.09007792879822038,
      "grad_norm": 20.25,
      "learning_rate": 1.995554558792311e-05,
      "loss": 7.7659,
      "step": 32030
    },
    {
      "epoch": 0.09010605178566909,
      "grad_norm": 26.5,
      "learning_rate": 1.995551784533973e-05,
      "loss": 8.111,
      "step": 32040
    },
    {
      "epoch": 0.0901341747731178,
      "grad_norm": 39.75,
      "learning_rate": 1.995549009412171e-05,
      "loss": 7.9312,
      "step": 32050
    },
    {
      "epoch": 0.09016229776056651,
      "grad_norm": 24.5,
      "learning_rate": 1.9955462334269083e-05,
      "loss": 8.0246,
      "step": 32060
    },
    {
      "epoch": 0.09019042074801523,
      "grad_norm": 28.5,
      "learning_rate": 1.9955434565781866e-05,
      "loss": 7.9773,
      "step": 32070
    },
    {
      "epoch": 0.09021854373546392,
      "grad_norm": 26.875,
      "learning_rate": 1.9955406788660087e-05,
      "loss": 8.0636,
      "step": 32080
    },
    {
      "epoch": 0.09024666672291264,
      "grad_norm": 22.875,
      "learning_rate": 1.995537900290377e-05,
      "loss": 8.456,
      "step": 32090
    },
    {
      "epoch": 0.09027478971036135,
      "grad_norm": 25.125,
      "learning_rate": 1.9955351208512933e-05,
      "loss": 7.2841,
      "step": 32100
    },
    {
      "epoch": 0.09030291269781006,
      "grad_norm": 67.0,
      "learning_rate": 1.9955323405487613e-05,
      "loss": 8.062,
      "step": 32110
    },
    {
      "epoch": 0.09033103568525877,
      "grad_norm": 40.75,
      "learning_rate": 1.9955295593827818e-05,
      "loss": 7.8764,
      "step": 32120
    },
    {
      "epoch": 0.09035915867270748,
      "grad_norm": 31.25,
      "learning_rate": 1.9955267773533586e-05,
      "loss": 8.1385,
      "step": 32130
    },
    {
      "epoch": 0.0903872816601562,
      "grad_norm": 26.5,
      "learning_rate": 1.9955239944604937e-05,
      "loss": 8.8775,
      "step": 32140
    },
    {
      "epoch": 0.0904154046476049,
      "grad_norm": 20.0,
      "learning_rate": 1.995521210704189e-05,
      "loss": 7.4645,
      "step": 32150
    },
    {
      "epoch": 0.09044352763505362,
      "grad_norm": 33.5,
      "learning_rate": 1.9955184260844475e-05,
      "loss": 7.7764,
      "step": 32160
    },
    {
      "epoch": 0.09047165062250233,
      "grad_norm": 67.0,
      "learning_rate": 1.9955156406012715e-05,
      "loss": 8.5735,
      "step": 32170
    },
    {
      "epoch": 0.09049977360995104,
      "grad_norm": 33.75,
      "learning_rate": 1.995512854254663e-05,
      "loss": 8.1308,
      "step": 32180
    },
    {
      "epoch": 0.09052789659739975,
      "grad_norm": 31.125,
      "learning_rate": 1.995510067044625e-05,
      "loss": 8.0062,
      "step": 32190
    },
    {
      "epoch": 0.09055601958484846,
      "grad_norm": 34.25,
      "learning_rate": 1.9955072789711597e-05,
      "loss": 7.9201,
      "step": 32200
    },
    {
      "epoch": 0.09058414257229717,
      "grad_norm": 36.0,
      "learning_rate": 1.9955044900342692e-05,
      "loss": 8.0047,
      "step": 32210
    },
    {
      "epoch": 0.09061226555974589,
      "grad_norm": 25.875,
      "learning_rate": 1.9955017002339564e-05,
      "loss": 7.7446,
      "step": 32220
    },
    {
      "epoch": 0.0906403885471946,
      "grad_norm": 31.5,
      "learning_rate": 1.9954989095702232e-05,
      "loss": 7.9312,
      "step": 32230
    },
    {
      "epoch": 0.0906685115346433,
      "grad_norm": 31.0,
      "learning_rate": 1.9954961180430728e-05,
      "loss": 7.9033,
      "step": 32240
    },
    {
      "epoch": 0.090696634522092,
      "grad_norm": 50.5,
      "learning_rate": 1.9954933256525068e-05,
      "loss": 7.6294,
      "step": 32250
    },
    {
      "epoch": 0.09072475750954072,
      "grad_norm": 27.875,
      "learning_rate": 1.9954905323985282e-05,
      "loss": 7.3763,
      "step": 32260
    },
    {
      "epoch": 0.09075288049698943,
      "grad_norm": 21.875,
      "learning_rate": 1.9954877382811392e-05,
      "loss": 7.7934,
      "step": 32270
    },
    {
      "epoch": 0.09078100348443814,
      "grad_norm": 23.5,
      "learning_rate": 1.995484943300342e-05,
      "loss": 8.4928,
      "step": 32280
    },
    {
      "epoch": 0.09080912647188685,
      "grad_norm": 24.5,
      "learning_rate": 1.9954821474561394e-05,
      "loss": 8.4223,
      "step": 32290
    },
    {
      "epoch": 0.09083724945933556,
      "grad_norm": 21.875,
      "learning_rate": 1.9954793507485338e-05,
      "loss": 8.5671,
      "step": 32300
    },
    {
      "epoch": 0.09086537244678428,
      "grad_norm": 27.5,
      "learning_rate": 1.9954765531775272e-05,
      "loss": 7.5427,
      "step": 32310
    },
    {
      "epoch": 0.09089349543423299,
      "grad_norm": 29.5,
      "learning_rate": 1.9954737547431226e-05,
      "loss": 7.5891,
      "step": 32320
    },
    {
      "epoch": 0.0909216184216817,
      "grad_norm": 24.25,
      "learning_rate": 1.995470955445322e-05,
      "loss": 8.3969,
      "step": 32330
    },
    {
      "epoch": 0.09094974140913041,
      "grad_norm": 25.875,
      "learning_rate": 1.995468155284128e-05,
      "loss": 7.6234,
      "step": 32340
    },
    {
      "epoch": 0.09097786439657912,
      "grad_norm": 24.25,
      "learning_rate": 1.995465354259543e-05,
      "loss": 7.8102,
      "step": 32350
    },
    {
      "epoch": 0.09100598738402783,
      "grad_norm": 25.125,
      "learning_rate": 1.9954625523715693e-05,
      "loss": 7.9179,
      "step": 32360
    },
    {
      "epoch": 0.09103411037147655,
      "grad_norm": 29.25,
      "learning_rate": 1.9954597496202098e-05,
      "loss": 8.1455,
      "step": 32370
    },
    {
      "epoch": 0.09106223335892526,
      "grad_norm": 23.875,
      "learning_rate": 1.9954569460054665e-05,
      "loss": 7.324,
      "step": 32380
    },
    {
      "epoch": 0.09109035634637397,
      "grad_norm": 26.25,
      "learning_rate": 1.9954541415273416e-05,
      "loss": 7.6981,
      "step": 32390
    },
    {
      "epoch": 0.09111847933382268,
      "grad_norm": 25.125,
      "learning_rate": 1.995451336185838e-05,
      "loss": 8.2768,
      "step": 32400
    },
    {
      "epoch": 0.09114660232127138,
      "grad_norm": 25.25,
      "learning_rate": 1.9954485299809584e-05,
      "loss": 7.4089,
      "step": 32410
    },
    {
      "epoch": 0.09117472530872009,
      "grad_norm": 34.25,
      "learning_rate": 1.9954457229127043e-05,
      "loss": 7.9298,
      "step": 32420
    },
    {
      "epoch": 0.0912028482961688,
      "grad_norm": 28.875,
      "learning_rate": 1.995442914981079e-05,
      "loss": 7.4125,
      "step": 32430
    },
    {
      "epoch": 0.09123097128361751,
      "grad_norm": 41.75,
      "learning_rate": 1.9954401061860845e-05,
      "loss": 8.3184,
      "step": 32440
    },
    {
      "epoch": 0.09125909427106622,
      "grad_norm": 29.625,
      "learning_rate": 1.9954372965277232e-05,
      "loss": 8.0944,
      "step": 32450
    },
    {
      "epoch": 0.09128721725851494,
      "grad_norm": 28.0,
      "learning_rate": 1.995434486005998e-05,
      "loss": 7.9146,
      "step": 32460
    },
    {
      "epoch": 0.09131534024596365,
      "grad_norm": 30.875,
      "learning_rate": 1.9954316746209108e-05,
      "loss": 7.5536,
      "step": 32470
    },
    {
      "epoch": 0.09134346323341236,
      "grad_norm": 24.375,
      "learning_rate": 1.9954288623724644e-05,
      "loss": 7.7397,
      "step": 32480
    },
    {
      "epoch": 0.09137158622086107,
      "grad_norm": 61.5,
      "learning_rate": 1.9954260492606607e-05,
      "loss": 8.1758,
      "step": 32490
    },
    {
      "epoch": 0.09139970920830978,
      "grad_norm": 21.125,
      "learning_rate": 1.9954232352855027e-05,
      "loss": 7.496,
      "step": 32500
    },
    {
      "epoch": 0.0914278321957585,
      "grad_norm": 37.0,
      "learning_rate": 1.9954204204469925e-05,
      "loss": 8.2438,
      "step": 32510
    },
    {
      "epoch": 0.0914559551832072,
      "grad_norm": 46.25,
      "learning_rate": 1.995417604745133e-05,
      "loss": 8.4312,
      "step": 32520
    },
    {
      "epoch": 0.09148407817065592,
      "grad_norm": 28.0,
      "learning_rate": 1.9954147881799267e-05,
      "loss": 7.7676,
      "step": 32530
    },
    {
      "epoch": 0.09151220115810463,
      "grad_norm": 37.5,
      "learning_rate": 1.9954119707513754e-05,
      "loss": 7.6462,
      "step": 32540
    },
    {
      "epoch": 0.09154032414555334,
      "grad_norm": 30.25,
      "learning_rate": 1.9954091524594814e-05,
      "loss": 8.7002,
      "step": 32550
    },
    {
      "epoch": 0.09156844713300205,
      "grad_norm": 40.75,
      "learning_rate": 1.995406333304248e-05,
      "loss": 7.591,
      "step": 32560
    },
    {
      "epoch": 0.09159657012045075,
      "grad_norm": 36.5,
      "learning_rate": 1.995403513285677e-05,
      "loss": 7.241,
      "step": 32570
    },
    {
      "epoch": 0.09162469310789946,
      "grad_norm": 28.25,
      "learning_rate": 1.9954006924037714e-05,
      "loss": 8.1166,
      "step": 32580
    },
    {
      "epoch": 0.09165281609534817,
      "grad_norm": 37.0,
      "learning_rate": 1.9953978706585332e-05,
      "loss": 7.65,
      "step": 32590
    },
    {
      "epoch": 0.09168093908279688,
      "grad_norm": 58.0,
      "learning_rate": 1.995395048049965e-05,
      "loss": 8.404,
      "step": 32600
    },
    {
      "epoch": 0.0917090620702456,
      "grad_norm": 21.625,
      "learning_rate": 1.995392224578069e-05,
      "loss": 7.1024,
      "step": 32610
    },
    {
      "epoch": 0.0917371850576943,
      "grad_norm": 36.5,
      "learning_rate": 1.9953894002428476e-05,
      "loss": 7.1803,
      "step": 32620
    },
    {
      "epoch": 0.09176530804514302,
      "grad_norm": 28.875,
      "learning_rate": 1.9953865750443042e-05,
      "loss": 8.5062,
      "step": 32630
    },
    {
      "epoch": 0.09179343103259173,
      "grad_norm": 30.375,
      "learning_rate": 1.99538374898244e-05,
      "loss": 8.2749,
      "step": 32640
    },
    {
      "epoch": 0.09182155402004044,
      "grad_norm": 17.875,
      "learning_rate": 1.9953809220572584e-05,
      "loss": 8.0172,
      "step": 32650
    },
    {
      "epoch": 0.09184967700748915,
      "grad_norm": 33.75,
      "learning_rate": 1.9953780942687613e-05,
      "loss": 8.1709,
      "step": 32660
    },
    {
      "epoch": 0.09187779999493786,
      "grad_norm": 28.25,
      "learning_rate": 1.9953752656169513e-05,
      "loss": 7.8763,
      "step": 32670
    },
    {
      "epoch": 0.09190592298238658,
      "grad_norm": 37.5,
      "learning_rate": 1.995372436101831e-05,
      "loss": 7.816,
      "step": 32680
    },
    {
      "epoch": 0.09193404596983529,
      "grad_norm": 31.125,
      "learning_rate": 1.9953696057234027e-05,
      "loss": 8.281,
      "step": 32690
    },
    {
      "epoch": 0.091962168957284,
      "grad_norm": 24.875,
      "learning_rate": 1.995366774481669e-05,
      "loss": 7.5432,
      "step": 32700
    },
    {
      "epoch": 0.09199029194473271,
      "grad_norm": 25.125,
      "learning_rate": 1.995363942376632e-05,
      "loss": 8.108,
      "step": 32710
    },
    {
      "epoch": 0.09201841493218142,
      "grad_norm": 23.125,
      "learning_rate": 1.9953611094082943e-05,
      "loss": 9.4395,
      "step": 32720
    },
    {
      "epoch": 0.09204653791963012,
      "grad_norm": 28.125,
      "learning_rate": 1.9953582755766586e-05,
      "loss": 7.9753,
      "step": 32730
    },
    {
      "epoch": 0.09207466090707883,
      "grad_norm": 24.5,
      "learning_rate": 1.9953554408817273e-05,
      "loss": 7.9057,
      "step": 32740
    },
    {
      "epoch": 0.09210278389452754,
      "grad_norm": 24.125,
      "learning_rate": 1.9953526053235027e-05,
      "loss": 6.9793,
      "step": 32750
    },
    {
      "epoch": 0.09213090688197625,
      "grad_norm": 25.875,
      "learning_rate": 1.9953497689019872e-05,
      "loss": 7.5584,
      "step": 32760
    },
    {
      "epoch": 0.09215902986942497,
      "grad_norm": 66.0,
      "learning_rate": 1.9953469316171836e-05,
      "loss": 8.645,
      "step": 32770
    },
    {
      "epoch": 0.09218715285687368,
      "grad_norm": 41.5,
      "learning_rate": 1.9953440934690937e-05,
      "loss": 8.3548,
      "step": 32780
    },
    {
      "epoch": 0.09221527584432239,
      "grad_norm": 52.25,
      "learning_rate": 1.995341254457721e-05,
      "loss": 7.9617,
      "step": 32790
    },
    {
      "epoch": 0.0922433988317711,
      "grad_norm": 26.5,
      "learning_rate": 1.995338414583067e-05,
      "loss": 7.9166,
      "step": 32800
    },
    {
      "epoch": 0.09227152181921981,
      "grad_norm": 36.0,
      "learning_rate": 1.9953355738451348e-05,
      "loss": 8.1291,
      "step": 32810
    },
    {
      "epoch": 0.09229964480666852,
      "grad_norm": 32.25,
      "learning_rate": 1.9953327322439264e-05,
      "loss": 8.5053,
      "step": 32820
    },
    {
      "epoch": 0.09232776779411724,
      "grad_norm": 21.5,
      "learning_rate": 1.9953298897794445e-05,
      "loss": 8.0581,
      "step": 32830
    },
    {
      "epoch": 0.09235589078156595,
      "grad_norm": 37.25,
      "learning_rate": 1.9953270464516916e-05,
      "loss": 8.0874,
      "step": 32840
    },
    {
      "epoch": 0.09238401376901466,
      "grad_norm": 26.125,
      "learning_rate": 1.9953242022606702e-05,
      "loss": 8.662,
      "step": 32850
    },
    {
      "epoch": 0.09241213675646337,
      "grad_norm": 41.0,
      "learning_rate": 1.9953213572063823e-05,
      "loss": 8.3718,
      "step": 32860
    },
    {
      "epoch": 0.09244025974391208,
      "grad_norm": 27.125,
      "learning_rate": 1.995318511288831e-05,
      "loss": 8.7256,
      "step": 32870
    },
    {
      "epoch": 0.0924683827313608,
      "grad_norm": 23.875,
      "learning_rate": 1.9953156645080184e-05,
      "loss": 7.5468,
      "step": 32880
    },
    {
      "epoch": 0.09249650571880949,
      "grad_norm": 26.5,
      "learning_rate": 1.995312816863947e-05,
      "loss": 7.3359,
      "step": 32890
    },
    {
      "epoch": 0.0925246287062582,
      "grad_norm": 37.25,
      "learning_rate": 1.9953099683566194e-05,
      "loss": 7.5172,
      "step": 32900
    },
    {
      "epoch": 0.09255275169370691,
      "grad_norm": 22.625,
      "learning_rate": 1.995307118986038e-05,
      "loss": 8.1141,
      "step": 32910
    },
    {
      "epoch": 0.09258087468115563,
      "grad_norm": 27.375,
      "learning_rate": 1.9953042687522056e-05,
      "loss": 7.7569,
      "step": 32920
    },
    {
      "epoch": 0.09260899766860434,
      "grad_norm": 33.25,
      "learning_rate": 1.995301417655124e-05,
      "loss": 7.5513,
      "step": 32930
    },
    {
      "epoch": 0.09263712065605305,
      "grad_norm": 27.75,
      "learning_rate": 1.9952985656947962e-05,
      "loss": 8.6064,
      "step": 32940
    },
    {
      "epoch": 0.09266524364350176,
      "grad_norm": 30.125,
      "learning_rate": 1.995295712871224e-05,
      "loss": 7.7238,
      "step": 32950
    },
    {
      "epoch": 0.09269336663095047,
      "grad_norm": 32.5,
      "learning_rate": 1.995292859184411e-05,
      "loss": 8.0163,
      "step": 32960
    },
    {
      "epoch": 0.09272148961839918,
      "grad_norm": 30.5,
      "learning_rate": 1.9952900046343585e-05,
      "loss": 7.7521,
      "step": 32970
    },
    {
      "epoch": 0.0927496126058479,
      "grad_norm": 21.875,
      "learning_rate": 1.99528714922107e-05,
      "loss": 7.9217,
      "step": 32980
    },
    {
      "epoch": 0.0927777355932966,
      "grad_norm": 35.25,
      "learning_rate": 1.9952842929445473e-05,
      "loss": 8.711,
      "step": 32990
    },
    {
      "epoch": 0.09280585858074532,
      "grad_norm": 40.25,
      "learning_rate": 1.9952814358047933e-05,
      "loss": 7.3418,
      "step": 33000
    },
    {
      "epoch": 0.09283398156819403,
      "grad_norm": 22.25,
      "learning_rate": 1.9952785778018097e-05,
      "loss": 8.1675,
      "step": 33010
    },
    {
      "epoch": 0.09286210455564274,
      "grad_norm": 41.0,
      "learning_rate": 1.9952757189356e-05,
      "loss": 7.9406,
      "step": 33020
    },
    {
      "epoch": 0.09289022754309145,
      "grad_norm": 26.25,
      "learning_rate": 1.995272859206166e-05,
      "loss": 7.7902,
      "step": 33030
    },
    {
      "epoch": 0.09291835053054016,
      "grad_norm": 27.875,
      "learning_rate": 1.9952699986135107e-05,
      "loss": 7.6033,
      "step": 33040
    },
    {
      "epoch": 0.09294647351798888,
      "grad_norm": 25.25,
      "learning_rate": 1.9952671371576356e-05,
      "loss": 7.7843,
      "step": 33050
    },
    {
      "epoch": 0.09297459650543757,
      "grad_norm": 26.25,
      "learning_rate": 1.9952642748385448e-05,
      "loss": 8.2589,
      "step": 33060
    },
    {
      "epoch": 0.09300271949288628,
      "grad_norm": 26.625,
      "learning_rate": 1.9952614116562394e-05,
      "loss": 7.9632,
      "step": 33070
    },
    {
      "epoch": 0.093030842480335,
      "grad_norm": 26.5,
      "learning_rate": 1.995258547610722e-05,
      "loss": 8.2487,
      "step": 33080
    },
    {
      "epoch": 0.09305896546778371,
      "grad_norm": 34.25,
      "learning_rate": 1.9952556827019956e-05,
      "loss": 8.5531,
      "step": 33090
    },
    {
      "epoch": 0.09308708845523242,
      "grad_norm": 34.75,
      "learning_rate": 1.9952528169300628e-05,
      "loss": 8.3147,
      "step": 33100
    },
    {
      "epoch": 0.09311521144268113,
      "grad_norm": 26.0,
      "learning_rate": 1.9952499502949255e-05,
      "loss": 7.9879,
      "step": 33110
    },
    {
      "epoch": 0.09314333443012984,
      "grad_norm": 45.25,
      "learning_rate": 1.9952470827965863e-05,
      "loss": 7.3487,
      "step": 33120
    },
    {
      "epoch": 0.09317145741757855,
      "grad_norm": 44.0,
      "learning_rate": 1.995244214435048e-05,
      "loss": 8.0107,
      "step": 33130
    },
    {
      "epoch": 0.09319958040502727,
      "grad_norm": 31.875,
      "learning_rate": 1.995241345210313e-05,
      "loss": 7.4307,
      "step": 33140
    },
    {
      "epoch": 0.09322770339247598,
      "grad_norm": 22.75,
      "learning_rate": 1.9952384751223836e-05,
      "loss": 7.8279,
      "step": 33150
    },
    {
      "epoch": 0.09325582637992469,
      "grad_norm": 27.625,
      "learning_rate": 1.9952356041712627e-05,
      "loss": 7.953,
      "step": 33160
    },
    {
      "epoch": 0.0932839493673734,
      "grad_norm": 44.5,
      "learning_rate": 1.9952327323569523e-05,
      "loss": 7.869,
      "step": 33170
    },
    {
      "epoch": 0.09331207235482211,
      "grad_norm": 31.625,
      "learning_rate": 1.995229859679455e-05,
      "loss": 8.1221,
      "step": 33180
    },
    {
      "epoch": 0.09334019534227082,
      "grad_norm": 36.25,
      "learning_rate": 1.9952269861387737e-05,
      "loss": 8.3212,
      "step": 33190
    },
    {
      "epoch": 0.09336831832971954,
      "grad_norm": 22.375,
      "learning_rate": 1.9952241117349104e-05,
      "loss": 7.9912,
      "step": 33200
    },
    {
      "epoch": 0.09339644131716825,
      "grad_norm": 35.25,
      "learning_rate": 1.9952212364678676e-05,
      "loss": 8.3282,
      "step": 33210
    },
    {
      "epoch": 0.09342456430461694,
      "grad_norm": 41.75,
      "learning_rate": 1.995218360337648e-05,
      "loss": 8.3339,
      "step": 33220
    },
    {
      "epoch": 0.09345268729206566,
      "grad_norm": 31.75,
      "learning_rate": 1.995215483344254e-05,
      "loss": 7.5689,
      "step": 33230
    },
    {
      "epoch": 0.09348081027951437,
      "grad_norm": 37.25,
      "learning_rate": 1.9952126054876886e-05,
      "loss": 7.9586,
      "step": 33240
    },
    {
      "epoch": 0.09350893326696308,
      "grad_norm": 37.0,
      "learning_rate": 1.9952097267679535e-05,
      "loss": 7.0005,
      "step": 33250
    },
    {
      "epoch": 0.09353705625441179,
      "grad_norm": 39.75,
      "learning_rate": 1.9952068471850516e-05,
      "loss": 7.9167,
      "step": 33260
    },
    {
      "epoch": 0.0935651792418605,
      "grad_norm": 22.75,
      "learning_rate": 1.9952039667389854e-05,
      "loss": 7.1487,
      "step": 33270
    },
    {
      "epoch": 0.09359330222930921,
      "grad_norm": 54.75,
      "learning_rate": 1.9952010854297575e-05,
      "loss": 8.1651,
      "step": 33280
    },
    {
      "epoch": 0.09362142521675793,
      "grad_norm": 24.75,
      "learning_rate": 1.99519820325737e-05,
      "loss": 7.9803,
      "step": 33290
    },
    {
      "epoch": 0.09364954820420664,
      "grad_norm": 30.125,
      "learning_rate": 1.9951953202218253e-05,
      "loss": 8.1252,
      "step": 33300
    },
    {
      "epoch": 0.09367767119165535,
      "grad_norm": 24.625,
      "learning_rate": 1.995192436323127e-05,
      "loss": 7.6334,
      "step": 33310
    },
    {
      "epoch": 0.09370579417910406,
      "grad_norm": 44.75,
      "learning_rate": 1.995189551561276e-05,
      "loss": 7.8594,
      "step": 33320
    },
    {
      "epoch": 0.09373391716655277,
      "grad_norm": 26.25,
      "learning_rate": 1.9951866659362764e-05,
      "loss": 7.9857,
      "step": 33330
    },
    {
      "epoch": 0.09376204015400148,
      "grad_norm": 31.5,
      "learning_rate": 1.9951837794481297e-05,
      "loss": 7.5463,
      "step": 33340
    },
    {
      "epoch": 0.0937901631414502,
      "grad_norm": 33.5,
      "learning_rate": 1.9951808920968386e-05,
      "loss": 8.3125,
      "step": 33350
    },
    {
      "epoch": 0.0938182861288989,
      "grad_norm": 36.5,
      "learning_rate": 1.9951780038824056e-05,
      "loss": 7.7852,
      "step": 33360
    },
    {
      "epoch": 0.09384640911634762,
      "grad_norm": 36.0,
      "learning_rate": 1.9951751148048335e-05,
      "loss": 8.0585,
      "step": 33370
    },
    {
      "epoch": 0.09387453210379632,
      "grad_norm": 22.5,
      "learning_rate": 1.9951722248641244e-05,
      "loss": 7.4017,
      "step": 33380
    },
    {
      "epoch": 0.09390265509124503,
      "grad_norm": 26.375,
      "learning_rate": 1.9951693340602812e-05,
      "loss": 7.4639,
      "step": 33390
    },
    {
      "epoch": 0.09393077807869374,
      "grad_norm": 29.375,
      "learning_rate": 1.995166442393306e-05,
      "loss": 7.6276,
      "step": 33400
    },
    {
      "epoch": 0.09395890106614245,
      "grad_norm": 25.625,
      "learning_rate": 1.9951635498632012e-05,
      "loss": 7.9188,
      "step": 33410
    },
    {
      "epoch": 0.09398702405359116,
      "grad_norm": 27.875,
      "learning_rate": 1.99516065646997e-05,
      "loss": 7.1325,
      "step": 33420
    },
    {
      "epoch": 0.09401514704103987,
      "grad_norm": 31.25,
      "learning_rate": 1.9951577622136146e-05,
      "loss": 7.869,
      "step": 33430
    },
    {
      "epoch": 0.09404327002848858,
      "grad_norm": 28.25,
      "learning_rate": 1.995154867094137e-05,
      "loss": 8.1573,
      "step": 33440
    },
    {
      "epoch": 0.0940713930159373,
      "grad_norm": 23.5,
      "learning_rate": 1.99515197111154e-05,
      "loss": 7.9537,
      "step": 33450
    },
    {
      "epoch": 0.09409951600338601,
      "grad_norm": 81.5,
      "learning_rate": 1.9951490742658267e-05,
      "loss": 8.169,
      "step": 33460
    },
    {
      "epoch": 0.09412763899083472,
      "grad_norm": 39.75,
      "learning_rate": 1.995146176556999e-05,
      "loss": 8.2499,
      "step": 33470
    },
    {
      "epoch": 0.09415576197828343,
      "grad_norm": 24.375,
      "learning_rate": 1.9951432779850598e-05,
      "loss": 8.3257,
      "step": 33480
    },
    {
      "epoch": 0.09418388496573214,
      "grad_norm": 29.5,
      "learning_rate": 1.9951403785500112e-05,
      "loss": 7.2528,
      "step": 33490
    },
    {
      "epoch": 0.09421200795318085,
      "grad_norm": 35.25,
      "learning_rate": 1.9951374782518556e-05,
      "loss": 8.024,
      "step": 33500
    },
    {
      "epoch": 0.09424013094062957,
      "grad_norm": 37.0,
      "learning_rate": 1.9951345770905962e-05,
      "loss": 8.3641,
      "step": 33510
    },
    {
      "epoch": 0.09426825392807828,
      "grad_norm": 44.75,
      "learning_rate": 1.995131675066235e-05,
      "loss": 8.0259,
      "step": 33520
    },
    {
      "epoch": 0.09429637691552699,
      "grad_norm": 35.0,
      "learning_rate": 1.9951287721787746e-05,
      "loss": 8.4489,
      "step": 33530
    },
    {
      "epoch": 0.09432449990297569,
      "grad_norm": 23.5,
      "learning_rate": 1.995125868428218e-05,
      "loss": 7.9927,
      "step": 33540
    },
    {
      "epoch": 0.0943526228904244,
      "grad_norm": 21.375,
      "learning_rate": 1.9951229638145663e-05,
      "loss": 7.3291,
      "step": 33550
    },
    {
      "epoch": 0.09438074587787311,
      "grad_norm": 21.25,
      "learning_rate": 1.9951200583378235e-05,
      "loss": 7.704,
      "step": 33560
    },
    {
      "epoch": 0.09440886886532182,
      "grad_norm": 64.0,
      "learning_rate": 1.995117151997992e-05,
      "loss": 7.7443,
      "step": 33570
    },
    {
      "epoch": 0.09443699185277053,
      "grad_norm": 25.875,
      "learning_rate": 1.9951142447950736e-05,
      "loss": 8.3682,
      "step": 33580
    },
    {
      "epoch": 0.09446511484021924,
      "grad_norm": 35.0,
      "learning_rate": 1.995111336729071e-05,
      "loss": 8.7455,
      "step": 33590
    },
    {
      "epoch": 0.09449323782766796,
      "grad_norm": 33.5,
      "learning_rate": 1.995108427799987e-05,
      "loss": 8.12,
      "step": 33600
    },
    {
      "epoch": 0.09452136081511667,
      "grad_norm": 25.0,
      "learning_rate": 1.9951055180078245e-05,
      "loss": 8.3369,
      "step": 33610
    },
    {
      "epoch": 0.09454948380256538,
      "grad_norm": 32.0,
      "learning_rate": 1.995102607352585e-05,
      "loss": 7.5879,
      "step": 33620
    },
    {
      "epoch": 0.09457760679001409,
      "grad_norm": 43.75,
      "learning_rate": 1.9950996958342717e-05,
      "loss": 7.9791,
      "step": 33630
    },
    {
      "epoch": 0.0946057297774628,
      "grad_norm": 30.875,
      "learning_rate": 1.995096783452887e-05,
      "loss": 8.7794,
      "step": 33640
    },
    {
      "epoch": 0.09463385276491151,
      "grad_norm": 28.0,
      "learning_rate": 1.9950938702084332e-05,
      "loss": 7.8254,
      "step": 33650
    },
    {
      "epoch": 0.09466197575236023,
      "grad_norm": 23.0,
      "learning_rate": 1.9950909561009132e-05,
      "loss": 8.8056,
      "step": 33660
    },
    {
      "epoch": 0.09469009873980894,
      "grad_norm": 40.5,
      "learning_rate": 1.9950880411303296e-05,
      "loss": 7.9687,
      "step": 33670
    },
    {
      "epoch": 0.09471822172725765,
      "grad_norm": 33.25,
      "learning_rate": 1.9950851252966842e-05,
      "loss": 8.2234,
      "step": 33680
    },
    {
      "epoch": 0.09474634471470636,
      "grad_norm": 34.0,
      "learning_rate": 1.9950822085999803e-05,
      "loss": 8.9285,
      "step": 33690
    },
    {
      "epoch": 0.09477446770215507,
      "grad_norm": 46.75,
      "learning_rate": 1.9950792910402203e-05,
      "loss": 7.8893,
      "step": 33700
    },
    {
      "epoch": 0.09480259068960377,
      "grad_norm": 50.5,
      "learning_rate": 1.995076372617406e-05,
      "loss": 7.6993,
      "step": 33710
    },
    {
      "epoch": 0.09483071367705248,
      "grad_norm": 30.75,
      "learning_rate": 1.995073453331541e-05,
      "loss": 8.126,
      "step": 33720
    },
    {
      "epoch": 0.09485883666450119,
      "grad_norm": 31.5,
      "learning_rate": 1.995070533182627e-05,
      "loss": 8.0762,
      "step": 33730
    },
    {
      "epoch": 0.0948869596519499,
      "grad_norm": 18.75,
      "learning_rate": 1.9950676121706674e-05,
      "loss": 7.4353,
      "step": 33740
    },
    {
      "epoch": 0.09491508263939862,
      "grad_norm": 50.25,
      "learning_rate": 1.9950646902956638e-05,
      "loss": 7.7661,
      "step": 33750
    },
    {
      "epoch": 0.09494320562684733,
      "grad_norm": 46.0,
      "learning_rate": 1.995061767557619e-05,
      "loss": 7.8698,
      "step": 33760
    },
    {
      "epoch": 0.09497132861429604,
      "grad_norm": 48.75,
      "learning_rate": 1.995058843956536e-05,
      "loss": 7.9607,
      "step": 33770
    },
    {
      "epoch": 0.09499945160174475,
      "grad_norm": 46.5,
      "learning_rate": 1.995055919492417e-05,
      "loss": 7.6457,
      "step": 33780
    },
    {
      "epoch": 0.09502757458919346,
      "grad_norm": 21.125,
      "learning_rate": 1.9950529941652647e-05,
      "loss": 7.9269,
      "step": 33790
    },
    {
      "epoch": 0.09505569757664217,
      "grad_norm": 34.25,
      "learning_rate": 1.9950500679750813e-05,
      "loss": 7.7117,
      "step": 33800
    },
    {
      "epoch": 0.09508382056409088,
      "grad_norm": 32.0,
      "learning_rate": 1.9950471409218695e-05,
      "loss": 8.1605,
      "step": 33810
    },
    {
      "epoch": 0.0951119435515396,
      "grad_norm": 31.5,
      "learning_rate": 1.9950442130056318e-05,
      "loss": 8.404,
      "step": 33820
    },
    {
      "epoch": 0.09514006653898831,
      "grad_norm": 34.25,
      "learning_rate": 1.9950412842263705e-05,
      "loss": 8.2777,
      "step": 33830
    },
    {
      "epoch": 0.09516818952643702,
      "grad_norm": 33.0,
      "learning_rate": 1.995038354584089e-05,
      "loss": 7.8773,
      "step": 33840
    },
    {
      "epoch": 0.09519631251388573,
      "grad_norm": 107.5,
      "learning_rate": 1.995035424078789e-05,
      "loss": 8.733,
      "step": 33850
    },
    {
      "epoch": 0.09522443550133444,
      "grad_norm": 45.5,
      "learning_rate": 1.9950324927104737e-05,
      "loss": 8.177,
      "step": 33860
    },
    {
      "epoch": 0.09525255848878314,
      "grad_norm": 48.5,
      "learning_rate": 1.995029560479145e-05,
      "loss": 8.4047,
      "step": 33870
    },
    {
      "epoch": 0.09528068147623185,
      "grad_norm": 31.875,
      "learning_rate": 1.9950266273848057e-05,
      "loss": 7.4535,
      "step": 33880
    },
    {
      "epoch": 0.09530880446368056,
      "grad_norm": 36.25,
      "learning_rate": 1.9950236934274582e-05,
      "loss": 8.4049,
      "step": 33890
    },
    {
      "epoch": 0.09533692745112927,
      "grad_norm": 41.25,
      "learning_rate": 1.9950207586071054e-05,
      "loss": 7.8892,
      "step": 33900
    },
    {
      "epoch": 0.09536505043857799,
      "grad_norm": 22.875,
      "learning_rate": 1.9950178229237494e-05,
      "loss": 7.4575,
      "step": 33910
    },
    {
      "epoch": 0.0953931734260267,
      "grad_norm": 42.75,
      "learning_rate": 1.9950148863773932e-05,
      "loss": 7.8154,
      "step": 33920
    },
    {
      "epoch": 0.09542129641347541,
      "grad_norm": 34.0,
      "learning_rate": 1.995011948968039e-05,
      "loss": 7.3913,
      "step": 33930
    },
    {
      "epoch": 0.09544941940092412,
      "grad_norm": 35.5,
      "learning_rate": 1.99500901069569e-05,
      "loss": 8.1025,
      "step": 33940
    },
    {
      "epoch": 0.09547754238837283,
      "grad_norm": 35.0,
      "learning_rate": 1.9950060715603478e-05,
      "loss": 7.7723,
      "step": 33950
    },
    {
      "epoch": 0.09550566537582154,
      "grad_norm": 25.875,
      "learning_rate": 1.995003131562015e-05,
      "loss": 8.2555,
      "step": 33960
    },
    {
      "epoch": 0.09553378836327026,
      "grad_norm": 23.875,
      "learning_rate": 1.9950001907006952e-05,
      "loss": 7.81,
      "step": 33970
    },
    {
      "epoch": 0.09556191135071897,
      "grad_norm": 40.0,
      "learning_rate": 1.99499724897639e-05,
      "loss": 8.7186,
      "step": 33980
    },
    {
      "epoch": 0.09559003433816768,
      "grad_norm": 39.0,
      "learning_rate": 1.9949943063891022e-05,
      "loss": 7.6267,
      "step": 33990
    },
    {
      "epoch": 0.09561815732561639,
      "grad_norm": 26.875,
      "learning_rate": 1.9949913629388346e-05,
      "loss": 7.2896,
      "step": 34000
    },
    {
      "epoch": 0.0956462803130651,
      "grad_norm": 28.625,
      "learning_rate": 1.9949884186255894e-05,
      "loss": 7.5311,
      "step": 34010
    },
    {
      "epoch": 0.09567440330051381,
      "grad_norm": 29.25,
      "learning_rate": 1.994985473449369e-05,
      "loss": 7.9879,
      "step": 34020
    },
    {
      "epoch": 0.09570252628796251,
      "grad_norm": 31.5,
      "learning_rate": 1.9949825274101768e-05,
      "loss": 8.336,
      "step": 34030
    },
    {
      "epoch": 0.09573064927541122,
      "grad_norm": 31.25,
      "learning_rate": 1.9949795805080144e-05,
      "loss": 8.4426,
      "step": 34040
    },
    {
      "epoch": 0.09575877226285993,
      "grad_norm": 24.75,
      "learning_rate": 1.9949766327428846e-05,
      "loss": 7.9934,
      "step": 34050
    },
    {
      "epoch": 0.09578689525030865,
      "grad_norm": 28.625,
      "learning_rate": 1.9949736841147906e-05,
      "loss": 7.1479,
      "step": 34060
    },
    {
      "epoch": 0.09581501823775736,
      "grad_norm": 23.5,
      "learning_rate": 1.9949707346237344e-05,
      "loss": 8.1959,
      "step": 34070
    },
    {
      "epoch": 0.09584314122520607,
      "grad_norm": 41.0,
      "learning_rate": 1.9949677842697186e-05,
      "loss": 8.5101,
      "step": 34080
    },
    {
      "epoch": 0.09587126421265478,
      "grad_norm": 32.75,
      "learning_rate": 1.9949648330527456e-05,
      "loss": 7.6801,
      "step": 34090
    },
    {
      "epoch": 0.09589938720010349,
      "grad_norm": 21.875,
      "learning_rate": 1.9949618809728182e-05,
      "loss": 8.6535,
      "step": 34100
    },
    {
      "epoch": 0.0959275101875522,
      "grad_norm": 27.0,
      "learning_rate": 1.9949589280299387e-05,
      "loss": 8.2399,
      "step": 34110
    },
    {
      "epoch": 0.09595563317500092,
      "grad_norm": 24.125,
      "learning_rate": 1.9949559742241105e-05,
      "loss": 7.6858,
      "step": 34120
    },
    {
      "epoch": 0.09598375616244963,
      "grad_norm": 29.0,
      "learning_rate": 1.994953019555335e-05,
      "loss": 8.0112,
      "step": 34130
    },
    {
      "epoch": 0.09601187914989834,
      "grad_norm": 31.5,
      "learning_rate": 1.9949500640236154e-05,
      "loss": 8.6509,
      "step": 34140
    },
    {
      "epoch": 0.09604000213734705,
      "grad_norm": 28.625,
      "learning_rate": 1.994947107628954e-05,
      "loss": 8.1553,
      "step": 34150
    },
    {
      "epoch": 0.09606812512479576,
      "grad_norm": 24.375,
      "learning_rate": 1.994944150371354e-05,
      "loss": 7.2941,
      "step": 34160
    },
    {
      "epoch": 0.09609624811224447,
      "grad_norm": 30.875,
      "learning_rate": 1.9949411922508173e-05,
      "loss": 7.7683,
      "step": 34170
    },
    {
      "epoch": 0.09612437109969318,
      "grad_norm": 30.625,
      "learning_rate": 1.9949382332673464e-05,
      "loss": 8.3982,
      "step": 34180
    },
    {
      "epoch": 0.09615249408714188,
      "grad_norm": 46.5,
      "learning_rate": 1.9949352734209444e-05,
      "loss": 7.7355,
      "step": 34190
    },
    {
      "epoch": 0.0961806170745906,
      "grad_norm": 23.125,
      "learning_rate": 1.9949323127116137e-05,
      "loss": 8.6584,
      "step": 34200
    },
    {
      "epoch": 0.0962087400620393,
      "grad_norm": 29.25,
      "learning_rate": 1.9949293511393563e-05,
      "loss": 7.6814,
      "step": 34210
    },
    {
      "epoch": 0.09623686304948802,
      "grad_norm": 23.125,
      "learning_rate": 1.9949263887041753e-05,
      "loss": 7.7171,
      "step": 34220
    },
    {
      "epoch": 0.09626498603693673,
      "grad_norm": 20.5,
      "learning_rate": 1.9949234254060738e-05,
      "loss": 7.921,
      "step": 34230
    },
    {
      "epoch": 0.09629310902438544,
      "grad_norm": 22.5,
      "learning_rate": 1.994920461245053e-05,
      "loss": 8.018,
      "step": 34240
    },
    {
      "epoch": 0.09632123201183415,
      "grad_norm": 47.25,
      "learning_rate": 1.9949174962211167e-05,
      "loss": 8.8086,
      "step": 34250
    },
    {
      "epoch": 0.09634935499928286,
      "grad_norm": 25.0,
      "learning_rate": 1.9949145303342666e-05,
      "loss": 8.3242,
      "step": 34260
    },
    {
      "epoch": 0.09637747798673157,
      "grad_norm": 26.0,
      "learning_rate": 1.994911563584506e-05,
      "loss": 8.0383,
      "step": 34270
    },
    {
      "epoch": 0.09640560097418029,
      "grad_norm": 29.75,
      "learning_rate": 1.994908595971837e-05,
      "loss": 7.8968,
      "step": 34280
    },
    {
      "epoch": 0.096433723961629,
      "grad_norm": 39.0,
      "learning_rate": 1.9949056274962627e-05,
      "loss": 7.9171,
      "step": 34290
    },
    {
      "epoch": 0.09646184694907771,
      "grad_norm": 34.25,
      "learning_rate": 1.994902658157785e-05,
      "loss": 8.4294,
      "step": 34300
    },
    {
      "epoch": 0.09648996993652642,
      "grad_norm": 17.0,
      "learning_rate": 1.9948996879564068e-05,
      "loss": 9.1109,
      "step": 34310
    },
    {
      "epoch": 0.09651809292397513,
      "grad_norm": 28.375,
      "learning_rate": 1.9948967168921305e-05,
      "loss": 7.9873,
      "step": 34320
    },
    {
      "epoch": 0.09654621591142384,
      "grad_norm": 23.75,
      "learning_rate": 1.994893744964959e-05,
      "loss": 7.8537,
      "step": 34330
    },
    {
      "epoch": 0.09657433889887256,
      "grad_norm": 24.5,
      "learning_rate": 1.9948907721748948e-05,
      "loss": 8.1767,
      "step": 34340
    },
    {
      "epoch": 0.09660246188632127,
      "grad_norm": 35.25,
      "learning_rate": 1.99488779852194e-05,
      "loss": 7.831,
      "step": 34350
    },
    {
      "epoch": 0.09663058487376996,
      "grad_norm": 37.0,
      "learning_rate": 1.994884824006098e-05,
      "loss": 7.8766,
      "step": 34360
    },
    {
      "epoch": 0.09665870786121868,
      "grad_norm": 30.0,
      "learning_rate": 1.9948818486273708e-05,
      "loss": 8.0223,
      "step": 34370
    },
    {
      "epoch": 0.09668683084866739,
      "grad_norm": 30.5,
      "learning_rate": 1.994878872385761e-05,
      "loss": 7.6153,
      "step": 34380
    },
    {
      "epoch": 0.0967149538361161,
      "grad_norm": 30.25,
      "learning_rate": 1.9948758952812715e-05,
      "loss": 8.1448,
      "step": 34390
    },
    {
      "epoch": 0.09674307682356481,
      "grad_norm": 30.625,
      "learning_rate": 1.9948729173139044e-05,
      "loss": 7.7981,
      "step": 34400
    },
    {
      "epoch": 0.09677119981101352,
      "grad_norm": 23.0,
      "learning_rate": 1.994869938483663e-05,
      "loss": 7.8277,
      "step": 34410
    },
    {
      "epoch": 0.09679932279846223,
      "grad_norm": 36.5,
      "learning_rate": 1.994866958790549e-05,
      "loss": 8.1864,
      "step": 34420
    },
    {
      "epoch": 0.09682744578591095,
      "grad_norm": 30.625,
      "learning_rate": 1.9948639782345655e-05,
      "loss": 7.3568,
      "step": 34430
    },
    {
      "epoch": 0.09685556877335966,
      "grad_norm": 35.0,
      "learning_rate": 1.9948609968157153e-05,
      "loss": 8.401,
      "step": 34440
    },
    {
      "epoch": 0.09688369176080837,
      "grad_norm": 25.75,
      "learning_rate": 1.9948580145340008e-05,
      "loss": 7.9404,
      "step": 34450
    },
    {
      "epoch": 0.09691181474825708,
      "grad_norm": 27.625,
      "learning_rate": 1.994855031389424e-05,
      "loss": 7.6216,
      "step": 34460
    },
    {
      "epoch": 0.09693993773570579,
      "grad_norm": 28.875,
      "learning_rate": 1.9948520473819882e-05,
      "loss": 8.0419,
      "step": 34470
    },
    {
      "epoch": 0.0969680607231545,
      "grad_norm": 26.5,
      "learning_rate": 1.994849062511696e-05,
      "loss": 7.6162,
      "step": 34480
    },
    {
      "epoch": 0.09699618371060321,
      "grad_norm": 36.25,
      "learning_rate": 1.9948460767785494e-05,
      "loss": 7.4739,
      "step": 34490
    },
    {
      "epoch": 0.09702430669805193,
      "grad_norm": 28.0,
      "learning_rate": 1.994843090182552e-05,
      "loss": 8.4236,
      "step": 34500
    },
    {
      "epoch": 0.09705242968550064,
      "grad_norm": 39.75,
      "learning_rate": 1.994840102723705e-05,
      "loss": 8.691,
      "step": 34510
    },
    {
      "epoch": 0.09708055267294934,
      "grad_norm": 29.75,
      "learning_rate": 1.994837114402012e-05,
      "loss": 7.315,
      "step": 34520
    },
    {
      "epoch": 0.09710867566039805,
      "grad_norm": 38.25,
      "learning_rate": 1.994834125217475e-05,
      "loss": 7.9799,
      "step": 34530
    },
    {
      "epoch": 0.09713679864784676,
      "grad_norm": 35.5,
      "learning_rate": 1.9948311351700975e-05,
      "loss": 7.4159,
      "step": 34540
    },
    {
      "epoch": 0.09716492163529547,
      "grad_norm": 39.0,
      "learning_rate": 1.9948281442598813e-05,
      "loss": 8.1024,
      "step": 34550
    },
    {
      "epoch": 0.09719304462274418,
      "grad_norm": 39.0,
      "learning_rate": 1.9948251524868292e-05,
      "loss": 7.6426,
      "step": 34560
    },
    {
      "epoch": 0.0972211676101929,
      "grad_norm": 24.5,
      "learning_rate": 1.9948221598509436e-05,
      "loss": 7.7878,
      "step": 34570
    },
    {
      "epoch": 0.0972492905976416,
      "grad_norm": 34.25,
      "learning_rate": 1.9948191663522275e-05,
      "loss": 7.7768,
      "step": 34580
    },
    {
      "epoch": 0.09727741358509032,
      "grad_norm": 28.75,
      "learning_rate": 1.994816171990683e-05,
      "loss": 7.858,
      "step": 34590
    },
    {
      "epoch": 0.09730553657253903,
      "grad_norm": 27.375,
      "learning_rate": 1.9948131767663136e-05,
      "loss": 8.1617,
      "step": 34600
    },
    {
      "epoch": 0.09733365955998774,
      "grad_norm": 27.75,
      "learning_rate": 1.9948101806791207e-05,
      "loss": 7.07,
      "step": 34610
    },
    {
      "epoch": 0.09736178254743645,
      "grad_norm": 45.25,
      "learning_rate": 1.9948071837291076e-05,
      "loss": 8.3011,
      "step": 34620
    },
    {
      "epoch": 0.09738990553488516,
      "grad_norm": 28.5,
      "learning_rate": 1.994804185916277e-05,
      "loss": 7.3544,
      "step": 34630
    },
    {
      "epoch": 0.09741802852233387,
      "grad_norm": 26.25,
      "learning_rate": 1.994801187240631e-05,
      "loss": 8.1194,
      "step": 34640
    },
    {
      "epoch": 0.09744615150978259,
      "grad_norm": 27.25,
      "learning_rate": 1.9947981877021726e-05,
      "loss": 8.4025,
      "step": 34650
    },
    {
      "epoch": 0.0974742744972313,
      "grad_norm": 27.625,
      "learning_rate": 1.9947951873009044e-05,
      "loss": 8.092,
      "step": 34660
    },
    {
      "epoch": 0.09750239748468001,
      "grad_norm": 23.125,
      "learning_rate": 1.994792186036829e-05,
      "loss": 8.1229,
      "step": 34670
    },
    {
      "epoch": 0.0975305204721287,
      "grad_norm": 36.25,
      "learning_rate": 1.9947891839099485e-05,
      "loss": 8.4437,
      "step": 34680
    },
    {
      "epoch": 0.09755864345957742,
      "grad_norm": 76.0,
      "learning_rate": 1.994786180920266e-05,
      "loss": 8.7042,
      "step": 34690
    },
    {
      "epoch": 0.09758676644702613,
      "grad_norm": 28.625,
      "learning_rate": 1.994783177067784e-05,
      "loss": 8.337,
      "step": 34700
    },
    {
      "epoch": 0.09761488943447484,
      "grad_norm": 27.0,
      "learning_rate": 1.9947801723525052e-05,
      "loss": 7.2786,
      "step": 34710
    },
    {
      "epoch": 0.09764301242192355,
      "grad_norm": 26.25,
      "learning_rate": 1.9947771667744322e-05,
      "loss": 8.52,
      "step": 34720
    },
    {
      "epoch": 0.09767113540937226,
      "grad_norm": 22.875,
      "learning_rate": 1.9947741603335674e-05,
      "loss": 7.8198,
      "step": 34730
    },
    {
      "epoch": 0.09769925839682098,
      "grad_norm": 32.75,
      "learning_rate": 1.9947711530299134e-05,
      "loss": 7.4021,
      "step": 34740
    },
    {
      "epoch": 0.09772738138426969,
      "grad_norm": 55.25,
      "learning_rate": 1.994768144863473e-05,
      "loss": 7.9327,
      "step": 34750
    },
    {
      "epoch": 0.0977555043717184,
      "grad_norm": 56.5,
      "learning_rate": 1.994765135834249e-05,
      "loss": 8.3904,
      "step": 34760
    },
    {
      "epoch": 0.09778362735916711,
      "grad_norm": 32.25,
      "learning_rate": 1.9947621259422437e-05,
      "loss": 8.3421,
      "step": 34770
    },
    {
      "epoch": 0.09781175034661582,
      "grad_norm": 33.75,
      "learning_rate": 1.9947591151874594e-05,
      "loss": 8.0124,
      "step": 34780
    },
    {
      "epoch": 0.09783987333406453,
      "grad_norm": 35.0,
      "learning_rate": 1.9947561035698995e-05,
      "loss": 8.6144,
      "step": 34790
    },
    {
      "epoch": 0.09786799632151325,
      "grad_norm": 30.0,
      "learning_rate": 1.994753091089566e-05,
      "loss": 7.5272,
      "step": 34800
    },
    {
      "epoch": 0.09789611930896196,
      "grad_norm": 26.75,
      "learning_rate": 1.9947500777464615e-05,
      "loss": 7.8444,
      "step": 34810
    },
    {
      "epoch": 0.09792424229641067,
      "grad_norm": 29.5,
      "learning_rate": 1.9947470635405892e-05,
      "loss": 7.2193,
      "step": 34820
    },
    {
      "epoch": 0.09795236528385938,
      "grad_norm": 28.875,
      "learning_rate": 1.9947440484719508e-05,
      "loss": 8.3753,
      "step": 34830
    },
    {
      "epoch": 0.09798048827130808,
      "grad_norm": 43.75,
      "learning_rate": 1.99474103254055e-05,
      "loss": 7.7153,
      "step": 34840
    },
    {
      "epoch": 0.09800861125875679,
      "grad_norm": 38.0,
      "learning_rate": 1.9947380157463886e-05,
      "loss": 8.1989,
      "step": 34850
    },
    {
      "epoch": 0.0980367342462055,
      "grad_norm": 28.25,
      "learning_rate": 1.9947349980894696e-05,
      "loss": 8.2828,
      "step": 34860
    },
    {
      "epoch": 0.09806485723365421,
      "grad_norm": 27.5,
      "learning_rate": 1.9947319795697953e-05,
      "loss": 7.9424,
      "step": 34870
    },
    {
      "epoch": 0.09809298022110292,
      "grad_norm": 27.0,
      "learning_rate": 1.9947289601873685e-05,
      "loss": 7.9443,
      "step": 34880
    },
    {
      "epoch": 0.09812110320855164,
      "grad_norm": 34.25,
      "learning_rate": 1.994725939942192e-05,
      "loss": 7.2726,
      "step": 34890
    },
    {
      "epoch": 0.09814922619600035,
      "grad_norm": 28.25,
      "learning_rate": 1.994722918834268e-05,
      "loss": 8.2475,
      "step": 34900
    },
    {
      "epoch": 0.09817734918344906,
      "grad_norm": 26.875,
      "learning_rate": 1.9947198968635995e-05,
      "loss": 8.0215,
      "step": 34910
    },
    {
      "epoch": 0.09820547217089777,
      "grad_norm": 22.5,
      "learning_rate": 1.9947168740301893e-05,
      "loss": 8.3231,
      "step": 34920
    },
    {
      "epoch": 0.09823359515834648,
      "grad_norm": 36.25,
      "learning_rate": 1.9947138503340392e-05,
      "loss": 7.8721,
      "step": 34930
    },
    {
      "epoch": 0.0982617181457952,
      "grad_norm": 40.0,
      "learning_rate": 1.9947108257751526e-05,
      "loss": 7.9962,
      "step": 34940
    },
    {
      "epoch": 0.0982898411332439,
      "grad_norm": 52.5,
      "learning_rate": 1.9947078003535316e-05,
      "loss": 7.9704,
      "step": 34950
    },
    {
      "epoch": 0.09831796412069262,
      "grad_norm": 21.0,
      "learning_rate": 1.9947047740691795e-05,
      "loss": 7.8284,
      "step": 34960
    },
    {
      "epoch": 0.09834608710814133,
      "grad_norm": 35.25,
      "learning_rate": 1.9947017469220983e-05,
      "loss": 7.5414,
      "step": 34970
    },
    {
      "epoch": 0.09837421009559004,
      "grad_norm": 23.875,
      "learning_rate": 1.994698718912291e-05,
      "loss": 7.8494,
      "step": 34980
    },
    {
      "epoch": 0.09840233308303875,
      "grad_norm": 33.0,
      "learning_rate": 1.99469569003976e-05,
      "loss": 8.441,
      "step": 34990
    },
    {
      "epoch": 0.09843045607048746,
      "grad_norm": 21.0,
      "learning_rate": 1.9946926603045074e-05,
      "loss": 6.9415,
      "step": 35000
    },
    {
      "epoch": 0.09845857905793616,
      "grad_norm": 23.0,
      "learning_rate": 1.994689629706537e-05,
      "loss": 7.4531,
      "step": 35010
    },
    {
      "epoch": 0.09848670204538487,
      "grad_norm": 27.375,
      "learning_rate": 1.9946865982458507e-05,
      "loss": 7.7361,
      "step": 35020
    },
    {
      "epoch": 0.09851482503283358,
      "grad_norm": 25.75,
      "learning_rate": 1.9946835659224512e-05,
      "loss": 8.5667,
      "step": 35030
    },
    {
      "epoch": 0.0985429480202823,
      "grad_norm": 38.0,
      "learning_rate": 1.994680532736341e-05,
      "loss": 7.8589,
      "step": 35040
    },
    {
      "epoch": 0.098571071007731,
      "grad_norm": 32.25,
      "learning_rate": 1.9946774986875236e-05,
      "loss": 7.6877,
      "step": 35050
    },
    {
      "epoch": 0.09859919399517972,
      "grad_norm": 24.625,
      "learning_rate": 1.9946744637760004e-05,
      "loss": 7.9455,
      "step": 35060
    },
    {
      "epoch": 0.09862731698262843,
      "grad_norm": 29.75,
      "learning_rate": 1.994671428001775e-05,
      "loss": 8.1593,
      "step": 35070
    },
    {
      "epoch": 0.09865543997007714,
      "grad_norm": 29.5,
      "learning_rate": 1.9946683913648494e-05,
      "loss": 7.8354,
      "step": 35080
    },
    {
      "epoch": 0.09868356295752585,
      "grad_norm": 28.75,
      "learning_rate": 1.9946653538652264e-05,
      "loss": 7.6561,
      "step": 35090
    },
    {
      "epoch": 0.09871168594497456,
      "grad_norm": 36.0,
      "learning_rate": 1.994662315502909e-05,
      "loss": 7.7363,
      "step": 35100
    },
    {
      "epoch": 0.09873980893242328,
      "grad_norm": 25.5,
      "learning_rate": 1.9946592762778988e-05,
      "loss": 7.9574,
      "step": 35110
    },
    {
      "epoch": 0.09876793191987199,
      "grad_norm": 23.625,
      "learning_rate": 1.9946562361902e-05,
      "loss": 8.0755,
      "step": 35120
    },
    {
      "epoch": 0.0987960549073207,
      "grad_norm": 24.5,
      "learning_rate": 1.9946531952398138e-05,
      "loss": 7.5878,
      "step": 35130
    },
    {
      "epoch": 0.09882417789476941,
      "grad_norm": 27.875,
      "learning_rate": 1.9946501534267436e-05,
      "loss": 7.3152,
      "step": 35140
    },
    {
      "epoch": 0.09885230088221812,
      "grad_norm": 30.0,
      "learning_rate": 1.994647110750992e-05,
      "loss": 7.9381,
      "step": 35150
    },
    {
      "epoch": 0.09888042386966683,
      "grad_norm": 30.125,
      "learning_rate": 1.9946440672125616e-05,
      "loss": 7.6638,
      "step": 35160
    },
    {
      "epoch": 0.09890854685711553,
      "grad_norm": 22.5,
      "learning_rate": 1.9946410228114548e-05,
      "loss": 7.8539,
      "step": 35170
    },
    {
      "epoch": 0.09893666984456424,
      "grad_norm": 31.125,
      "learning_rate": 1.9946379775476744e-05,
      "loss": 7.1999,
      "step": 35180
    },
    {
      "epoch": 0.09896479283201295,
      "grad_norm": 20.75,
      "learning_rate": 1.994634931421223e-05,
      "loss": 7.997,
      "step": 35190
    },
    {
      "epoch": 0.09899291581946167,
      "grad_norm": 43.5,
      "learning_rate": 1.9946318844321034e-05,
      "loss": 8.3306,
      "step": 35200
    },
    {
      "epoch": 0.09902103880691038,
      "grad_norm": 31.875,
      "learning_rate": 1.9946288365803182e-05,
      "loss": 7.5004,
      "step": 35210
    },
    {
      "epoch": 0.09904916179435909,
      "grad_norm": 22.25,
      "learning_rate": 1.9946257878658696e-05,
      "loss": 8.2654,
      "step": 35220
    },
    {
      "epoch": 0.0990772847818078,
      "grad_norm": 38.75,
      "learning_rate": 1.994622738288761e-05,
      "loss": 8.0553,
      "step": 35230
    },
    {
      "epoch": 0.09910540776925651,
      "grad_norm": 31.625,
      "learning_rate": 1.9946196878489946e-05,
      "loss": 7.9849,
      "step": 35240
    },
    {
      "epoch": 0.09913353075670522,
      "grad_norm": 23.25,
      "learning_rate": 1.9946166365465727e-05,
      "loss": 8.2555,
      "step": 35250
    },
    {
      "epoch": 0.09916165374415394,
      "grad_norm": 30.75,
      "learning_rate": 1.9946135843814985e-05,
      "loss": 8.1368,
      "step": 35260
    },
    {
      "epoch": 0.09918977673160265,
      "grad_norm": 26.0,
      "learning_rate": 1.994610531353775e-05,
      "loss": 8.1076,
      "step": 35270
    },
    {
      "epoch": 0.09921789971905136,
      "grad_norm": 28.875,
      "learning_rate": 1.994607477463404e-05,
      "loss": 7.6869,
      "step": 35280
    },
    {
      "epoch": 0.09924602270650007,
      "grad_norm": 31.75,
      "learning_rate": 1.9946044227103884e-05,
      "loss": 7.9467,
      "step": 35290
    },
    {
      "epoch": 0.09927414569394878,
      "grad_norm": 44.75,
      "learning_rate": 1.994601367094731e-05,
      "loss": 8.0237,
      "step": 35300
    },
    {
      "epoch": 0.0993022686813975,
      "grad_norm": 41.0,
      "learning_rate": 1.9945983106164343e-05,
      "loss": 8.5538,
      "step": 35310
    },
    {
      "epoch": 0.0993303916688462,
      "grad_norm": 33.0,
      "learning_rate": 1.9945952532755013e-05,
      "loss": 7.9339,
      "step": 35320
    },
    {
      "epoch": 0.0993585146562949,
      "grad_norm": 45.0,
      "learning_rate": 1.9945921950719344e-05,
      "loss": 7.7516,
      "step": 35330
    },
    {
      "epoch": 0.09938663764374361,
      "grad_norm": 24.125,
      "learning_rate": 1.994589136005736e-05,
      "loss": 7.5385,
      "step": 35340
    },
    {
      "epoch": 0.09941476063119233,
      "grad_norm": 21.0,
      "learning_rate": 1.9945860760769088e-05,
      "loss": 7.6339,
      "step": 35350
    },
    {
      "epoch": 0.09944288361864104,
      "grad_norm": 26.875,
      "learning_rate": 1.9945830152854563e-05,
      "loss": 8.0207,
      "step": 35360
    },
    {
      "epoch": 0.09947100660608975,
      "grad_norm": 28.875,
      "learning_rate": 1.99457995363138e-05,
      "loss": 7.8396,
      "step": 35370
    },
    {
      "epoch": 0.09949912959353846,
      "grad_norm": 34.25,
      "learning_rate": 1.9945768911146833e-05,
      "loss": 8.5064,
      "step": 35380
    },
    {
      "epoch": 0.09952725258098717,
      "grad_norm": 23.5,
      "learning_rate": 1.9945738277353683e-05,
      "loss": 8.0252,
      "step": 35390
    },
    {
      "epoch": 0.09955537556843588,
      "grad_norm": 25.5,
      "learning_rate": 1.9945707634934383e-05,
      "loss": 8.0851,
      "step": 35400
    },
    {
      "epoch": 0.0995834985558846,
      "grad_norm": 21.125,
      "learning_rate": 1.9945676983888957e-05,
      "loss": 8.1571,
      "step": 35410
    },
    {
      "epoch": 0.0996116215433333,
      "grad_norm": 23.5,
      "learning_rate": 1.994564632421743e-05,
      "loss": 7.6056,
      "step": 35420
    },
    {
      "epoch": 0.09963974453078202,
      "grad_norm": 29.5,
      "learning_rate": 1.994561565591983e-05,
      "loss": 8.536,
      "step": 35430
    },
    {
      "epoch": 0.09966786751823073,
      "grad_norm": 37.0,
      "learning_rate": 1.9945584978996178e-05,
      "loss": 8.1186,
      "step": 35440
    },
    {
      "epoch": 0.09969599050567944,
      "grad_norm": 25.0,
      "learning_rate": 1.9945554293446513e-05,
      "loss": 8.1453,
      "step": 35450
    },
    {
      "epoch": 0.09972411349312815,
      "grad_norm": 22.125,
      "learning_rate": 1.9945523599270847e-05,
      "loss": 7.6347,
      "step": 35460
    },
    {
      "epoch": 0.09975223648057686,
      "grad_norm": 28.5,
      "learning_rate": 1.994549289646922e-05,
      "loss": 7.6964,
      "step": 35470
    },
    {
      "epoch": 0.09978035946802558,
      "grad_norm": 32.0,
      "learning_rate": 1.9945462185041652e-05,
      "loss": 8.6327,
      "step": 35480
    },
    {
      "epoch": 0.09980848245547427,
      "grad_norm": 23.0,
      "learning_rate": 1.994543146498817e-05,
      "loss": 8.5013,
      "step": 35490
    },
    {
      "epoch": 0.09983660544292298,
      "grad_norm": 23.125,
      "learning_rate": 1.9945400736308797e-05,
      "loss": 7.8948,
      "step": 35500
    },
    {
      "epoch": 0.0998647284303717,
      "grad_norm": 22.875,
      "learning_rate": 1.994536999900357e-05,
      "loss": 8.0567,
      "step": 35510
    },
    {
      "epoch": 0.09989285141782041,
      "grad_norm": 30.875,
      "learning_rate": 1.9945339253072503e-05,
      "loss": 7.4077,
      "step": 35520
    },
    {
      "epoch": 0.09992097440526912,
      "grad_norm": 62.25,
      "learning_rate": 1.9945308498515633e-05,
      "loss": 7.7931,
      "step": 35530
    },
    {
      "epoch": 0.09994909739271783,
      "grad_norm": 38.5,
      "learning_rate": 1.994527773533298e-05,
      "loss": 7.4245,
      "step": 35540
    },
    {
      "epoch": 0.09997722038016654,
      "grad_norm": 75.0,
      "learning_rate": 1.9945246963524575e-05,
      "loss": 8.6513,
      "step": 35550
    },
    {
      "epoch": 0.10000534336761525,
      "grad_norm": 20.25,
      "learning_rate": 1.9945216183090438e-05,
      "loss": 8.1036,
      "step": 35560
    },
    {
      "epoch": 0.10003346635506397,
      "grad_norm": 42.5,
      "learning_rate": 1.994518539403061e-05,
      "loss": 8.4752,
      "step": 35570
    },
    {
      "epoch": 0.10006158934251268,
      "grad_norm": 21.75,
      "learning_rate": 1.99451545963451e-05,
      "loss": 8.1236,
      "step": 35580
    },
    {
      "epoch": 0.10008971232996139,
      "grad_norm": 29.125,
      "learning_rate": 1.9945123790033945e-05,
      "loss": 8.6495,
      "step": 35590
    },
    {
      "epoch": 0.1001178353174101,
      "grad_norm": 21.5,
      "learning_rate": 1.994509297509717e-05,
      "loss": 7.7275,
      "step": 35600
    },
    {
      "epoch": 0.10014595830485881,
      "grad_norm": 25.875,
      "learning_rate": 1.9945062151534803e-05,
      "loss": 7.8538,
      "step": 35610
    },
    {
      "epoch": 0.10017408129230752,
      "grad_norm": 21.375,
      "learning_rate": 1.994503131934687e-05,
      "loss": 8.1354,
      "step": 35620
    },
    {
      "epoch": 0.10020220427975624,
      "grad_norm": 39.25,
      "learning_rate": 1.9945000478533394e-05,
      "loss": 7.7368,
      "step": 35630
    },
    {
      "epoch": 0.10023032726720495,
      "grad_norm": 43.25,
      "learning_rate": 1.9944969629094407e-05,
      "loss": 8.6796,
      "step": 35640
    },
    {
      "epoch": 0.10025845025465366,
      "grad_norm": 60.0,
      "learning_rate": 1.9944938771029928e-05,
      "loss": 7.421,
      "step": 35650
    },
    {
      "epoch": 0.10028657324210236,
      "grad_norm": 39.5,
      "learning_rate": 1.9944907904339997e-05,
      "loss": 7.9019,
      "step": 35660
    },
    {
      "epoch": 0.10031469622955107,
      "grad_norm": 23.75,
      "learning_rate": 1.994487702902463e-05,
      "loss": 7.9089,
      "step": 35670
    },
    {
      "epoch": 0.10034281921699978,
      "grad_norm": 27.125,
      "learning_rate": 1.9944846145083857e-05,
      "loss": 7.4266,
      "step": 35680
    },
    {
      "epoch": 0.10037094220444849,
      "grad_norm": 25.875,
      "learning_rate": 1.9944815252517703e-05,
      "loss": 8.6505,
      "step": 35690
    },
    {
      "epoch": 0.1003990651918972,
      "grad_norm": 34.25,
      "learning_rate": 1.9944784351326195e-05,
      "loss": 7.5449,
      "step": 35700
    },
    {
      "epoch": 0.10042718817934591,
      "grad_norm": 31.875,
      "learning_rate": 1.9944753441509365e-05,
      "loss": 7.5257,
      "step": 35710
    },
    {
      "epoch": 0.10045531116679463,
      "grad_norm": 39.25,
      "learning_rate": 1.9944722523067232e-05,
      "loss": 8.2313,
      "step": 35720
    },
    {
      "epoch": 0.10048343415424334,
      "grad_norm": 30.875,
      "learning_rate": 1.994469159599983e-05,
      "loss": 7.1398,
      "step": 35730
    },
    {
      "epoch": 0.10051155714169205,
      "grad_norm": 33.75,
      "learning_rate": 1.994466066030718e-05,
      "loss": 8.1076,
      "step": 35740
    },
    {
      "epoch": 0.10053968012914076,
      "grad_norm": 29.75,
      "learning_rate": 1.9944629715989315e-05,
      "loss": 7.6846,
      "step": 35750
    },
    {
      "epoch": 0.10056780311658947,
      "grad_norm": 30.5,
      "learning_rate": 1.9944598763046256e-05,
      "loss": 7.6026,
      "step": 35760
    },
    {
      "epoch": 0.10059592610403818,
      "grad_norm": 36.0,
      "learning_rate": 1.994456780147803e-05,
      "loss": 7.1352,
      "step": 35770
    },
    {
      "epoch": 0.1006240490914869,
      "grad_norm": 33.75,
      "learning_rate": 1.9944536831284667e-05,
      "loss": 7.9691,
      "step": 35780
    },
    {
      "epoch": 0.1006521720789356,
      "grad_norm": 28.875,
      "learning_rate": 1.9944505852466195e-05,
      "loss": 8.5828,
      "step": 35790
    },
    {
      "epoch": 0.10068029506638432,
      "grad_norm": 45.5,
      "learning_rate": 1.9944474865022637e-05,
      "loss": 7.9256,
      "step": 35800
    },
    {
      "epoch": 0.10070841805383303,
      "grad_norm": 28.25,
      "learning_rate": 1.9944443868954024e-05,
      "loss": 7.6761,
      "step": 35810
    },
    {
      "epoch": 0.10073654104128173,
      "grad_norm": 43.5,
      "learning_rate": 1.994441286426038e-05,
      "loss": 8.0824,
      "step": 35820
    },
    {
      "epoch": 0.10076466402873044,
      "grad_norm": 30.5,
      "learning_rate": 1.994438185094173e-05,
      "loss": 8.0222,
      "step": 35830
    },
    {
      "epoch": 0.10079278701617915,
      "grad_norm": 25.625,
      "learning_rate": 1.9944350828998106e-05,
      "loss": 8.3097,
      "step": 35840
    },
    {
      "epoch": 0.10082091000362786,
      "grad_norm": 47.75,
      "learning_rate": 1.994431979842953e-05,
      "loss": 7.1629,
      "step": 35850
    },
    {
      "epoch": 0.10084903299107657,
      "grad_norm": 47.0,
      "learning_rate": 1.994428875923603e-05,
      "loss": 8.1664,
      "step": 35860
    },
    {
      "epoch": 0.10087715597852528,
      "grad_norm": 46.0,
      "learning_rate": 1.9944257711417637e-05,
      "loss": 8.706,
      "step": 35870
    },
    {
      "epoch": 0.100905278965974,
      "grad_norm": 29.25,
      "learning_rate": 1.994422665497437e-05,
      "loss": 7.677,
      "step": 35880
    },
    {
      "epoch": 0.10093340195342271,
      "grad_norm": 46.25,
      "learning_rate": 1.9944195589906265e-05,
      "loss": 8.384,
      "step": 35890
    },
    {
      "epoch": 0.10096152494087142,
      "grad_norm": 31.125,
      "learning_rate": 1.9944164516213343e-05,
      "loss": 8.1885,
      "step": 35900
    },
    {
      "epoch": 0.10098964792832013,
      "grad_norm": 28.25,
      "learning_rate": 1.9944133433895634e-05,
      "loss": 8.0183,
      "step": 35910
    },
    {
      "epoch": 0.10101777091576884,
      "grad_norm": 28.5,
      "learning_rate": 1.9944102342953164e-05,
      "loss": 8.1459,
      "step": 35920
    },
    {
      "epoch": 0.10104589390321755,
      "grad_norm": 54.5,
      "learning_rate": 1.9944071243385958e-05,
      "loss": 8.0077,
      "step": 35930
    },
    {
      "epoch": 0.10107401689066627,
      "grad_norm": 30.5,
      "learning_rate": 1.9944040135194048e-05,
      "loss": 8.3934,
      "step": 35940
    },
    {
      "epoch": 0.10110213987811498,
      "grad_norm": 29.625,
      "learning_rate": 1.9944009018377453e-05,
      "loss": 8.2555,
      "step": 35950
    },
    {
      "epoch": 0.10113026286556369,
      "grad_norm": 25.5,
      "learning_rate": 1.9943977892936205e-05,
      "loss": 7.8803,
      "step": 35960
    },
    {
      "epoch": 0.1011583858530124,
      "grad_norm": 40.25,
      "learning_rate": 1.9943946758870332e-05,
      "loss": 8.1515,
      "step": 35970
    },
    {
      "epoch": 0.1011865088404611,
      "grad_norm": 28.375,
      "learning_rate": 1.9943915616179862e-05,
      "loss": 7.8195,
      "step": 35980
    },
    {
      "epoch": 0.10121463182790981,
      "grad_norm": 22.25,
      "learning_rate": 1.9943884464864814e-05,
      "loss": 8.0787,
      "step": 35990
    },
    {
      "epoch": 0.10124275481535852,
      "grad_norm": 30.75,
      "learning_rate": 1.9943853304925226e-05,
      "loss": 7.0991,
      "step": 36000
    },
    {
      "epoch": 0.10127087780280723,
      "grad_norm": 30.75,
      "learning_rate": 1.994382213636112e-05,
      "loss": 7.5796,
      "step": 36010
    },
    {
      "epoch": 0.10129900079025594,
      "grad_norm": 27.625,
      "learning_rate": 1.994379095917252e-05,
      "loss": 6.7867,
      "step": 36020
    },
    {
      "epoch": 0.10132712377770466,
      "grad_norm": 53.75,
      "learning_rate": 1.9943759773359456e-05,
      "loss": 7.9833,
      "step": 36030
    },
    {
      "epoch": 0.10135524676515337,
      "grad_norm": 27.0,
      "learning_rate": 1.9943728578921954e-05,
      "loss": 8.0137,
      "step": 36040
    },
    {
      "epoch": 0.10138336975260208,
      "grad_norm": 55.5,
      "learning_rate": 1.9943697375860044e-05,
      "loss": 8.0634,
      "step": 36050
    },
    {
      "epoch": 0.10141149274005079,
      "grad_norm": 32.25,
      "learning_rate": 1.994366616417375e-05,
      "loss": 8.5468,
      "step": 36060
    },
    {
      "epoch": 0.1014396157274995,
      "grad_norm": 33.0,
      "learning_rate": 1.9943634943863097e-05,
      "loss": 8.3418,
      "step": 36070
    },
    {
      "epoch": 0.10146773871494821,
      "grad_norm": 33.75,
      "learning_rate": 1.9943603714928118e-05,
      "loss": 8.2579,
      "step": 36080
    },
    {
      "epoch": 0.10149586170239693,
      "grad_norm": 25.375,
      "learning_rate": 1.9943572477368836e-05,
      "loss": 8.0526,
      "step": 36090
    },
    {
      "epoch": 0.10152398468984564,
      "grad_norm": 25.375,
      "learning_rate": 1.994354123118528e-05,
      "loss": 8.4065,
      "step": 36100
    },
    {
      "epoch": 0.10155210767729435,
      "grad_norm": 38.75,
      "learning_rate": 1.9943509976377478e-05,
      "loss": 8.1438,
      "step": 36110
    },
    {
      "epoch": 0.10158023066474306,
      "grad_norm": 30.125,
      "learning_rate": 1.9943478712945453e-05,
      "loss": 7.3432,
      "step": 36120
    },
    {
      "epoch": 0.10160835365219177,
      "grad_norm": 49.25,
      "learning_rate": 1.9943447440889233e-05,
      "loss": 8.7826,
      "step": 36130
    },
    {
      "epoch": 0.10163647663964047,
      "grad_norm": 25.125,
      "learning_rate": 1.9943416160208852e-05,
      "loss": 7.9611,
      "step": 36140
    },
    {
      "epoch": 0.10166459962708918,
      "grad_norm": 47.75,
      "learning_rate": 1.9943384870904327e-05,
      "loss": 7.9598,
      "step": 36150
    },
    {
      "epoch": 0.10169272261453789,
      "grad_norm": 35.25,
      "learning_rate": 1.9943353572975692e-05,
      "loss": 8.4308,
      "step": 36160
    },
    {
      "epoch": 0.1017208456019866,
      "grad_norm": 29.375,
      "learning_rate": 1.994332226642297e-05,
      "loss": 8.091,
      "step": 36170
    },
    {
      "epoch": 0.10174896858943532,
      "grad_norm": 33.5,
      "learning_rate": 1.9943290951246194e-05,
      "loss": 8.5679,
      "step": 36180
    },
    {
      "epoch": 0.10177709157688403,
      "grad_norm": 35.0,
      "learning_rate": 1.9943259627445385e-05,
      "loss": 7.8919,
      "step": 36190
    },
    {
      "epoch": 0.10180521456433274,
      "grad_norm": 26.875,
      "learning_rate": 1.9943228295020575e-05,
      "loss": 7.7832,
      "step": 36200
    },
    {
      "epoch": 0.10183333755178145,
      "grad_norm": 43.75,
      "learning_rate": 1.9943196953971783e-05,
      "loss": 7.6631,
      "step": 36210
    },
    {
      "epoch": 0.10186146053923016,
      "grad_norm": 31.625,
      "learning_rate": 1.9943165604299048e-05,
      "loss": 9.0123,
      "step": 36220
    },
    {
      "epoch": 0.10188958352667887,
      "grad_norm": 36.25,
      "learning_rate": 1.9943134246002388e-05,
      "loss": 7.4637,
      "step": 36230
    },
    {
      "epoch": 0.10191770651412758,
      "grad_norm": 29.625,
      "learning_rate": 1.9943102879081832e-05,
      "loss": 7.6421,
      "step": 36240
    },
    {
      "epoch": 0.1019458295015763,
      "grad_norm": 31.0,
      "learning_rate": 1.9943071503537414e-05,
      "loss": 7.6428,
      "step": 36250
    },
    {
      "epoch": 0.10197395248902501,
      "grad_norm": 29.375,
      "learning_rate": 1.994304011936915e-05,
      "loss": 7.9238,
      "step": 36260
    },
    {
      "epoch": 0.10200207547647372,
      "grad_norm": 31.125,
      "learning_rate": 1.9943008726577077e-05,
      "loss": 7.7321,
      "step": 36270
    },
    {
      "epoch": 0.10203019846392243,
      "grad_norm": 27.875,
      "learning_rate": 1.9942977325161213e-05,
      "loss": 8.0739,
      "step": 36280
    },
    {
      "epoch": 0.10205832145137114,
      "grad_norm": 34.0,
      "learning_rate": 1.9942945915121594e-05,
      "loss": 7.6602,
      "step": 36290
    },
    {
      "epoch": 0.10208644443881985,
      "grad_norm": 38.0,
      "learning_rate": 1.994291449645824e-05,
      "loss": 7.6828,
      "step": 36300
    },
    {
      "epoch": 0.10211456742626855,
      "grad_norm": 42.0,
      "learning_rate": 1.994288306917119e-05,
      "loss": 7.5799,
      "step": 36310
    },
    {
      "epoch": 0.10214269041371726,
      "grad_norm": 36.0,
      "learning_rate": 1.9942851633260457e-05,
      "loss": 7.3619,
      "step": 36320
    },
    {
      "epoch": 0.10217081340116597,
      "grad_norm": 28.125,
      "learning_rate": 1.9942820188726076e-05,
      "loss": 7.4803,
      "step": 36330
    },
    {
      "epoch": 0.10219893638861469,
      "grad_norm": 38.0,
      "learning_rate": 1.9942788735568072e-05,
      "loss": 8.5912,
      "step": 36340
    },
    {
      "epoch": 0.1022270593760634,
      "grad_norm": 26.125,
      "learning_rate": 1.9942757273786474e-05,
      "loss": 7.8641,
      "step": 36350
    },
    {
      "epoch": 0.10225518236351211,
      "grad_norm": 33.5,
      "learning_rate": 1.9942725803381307e-05,
      "loss": 7.8217,
      "step": 36360
    },
    {
      "epoch": 0.10228330535096082,
      "grad_norm": 30.375,
      "learning_rate": 1.9942694324352597e-05,
      "loss": 7.5193,
      "step": 36370
    },
    {
      "epoch": 0.10231142833840953,
      "grad_norm": 27.5,
      "learning_rate": 1.994266283670038e-05,
      "loss": 7.9239,
      "step": 36380
    },
    {
      "epoch": 0.10233955132585824,
      "grad_norm": 30.125,
      "learning_rate": 1.9942631340424675e-05,
      "loss": 7.5537,
      "step": 36390
    },
    {
      "epoch": 0.10236767431330696,
      "grad_norm": 36.5,
      "learning_rate": 1.9942599835525512e-05,
      "loss": 7.45,
      "step": 36400
    },
    {
      "epoch": 0.10239579730075567,
      "grad_norm": 35.5,
      "learning_rate": 1.9942568322002916e-05,
      "loss": 7.8622,
      "step": 36410
    },
    {
      "epoch": 0.10242392028820438,
      "grad_norm": 25.0,
      "learning_rate": 1.9942536799856915e-05,
      "loss": 8.0497,
      "step": 36420
    },
    {
      "epoch": 0.10245204327565309,
      "grad_norm": 29.375,
      "learning_rate": 1.994250526908754e-05,
      "loss": 8.5126,
      "step": 36430
    },
    {
      "epoch": 0.1024801662631018,
      "grad_norm": 38.0,
      "learning_rate": 1.9942473729694812e-05,
      "loss": 8.2699,
      "step": 36440
    },
    {
      "epoch": 0.10250828925055051,
      "grad_norm": 25.5,
      "learning_rate": 1.9942442181678768e-05,
      "loss": 7.8617,
      "step": 36450
    },
    {
      "epoch": 0.10253641223799922,
      "grad_norm": 29.25,
      "learning_rate": 1.9942410625039427e-05,
      "loss": 8.0625,
      "step": 36460
    },
    {
      "epoch": 0.10256453522544792,
      "grad_norm": 40.75,
      "learning_rate": 1.9942379059776816e-05,
      "loss": 7.8067,
      "step": 36470
    },
    {
      "epoch": 0.10259265821289663,
      "grad_norm": 34.75,
      "learning_rate": 1.9942347485890966e-05,
      "loss": 7.7717,
      "step": 36480
    },
    {
      "epoch": 0.10262078120034535,
      "grad_norm": 29.125,
      "learning_rate": 1.9942315903381907e-05,
      "loss": 7.9271,
      "step": 36490
    },
    {
      "epoch": 0.10264890418779406,
      "grad_norm": 36.25,
      "learning_rate": 1.9942284312249663e-05,
      "loss": 8.7874,
      "step": 36500
    },
    {
      "epoch": 0.10267702717524277,
      "grad_norm": 28.75,
      "learning_rate": 1.9942252712494258e-05,
      "loss": 7.8891,
      "step": 36510
    },
    {
      "epoch": 0.10270515016269148,
      "grad_norm": 27.0,
      "learning_rate": 1.9942221104115726e-05,
      "loss": 8.2818,
      "step": 36520
    },
    {
      "epoch": 0.10273327315014019,
      "grad_norm": 37.0,
      "learning_rate": 1.9942189487114093e-05,
      "loss": 7.7552,
      "step": 36530
    },
    {
      "epoch": 0.1027613961375889,
      "grad_norm": 19.875,
      "learning_rate": 1.994215786148938e-05,
      "loss": 9.1373,
      "step": 36540
    },
    {
      "epoch": 0.10278951912503762,
      "grad_norm": 24.375,
      "learning_rate": 1.994212622724162e-05,
      "loss": 7.2317,
      "step": 36550
    },
    {
      "epoch": 0.10281764211248633,
      "grad_norm": 37.0,
      "learning_rate": 1.9942094584370843e-05,
      "loss": 8.384,
      "step": 36560
    },
    {
      "epoch": 0.10284576509993504,
      "grad_norm": 33.75,
      "learning_rate": 1.994206293287707e-05,
      "loss": 7.95,
      "step": 36570
    },
    {
      "epoch": 0.10287388808738375,
      "grad_norm": 48.75,
      "learning_rate": 1.994203127276033e-05,
      "loss": 7.9459,
      "step": 36580
    },
    {
      "epoch": 0.10290201107483246,
      "grad_norm": 52.0,
      "learning_rate": 1.9941999604020655e-05,
      "loss": 7.3141,
      "step": 36590
    },
    {
      "epoch": 0.10293013406228117,
      "grad_norm": 29.25,
      "learning_rate": 1.994196792665807e-05,
      "loss": 8.5512,
      "step": 36600
    },
    {
      "epoch": 0.10295825704972988,
      "grad_norm": 25.25,
      "learning_rate": 1.99419362406726e-05,
      "loss": 8.3649,
      "step": 36610
    },
    {
      "epoch": 0.1029863800371786,
      "grad_norm": 26.625,
      "learning_rate": 1.9941904546064276e-05,
      "loss": 8.5234,
      "step": 36620
    },
    {
      "epoch": 0.1030145030246273,
      "grad_norm": 26.0,
      "learning_rate": 1.994187284283312e-05,
      "loss": 7.7746,
      "step": 36630
    },
    {
      "epoch": 0.103042626012076,
      "grad_norm": 46.75,
      "learning_rate": 1.9941841130979166e-05,
      "loss": 8.1299,
      "step": 36640
    },
    {
      "epoch": 0.10307074899952472,
      "grad_norm": 22.875,
      "learning_rate": 1.9941809410502436e-05,
      "loss": 8.0411,
      "step": 36650
    },
    {
      "epoch": 0.10309887198697343,
      "grad_norm": 27.625,
      "learning_rate": 1.9941777681402966e-05,
      "loss": 7.9705,
      "step": 36660
    },
    {
      "epoch": 0.10312699497442214,
      "grad_norm": 31.125,
      "learning_rate": 1.9941745943680774e-05,
      "loss": 7.3433,
      "step": 36670
    },
    {
      "epoch": 0.10315511796187085,
      "grad_norm": 27.75,
      "learning_rate": 1.9941714197335892e-05,
      "loss": 8.0139,
      "step": 36680
    },
    {
      "epoch": 0.10318324094931956,
      "grad_norm": 33.75,
      "learning_rate": 1.9941682442368347e-05,
      "loss": 7.449,
      "step": 36690
    },
    {
      "epoch": 0.10321136393676827,
      "grad_norm": 30.75,
      "learning_rate": 1.994165067877817e-05,
      "loss": 7.9531,
      "step": 36700
    },
    {
      "epoch": 0.10323948692421699,
      "grad_norm": 34.5,
      "learning_rate": 1.9941618906565382e-05,
      "loss": 7.9519,
      "step": 36710
    },
    {
      "epoch": 0.1032676099116657,
      "grad_norm": 30.375,
      "learning_rate": 1.994158712573001e-05,
      "loss": 7.8351,
      "step": 36720
    },
    {
      "epoch": 0.10329573289911441,
      "grad_norm": 33.5,
      "learning_rate": 1.994155533627209e-05,
      "loss": 8.7328,
      "step": 36730
    },
    {
      "epoch": 0.10332385588656312,
      "grad_norm": 31.25,
      "learning_rate": 1.994152353819164e-05,
      "loss": 7.5858,
      "step": 36740
    },
    {
      "epoch": 0.10335197887401183,
      "grad_norm": 29.125,
      "learning_rate": 1.99414917314887e-05,
      "loss": 7.6278,
      "step": 36750
    },
    {
      "epoch": 0.10338010186146054,
      "grad_norm": 21.0,
      "learning_rate": 1.9941459916163284e-05,
      "loss": 7.8423,
      "step": 36760
    },
    {
      "epoch": 0.10340822484890926,
      "grad_norm": 31.5,
      "learning_rate": 1.994142809221543e-05,
      "loss": 7.5101,
      "step": 36770
    },
    {
      "epoch": 0.10343634783635797,
      "grad_norm": 24.75,
      "learning_rate": 1.9941396259645157e-05,
      "loss": 7.8242,
      "step": 36780
    },
    {
      "epoch": 0.10346447082380666,
      "grad_norm": 22.5,
      "learning_rate": 1.9941364418452497e-05,
      "loss": 8.0069,
      "step": 36790
    },
    {
      "epoch": 0.10349259381125538,
      "grad_norm": 31.0,
      "learning_rate": 1.9941332568637476e-05,
      "loss": 7.8419,
      "step": 36800
    },
    {
      "epoch": 0.10352071679870409,
      "grad_norm": 32.75,
      "learning_rate": 1.9941300710200123e-05,
      "loss": 7.4385,
      "step": 36810
    },
    {
      "epoch": 0.1035488397861528,
      "grad_norm": 22.0,
      "learning_rate": 1.9941268843140473e-05,
      "loss": 7.753,
      "step": 36820
    },
    {
      "epoch": 0.10357696277360151,
      "grad_norm": 25.875,
      "learning_rate": 1.994123696745854e-05,
      "loss": 7.7354,
      "step": 36830
    },
    {
      "epoch": 0.10360508576105022,
      "grad_norm": 43.75,
      "learning_rate": 1.9941205083154358e-05,
      "loss": 7.4854,
      "step": 36840
    },
    {
      "epoch": 0.10363320874849893,
      "grad_norm": 26.375,
      "learning_rate": 1.9941173190227954e-05,
      "loss": 7.4266,
      "step": 36850
    },
    {
      "epoch": 0.10366133173594765,
      "grad_norm": 24.375,
      "learning_rate": 1.994114128867936e-05,
      "loss": 7.648,
      "step": 36860
    },
    {
      "epoch": 0.10368945472339636,
      "grad_norm": 24.875,
      "learning_rate": 1.9941109378508594e-05,
      "loss": 8.2916,
      "step": 36870
    },
    {
      "epoch": 0.10371757771084507,
      "grad_norm": 23.125,
      "learning_rate": 1.9941077459715692e-05,
      "loss": 7.9781,
      "step": 36880
    },
    {
      "epoch": 0.10374570069829378,
      "grad_norm": 27.625,
      "learning_rate": 1.994104553230068e-05,
      "loss": 8.0527,
      "step": 36890
    },
    {
      "epoch": 0.10377382368574249,
      "grad_norm": 43.25,
      "learning_rate": 1.9941013596263585e-05,
      "loss": 8.6009,
      "step": 36900
    },
    {
      "epoch": 0.1038019466731912,
      "grad_norm": 24.375,
      "learning_rate": 1.9940981651604434e-05,
      "loss": 7.3807,
      "step": 36910
    },
    {
      "epoch": 0.10383006966063991,
      "grad_norm": 26.0,
      "learning_rate": 1.9940949698323257e-05,
      "loss": 8.1778,
      "step": 36920
    },
    {
      "epoch": 0.10385819264808863,
      "grad_norm": 54.5,
      "learning_rate": 1.9940917736420074e-05,
      "loss": 8.0029,
      "step": 36930
    },
    {
      "epoch": 0.10388631563553734,
      "grad_norm": 32.25,
      "learning_rate": 1.9940885765894926e-05,
      "loss": 7.7953,
      "step": 36940
    },
    {
      "epoch": 0.10391443862298605,
      "grad_norm": 39.0,
      "learning_rate": 1.9940853786747832e-05,
      "loss": 7.3966,
      "step": 36950
    },
    {
      "epoch": 0.10394256161043475,
      "grad_norm": 63.0,
      "learning_rate": 1.9940821798978818e-05,
      "loss": 7.5035,
      "step": 36960
    },
    {
      "epoch": 0.10397068459788346,
      "grad_norm": 31.25,
      "learning_rate": 1.9940789802587916e-05,
      "loss": 7.3781,
      "step": 36970
    },
    {
      "epoch": 0.10399880758533217,
      "grad_norm": 34.75,
      "learning_rate": 1.9940757797575157e-05,
      "loss": 8.181,
      "step": 36980
    },
    {
      "epoch": 0.10402693057278088,
      "grad_norm": 30.875,
      "learning_rate": 1.994072578394056e-05,
      "loss": 7.4985,
      "step": 36990
    },
    {
      "epoch": 0.1040550535602296,
      "grad_norm": 38.5,
      "learning_rate": 1.9940693761684157e-05,
      "loss": 6.9427,
      "step": 37000
    },
    {
      "epoch": 0.1040831765476783,
      "grad_norm": 34.0,
      "learning_rate": 1.9940661730805977e-05,
      "loss": 7.6136,
      "step": 37010
    },
    {
      "epoch": 0.10411129953512702,
      "grad_norm": 29.375,
      "learning_rate": 1.9940629691306048e-05,
      "loss": 7.9447,
      "step": 37020
    },
    {
      "epoch": 0.10413942252257573,
      "grad_norm": 26.875,
      "learning_rate": 1.9940597643184394e-05,
      "loss": 8.3829,
      "step": 37030
    },
    {
      "epoch": 0.10416754551002444,
      "grad_norm": 32.75,
      "learning_rate": 1.994056558644105e-05,
      "loss": 8.5726,
      "step": 37040
    },
    {
      "epoch": 0.10419566849747315,
      "grad_norm": 32.25,
      "learning_rate": 1.9940533521076035e-05,
      "loss": 8.2909,
      "step": 37050
    },
    {
      "epoch": 0.10422379148492186,
      "grad_norm": 31.625,
      "learning_rate": 1.994050144708938e-05,
      "loss": 7.454,
      "step": 37060
    },
    {
      "epoch": 0.10425191447237057,
      "grad_norm": 30.75,
      "learning_rate": 1.9940469364481115e-05,
      "loss": 8.4485,
      "step": 37070
    },
    {
      "epoch": 0.10428003745981929,
      "grad_norm": 23.875,
      "learning_rate": 1.994043727325127e-05,
      "loss": 7.6704,
      "step": 37080
    },
    {
      "epoch": 0.104308160447268,
      "grad_norm": 30.25,
      "learning_rate": 1.9940405173399865e-05,
      "loss": 7.3739,
      "step": 37090
    },
    {
      "epoch": 0.10433628343471671,
      "grad_norm": 38.5,
      "learning_rate": 1.9940373064926938e-05,
      "loss": 8.6649,
      "step": 37100
    },
    {
      "epoch": 0.10436440642216542,
      "grad_norm": 41.25,
      "learning_rate": 1.9940340947832503e-05,
      "loss": 7.579,
      "step": 37110
    },
    {
      "epoch": 0.10439252940961412,
      "grad_norm": 35.5,
      "learning_rate": 1.9940308822116604e-05,
      "loss": 7.8713,
      "step": 37120
    },
    {
      "epoch": 0.10442065239706283,
      "grad_norm": 28.125,
      "learning_rate": 1.9940276687779258e-05,
      "loss": 7.9107,
      "step": 37130
    },
    {
      "epoch": 0.10444877538451154,
      "grad_norm": 33.5,
      "learning_rate": 1.9940244544820492e-05,
      "loss": 8.5009,
      "step": 37140
    },
    {
      "epoch": 0.10447689837196025,
      "grad_norm": 20.125,
      "learning_rate": 1.994021239324034e-05,
      "loss": 8.3403,
      "step": 37150
    },
    {
      "epoch": 0.10450502135940896,
      "grad_norm": 50.75,
      "learning_rate": 1.994018023303883e-05,
      "loss": 8.3618,
      "step": 37160
    },
    {
      "epoch": 0.10453314434685768,
      "grad_norm": 24.0,
      "learning_rate": 1.9940148064215984e-05,
      "loss": 7.5348,
      "step": 37170
    },
    {
      "epoch": 0.10456126733430639,
      "grad_norm": 56.25,
      "learning_rate": 1.9940115886771833e-05,
      "loss": 8.145,
      "step": 37180
    },
    {
      "epoch": 0.1045893903217551,
      "grad_norm": 29.875,
      "learning_rate": 1.994008370070641e-05,
      "loss": 8.2811,
      "step": 37190
    },
    {
      "epoch": 0.10461751330920381,
      "grad_norm": 33.25,
      "learning_rate": 1.9940051506019734e-05,
      "loss": 7.3916,
      "step": 37200
    },
    {
      "epoch": 0.10464563629665252,
      "grad_norm": 27.625,
      "learning_rate": 1.9940019302711836e-05,
      "loss": 7.7962,
      "step": 37210
    },
    {
      "epoch": 0.10467375928410123,
      "grad_norm": 23.375,
      "learning_rate": 1.9939987090782748e-05,
      "loss": 8.1145,
      "step": 37220
    },
    {
      "epoch": 0.10470188227154995,
      "grad_norm": 41.0,
      "learning_rate": 1.9939954870232495e-05,
      "loss": 8.0341,
      "step": 37230
    },
    {
      "epoch": 0.10473000525899866,
      "grad_norm": 40.25,
      "learning_rate": 1.99399226410611e-05,
      "loss": 8.7699,
      "step": 37240
    },
    {
      "epoch": 0.10475812824644737,
      "grad_norm": 28.0,
      "learning_rate": 1.99398904032686e-05,
      "loss": 7.3443,
      "step": 37250
    },
    {
      "epoch": 0.10478625123389608,
      "grad_norm": 24.125,
      "learning_rate": 1.993985815685502e-05,
      "loss": 7.2997,
      "step": 37260
    },
    {
      "epoch": 0.10481437422134479,
      "grad_norm": 24.625,
      "learning_rate": 1.9939825901820384e-05,
      "loss": 8.0188,
      "step": 37270
    },
    {
      "epoch": 0.10484249720879349,
      "grad_norm": 30.125,
      "learning_rate": 1.9939793638164722e-05,
      "loss": 7.9894,
      "step": 37280
    },
    {
      "epoch": 0.1048706201962422,
      "grad_norm": 38.75,
      "learning_rate": 1.9939761365888067e-05,
      "loss": 7.5391,
      "step": 37290
    },
    {
      "epoch": 0.10489874318369091,
      "grad_norm": 42.25,
      "learning_rate": 1.9939729084990436e-05,
      "loss": 8.465,
      "step": 37300
    },
    {
      "epoch": 0.10492686617113962,
      "grad_norm": 29.125,
      "learning_rate": 1.993969679547187e-05,
      "loss": 7.17,
      "step": 37310
    },
    {
      "epoch": 0.10495498915858834,
      "grad_norm": 28.75,
      "learning_rate": 1.9939664497332386e-05,
      "loss": 8.027,
      "step": 37320
    },
    {
      "epoch": 0.10498311214603705,
      "grad_norm": 25.875,
      "learning_rate": 1.9939632190572024e-05,
      "loss": 7.7895,
      "step": 37330
    },
    {
      "epoch": 0.10501123513348576,
      "grad_norm": 31.875,
      "learning_rate": 1.9939599875190798e-05,
      "loss": 7.9384,
      "step": 37340
    },
    {
      "epoch": 0.10503935812093447,
      "grad_norm": 22.375,
      "learning_rate": 1.9939567551188745e-05,
      "loss": 7.5696,
      "step": 37350
    },
    {
      "epoch": 0.10506748110838318,
      "grad_norm": 25.5,
      "learning_rate": 1.9939535218565888e-05,
      "loss": 8.0689,
      "step": 37360
    },
    {
      "epoch": 0.1050956040958319,
      "grad_norm": 27.125,
      "learning_rate": 1.9939502877322263e-05,
      "loss": 7.6621,
      "step": 37370
    },
    {
      "epoch": 0.1051237270832806,
      "grad_norm": 32.25,
      "learning_rate": 1.9939470527457888e-05,
      "loss": 7.7695,
      "step": 37380
    },
    {
      "epoch": 0.10515185007072932,
      "grad_norm": 43.25,
      "learning_rate": 1.9939438168972795e-05,
      "loss": 7.6483,
      "step": 37390
    },
    {
      "epoch": 0.10517997305817803,
      "grad_norm": 24.0,
      "learning_rate": 1.9939405801867017e-05,
      "loss": 8.0962,
      "step": 37400
    },
    {
      "epoch": 0.10520809604562674,
      "grad_norm": 23.0,
      "learning_rate": 1.9939373426140575e-05,
      "loss": 8.2683,
      "step": 37410
    },
    {
      "epoch": 0.10523621903307545,
      "grad_norm": 32.5,
      "learning_rate": 1.9939341041793506e-05,
      "loss": 8.4331,
      "step": 37420
    },
    {
      "epoch": 0.10526434202052416,
      "grad_norm": 31.5,
      "learning_rate": 1.9939308648825827e-05,
      "loss": 7.7573,
      "step": 37430
    },
    {
      "epoch": 0.10529246500797286,
      "grad_norm": 20.25,
      "learning_rate": 1.9939276247237572e-05,
      "loss": 8.1228,
      "step": 37440
    },
    {
      "epoch": 0.10532058799542157,
      "grad_norm": 39.25,
      "learning_rate": 1.993924383702877e-05,
      "loss": 8.1064,
      "step": 37450
    },
    {
      "epoch": 0.10534871098287028,
      "grad_norm": 36.5,
      "learning_rate": 1.9939211418199445e-05,
      "loss": 8.1962,
      "step": 37460
    },
    {
      "epoch": 0.105376833970319,
      "grad_norm": 32.5,
      "learning_rate": 1.9939178990749627e-05,
      "loss": 8.1945,
      "step": 37470
    },
    {
      "epoch": 0.1054049569577677,
      "grad_norm": 25.0,
      "learning_rate": 1.993914655467935e-05,
      "loss": 7.8759,
      "step": 37480
    },
    {
      "epoch": 0.10543307994521642,
      "grad_norm": 33.25,
      "learning_rate": 1.9939114109988633e-05,
      "loss": 8.132,
      "step": 37490
    },
    {
      "epoch": 0.10546120293266513,
      "grad_norm": 22.5,
      "learning_rate": 1.9939081656677507e-05,
      "loss": 8.2575,
      "step": 37500
    },
    {
      "epoch": 0.10548932592011384,
      "grad_norm": 23.5,
      "learning_rate": 1.9939049194746004e-05,
      "loss": 8.2233,
      "step": 37510
    },
    {
      "epoch": 0.10551744890756255,
      "grad_norm": 24.0,
      "learning_rate": 1.993901672419415e-05,
      "loss": 7.0798,
      "step": 37520
    },
    {
      "epoch": 0.10554557189501126,
      "grad_norm": 22.875,
      "learning_rate": 1.993898424502197e-05,
      "loss": 7.7662,
      "step": 37530
    },
    {
      "epoch": 0.10557369488245998,
      "grad_norm": 28.375,
      "learning_rate": 1.9938951757229494e-05,
      "loss": 8.581,
      "step": 37540
    },
    {
      "epoch": 0.10560181786990869,
      "grad_norm": 38.25,
      "learning_rate": 1.9938919260816754e-05,
      "loss": 8.3698,
      "step": 37550
    },
    {
      "epoch": 0.1056299408573574,
      "grad_norm": 25.125,
      "learning_rate": 1.9938886755783773e-05,
      "loss": 7.6149,
      "step": 37560
    },
    {
      "epoch": 0.10565806384480611,
      "grad_norm": 26.625,
      "learning_rate": 1.9938854242130582e-05,
      "loss": 7.6656,
      "step": 37570
    },
    {
      "epoch": 0.10568618683225482,
      "grad_norm": 60.75,
      "learning_rate": 1.993882171985721e-05,
      "loss": 7.4342,
      "step": 37580
    },
    {
      "epoch": 0.10571430981970353,
      "grad_norm": 38.0,
      "learning_rate": 1.9938789188963682e-05,
      "loss": 7.4924,
      "step": 37590
    },
    {
      "epoch": 0.10574243280715223,
      "grad_norm": 30.0,
      "learning_rate": 1.9938756649450026e-05,
      "loss": 8.1272,
      "step": 37600
    },
    {
      "epoch": 0.10577055579460094,
      "grad_norm": 21.375,
      "learning_rate": 1.9938724101316272e-05,
      "loss": 7.9996,
      "step": 37610
    },
    {
      "epoch": 0.10579867878204965,
      "grad_norm": 22.25,
      "learning_rate": 1.993869154456245e-05,
      "loss": 7.8753,
      "step": 37620
    },
    {
      "epoch": 0.10582680176949837,
      "grad_norm": 24.875,
      "learning_rate": 1.993865897918859e-05,
      "loss": 8.1137,
      "step": 37630
    },
    {
      "epoch": 0.10585492475694708,
      "grad_norm": 36.75,
      "learning_rate": 1.993862640519471e-05,
      "loss": 7.5714,
      "step": 37640
    },
    {
      "epoch": 0.10588304774439579,
      "grad_norm": 40.75,
      "learning_rate": 1.993859382258085e-05,
      "loss": 8.6856,
      "step": 37650
    },
    {
      "epoch": 0.1059111707318445,
      "grad_norm": 43.5,
      "learning_rate": 1.993856123134703e-05,
      "loss": 8.0123,
      "step": 37660
    },
    {
      "epoch": 0.10593929371929321,
      "grad_norm": 25.875,
      "learning_rate": 1.9938528631493287e-05,
      "loss": 7.2639,
      "step": 37670
    },
    {
      "epoch": 0.10596741670674192,
      "grad_norm": 32.75,
      "learning_rate": 1.993849602301964e-05,
      "loss": 8.0832,
      "step": 37680
    },
    {
      "epoch": 0.10599553969419064,
      "grad_norm": 30.75,
      "learning_rate": 1.993846340592612e-05,
      "loss": 7.6851,
      "step": 37690
    },
    {
      "epoch": 0.10602366268163935,
      "grad_norm": 29.875,
      "learning_rate": 1.993843078021276e-05,
      "loss": 7.7075,
      "step": 37700
    },
    {
      "epoch": 0.10605178566908806,
      "grad_norm": 50.0,
      "learning_rate": 1.9938398145879582e-05,
      "loss": 7.9974,
      "step": 37710
    },
    {
      "epoch": 0.10607990865653677,
      "grad_norm": 28.625,
      "learning_rate": 1.9938365502926615e-05,
      "loss": 7.8758,
      "step": 37720
    },
    {
      "epoch": 0.10610803164398548,
      "grad_norm": 27.625,
      "learning_rate": 1.9938332851353892e-05,
      "loss": 8.1574,
      "step": 37730
    },
    {
      "epoch": 0.10613615463143419,
      "grad_norm": 30.0,
      "learning_rate": 1.993830019116144e-05,
      "loss": 8.2086,
      "step": 37740
    },
    {
      "epoch": 0.1061642776188829,
      "grad_norm": 32.75,
      "learning_rate": 1.9938267522349282e-05,
      "loss": 8.6949,
      "step": 37750
    },
    {
      "epoch": 0.10619240060633162,
      "grad_norm": 27.75,
      "learning_rate": 1.9938234844917453e-05,
      "loss": 7.2705,
      "step": 37760
    },
    {
      "epoch": 0.10622052359378031,
      "grad_norm": 24.75,
      "learning_rate": 1.993820215886598e-05,
      "loss": 8.1567,
      "step": 37770
    },
    {
      "epoch": 0.10624864658122903,
      "grad_norm": 44.75,
      "learning_rate": 1.993816946419489e-05,
      "loss": 8.4836,
      "step": 37780
    },
    {
      "epoch": 0.10627676956867774,
      "grad_norm": 31.25,
      "learning_rate": 1.9938136760904208e-05,
      "loss": 7.6923,
      "step": 37790
    },
    {
      "epoch": 0.10630489255612645,
      "grad_norm": 40.75,
      "learning_rate": 1.9938104048993967e-05,
      "loss": 8.018,
      "step": 37800
    },
    {
      "epoch": 0.10633301554357516,
      "grad_norm": 30.75,
      "learning_rate": 1.9938071328464193e-05,
      "loss": 7.7576,
      "step": 37810
    },
    {
      "epoch": 0.10636113853102387,
      "grad_norm": 60.0,
      "learning_rate": 1.9938038599314918e-05,
      "loss": 7.9061,
      "step": 37820
    },
    {
      "epoch": 0.10638926151847258,
      "grad_norm": 26.125,
      "learning_rate": 1.9938005861546167e-05,
      "loss": 8.0266,
      "step": 37830
    },
    {
      "epoch": 0.1064173845059213,
      "grad_norm": 31.125,
      "learning_rate": 1.9937973115157972e-05,
      "loss": 7.648,
      "step": 37840
    },
    {
      "epoch": 0.10644550749337,
      "grad_norm": 29.5,
      "learning_rate": 1.9937940360150353e-05,
      "loss": 8.1198,
      "step": 37850
    },
    {
      "epoch": 0.10647363048081872,
      "grad_norm": 41.75,
      "learning_rate": 1.9937907596523346e-05,
      "loss": 7.8347,
      "step": 37860
    },
    {
      "epoch": 0.10650175346826743,
      "grad_norm": 24.5,
      "learning_rate": 1.993787482427698e-05,
      "loss": 8.3854,
      "step": 37870
    },
    {
      "epoch": 0.10652987645571614,
      "grad_norm": 51.75,
      "learning_rate": 1.993784204341128e-05,
      "loss": 7.7123,
      "step": 37880
    },
    {
      "epoch": 0.10655799944316485,
      "grad_norm": 26.625,
      "learning_rate": 1.9937809253926273e-05,
      "loss": 7.8755,
      "step": 37890
    },
    {
      "epoch": 0.10658612243061356,
      "grad_norm": 27.875,
      "learning_rate": 1.9937776455821994e-05,
      "loss": 7.5196,
      "step": 37900
    },
    {
      "epoch": 0.10661424541806228,
      "grad_norm": 34.0,
      "learning_rate": 1.9937743649098462e-05,
      "loss": 7.767,
      "step": 37910
    },
    {
      "epoch": 0.10664236840551099,
      "grad_norm": 41.25,
      "learning_rate": 1.9937710833755717e-05,
      "loss": 7.3884,
      "step": 37920
    },
    {
      "epoch": 0.10667049139295968,
      "grad_norm": 29.625,
      "learning_rate": 1.9937678009793778e-05,
      "loss": 7.2371,
      "step": 37930
    },
    {
      "epoch": 0.1066986143804084,
      "grad_norm": 43.75,
      "learning_rate": 1.9937645177212676e-05,
      "loss": 7.7799,
      "step": 37940
    },
    {
      "epoch": 0.10672673736785711,
      "grad_norm": 31.375,
      "learning_rate": 1.993761233601244e-05,
      "loss": 7.306,
      "step": 37950
    },
    {
      "epoch": 0.10675486035530582,
      "grad_norm": 26.625,
      "learning_rate": 1.9937579486193096e-05,
      "loss": 7.4619,
      "step": 37960
    },
    {
      "epoch": 0.10678298334275453,
      "grad_norm": 24.5,
      "learning_rate": 1.993754662775468e-05,
      "loss": 7.2675,
      "step": 37970
    },
    {
      "epoch": 0.10681110633020324,
      "grad_norm": 36.75,
      "learning_rate": 1.9937513760697212e-05,
      "loss": 7.9504,
      "step": 37980
    },
    {
      "epoch": 0.10683922931765195,
      "grad_norm": 59.25,
      "learning_rate": 1.9937480885020727e-05,
      "loss": 7.9405,
      "step": 37990
    },
    {
      "epoch": 0.10686735230510067,
      "grad_norm": 26.5,
      "learning_rate": 1.9937448000725248e-05,
      "loss": 7.982,
      "step": 38000
    },
    {
      "epoch": 0.10689547529254938,
      "grad_norm": 26.875,
      "learning_rate": 1.993741510781081e-05,
      "loss": 8.093,
      "step": 38010
    },
    {
      "epoch": 0.10692359827999809,
      "grad_norm": 26.25,
      "learning_rate": 1.9937382206277433e-05,
      "loss": 7.3795,
      "step": 38020
    },
    {
      "epoch": 0.1069517212674468,
      "grad_norm": 22.75,
      "learning_rate": 1.993734929612515e-05,
      "loss": 7.485,
      "step": 38030
    },
    {
      "epoch": 0.10697984425489551,
      "grad_norm": 22.5,
      "learning_rate": 1.993731637735399e-05,
      "loss": 7.463,
      "step": 38040
    },
    {
      "epoch": 0.10700796724234422,
      "grad_norm": 46.25,
      "learning_rate": 1.9937283449963986e-05,
      "loss": 7.8458,
      "step": 38050
    },
    {
      "epoch": 0.10703609022979294,
      "grad_norm": 30.875,
      "learning_rate": 1.993725051395516e-05,
      "loss": 8.2385,
      "step": 38060
    },
    {
      "epoch": 0.10706421321724165,
      "grad_norm": 36.25,
      "learning_rate": 1.9937217569327537e-05,
      "loss": 8.235,
      "step": 38070
    },
    {
      "epoch": 0.10709233620469036,
      "grad_norm": 25.25,
      "learning_rate": 1.9937184616081156e-05,
      "loss": 7.8194,
      "step": 38080
    },
    {
      "epoch": 0.10712045919213906,
      "grad_norm": 39.75,
      "learning_rate": 1.993715165421604e-05,
      "loss": 7.691,
      "step": 38090
    },
    {
      "epoch": 0.10714858217958777,
      "grad_norm": 49.5,
      "learning_rate": 1.9937118683732217e-05,
      "loss": 7.9876,
      "step": 38100
    },
    {
      "epoch": 0.10717670516703648,
      "grad_norm": 39.0,
      "learning_rate": 1.993708570462972e-05,
      "loss": 8.2332,
      "step": 38110
    },
    {
      "epoch": 0.10720482815448519,
      "grad_norm": 44.0,
      "learning_rate": 1.9937052716908573e-05,
      "loss": 7.9076,
      "step": 38120
    },
    {
      "epoch": 0.1072329511419339,
      "grad_norm": 22.25,
      "learning_rate": 1.99370197205688e-05,
      "loss": 8.1853,
      "step": 38130
    },
    {
      "epoch": 0.10726107412938261,
      "grad_norm": 23.5,
      "learning_rate": 1.9936986715610443e-05,
      "loss": 8.5357,
      "step": 38140
    },
    {
      "epoch": 0.10728919711683133,
      "grad_norm": 40.75,
      "learning_rate": 1.993695370203352e-05,
      "loss": 7.591,
      "step": 38150
    },
    {
      "epoch": 0.10731732010428004,
      "grad_norm": 20.125,
      "learning_rate": 1.9936920679838066e-05,
      "loss": 7.7734,
      "step": 38160
    },
    {
      "epoch": 0.10734544309172875,
      "grad_norm": 28.0,
      "learning_rate": 1.99368876490241e-05,
      "loss": 7.7951,
      "step": 38170
    },
    {
      "epoch": 0.10737356607917746,
      "grad_norm": 21.125,
      "learning_rate": 1.993685460959166e-05,
      "loss": 8.2128,
      "step": 38180
    },
    {
      "epoch": 0.10740168906662617,
      "grad_norm": 46.0,
      "learning_rate": 1.9936821561540776e-05,
      "loss": 7.8836,
      "step": 38190
    },
    {
      "epoch": 0.10742981205407488,
      "grad_norm": 24.5,
      "learning_rate": 1.993678850487147e-05,
      "loss": 7.9884,
      "step": 38200
    },
    {
      "epoch": 0.1074579350415236,
      "grad_norm": 36.5,
      "learning_rate": 1.9936755439583773e-05,
      "loss": 7.1352,
      "step": 38210
    },
    {
      "epoch": 0.1074860580289723,
      "grad_norm": 62.75,
      "learning_rate": 1.9936722365677708e-05,
      "loss": 8.0675,
      "step": 38220
    },
    {
      "epoch": 0.10751418101642102,
      "grad_norm": 24.375,
      "learning_rate": 1.9936689283153317e-05,
      "loss": 7.6122,
      "step": 38230
    },
    {
      "epoch": 0.10754230400386973,
      "grad_norm": 27.25,
      "learning_rate": 1.993665619201062e-05,
      "loss": 7.8434,
      "step": 38240
    },
    {
      "epoch": 0.10757042699131843,
      "grad_norm": 28.375,
      "learning_rate": 1.9936623092249644e-05,
      "loss": 7.8096,
      "step": 38250
    },
    {
      "epoch": 0.10759854997876714,
      "grad_norm": 24.75,
      "learning_rate": 1.9936589983870423e-05,
      "loss": 8.1659,
      "step": 38260
    },
    {
      "epoch": 0.10762667296621585,
      "grad_norm": 33.5,
      "learning_rate": 1.9936556866872984e-05,
      "loss": 8.0282,
      "step": 38270
    },
    {
      "epoch": 0.10765479595366456,
      "grad_norm": 24.5,
      "learning_rate": 1.993652374125735e-05,
      "loss": 7.9018,
      "step": 38280
    },
    {
      "epoch": 0.10768291894111327,
      "grad_norm": 24.0,
      "learning_rate": 1.993649060702356e-05,
      "loss": 7.9096,
      "step": 38290
    },
    {
      "epoch": 0.10771104192856198,
      "grad_norm": 35.75,
      "learning_rate": 1.9936457464171636e-05,
      "loss": 7.3442,
      "step": 38300
    },
    {
      "epoch": 0.1077391649160107,
      "grad_norm": 31.0,
      "learning_rate": 1.9936424312701606e-05,
      "loss": 8.0277,
      "step": 38310
    },
    {
      "epoch": 0.10776728790345941,
      "grad_norm": 30.375,
      "learning_rate": 1.9936391152613504e-05,
      "loss": 7.9162,
      "step": 38320
    },
    {
      "epoch": 0.10779541089090812,
      "grad_norm": 40.0,
      "learning_rate": 1.9936357983907353e-05,
      "loss": 8.5291,
      "step": 38330
    },
    {
      "epoch": 0.10782353387835683,
      "grad_norm": 34.75,
      "learning_rate": 1.9936324806583187e-05,
      "loss": 8.0942,
      "step": 38340
    },
    {
      "epoch": 0.10785165686580554,
      "grad_norm": 22.125,
      "learning_rate": 1.993629162064103e-05,
      "loss": 7.735,
      "step": 38350
    },
    {
      "epoch": 0.10787977985325425,
      "grad_norm": 24.875,
      "learning_rate": 1.9936258426080916e-05,
      "loss": 7.7416,
      "step": 38360
    },
    {
      "epoch": 0.10790790284070297,
      "grad_norm": 33.75,
      "learning_rate": 1.993622522290287e-05,
      "loss": 7.2172,
      "step": 38370
    },
    {
      "epoch": 0.10793602582815168,
      "grad_norm": 23.5,
      "learning_rate": 1.993619201110692e-05,
      "loss": 7.8767,
      "step": 38380
    },
    {
      "epoch": 0.10796414881560039,
      "grad_norm": 43.0,
      "learning_rate": 1.9936158790693095e-05,
      "loss": 7.7351,
      "step": 38390
    },
    {
      "epoch": 0.1079922718030491,
      "grad_norm": 46.0,
      "learning_rate": 1.9936125561661427e-05,
      "loss": 8.4808,
      "step": 38400
    },
    {
      "epoch": 0.10802039479049781,
      "grad_norm": 28.75,
      "learning_rate": 1.9936092324011948e-05,
      "loss": 7.7932,
      "step": 38410
    },
    {
      "epoch": 0.10804851777794651,
      "grad_norm": 27.25,
      "learning_rate": 1.9936059077744673e-05,
      "loss": 7.4869,
      "step": 38420
    },
    {
      "epoch": 0.10807664076539522,
      "grad_norm": 29.5,
      "learning_rate": 1.9936025822859645e-05,
      "loss": 8.0268,
      "step": 38430
    },
    {
      "epoch": 0.10810476375284393,
      "grad_norm": 25.875,
      "learning_rate": 1.993599255935689e-05,
      "loss": 8.8957,
      "step": 38440
    },
    {
      "epoch": 0.10813288674029264,
      "grad_norm": 26.375,
      "learning_rate": 1.993595928723643e-05,
      "loss": 7.9603,
      "step": 38450
    },
    {
      "epoch": 0.10816100972774136,
      "grad_norm": 32.0,
      "learning_rate": 1.99359260064983e-05,
      "loss": 7.707,
      "step": 38460
    },
    {
      "epoch": 0.10818913271519007,
      "grad_norm": 34.0,
      "learning_rate": 1.9935892717142527e-05,
      "loss": 8.0403,
      "step": 38470
    },
    {
      "epoch": 0.10821725570263878,
      "grad_norm": 31.375,
      "learning_rate": 1.993585941916914e-05,
      "loss": 8.6465,
      "step": 38480
    },
    {
      "epoch": 0.10824537869008749,
      "grad_norm": 23.625,
      "learning_rate": 1.9935826112578164e-05,
      "loss": 7.9494,
      "step": 38490
    },
    {
      "epoch": 0.1082735016775362,
      "grad_norm": 29.5,
      "learning_rate": 1.9935792797369638e-05,
      "loss": 6.708,
      "step": 38500
    },
    {
      "epoch": 0.10830162466498491,
      "grad_norm": 28.25,
      "learning_rate": 1.9935759473543582e-05,
      "loss": 7.4881,
      "step": 38510
    },
    {
      "epoch": 0.10832974765243363,
      "grad_norm": 27.125,
      "learning_rate": 1.9935726141100026e-05,
      "loss": 7.9426,
      "step": 38520
    },
    {
      "epoch": 0.10835787063988234,
      "grad_norm": 29.375,
      "learning_rate": 1.9935692800039005e-05,
      "loss": 8.0752,
      "step": 38530
    },
    {
      "epoch": 0.10838599362733105,
      "grad_norm": 36.75,
      "learning_rate": 1.9935659450360538e-05,
      "loss": 7.7482,
      "step": 38540
    },
    {
      "epoch": 0.10841411661477976,
      "grad_norm": 48.25,
      "learning_rate": 1.9935626092064665e-05,
      "loss": 7.7859,
      "step": 38550
    },
    {
      "epoch": 0.10844223960222847,
      "grad_norm": 31.25,
      "learning_rate": 1.9935592725151405e-05,
      "loss": 7.3673,
      "step": 38560
    },
    {
      "epoch": 0.10847036258967718,
      "grad_norm": 33.25,
      "learning_rate": 1.993555934962079e-05,
      "loss": 8.2916,
      "step": 38570
    },
    {
      "epoch": 0.10849848557712588,
      "grad_norm": 29.625,
      "learning_rate": 1.9935525965472854e-05,
      "loss": 8.0459,
      "step": 38580
    },
    {
      "epoch": 0.10852660856457459,
      "grad_norm": 25.5,
      "learning_rate": 1.9935492572707623e-05,
      "loss": 8.422,
      "step": 38590
    },
    {
      "epoch": 0.1085547315520233,
      "grad_norm": 26.625,
      "learning_rate": 1.9935459171325123e-05,
      "loss": 7.9267,
      "step": 38600
    },
    {
      "epoch": 0.10858285453947202,
      "grad_norm": 24.375,
      "learning_rate": 1.9935425761325384e-05,
      "loss": 8.7817,
      "step": 38610
    },
    {
      "epoch": 0.10861097752692073,
      "grad_norm": 23.125,
      "learning_rate": 1.9935392342708436e-05,
      "loss": 7.4169,
      "step": 38620
    },
    {
      "epoch": 0.10863910051436944,
      "grad_norm": 23.125,
      "learning_rate": 1.993535891547431e-05,
      "loss": 7.3181,
      "step": 38630
    },
    {
      "epoch": 0.10866722350181815,
      "grad_norm": 32.25,
      "learning_rate": 1.9935325479623032e-05,
      "loss": 7.9157,
      "step": 38640
    },
    {
      "epoch": 0.10869534648926686,
      "grad_norm": 37.75,
      "learning_rate": 1.9935292035154632e-05,
      "loss": 8.0658,
      "step": 38650
    },
    {
      "epoch": 0.10872346947671557,
      "grad_norm": 43.5,
      "learning_rate": 1.993525858206914e-05,
      "loss": 7.5357,
      "step": 38660
    },
    {
      "epoch": 0.10875159246416428,
      "grad_norm": 28.875,
      "learning_rate": 1.993522512036658e-05,
      "loss": 7.9626,
      "step": 38670
    },
    {
      "epoch": 0.108779715451613,
      "grad_norm": 28.125,
      "learning_rate": 1.993519165004699e-05,
      "loss": 8.3615,
      "step": 38680
    },
    {
      "epoch": 0.10880783843906171,
      "grad_norm": 36.75,
      "learning_rate": 1.9935158171110393e-05,
      "loss": 7.46,
      "step": 38690
    },
    {
      "epoch": 0.10883596142651042,
      "grad_norm": 41.25,
      "learning_rate": 1.993512468355682e-05,
      "loss": 7.8048,
      "step": 38700
    },
    {
      "epoch": 0.10886408441395913,
      "grad_norm": 50.0,
      "learning_rate": 1.9935091187386294e-05,
      "loss": 7.4619,
      "step": 38710
    },
    {
      "epoch": 0.10889220740140784,
      "grad_norm": 23.125,
      "learning_rate": 1.9935057682598855e-05,
      "loss": 7.8125,
      "step": 38720
    },
    {
      "epoch": 0.10892033038885655,
      "grad_norm": 33.5,
      "learning_rate": 1.9935024169194523e-05,
      "loss": 7.2023,
      "step": 38730
    },
    {
      "epoch": 0.10894845337630525,
      "grad_norm": 25.625,
      "learning_rate": 1.9934990647173332e-05,
      "loss": 7.8011,
      "step": 38740
    },
    {
      "epoch": 0.10897657636375396,
      "grad_norm": 31.5,
      "learning_rate": 1.993495711653531e-05,
      "loss": 7.6673,
      "step": 38750
    },
    {
      "epoch": 0.10900469935120267,
      "grad_norm": 28.5,
      "learning_rate": 1.9934923577280483e-05,
      "loss": 8.1333,
      "step": 38760
    },
    {
      "epoch": 0.10903282233865139,
      "grad_norm": 29.5,
      "learning_rate": 1.9934890029408883e-05,
      "loss": 7.2259,
      "step": 38770
    },
    {
      "epoch": 0.1090609453261001,
      "grad_norm": 24.125,
      "learning_rate": 1.9934856472920542e-05,
      "loss": 8.1442,
      "step": 38780
    },
    {
      "epoch": 0.10908906831354881,
      "grad_norm": 34.75,
      "learning_rate": 1.993482290781548e-05,
      "loss": 7.69,
      "step": 38790
    },
    {
      "epoch": 0.10911719130099752,
      "grad_norm": 37.25,
      "learning_rate": 1.9934789334093736e-05,
      "loss": 8.0078,
      "step": 38800
    },
    {
      "epoch": 0.10914531428844623,
      "grad_norm": 29.875,
      "learning_rate": 1.993475575175534e-05,
      "loss": 7.8476,
      "step": 38810
    },
    {
      "epoch": 0.10917343727589494,
      "grad_norm": 28.375,
      "learning_rate": 1.9934722160800305e-05,
      "loss": 8.8081,
      "step": 38820
    },
    {
      "epoch": 0.10920156026334366,
      "grad_norm": 25.375,
      "learning_rate": 1.993468856122868e-05,
      "loss": 7.4624,
      "step": 38830
    },
    {
      "epoch": 0.10922968325079237,
      "grad_norm": 26.375,
      "learning_rate": 1.993465495304048e-05,
      "loss": 8.3472,
      "step": 38840
    },
    {
      "epoch": 0.10925780623824108,
      "grad_norm": 20.125,
      "learning_rate": 1.9934621336235744e-05,
      "loss": 7.6499,
      "step": 38850
    },
    {
      "epoch": 0.10928592922568979,
      "grad_norm": 39.75,
      "learning_rate": 1.9934587710814493e-05,
      "loss": 8.4416,
      "step": 38860
    },
    {
      "epoch": 0.1093140522131385,
      "grad_norm": 22.125,
      "learning_rate": 1.9934554076776762e-05,
      "loss": 7.7076,
      "step": 38870
    },
    {
      "epoch": 0.10934217520058721,
      "grad_norm": 54.75,
      "learning_rate": 1.9934520434122575e-05,
      "loss": 7.9891,
      "step": 38880
    },
    {
      "epoch": 0.10937029818803592,
      "grad_norm": 32.0,
      "learning_rate": 1.993448678285197e-05,
      "loss": 7.6102,
      "step": 38890
    },
    {
      "epoch": 0.10939842117548462,
      "grad_norm": 20.5,
      "learning_rate": 1.9934453122964964e-05,
      "loss": 7.247,
      "step": 38900
    },
    {
      "epoch": 0.10942654416293333,
      "grad_norm": 22.0,
      "learning_rate": 1.9934419454461595e-05,
      "loss": 9.1693,
      "step": 38910
    },
    {
      "epoch": 0.10945466715038205,
      "grad_norm": 51.75,
      "learning_rate": 1.993438577734189e-05,
      "loss": 7.5433,
      "step": 38920
    },
    {
      "epoch": 0.10948279013783076,
      "grad_norm": 44.5,
      "learning_rate": 1.993435209160588e-05,
      "loss": 8.2855,
      "step": 38930
    },
    {
      "epoch": 0.10951091312527947,
      "grad_norm": 52.5,
      "learning_rate": 1.9934318397253593e-05,
      "loss": 7.7567,
      "step": 38940
    },
    {
      "epoch": 0.10953903611272818,
      "grad_norm": 31.75,
      "learning_rate": 1.9934284694285052e-05,
      "loss": 8.002,
      "step": 38950
    },
    {
      "epoch": 0.10956715910017689,
      "grad_norm": 28.0,
      "learning_rate": 1.99342509827003e-05,
      "loss": 7.9833,
      "step": 38960
    },
    {
      "epoch": 0.1095952820876256,
      "grad_norm": 27.75,
      "learning_rate": 1.9934217262499352e-05,
      "loss": 8.8831,
      "step": 38970
    },
    {
      "epoch": 0.10962340507507431,
      "grad_norm": 38.25,
      "learning_rate": 1.9934183533682243e-05,
      "loss": 8.0133,
      "step": 38980
    },
    {
      "epoch": 0.10965152806252303,
      "grad_norm": 41.25,
      "learning_rate": 1.9934149796249002e-05,
      "loss": 8.0578,
      "step": 38990
    },
    {
      "epoch": 0.10967965104997174,
      "grad_norm": 36.75,
      "learning_rate": 1.993411605019966e-05,
      "loss": 7.9605,
      "step": 39000
    },
    {
      "epoch": 0.10970777403742045,
      "grad_norm": 22.25,
      "learning_rate": 1.9934082295534244e-05,
      "loss": 7.3487,
      "step": 39010
    },
    {
      "epoch": 0.10973589702486916,
      "grad_norm": 26.875,
      "learning_rate": 1.9934048532252787e-05,
      "loss": 8.9964,
      "step": 39020
    },
    {
      "epoch": 0.10976402001231787,
      "grad_norm": 23.375,
      "learning_rate": 1.9934014760355314e-05,
      "loss": 8.1482,
      "step": 39030
    },
    {
      "epoch": 0.10979214299976658,
      "grad_norm": 30.75,
      "learning_rate": 1.9933980979841857e-05,
      "loss": 7.5367,
      "step": 39040
    },
    {
      "epoch": 0.1098202659872153,
      "grad_norm": 29.25,
      "learning_rate": 1.9933947190712444e-05,
      "loss": 7.764,
      "step": 39050
    },
    {
      "epoch": 0.10984838897466401,
      "grad_norm": 21.5,
      "learning_rate": 1.9933913392967103e-05,
      "loss": 8.0977,
      "step": 39060
    },
    {
      "epoch": 0.1098765119621127,
      "grad_norm": 22.875,
      "learning_rate": 1.9933879586605866e-05,
      "loss": 8.1165,
      "step": 39070
    },
    {
      "epoch": 0.10990463494956142,
      "grad_norm": 23.125,
      "learning_rate": 1.9933845771628758e-05,
      "loss": 7.4995,
      "step": 39080
    },
    {
      "epoch": 0.10993275793701013,
      "grad_norm": 26.5,
      "learning_rate": 1.9933811948035814e-05,
      "loss": 8.1445,
      "step": 39090
    },
    {
      "epoch": 0.10996088092445884,
      "grad_norm": 23.375,
      "learning_rate": 1.993377811582706e-05,
      "loss": 8.1046,
      "step": 39100
    },
    {
      "epoch": 0.10998900391190755,
      "grad_norm": 28.25,
      "learning_rate": 1.9933744275002527e-05,
      "loss": 8.1115,
      "step": 39110
    },
    {
      "epoch": 0.11001712689935626,
      "grad_norm": 21.5,
      "learning_rate": 1.9933710425562244e-05,
      "loss": 8.5553,
      "step": 39120
    },
    {
      "epoch": 0.11004524988680497,
      "grad_norm": 28.75,
      "learning_rate": 1.993367656750624e-05,
      "loss": 7.7598,
      "step": 39130
    },
    {
      "epoch": 0.11007337287425369,
      "grad_norm": 29.875,
      "learning_rate": 1.9933642700834542e-05,
      "loss": 7.7913,
      "step": 39140
    },
    {
      "epoch": 0.1101014958617024,
      "grad_norm": 23.375,
      "learning_rate": 1.993360882554718e-05,
      "loss": 8.078,
      "step": 39150
    },
    {
      "epoch": 0.11012961884915111,
      "grad_norm": 45.5,
      "learning_rate": 1.993357494164419e-05,
      "loss": 7.9046,
      "step": 39160
    },
    {
      "epoch": 0.11015774183659982,
      "grad_norm": 37.25,
      "learning_rate": 1.9933541049125594e-05,
      "loss": 7.7218,
      "step": 39170
    },
    {
      "epoch": 0.11018586482404853,
      "grad_norm": 33.5,
      "learning_rate": 1.9933507147991422e-05,
      "loss": 8.1146,
      "step": 39180
    },
    {
      "epoch": 0.11021398781149724,
      "grad_norm": 34.75,
      "learning_rate": 1.9933473238241705e-05,
      "loss": 8.322,
      "step": 39190
    },
    {
      "epoch": 0.11024211079894596,
      "grad_norm": 26.375,
      "learning_rate": 1.9933439319876475e-05,
      "loss": 7.8498,
      "step": 39200
    },
    {
      "epoch": 0.11027023378639467,
      "grad_norm": 44.25,
      "learning_rate": 1.9933405392895756e-05,
      "loss": 7.5955,
      "step": 39210
    },
    {
      "epoch": 0.11029835677384338,
      "grad_norm": 27.0,
      "learning_rate": 1.9933371457299585e-05,
      "loss": 7.3886,
      "step": 39220
    },
    {
      "epoch": 0.11032647976129208,
      "grad_norm": 23.875,
      "learning_rate": 1.9933337513087984e-05,
      "loss": 7.68,
      "step": 39230
    },
    {
      "epoch": 0.11035460274874079,
      "grad_norm": 46.0,
      "learning_rate": 1.9933303560260984e-05,
      "loss": 8.0385,
      "step": 39240
    },
    {
      "epoch": 0.1103827257361895,
      "grad_norm": 31.375,
      "learning_rate": 1.9933269598818615e-05,
      "loss": 7.9032,
      "step": 39250
    },
    {
      "epoch": 0.11041084872363821,
      "grad_norm": 31.125,
      "learning_rate": 1.993323562876091e-05,
      "loss": 8.1931,
      "step": 39260
    },
    {
      "epoch": 0.11043897171108692,
      "grad_norm": 33.0,
      "learning_rate": 1.9933201650087894e-05,
      "loss": 7.5755,
      "step": 39270
    },
    {
      "epoch": 0.11046709469853563,
      "grad_norm": 22.625,
      "learning_rate": 1.99331676627996e-05,
      "loss": 7.5847,
      "step": 39280
    },
    {
      "epoch": 0.11049521768598435,
      "grad_norm": 27.25,
      "learning_rate": 1.993313366689605e-05,
      "loss": 8.3142,
      "step": 39290
    },
    {
      "epoch": 0.11052334067343306,
      "grad_norm": 31.375,
      "learning_rate": 1.9933099662377286e-05,
      "loss": 7.5893,
      "step": 39300
    },
    {
      "epoch": 0.11055146366088177,
      "grad_norm": 26.5,
      "learning_rate": 1.993306564924333e-05,
      "loss": 7.9664,
      "step": 39310
    },
    {
      "epoch": 0.11057958664833048,
      "grad_norm": 61.25,
      "learning_rate": 1.9933031627494207e-05,
      "loss": 7.7768,
      "step": 39320
    },
    {
      "epoch": 0.11060770963577919,
      "grad_norm": 28.625,
      "learning_rate": 1.9932997597129953e-05,
      "loss": 7.6146,
      "step": 39330
    },
    {
      "epoch": 0.1106358326232279,
      "grad_norm": 40.5,
      "learning_rate": 1.9932963558150598e-05,
      "loss": 8.3587,
      "step": 39340
    },
    {
      "epoch": 0.11066395561067661,
      "grad_norm": 33.75,
      "learning_rate": 1.993292951055617e-05,
      "loss": 8.1968,
      "step": 39350
    },
    {
      "epoch": 0.11069207859812533,
      "grad_norm": 35.0,
      "learning_rate": 1.9932895454346694e-05,
      "loss": 8.0016,
      "step": 39360
    },
    {
      "epoch": 0.11072020158557404,
      "grad_norm": 37.75,
      "learning_rate": 1.993286138952221e-05,
      "loss": 7.9206,
      "step": 39370
    },
    {
      "epoch": 0.11074832457302275,
      "grad_norm": 30.25,
      "learning_rate": 1.9932827316082735e-05,
      "loss": 7.5879,
      "step": 39380
    },
    {
      "epoch": 0.11077644756047145,
      "grad_norm": 33.5,
      "learning_rate": 1.9932793234028306e-05,
      "loss": 8.7761,
      "step": 39390
    },
    {
      "epoch": 0.11080457054792016,
      "grad_norm": 19.5,
      "learning_rate": 1.9932759143358955e-05,
      "loss": 8.1826,
      "step": 39400
    },
    {
      "epoch": 0.11083269353536887,
      "grad_norm": 22.875,
      "learning_rate": 1.9932725044074707e-05,
      "loss": 7.3338,
      "step": 39410
    },
    {
      "epoch": 0.11086081652281758,
      "grad_norm": 22.75,
      "learning_rate": 1.9932690936175588e-05,
      "loss": 7.8626,
      "step": 39420
    },
    {
      "epoch": 0.1108889395102663,
      "grad_norm": 31.125,
      "learning_rate": 1.9932656819661636e-05,
      "loss": 8.351,
      "step": 39430
    },
    {
      "epoch": 0.110917062497715,
      "grad_norm": 41.5,
      "learning_rate": 1.9932622694532874e-05,
      "loss": 7.4187,
      "step": 39440
    },
    {
      "epoch": 0.11094518548516372,
      "grad_norm": 26.0,
      "learning_rate": 1.9932588560789337e-05,
      "loss": 8.1838,
      "step": 39450
    },
    {
      "epoch": 0.11097330847261243,
      "grad_norm": 31.625,
      "learning_rate": 1.993255441843105e-05,
      "loss": 7.7531,
      "step": 39460
    },
    {
      "epoch": 0.11100143146006114,
      "grad_norm": 30.5,
      "learning_rate": 1.9932520267458043e-05,
      "loss": 7.8584,
      "step": 39470
    },
    {
      "epoch": 0.11102955444750985,
      "grad_norm": 55.75,
      "learning_rate": 1.993248610787035e-05,
      "loss": 8.2651,
      "step": 39480
    },
    {
      "epoch": 0.11105767743495856,
      "grad_norm": 25.375,
      "learning_rate": 1.9932451939668e-05,
      "loss": 8.1219,
      "step": 39490
    },
    {
      "epoch": 0.11108580042240727,
      "grad_norm": 37.0,
      "learning_rate": 1.9932417762851014e-05,
      "loss": 8.2001,
      "step": 39500
    },
    {
      "epoch": 0.11111392340985599,
      "grad_norm": 35.75,
      "learning_rate": 1.9932383577419432e-05,
      "loss": 7.2401,
      "step": 39510
    },
    {
      "epoch": 0.1111420463973047,
      "grad_norm": 32.5,
      "learning_rate": 1.9932349383373278e-05,
      "loss": 7.6313,
      "step": 39520
    },
    {
      "epoch": 0.11117016938475341,
      "grad_norm": 65.5,
      "learning_rate": 1.9932315180712582e-05,
      "loss": 8.0156,
      "step": 39530
    },
    {
      "epoch": 0.11119829237220212,
      "grad_norm": 40.25,
      "learning_rate": 1.9932280969437378e-05,
      "loss": 8.2392,
      "step": 39540
    },
    {
      "epoch": 0.11122641535965082,
      "grad_norm": 26.375,
      "learning_rate": 1.9932246749547693e-05,
      "loss": 8.1649,
      "step": 39550
    },
    {
      "epoch": 0.11125453834709953,
      "grad_norm": 36.75,
      "learning_rate": 1.9932212521043554e-05,
      "loss": 8.0057,
      "step": 39560
    },
    {
      "epoch": 0.11128266133454824,
      "grad_norm": 31.375,
      "learning_rate": 1.9932178283924992e-05,
      "loss": 8.1593,
      "step": 39570
    },
    {
      "epoch": 0.11131078432199695,
      "grad_norm": 27.125,
      "learning_rate": 1.993214403819204e-05,
      "loss": 8.1344,
      "step": 39580
    },
    {
      "epoch": 0.11133890730944566,
      "grad_norm": 20.875,
      "learning_rate": 1.9932109783844723e-05,
      "loss": 7.6902,
      "step": 39590
    },
    {
      "epoch": 0.11136703029689438,
      "grad_norm": 35.25,
      "learning_rate": 1.9932075520883077e-05,
      "loss": 6.4047,
      "step": 39600
    },
    {
      "epoch": 0.11139515328434309,
      "grad_norm": 37.25,
      "learning_rate": 1.993204124930712e-05,
      "loss": 7.9469,
      "step": 39610
    },
    {
      "epoch": 0.1114232762717918,
      "grad_norm": 62.5,
      "learning_rate": 1.9932006969116898e-05,
      "loss": 6.9958,
      "step": 39620
    },
    {
      "epoch": 0.11145139925924051,
      "grad_norm": 24.0,
      "learning_rate": 1.993197268031243e-05,
      "loss": 7.4426,
      "step": 39630
    },
    {
      "epoch": 0.11147952224668922,
      "grad_norm": 27.625,
      "learning_rate": 1.9931938382893744e-05,
      "loss": 7.5213,
      "step": 39640
    },
    {
      "epoch": 0.11150764523413793,
      "grad_norm": 22.625,
      "learning_rate": 1.9931904076860877e-05,
      "loss": 8.2028,
      "step": 39650
    },
    {
      "epoch": 0.11153576822158665,
      "grad_norm": 34.25,
      "learning_rate": 1.9931869762213856e-05,
      "loss": 7.0324,
      "step": 39660
    },
    {
      "epoch": 0.11156389120903536,
      "grad_norm": 25.375,
      "learning_rate": 1.9931835438952707e-05,
      "loss": 8.0254,
      "step": 39670
    },
    {
      "epoch": 0.11159201419648407,
      "grad_norm": 40.75,
      "learning_rate": 1.9931801107077467e-05,
      "loss": 8.0834,
      "step": 39680
    },
    {
      "epoch": 0.11162013718393278,
      "grad_norm": 27.125,
      "learning_rate": 1.993176676658816e-05,
      "loss": 8.0478,
      "step": 39690
    },
    {
      "epoch": 0.11164826017138149,
      "grad_norm": 23.0,
      "learning_rate": 1.9931732417484818e-05,
      "loss": 7.8568,
      "step": 39700
    },
    {
      "epoch": 0.1116763831588302,
      "grad_norm": 35.5,
      "learning_rate": 1.993169805976747e-05,
      "loss": 8.0158,
      "step": 39710
    },
    {
      "epoch": 0.1117045061462789,
      "grad_norm": 28.75,
      "learning_rate": 1.9931663693436146e-05,
      "loss": 7.6441,
      "step": 39720
    },
    {
      "epoch": 0.11173262913372761,
      "grad_norm": 32.75,
      "learning_rate": 1.9931629318490876e-05,
      "loss": 8.2805,
      "step": 39730
    },
    {
      "epoch": 0.11176075212117632,
      "grad_norm": 27.375,
      "learning_rate": 1.9931594934931692e-05,
      "loss": 7.3246,
      "step": 39740
    },
    {
      "epoch": 0.11178887510862504,
      "grad_norm": 43.0,
      "learning_rate": 1.9931560542758616e-05,
      "loss": 7.8646,
      "step": 39750
    },
    {
      "epoch": 0.11181699809607375,
      "grad_norm": 26.125,
      "learning_rate": 1.9931526141971685e-05,
      "loss": 8.6715,
      "step": 39760
    },
    {
      "epoch": 0.11184512108352246,
      "grad_norm": 42.0,
      "learning_rate": 1.993149173257093e-05,
      "loss": 7.5787,
      "step": 39770
    },
    {
      "epoch": 0.11187324407097117,
      "grad_norm": 25.875,
      "learning_rate": 1.993145731455638e-05,
      "loss": 7.7623,
      "step": 39780
    },
    {
      "epoch": 0.11190136705841988,
      "grad_norm": 31.0,
      "learning_rate": 1.993142288792806e-05,
      "loss": 7.8898,
      "step": 39790
    },
    {
      "epoch": 0.1119294900458686,
      "grad_norm": 29.75,
      "learning_rate": 1.9931388452686002e-05,
      "loss": 7.6983,
      "step": 39800
    },
    {
      "epoch": 0.1119576130333173,
      "grad_norm": 33.25,
      "learning_rate": 1.9931354008830238e-05,
      "loss": 8.3527,
      "step": 39810
    },
    {
      "epoch": 0.11198573602076602,
      "grad_norm": 22.125,
      "learning_rate": 1.9931319556360797e-05,
      "loss": 8.0936,
      "step": 39820
    },
    {
      "epoch": 0.11201385900821473,
      "grad_norm": 22.625,
      "learning_rate": 1.9931285095277707e-05,
      "loss": 7.8523,
      "step": 39830
    },
    {
      "epoch": 0.11204198199566344,
      "grad_norm": 24.25,
      "learning_rate": 1.9931250625581e-05,
      "loss": 8.0918,
      "step": 39840
    },
    {
      "epoch": 0.11207010498311215,
      "grad_norm": 34.5,
      "learning_rate": 1.9931216147270705e-05,
      "loss": 8.219,
      "step": 39850
    },
    {
      "epoch": 0.11209822797056086,
      "grad_norm": 29.25,
      "learning_rate": 1.9931181660346853e-05,
      "loss": 7.4523,
      "step": 39860
    },
    {
      "epoch": 0.11212635095800957,
      "grad_norm": 22.625,
      "learning_rate": 1.993114716480947e-05,
      "loss": 7.7385,
      "step": 39870
    },
    {
      "epoch": 0.11215447394545827,
      "grad_norm": 25.375,
      "learning_rate": 1.9931112660658593e-05,
      "loss": 8.9317,
      "step": 39880
    },
    {
      "epoch": 0.11218259693290698,
      "grad_norm": 47.25,
      "learning_rate": 1.9931078147894246e-05,
      "loss": 7.8708,
      "step": 39890
    },
    {
      "epoch": 0.1122107199203557,
      "grad_norm": 22.875,
      "learning_rate": 1.993104362651646e-05,
      "loss": 7.8496,
      "step": 39900
    },
    {
      "epoch": 0.1122388429078044,
      "grad_norm": 35.5,
      "learning_rate": 1.993100909652527e-05,
      "loss": 7.8997,
      "step": 39910
    },
    {
      "epoch": 0.11226696589525312,
      "grad_norm": 21.125,
      "learning_rate": 1.9930974557920695e-05,
      "loss": 8.1541,
      "step": 39920
    },
    {
      "epoch": 0.11229508888270183,
      "grad_norm": 34.5,
      "learning_rate": 1.9930940010702775e-05,
      "loss": 8.8548,
      "step": 39930
    },
    {
      "epoch": 0.11232321187015054,
      "grad_norm": 25.125,
      "learning_rate": 1.993090545487154e-05,
      "loss": 8.0927,
      "step": 39940
    },
    {
      "epoch": 0.11235133485759925,
      "grad_norm": 28.75,
      "learning_rate": 1.993087089042701e-05,
      "loss": 7.729,
      "step": 39950
    },
    {
      "epoch": 0.11237945784504796,
      "grad_norm": 32.5,
      "learning_rate": 1.9930836317369226e-05,
      "loss": 7.6121,
      "step": 39960
    },
    {
      "epoch": 0.11240758083249668,
      "grad_norm": 31.5,
      "learning_rate": 1.9930801735698213e-05,
      "loss": 7.4255,
      "step": 39970
    },
    {
      "epoch": 0.11243570381994539,
      "grad_norm": 54.25,
      "learning_rate": 1.9930767145414e-05,
      "loss": 8.117,
      "step": 39980
    },
    {
      "epoch": 0.1124638268073941,
      "grad_norm": 26.25,
      "learning_rate": 1.993073254651662e-05,
      "loss": 7.7727,
      "step": 39990
    },
    {
      "epoch": 0.11249194979484281,
      "grad_norm": 25.75,
      "learning_rate": 1.99306979390061e-05,
      "loss": 7.6773,
      "step": 40000
    },
    {
      "epoch": 0.11252007278229152,
      "grad_norm": 36.0,
      "learning_rate": 1.9930663322882476e-05,
      "loss": 7.8418,
      "step": 40010
    },
    {
      "epoch": 0.11254819576974023,
      "grad_norm": 25.0,
      "learning_rate": 1.993062869814577e-05,
      "loss": 8.5814,
      "step": 40020
    },
    {
      "epoch": 0.11257631875718895,
      "grad_norm": 24.0,
      "learning_rate": 1.9930594064796013e-05,
      "loss": 7.6275,
      "step": 40030
    },
    {
      "epoch": 0.11260444174463764,
      "grad_norm": 39.75,
      "learning_rate": 1.9930559422833243e-05,
      "loss": 7.8084,
      "step": 40040
    },
    {
      "epoch": 0.11263256473208635,
      "grad_norm": 29.375,
      "learning_rate": 1.993052477225748e-05,
      "loss": 8.169,
      "step": 40050
    },
    {
      "epoch": 0.11266068771953507,
      "grad_norm": 32.0,
      "learning_rate": 1.9930490113068762e-05,
      "loss": 7.5766,
      "step": 40060
    },
    {
      "epoch": 0.11268881070698378,
      "grad_norm": 33.0,
      "learning_rate": 1.9930455445267113e-05,
      "loss": 7.9427,
      "step": 40070
    },
    {
      "epoch": 0.11271693369443249,
      "grad_norm": 31.5,
      "learning_rate": 1.9930420768852568e-05,
      "loss": 7.2257,
      "step": 40080
    },
    {
      "epoch": 0.1127450566818812,
      "grad_norm": 23.125,
      "learning_rate": 1.993038608382516e-05,
      "loss": 6.6947,
      "step": 40090
    },
    {
      "epoch": 0.11277317966932991,
      "grad_norm": 26.125,
      "learning_rate": 1.9930351390184905e-05,
      "loss": 8.1295,
      "step": 40100
    },
    {
      "epoch": 0.11280130265677862,
      "grad_norm": 31.0,
      "learning_rate": 1.9930316687931847e-05,
      "loss": 7.7295,
      "step": 40110
    },
    {
      "epoch": 0.11282942564422734,
      "grad_norm": 44.75,
      "learning_rate": 1.993028197706601e-05,
      "loss": 8.4976,
      "step": 40120
    },
    {
      "epoch": 0.11285754863167605,
      "grad_norm": 28.375,
      "learning_rate": 1.993024725758743e-05,
      "loss": 8.7573,
      "step": 40130
    },
    {
      "epoch": 0.11288567161912476,
      "grad_norm": 28.25,
      "learning_rate": 1.9930212529496122e-05,
      "loss": 8.0468,
      "step": 40140
    },
    {
      "epoch": 0.11291379460657347,
      "grad_norm": 44.5,
      "learning_rate": 1.9930177792792133e-05,
      "loss": 8.1924,
      "step": 40150
    },
    {
      "epoch": 0.11294191759402218,
      "grad_norm": 29.5,
      "learning_rate": 1.9930143047475487e-05,
      "loss": 7.9185,
      "step": 40160
    },
    {
      "epoch": 0.11297004058147089,
      "grad_norm": 30.875,
      "learning_rate": 1.9930108293546216e-05,
      "loss": 8.0364,
      "step": 40170
    },
    {
      "epoch": 0.1129981635689196,
      "grad_norm": 27.0,
      "learning_rate": 1.9930073531004345e-05,
      "loss": 7.8107,
      "step": 40180
    },
    {
      "epoch": 0.11302628655636832,
      "grad_norm": 34.5,
      "learning_rate": 1.9930038759849907e-05,
      "loss": 7.8012,
      "step": 40190
    },
    {
      "epoch": 0.11305440954381701,
      "grad_norm": 24.0,
      "learning_rate": 1.993000398008293e-05,
      "loss": 7.6009,
      "step": 40200
    },
    {
      "epoch": 0.11308253253126573,
      "grad_norm": 26.375,
      "learning_rate": 1.992996919170345e-05,
      "loss": 7.8997,
      "step": 40210
    },
    {
      "epoch": 0.11311065551871444,
      "grad_norm": 24.125,
      "learning_rate": 1.9929934394711492e-05,
      "loss": 8.0802,
      "step": 40220
    },
    {
      "epoch": 0.11313877850616315,
      "grad_norm": 45.5,
      "learning_rate": 1.992989958910709e-05,
      "loss": 7.4456,
      "step": 40230
    },
    {
      "epoch": 0.11316690149361186,
      "grad_norm": 28.625,
      "learning_rate": 1.992986477489027e-05,
      "loss": 8.2285,
      "step": 40240
    },
    {
      "epoch": 0.11319502448106057,
      "grad_norm": 24.75,
      "learning_rate": 1.9929829952061065e-05,
      "loss": 8.2793,
      "step": 40250
    },
    {
      "epoch": 0.11322314746850928,
      "grad_norm": 80.5,
      "learning_rate": 1.9929795120619504e-05,
      "loss": 7.9103,
      "step": 40260
    },
    {
      "epoch": 0.113251270455958,
      "grad_norm": 23.5,
      "learning_rate": 1.9929760280565617e-05,
      "loss": 8.1574,
      "step": 40270
    },
    {
      "epoch": 0.1132793934434067,
      "grad_norm": 32.25,
      "learning_rate": 1.9929725431899435e-05,
      "loss": 8.1167,
      "step": 40280
    },
    {
      "epoch": 0.11330751643085542,
      "grad_norm": 29.375,
      "learning_rate": 1.992969057462099e-05,
      "loss": 7.6598,
      "step": 40290
    },
    {
      "epoch": 0.11333563941830413,
      "grad_norm": 23.625,
      "learning_rate": 1.992965570873031e-05,
      "loss": 8.3943,
      "step": 40300
    },
    {
      "epoch": 0.11336376240575284,
      "grad_norm": 20.375,
      "learning_rate": 1.9929620834227423e-05,
      "loss": 8.2627,
      "step": 40310
    },
    {
      "epoch": 0.11339188539320155,
      "grad_norm": 31.25,
      "learning_rate": 1.9929585951112363e-05,
      "loss": 8.5344,
      "step": 40320
    },
    {
      "epoch": 0.11342000838065026,
      "grad_norm": 29.625,
      "learning_rate": 1.992955105938516e-05,
      "loss": 7.991,
      "step": 40330
    },
    {
      "epoch": 0.11344813136809898,
      "grad_norm": 24.375,
      "learning_rate": 1.992951615904584e-05,
      "loss": 7.4237,
      "step": 40340
    },
    {
      "epoch": 0.11347625435554769,
      "grad_norm": 32.25,
      "learning_rate": 1.992948125009444e-05,
      "loss": 7.5065,
      "step": 40350
    },
    {
      "epoch": 0.1135043773429964,
      "grad_norm": 29.0,
      "learning_rate": 1.9929446332530985e-05,
      "loss": 8.0293,
      "step": 40360
    },
    {
      "epoch": 0.1135325003304451,
      "grad_norm": 22.75,
      "learning_rate": 1.992941140635551e-05,
      "loss": 7.3145,
      "step": 40370
    },
    {
      "epoch": 0.11356062331789381,
      "grad_norm": 26.625,
      "learning_rate": 1.992937647156804e-05,
      "loss": 7.7182,
      "step": 40380
    },
    {
      "epoch": 0.11358874630534252,
      "grad_norm": 33.25,
      "learning_rate": 1.992934152816861e-05,
      "loss": 8.6878,
      "step": 40390
    },
    {
      "epoch": 0.11361686929279123,
      "grad_norm": 37.5,
      "learning_rate": 1.9929306576157244e-05,
      "loss": 8.2749,
      "step": 40400
    },
    {
      "epoch": 0.11364499228023994,
      "grad_norm": 25.375,
      "learning_rate": 1.9929271615533977e-05,
      "loss": 8.1686,
      "step": 40410
    },
    {
      "epoch": 0.11367311526768865,
      "grad_norm": 26.0,
      "learning_rate": 1.9929236646298845e-05,
      "loss": 8.0307,
      "step": 40420
    },
    {
      "epoch": 0.11370123825513737,
      "grad_norm": 32.75,
      "learning_rate": 1.9929201668451864e-05,
      "loss": 7.8786,
      "step": 40430
    },
    {
      "epoch": 0.11372936124258608,
      "grad_norm": 52.25,
      "learning_rate": 1.9929166681993076e-05,
      "loss": 7.8957,
      "step": 40440
    },
    {
      "epoch": 0.11375748423003479,
      "grad_norm": 30.75,
      "learning_rate": 1.9929131686922507e-05,
      "loss": 8.1903,
      "step": 40450
    },
    {
      "epoch": 0.1137856072174835,
      "grad_norm": 40.25,
      "learning_rate": 1.992909668324019e-05,
      "loss": 8.153,
      "step": 40460
    },
    {
      "epoch": 0.11381373020493221,
      "grad_norm": 22.5,
      "learning_rate": 1.992906167094615e-05,
      "loss": 8.0373,
      "step": 40470
    },
    {
      "epoch": 0.11384185319238092,
      "grad_norm": 35.5,
      "learning_rate": 1.9929026650040422e-05,
      "loss": 6.9035,
      "step": 40480
    },
    {
      "epoch": 0.11386997617982964,
      "grad_norm": 25.75,
      "learning_rate": 1.9928991620523036e-05,
      "loss": 7.6534,
      "step": 40490
    },
    {
      "epoch": 0.11389809916727835,
      "grad_norm": 29.375,
      "learning_rate": 1.992895658239402e-05,
      "loss": 8.3205,
      "step": 40500
    },
    {
      "epoch": 0.11392622215472706,
      "grad_norm": 23.25,
      "learning_rate": 1.992892153565341e-05,
      "loss": 7.5767,
      "step": 40510
    },
    {
      "epoch": 0.11395434514217577,
      "grad_norm": 28.25,
      "learning_rate": 1.9928886480301227e-05,
      "loss": 8.2087,
      "step": 40520
    },
    {
      "epoch": 0.11398246812962447,
      "grad_norm": 19.75,
      "learning_rate": 1.9928851416337508e-05,
      "loss": 8.4736,
      "step": 40530
    },
    {
      "epoch": 0.11401059111707318,
      "grad_norm": 35.5,
      "learning_rate": 1.9928816343762282e-05,
      "loss": 8.0914,
      "step": 40540
    },
    {
      "epoch": 0.11403871410452189,
      "grad_norm": 27.25,
      "learning_rate": 1.992878126257558e-05,
      "loss": 8.4895,
      "step": 40550
    },
    {
      "epoch": 0.1140668370919706,
      "grad_norm": 55.25,
      "learning_rate": 1.9928746172777432e-05,
      "loss": 7.4542,
      "step": 40560
    },
    {
      "epoch": 0.11409496007941931,
      "grad_norm": 30.75,
      "learning_rate": 1.992871107436787e-05,
      "loss": 7.4491,
      "step": 40570
    },
    {
      "epoch": 0.11412308306686803,
      "grad_norm": 67.0,
      "learning_rate": 1.9928675967346923e-05,
      "loss": 8.521,
      "step": 40580
    },
    {
      "epoch": 0.11415120605431674,
      "grad_norm": 26.625,
      "learning_rate": 1.992864085171462e-05,
      "loss": 8.2914,
      "step": 40590
    },
    {
      "epoch": 0.11417932904176545,
      "grad_norm": 21.375,
      "learning_rate": 1.9928605727470992e-05,
      "loss": 7.3769,
      "step": 40600
    },
    {
      "epoch": 0.11420745202921416,
      "grad_norm": 28.0,
      "learning_rate": 1.992857059461607e-05,
      "loss": 7.8811,
      "step": 40610
    },
    {
      "epoch": 0.11423557501666287,
      "grad_norm": 30.375,
      "learning_rate": 1.9928535453149887e-05,
      "loss": 7.9895,
      "step": 40620
    },
    {
      "epoch": 0.11426369800411158,
      "grad_norm": 20.625,
      "learning_rate": 1.992850030307247e-05,
      "loss": 8.1156,
      "step": 40630
    },
    {
      "epoch": 0.1142918209915603,
      "grad_norm": 31.625,
      "learning_rate": 1.9928465144383852e-05,
      "loss": 7.5317,
      "step": 40640
    },
    {
      "epoch": 0.114319943979009,
      "grad_norm": 39.25,
      "learning_rate": 1.9928429977084058e-05,
      "loss": 8.3036,
      "step": 40650
    },
    {
      "epoch": 0.11434806696645772,
      "grad_norm": 32.75,
      "learning_rate": 1.9928394801173124e-05,
      "loss": 7.7765,
      "step": 40660
    },
    {
      "epoch": 0.11437618995390643,
      "grad_norm": 25.625,
      "learning_rate": 1.992835961665108e-05,
      "loss": 8.5802,
      "step": 40670
    },
    {
      "epoch": 0.11440431294135514,
      "grad_norm": 38.0,
      "learning_rate": 1.9928324423517957e-05,
      "loss": 7.9659,
      "step": 40680
    },
    {
      "epoch": 0.11443243592880384,
      "grad_norm": 25.5,
      "learning_rate": 1.9928289221773785e-05,
      "loss": 7.1605,
      "step": 40690
    },
    {
      "epoch": 0.11446055891625255,
      "grad_norm": 38.25,
      "learning_rate": 1.992825401141859e-05,
      "loss": 7.8393,
      "step": 40700
    },
    {
      "epoch": 0.11448868190370126,
      "grad_norm": 36.5,
      "learning_rate": 1.992821879245241e-05,
      "loss": 7.6257,
      "step": 40710
    },
    {
      "epoch": 0.11451680489114997,
      "grad_norm": 27.875,
      "learning_rate": 1.9928183564875268e-05,
      "loss": 8.8567,
      "step": 40720
    },
    {
      "epoch": 0.11454492787859868,
      "grad_norm": 37.5,
      "learning_rate": 1.99281483286872e-05,
      "loss": 8.3439,
      "step": 40730
    },
    {
      "epoch": 0.1145730508660474,
      "grad_norm": 44.75,
      "learning_rate": 1.9928113083888238e-05,
      "loss": 8.1339,
      "step": 40740
    },
    {
      "epoch": 0.11460117385349611,
      "grad_norm": 27.125,
      "learning_rate": 1.9928077830478403e-05,
      "loss": 7.9466,
      "step": 40750
    },
    {
      "epoch": 0.11462929684094482,
      "grad_norm": 28.375,
      "learning_rate": 1.9928042568457737e-05,
      "loss": 7.523,
      "step": 40760
    },
    {
      "epoch": 0.11465741982839353,
      "grad_norm": 46.5,
      "learning_rate": 1.9928007297826263e-05,
      "loss": 7.9849,
      "step": 40770
    },
    {
      "epoch": 0.11468554281584224,
      "grad_norm": 28.0,
      "learning_rate": 1.9927972018584018e-05,
      "loss": 8.7341,
      "step": 40780
    },
    {
      "epoch": 0.11471366580329095,
      "grad_norm": 29.375,
      "learning_rate": 1.9927936730731025e-05,
      "loss": 7.8543,
      "step": 40790
    },
    {
      "epoch": 0.11474178879073967,
      "grad_norm": 26.75,
      "learning_rate": 1.992790143426732e-05,
      "loss": 8.2812,
      "step": 40800
    },
    {
      "epoch": 0.11476991177818838,
      "grad_norm": 26.625,
      "learning_rate": 1.992786612919293e-05,
      "loss": 7.7779,
      "step": 40810
    },
    {
      "epoch": 0.11479803476563709,
      "grad_norm": 26.0,
      "learning_rate": 1.992783081550789e-05,
      "loss": 7.4658,
      "step": 40820
    },
    {
      "epoch": 0.1148261577530858,
      "grad_norm": 29.5,
      "learning_rate": 1.992779549321223e-05,
      "loss": 8.4586,
      "step": 40830
    },
    {
      "epoch": 0.11485428074053451,
      "grad_norm": 35.25,
      "learning_rate": 1.9927760162305976e-05,
      "loss": 7.5136,
      "step": 40840
    },
    {
      "epoch": 0.11488240372798321,
      "grad_norm": 21.0,
      "learning_rate": 1.992772482278916e-05,
      "loss": 7.7615,
      "step": 40850
    },
    {
      "epoch": 0.11491052671543192,
      "grad_norm": 32.5,
      "learning_rate": 1.9927689474661816e-05,
      "loss": 7.246,
      "step": 40860
    },
    {
      "epoch": 0.11493864970288063,
      "grad_norm": 26.125,
      "learning_rate": 1.9927654117923974e-05,
      "loss": 8.2247,
      "step": 40870
    },
    {
      "epoch": 0.11496677269032934,
      "grad_norm": 31.375,
      "learning_rate": 1.9927618752575663e-05,
      "loss": 8.0046,
      "step": 40880
    },
    {
      "epoch": 0.11499489567777806,
      "grad_norm": 42.0,
      "learning_rate": 1.9927583378616916e-05,
      "loss": 7.6489,
      "step": 40890
    },
    {
      "epoch": 0.11502301866522677,
      "grad_norm": 41.75,
      "learning_rate": 1.992754799604776e-05,
      "loss": 7.8937,
      "step": 40900
    },
    {
      "epoch": 0.11505114165267548,
      "grad_norm": 39.5,
      "learning_rate": 1.9927512604868228e-05,
      "loss": 8.6107,
      "step": 40910
    },
    {
      "epoch": 0.11507926464012419,
      "grad_norm": 25.0,
      "learning_rate": 1.992747720507835e-05,
      "loss": 8.4515,
      "step": 40920
    },
    {
      "epoch": 0.1151073876275729,
      "grad_norm": 46.75,
      "learning_rate": 1.9927441796678154e-05,
      "loss": 8.0222,
      "step": 40930
    },
    {
      "epoch": 0.11513551061502161,
      "grad_norm": 27.25,
      "learning_rate": 1.9927406379667677e-05,
      "loss": 8.3719,
      "step": 40940
    },
    {
      "epoch": 0.11516363360247032,
      "grad_norm": 39.0,
      "learning_rate": 1.9927370954046947e-05,
      "loss": 7.949,
      "step": 40950
    },
    {
      "epoch": 0.11519175658991904,
      "grad_norm": 35.0,
      "learning_rate": 1.9927335519815992e-05,
      "loss": 7.863,
      "step": 40960
    },
    {
      "epoch": 0.11521987957736775,
      "grad_norm": 24.75,
      "learning_rate": 1.9927300076974846e-05,
      "loss": 8.3,
      "step": 40970
    },
    {
      "epoch": 0.11524800256481646,
      "grad_norm": 33.0,
      "learning_rate": 1.9927264625523538e-05,
      "loss": 8.0492,
      "step": 40980
    },
    {
      "epoch": 0.11527612555226517,
      "grad_norm": 26.5,
      "learning_rate": 1.9927229165462103e-05,
      "loss": 7.7945,
      "step": 40990
    },
    {
      "epoch": 0.11530424853971388,
      "grad_norm": 20.375,
      "learning_rate": 1.992719369679056e-05,
      "loss": 7.9621,
      "step": 41000
    },
    {
      "epoch": 0.1153323715271626,
      "grad_norm": 28.125,
      "learning_rate": 1.9927158219508954e-05,
      "loss": 7.8373,
      "step": 41010
    },
    {
      "epoch": 0.11536049451461129,
      "grad_norm": 43.25,
      "learning_rate": 1.9927122733617314e-05,
      "loss": 8.271,
      "step": 41020
    },
    {
      "epoch": 0.11538861750206,
      "grad_norm": 33.75,
      "learning_rate": 1.9927087239115658e-05,
      "loss": 7.6339,
      "step": 41030
    },
    {
      "epoch": 0.11541674048950872,
      "grad_norm": 28.5,
      "learning_rate": 1.9927051736004028e-05,
      "loss": 7.6038,
      "step": 41040
    },
    {
      "epoch": 0.11544486347695743,
      "grad_norm": 30.0,
      "learning_rate": 1.9927016224282454e-05,
      "loss": 7.4886,
      "step": 41050
    },
    {
      "epoch": 0.11547298646440614,
      "grad_norm": 24.375,
      "learning_rate": 1.9926980703950962e-05,
      "loss": 7.3445,
      "step": 41060
    },
    {
      "epoch": 0.11550110945185485,
      "grad_norm": 27.625,
      "learning_rate": 1.9926945175009587e-05,
      "loss": 7.8267,
      "step": 41070
    },
    {
      "epoch": 0.11552923243930356,
      "grad_norm": 27.875,
      "learning_rate": 1.9926909637458357e-05,
      "loss": 7.1621,
      "step": 41080
    },
    {
      "epoch": 0.11555735542675227,
      "grad_norm": 32.5,
      "learning_rate": 1.9926874091297307e-05,
      "loss": 8.0609,
      "step": 41090
    },
    {
      "epoch": 0.11558547841420098,
      "grad_norm": 21.75,
      "learning_rate": 1.9926838536526462e-05,
      "loss": 7.9371,
      "step": 41100
    },
    {
      "epoch": 0.1156136014016497,
      "grad_norm": 27.125,
      "learning_rate": 1.992680297314586e-05,
      "loss": 7.468,
      "step": 41110
    },
    {
      "epoch": 0.11564172438909841,
      "grad_norm": 48.75,
      "learning_rate": 1.9926767401155526e-05,
      "loss": 7.7543,
      "step": 41120
    },
    {
      "epoch": 0.11566984737654712,
      "grad_norm": 26.875,
      "learning_rate": 1.992673182055549e-05,
      "loss": 7.8023,
      "step": 41130
    },
    {
      "epoch": 0.11569797036399583,
      "grad_norm": 32.25,
      "learning_rate": 1.9926696231345785e-05,
      "loss": 7.9129,
      "step": 41140
    },
    {
      "epoch": 0.11572609335144454,
      "grad_norm": 36.0,
      "learning_rate": 1.9926660633526445e-05,
      "loss": 7.9453,
      "step": 41150
    },
    {
      "epoch": 0.11575421633889325,
      "grad_norm": 21.125,
      "learning_rate": 1.99266250270975e-05,
      "loss": 8.0921,
      "step": 41160
    },
    {
      "epoch": 0.11578233932634197,
      "grad_norm": 38.25,
      "learning_rate": 1.9926589412058974e-05,
      "loss": 7.6832,
      "step": 41170
    },
    {
      "epoch": 0.11581046231379066,
      "grad_norm": 47.25,
      "learning_rate": 1.9926553788410906e-05,
      "loss": 8.3778,
      "step": 41180
    },
    {
      "epoch": 0.11583858530123937,
      "grad_norm": 23.625,
      "learning_rate": 1.9926518156153328e-05,
      "loss": 7.8141,
      "step": 41190
    },
    {
      "epoch": 0.11586670828868809,
      "grad_norm": 26.75,
      "learning_rate": 1.992648251528626e-05,
      "loss": 7.8312,
      "step": 41200
    },
    {
      "epoch": 0.1158948312761368,
      "grad_norm": 22.875,
      "learning_rate": 1.9926446865809742e-05,
      "loss": 8.0516,
      "step": 41210
    },
    {
      "epoch": 0.11592295426358551,
      "grad_norm": 23.5,
      "learning_rate": 1.9926411207723805e-05,
      "loss": 8.4521,
      "step": 41220
    },
    {
      "epoch": 0.11595107725103422,
      "grad_norm": 25.5,
      "learning_rate": 1.9926375541028476e-05,
      "loss": 7.3135,
      "step": 41230
    },
    {
      "epoch": 0.11597920023848293,
      "grad_norm": 43.0,
      "learning_rate": 1.9926339865723786e-05,
      "loss": 7.5481,
      "step": 41240
    },
    {
      "epoch": 0.11600732322593164,
      "grad_norm": 38.75,
      "learning_rate": 1.9926304181809768e-05,
      "loss": 7.3959,
      "step": 41250
    },
    {
      "epoch": 0.11603544621338036,
      "grad_norm": 24.75,
      "learning_rate": 1.9926268489286453e-05,
      "loss": 7.3439,
      "step": 41260
    },
    {
      "epoch": 0.11606356920082907,
      "grad_norm": 36.0,
      "learning_rate": 1.992623278815387e-05,
      "loss": 8.1905,
      "step": 41270
    },
    {
      "epoch": 0.11609169218827778,
      "grad_norm": 36.0,
      "learning_rate": 1.9926197078412053e-05,
      "loss": 8.0033,
      "step": 41280
    },
    {
      "epoch": 0.11611981517572649,
      "grad_norm": 25.875,
      "learning_rate": 1.992616136006103e-05,
      "loss": 8.1932,
      "step": 41290
    },
    {
      "epoch": 0.1161479381631752,
      "grad_norm": 22.125,
      "learning_rate": 1.9926125633100835e-05,
      "loss": 7.7511,
      "step": 41300
    },
    {
      "epoch": 0.11617606115062391,
      "grad_norm": 30.75,
      "learning_rate": 1.9926089897531495e-05,
      "loss": 7.9502,
      "step": 41310
    },
    {
      "epoch": 0.11620418413807262,
      "grad_norm": 49.75,
      "learning_rate": 1.9926054153353045e-05,
      "loss": 8.1812,
      "step": 41320
    },
    {
      "epoch": 0.11623230712552134,
      "grad_norm": 27.875,
      "learning_rate": 1.992601840056551e-05,
      "loss": 7.9953,
      "step": 41330
    },
    {
      "epoch": 0.11626043011297003,
      "grad_norm": 21.75,
      "learning_rate": 1.9925982639168934e-05,
      "loss": 7.399,
      "step": 41340
    },
    {
      "epoch": 0.11628855310041875,
      "grad_norm": 30.0,
      "learning_rate": 1.9925946869163334e-05,
      "loss": 8.3527,
      "step": 41350
    },
    {
      "epoch": 0.11631667608786746,
      "grad_norm": 57.25,
      "learning_rate": 1.9925911090548746e-05,
      "loss": 8.0927,
      "step": 41360
    },
    {
      "epoch": 0.11634479907531617,
      "grad_norm": 29.125,
      "learning_rate": 1.9925875303325202e-05,
      "loss": 7.8971,
      "step": 41370
    },
    {
      "epoch": 0.11637292206276488,
      "grad_norm": 24.875,
      "learning_rate": 1.992583950749273e-05,
      "loss": 8.6742,
      "step": 41380
    },
    {
      "epoch": 0.11640104505021359,
      "grad_norm": 26.375,
      "learning_rate": 1.9925803703051366e-05,
      "loss": 7.686,
      "step": 41390
    },
    {
      "epoch": 0.1164291680376623,
      "grad_norm": 26.375,
      "learning_rate": 1.9925767890001138e-05,
      "loss": 7.2153,
      "step": 41400
    },
    {
      "epoch": 0.11645729102511101,
      "grad_norm": 32.75,
      "learning_rate": 1.992573206834208e-05,
      "loss": 8.8144,
      "step": 41410
    },
    {
      "epoch": 0.11648541401255973,
      "grad_norm": 23.375,
      "learning_rate": 1.9925696238074216e-05,
      "loss": 7.6394,
      "step": 41420
    },
    {
      "epoch": 0.11651353700000844,
      "grad_norm": 30.875,
      "learning_rate": 1.9925660399197584e-05,
      "loss": 7.4493,
      "step": 41430
    },
    {
      "epoch": 0.11654165998745715,
      "grad_norm": 31.0,
      "learning_rate": 1.9925624551712213e-05,
      "loss": 8.1826,
      "step": 41440
    },
    {
      "epoch": 0.11656978297490586,
      "grad_norm": 33.75,
      "learning_rate": 1.9925588695618132e-05,
      "loss": 7.5433,
      "step": 41450
    },
    {
      "epoch": 0.11659790596235457,
      "grad_norm": 45.0,
      "learning_rate": 1.9925552830915377e-05,
      "loss": 8.6094,
      "step": 41460
    },
    {
      "epoch": 0.11662602894980328,
      "grad_norm": 33.5,
      "learning_rate": 1.9925516957603972e-05,
      "loss": 7.1974,
      "step": 41470
    },
    {
      "epoch": 0.116654151937252,
      "grad_norm": 30.375,
      "learning_rate": 1.9925481075683955e-05,
      "loss": 7.5409,
      "step": 41480
    },
    {
      "epoch": 0.11668227492470071,
      "grad_norm": 39.5,
      "learning_rate": 1.992544518515535e-05,
      "loss": 8.2786,
      "step": 41490
    },
    {
      "epoch": 0.1167103979121494,
      "grad_norm": 31.0,
      "learning_rate": 1.99254092860182e-05,
      "loss": 7.5049,
      "step": 41500
    },
    {
      "epoch": 0.11673852089959812,
      "grad_norm": 26.625,
      "learning_rate": 1.9925373378272524e-05,
      "loss": 7.8198,
      "step": 41510
    },
    {
      "epoch": 0.11676664388704683,
      "grad_norm": 52.5,
      "learning_rate": 1.992533746191836e-05,
      "loss": 7.7909,
      "step": 41520
    },
    {
      "epoch": 0.11679476687449554,
      "grad_norm": 28.125,
      "learning_rate": 1.9925301536955734e-05,
      "loss": 7.5859,
      "step": 41530
    },
    {
      "epoch": 0.11682288986194425,
      "grad_norm": 37.0,
      "learning_rate": 1.992526560338468e-05,
      "loss": 7.2414,
      "step": 41540
    },
    {
      "epoch": 0.11685101284939296,
      "grad_norm": 25.75,
      "learning_rate": 1.992522966120523e-05,
      "loss": 7.469,
      "step": 41550
    },
    {
      "epoch": 0.11687913583684167,
      "grad_norm": 32.5,
      "learning_rate": 1.9925193710417417e-05,
      "loss": 7.614,
      "step": 41560
    },
    {
      "epoch": 0.11690725882429039,
      "grad_norm": 34.0,
      "learning_rate": 1.9925157751021267e-05,
      "loss": 7.964,
      "step": 41570
    },
    {
      "epoch": 0.1169353818117391,
      "grad_norm": 20.875,
      "learning_rate": 1.9925121783016814e-05,
      "loss": 7.3999,
      "step": 41580
    },
    {
      "epoch": 0.11696350479918781,
      "grad_norm": 22.5,
      "learning_rate": 1.9925085806404086e-05,
      "loss": 7.9171,
      "step": 41590
    },
    {
      "epoch": 0.11699162778663652,
      "grad_norm": 48.5,
      "learning_rate": 1.992504982118312e-05,
      "loss": 8.326,
      "step": 41600
    },
    {
      "epoch": 0.11701975077408523,
      "grad_norm": 32.75,
      "learning_rate": 1.9925013827353945e-05,
      "loss": 8.3404,
      "step": 41610
    },
    {
      "epoch": 0.11704787376153394,
      "grad_norm": 30.375,
      "learning_rate": 1.9924977824916592e-05,
      "loss": 8.084,
      "step": 41620
    },
    {
      "epoch": 0.11707599674898266,
      "grad_norm": 33.0,
      "learning_rate": 1.9924941813871087e-05,
      "loss": 7.9195,
      "step": 41630
    },
    {
      "epoch": 0.11710411973643137,
      "grad_norm": 51.25,
      "learning_rate": 1.992490579421747e-05,
      "loss": 7.9102,
      "step": 41640
    },
    {
      "epoch": 0.11713224272388008,
      "grad_norm": 35.25,
      "learning_rate": 1.992486976595577e-05,
      "loss": 8.1494,
      "step": 41650
    },
    {
      "epoch": 0.11716036571132879,
      "grad_norm": 26.875,
      "learning_rate": 1.9924833729086015e-05,
      "loss": 7.658,
      "step": 41660
    },
    {
      "epoch": 0.11718848869877749,
      "grad_norm": 25.75,
      "learning_rate": 1.9924797683608234e-05,
      "loss": 8.2779,
      "step": 41670
    },
    {
      "epoch": 0.1172166116862262,
      "grad_norm": 37.0,
      "learning_rate": 1.9924761629522465e-05,
      "loss": 8.0279,
      "step": 41680
    },
    {
      "epoch": 0.11724473467367491,
      "grad_norm": 30.625,
      "learning_rate": 1.9924725566828737e-05,
      "loss": 8.3131,
      "step": 41690
    },
    {
      "epoch": 0.11727285766112362,
      "grad_norm": 27.125,
      "learning_rate": 1.992468949552708e-05,
      "loss": 7.818,
      "step": 41700
    },
    {
      "epoch": 0.11730098064857233,
      "grad_norm": 23.75,
      "learning_rate": 1.9924653415617526e-05,
      "loss": 7.8327,
      "step": 41710
    },
    {
      "epoch": 0.11732910363602105,
      "grad_norm": 29.5,
      "learning_rate": 1.9924617327100108e-05,
      "loss": 7.4355,
      "step": 41720
    },
    {
      "epoch": 0.11735722662346976,
      "grad_norm": 31.625,
      "learning_rate": 1.992458122997485e-05,
      "loss": 8.3332,
      "step": 41730
    },
    {
      "epoch": 0.11738534961091847,
      "grad_norm": 32.75,
      "learning_rate": 1.992454512424179e-05,
      "loss": 8.936,
      "step": 41740
    },
    {
      "epoch": 0.11741347259836718,
      "grad_norm": 37.25,
      "learning_rate": 1.992450900990096e-05,
      "loss": 8.6034,
      "step": 41750
    },
    {
      "epoch": 0.11744159558581589,
      "grad_norm": 33.5,
      "learning_rate": 1.9924472886952393e-05,
      "loss": 9.0471,
      "step": 41760
    },
    {
      "epoch": 0.1174697185732646,
      "grad_norm": 25.875,
      "learning_rate": 1.9924436755396112e-05,
      "loss": 8.03,
      "step": 41770
    },
    {
      "epoch": 0.11749784156071331,
      "grad_norm": 30.125,
      "learning_rate": 1.9924400615232154e-05,
      "loss": 7.6247,
      "step": 41780
    },
    {
      "epoch": 0.11752596454816203,
      "grad_norm": 29.375,
      "learning_rate": 1.992436446646055e-05,
      "loss": 8.211,
      "step": 41790
    },
    {
      "epoch": 0.11755408753561074,
      "grad_norm": 54.25,
      "learning_rate": 1.9924328309081328e-05,
      "loss": 7.8395,
      "step": 41800
    },
    {
      "epoch": 0.11758221052305945,
      "grad_norm": 24.5,
      "learning_rate": 1.9924292143094527e-05,
      "loss": 8.0087,
      "step": 41810
    },
    {
      "epoch": 0.11761033351050816,
      "grad_norm": 33.0,
      "learning_rate": 1.992425596850017e-05,
      "loss": 7.6672,
      "step": 41820
    },
    {
      "epoch": 0.11763845649795686,
      "grad_norm": 29.375,
      "learning_rate": 1.9924219785298293e-05,
      "loss": 7.7278,
      "step": 41830
    },
    {
      "epoch": 0.11766657948540557,
      "grad_norm": 32.75,
      "learning_rate": 1.9924183593488923e-05,
      "loss": 7.6387,
      "step": 41840
    },
    {
      "epoch": 0.11769470247285428,
      "grad_norm": 32.5,
      "learning_rate": 1.99241473930721e-05,
      "loss": 8.3007,
      "step": 41850
    },
    {
      "epoch": 0.117722825460303,
      "grad_norm": 29.0,
      "learning_rate": 1.9924111184047847e-05,
      "loss": 8.5404,
      "step": 41860
    },
    {
      "epoch": 0.1177509484477517,
      "grad_norm": 32.5,
      "learning_rate": 1.9924074966416197e-05,
      "loss": 8.1067,
      "step": 41870
    },
    {
      "epoch": 0.11777907143520042,
      "grad_norm": 28.0,
      "learning_rate": 1.9924038740177185e-05,
      "loss": 7.7048,
      "step": 41880
    },
    {
      "epoch": 0.11780719442264913,
      "grad_norm": 28.125,
      "learning_rate": 1.992400250533084e-05,
      "loss": 8.0787,
      "step": 41890
    },
    {
      "epoch": 0.11783531741009784,
      "grad_norm": 21.375,
      "learning_rate": 1.992396626187719e-05,
      "loss": 8.2071,
      "step": 41900
    },
    {
      "epoch": 0.11786344039754655,
      "grad_norm": 24.75,
      "learning_rate": 1.9923930009816278e-05,
      "loss": 8.5741,
      "step": 41910
    },
    {
      "epoch": 0.11789156338499526,
      "grad_norm": 25.375,
      "learning_rate": 1.992389374914812e-05,
      "loss": 7.8593,
      "step": 41920
    },
    {
      "epoch": 0.11791968637244397,
      "grad_norm": 26.875,
      "learning_rate": 1.992385747987276e-05,
      "loss": 7.4798,
      "step": 41930
    },
    {
      "epoch": 0.11794780935989269,
      "grad_norm": 27.0,
      "learning_rate": 1.992382120199022e-05,
      "loss": 6.99,
      "step": 41940
    },
    {
      "epoch": 0.1179759323473414,
      "grad_norm": 37.25,
      "learning_rate": 1.992378491550054e-05,
      "loss": 7.0314,
      "step": 41950
    },
    {
      "epoch": 0.11800405533479011,
      "grad_norm": 28.5,
      "learning_rate": 1.9923748620403748e-05,
      "loss": 8.5149,
      "step": 41960
    },
    {
      "epoch": 0.11803217832223882,
      "grad_norm": 34.5,
      "learning_rate": 1.9923712316699873e-05,
      "loss": 8.0895,
      "step": 41970
    },
    {
      "epoch": 0.11806030130968753,
      "grad_norm": 28.75,
      "learning_rate": 1.9923676004388947e-05,
      "loss": 7.6601,
      "step": 41980
    },
    {
      "epoch": 0.11808842429713623,
      "grad_norm": 52.25,
      "learning_rate": 1.9923639683471004e-05,
      "loss": 7.2399,
      "step": 41990
    },
    {
      "epoch": 0.11811654728458494,
      "grad_norm": 28.25,
      "learning_rate": 1.9923603353946075e-05,
      "loss": 7.6709,
      "step": 42000
    },
    {
      "epoch": 0.11814467027203365,
      "grad_norm": 37.0,
      "learning_rate": 1.992356701581419e-05,
      "loss": 7.6057,
      "step": 42010
    },
    {
      "epoch": 0.11817279325948236,
      "grad_norm": 24.875,
      "learning_rate": 1.992353066907538e-05,
      "loss": 7.3972,
      "step": 42020
    },
    {
      "epoch": 0.11820091624693108,
      "grad_norm": 42.25,
      "learning_rate": 1.992349431372968e-05,
      "loss": 7.9869,
      "step": 42030
    },
    {
      "epoch": 0.11822903923437979,
      "grad_norm": 27.5,
      "learning_rate": 1.9923457949777118e-05,
      "loss": 7.9521,
      "step": 42040
    },
    {
      "epoch": 0.1182571622218285,
      "grad_norm": 48.75,
      "learning_rate": 1.992342157721773e-05,
      "loss": 7.7016,
      "step": 42050
    },
    {
      "epoch": 0.11828528520927721,
      "grad_norm": 22.125,
      "learning_rate": 1.992338519605154e-05,
      "loss": 7.7236,
      "step": 42060
    },
    {
      "epoch": 0.11831340819672592,
      "grad_norm": 25.75,
      "learning_rate": 1.9923348806278588e-05,
      "loss": 7.9081,
      "step": 42070
    },
    {
      "epoch": 0.11834153118417463,
      "grad_norm": 37.5,
      "learning_rate": 1.99233124078989e-05,
      "loss": 7.9078,
      "step": 42080
    },
    {
      "epoch": 0.11836965417162335,
      "grad_norm": 26.25,
      "learning_rate": 1.992327600091251e-05,
      "loss": 7.6559,
      "step": 42090
    },
    {
      "epoch": 0.11839777715907206,
      "grad_norm": 32.75,
      "learning_rate": 1.9923239585319447e-05,
      "loss": 8.2386,
      "step": 42100
    },
    {
      "epoch": 0.11842590014652077,
      "grad_norm": 31.625,
      "learning_rate": 1.9923203161119745e-05,
      "loss": 8.2743,
      "step": 42110
    },
    {
      "epoch": 0.11845402313396948,
      "grad_norm": 27.125,
      "learning_rate": 1.9923166728313436e-05,
      "loss": 8.4952,
      "step": 42120
    },
    {
      "epoch": 0.11848214612141819,
      "grad_norm": 26.5,
      "learning_rate": 1.992313028690055e-05,
      "loss": 7.0648,
      "step": 42130
    },
    {
      "epoch": 0.1185102691088669,
      "grad_norm": 26.125,
      "learning_rate": 1.992309383688112e-05,
      "loss": 7.5642,
      "step": 42140
    },
    {
      "epoch": 0.1185383920963156,
      "grad_norm": 33.25,
      "learning_rate": 1.9923057378255175e-05,
      "loss": 7.7963,
      "step": 42150
    },
    {
      "epoch": 0.11856651508376431,
      "grad_norm": 32.75,
      "learning_rate": 1.992302091102275e-05,
      "loss": 7.4376,
      "step": 42160
    },
    {
      "epoch": 0.11859463807121302,
      "grad_norm": 24.625,
      "learning_rate": 1.9922984435183873e-05,
      "loss": 8.1368,
      "step": 42170
    },
    {
      "epoch": 0.11862276105866174,
      "grad_norm": 23.875,
      "learning_rate": 1.992294795073858e-05,
      "loss": 7.7827,
      "step": 42180
    },
    {
      "epoch": 0.11865088404611045,
      "grad_norm": 26.5,
      "learning_rate": 1.99229114576869e-05,
      "loss": 7.9585,
      "step": 42190
    },
    {
      "epoch": 0.11867900703355916,
      "grad_norm": 43.75,
      "learning_rate": 1.9922874956028863e-05,
      "loss": 7.7376,
      "step": 42200
    },
    {
      "epoch": 0.11870713002100787,
      "grad_norm": 25.375,
      "learning_rate": 1.9922838445764504e-05,
      "loss": 8.3933,
      "step": 42210
    },
    {
      "epoch": 0.11873525300845658,
      "grad_norm": 40.75,
      "learning_rate": 1.9922801926893854e-05,
      "loss": 7.8417,
      "step": 42220
    },
    {
      "epoch": 0.1187633759959053,
      "grad_norm": 28.125,
      "learning_rate": 1.9922765399416942e-05,
      "loss": 7.7912,
      "step": 42230
    },
    {
      "epoch": 0.118791498983354,
      "grad_norm": 31.0,
      "learning_rate": 1.9922728863333803e-05,
      "loss": 8.3337,
      "step": 42240
    },
    {
      "epoch": 0.11881962197080272,
      "grad_norm": 26.5,
      "learning_rate": 1.9922692318644467e-05,
      "loss": 8.4044,
      "step": 42250
    },
    {
      "epoch": 0.11884774495825143,
      "grad_norm": 37.0,
      "learning_rate": 1.9922655765348968e-05,
      "loss": 7.9306,
      "step": 42260
    },
    {
      "epoch": 0.11887586794570014,
      "grad_norm": 28.75,
      "learning_rate": 1.9922619203447332e-05,
      "loss": 7.5269,
      "step": 42270
    },
    {
      "epoch": 0.11890399093314885,
      "grad_norm": 32.5,
      "learning_rate": 1.99225826329396e-05,
      "loss": 7.6651,
      "step": 42280
    },
    {
      "epoch": 0.11893211392059756,
      "grad_norm": 22.875,
      "learning_rate": 1.992254605382579e-05,
      "loss": 7.6543,
      "step": 42290
    },
    {
      "epoch": 0.11896023690804627,
      "grad_norm": 38.75,
      "learning_rate": 1.9922509466105947e-05,
      "loss": 8.4023,
      "step": 42300
    },
    {
      "epoch": 0.11898835989549497,
      "grad_norm": 30.25,
      "learning_rate": 1.99224728697801e-05,
      "loss": 8.0359,
      "step": 42310
    },
    {
      "epoch": 0.11901648288294368,
      "grad_norm": 50.0,
      "learning_rate": 1.9922436264848272e-05,
      "loss": 8.1061,
      "step": 42320
    },
    {
      "epoch": 0.1190446058703924,
      "grad_norm": 28.375,
      "learning_rate": 1.9922399651310506e-05,
      "loss": 7.6448,
      "step": 42330
    },
    {
      "epoch": 0.1190727288578411,
      "grad_norm": 22.125,
      "learning_rate": 1.9922363029166827e-05,
      "loss": 7.9373,
      "step": 42340
    },
    {
      "epoch": 0.11910085184528982,
      "grad_norm": 47.25,
      "learning_rate": 1.992232639841727e-05,
      "loss": 8.4119,
      "step": 42350
    },
    {
      "epoch": 0.11912897483273853,
      "grad_norm": 27.625,
      "learning_rate": 1.992228975906186e-05,
      "loss": 6.976,
      "step": 42360
    },
    {
      "epoch": 0.11915709782018724,
      "grad_norm": 31.625,
      "learning_rate": 1.992225311110064e-05,
      "loss": 7.878,
      "step": 42370
    },
    {
      "epoch": 0.11918522080763595,
      "grad_norm": 62.75,
      "learning_rate": 1.9922216454533635e-05,
      "loss": 8.2264,
      "step": 42380
    },
    {
      "epoch": 0.11921334379508466,
      "grad_norm": 40.0,
      "learning_rate": 1.992217978936088e-05,
      "loss": 7.9325,
      "step": 42390
    },
    {
      "epoch": 0.11924146678253338,
      "grad_norm": 34.0,
      "learning_rate": 1.99221431155824e-05,
      "loss": 8.1091,
      "step": 42400
    },
    {
      "epoch": 0.11926958976998209,
      "grad_norm": 35.0,
      "learning_rate": 1.9922106433198233e-05,
      "loss": 8.1991,
      "step": 42410
    },
    {
      "epoch": 0.1192977127574308,
      "grad_norm": 43.25,
      "learning_rate": 1.992206974220841e-05,
      "loss": 7.6998,
      "step": 42420
    },
    {
      "epoch": 0.11932583574487951,
      "grad_norm": 29.375,
      "learning_rate": 1.992203304261296e-05,
      "loss": 8.1427,
      "step": 42430
    },
    {
      "epoch": 0.11935395873232822,
      "grad_norm": 22.125,
      "learning_rate": 1.992199633441192e-05,
      "loss": 7.4231,
      "step": 42440
    },
    {
      "epoch": 0.11938208171977693,
      "grad_norm": 40.5,
      "learning_rate": 1.9921959617605316e-05,
      "loss": 8.3654,
      "step": 42450
    },
    {
      "epoch": 0.11941020470722565,
      "grad_norm": 22.875,
      "learning_rate": 1.9921922892193183e-05,
      "loss": 8.5591,
      "step": 42460
    },
    {
      "epoch": 0.11943832769467436,
      "grad_norm": 30.625,
      "learning_rate": 1.992188615817555e-05,
      "loss": 7.637,
      "step": 42470
    },
    {
      "epoch": 0.11946645068212305,
      "grad_norm": 48.75,
      "learning_rate": 1.9921849415552453e-05,
      "loss": 8.2896,
      "step": 42480
    },
    {
      "epoch": 0.11949457366957177,
      "grad_norm": 28.0,
      "learning_rate": 1.9921812664323928e-05,
      "loss": 7.3377,
      "step": 42490
    },
    {
      "epoch": 0.11952269665702048,
      "grad_norm": 42.25,
      "learning_rate": 1.9921775904489995e-05,
      "loss": 7.9522,
      "step": 42500
    },
    {
      "epoch": 0.11955081964446919,
      "grad_norm": 24.0,
      "learning_rate": 1.9921739136050692e-05,
      "loss": 7.8872,
      "step": 42510
    },
    {
      "epoch": 0.1195789426319179,
      "grad_norm": 76.0,
      "learning_rate": 1.9921702359006053e-05,
      "loss": 8.4018,
      "step": 42520
    },
    {
      "epoch": 0.11960706561936661,
      "grad_norm": 22.5,
      "learning_rate": 1.9921665573356104e-05,
      "loss": 8.0526,
      "step": 42530
    },
    {
      "epoch": 0.11963518860681532,
      "grad_norm": 41.75,
      "learning_rate": 1.9921628779100884e-05,
      "loss": 7.7303,
      "step": 42540
    },
    {
      "epoch": 0.11966331159426404,
      "grad_norm": 20.375,
      "learning_rate": 1.9921591976240418e-05,
      "loss": 7.2234,
      "step": 42550
    },
    {
      "epoch": 0.11969143458171275,
      "grad_norm": 37.0,
      "learning_rate": 1.9921555164774745e-05,
      "loss": 8.2108,
      "step": 42560
    },
    {
      "epoch": 0.11971955756916146,
      "grad_norm": 28.625,
      "learning_rate": 1.992151834470389e-05,
      "loss": 8.2093,
      "step": 42570
    },
    {
      "epoch": 0.11974768055661017,
      "grad_norm": 31.125,
      "learning_rate": 1.992148151602789e-05,
      "loss": 8.0135,
      "step": 42580
    },
    {
      "epoch": 0.11977580354405888,
      "grad_norm": 32.5,
      "learning_rate": 1.9921444678746778e-05,
      "loss": 8.3989,
      "step": 42590
    },
    {
      "epoch": 0.11980392653150759,
      "grad_norm": 31.125,
      "learning_rate": 1.992140783286058e-05,
      "loss": 8.0066,
      "step": 42600
    },
    {
      "epoch": 0.1198320495189563,
      "grad_norm": 35.75,
      "learning_rate": 1.992137097836933e-05,
      "loss": 8.2794,
      "step": 42610
    },
    {
      "epoch": 0.11986017250640502,
      "grad_norm": 48.25,
      "learning_rate": 1.9921334115273064e-05,
      "loss": 7.5298,
      "step": 42620
    },
    {
      "epoch": 0.11988829549385373,
      "grad_norm": 28.5,
      "learning_rate": 1.992129724357181e-05,
      "loss": 7.2895,
      "step": 42630
    },
    {
      "epoch": 0.11991641848130243,
      "grad_norm": 27.625,
      "learning_rate": 1.99212603632656e-05,
      "loss": 8.1777,
      "step": 42640
    },
    {
      "epoch": 0.11994454146875114,
      "grad_norm": 29.375,
      "learning_rate": 1.9921223474354467e-05,
      "loss": 7.1873,
      "step": 42650
    },
    {
      "epoch": 0.11997266445619985,
      "grad_norm": 38.75,
      "learning_rate": 1.9921186576838446e-05,
      "loss": 8.0276,
      "step": 42660
    },
    {
      "epoch": 0.12000078744364856,
      "grad_norm": 32.5,
      "learning_rate": 1.992114967071756e-05,
      "loss": 8.1141,
      "step": 42670
    },
    {
      "epoch": 0.12002891043109727,
      "grad_norm": 32.0,
      "learning_rate": 1.992111275599185e-05,
      "loss": 8.3714,
      "step": 42680
    },
    {
      "epoch": 0.12005703341854598,
      "grad_norm": 22.125,
      "learning_rate": 1.9921075832661346e-05,
      "loss": 8.3235,
      "step": 42690
    },
    {
      "epoch": 0.1200851564059947,
      "grad_norm": 27.0,
      "learning_rate": 1.992103890072608e-05,
      "loss": 8.3359,
      "step": 42700
    },
    {
      "epoch": 0.1201132793934434,
      "grad_norm": 23.125,
      "learning_rate": 1.992100196018608e-05,
      "loss": 6.9898,
      "step": 42710
    },
    {
      "epoch": 0.12014140238089212,
      "grad_norm": 24.125,
      "learning_rate": 1.9920965011041385e-05,
      "loss": 7.6583,
      "step": 42720
    },
    {
      "epoch": 0.12016952536834083,
      "grad_norm": 38.5,
      "learning_rate": 1.992092805329202e-05,
      "loss": 7.0192,
      "step": 42730
    },
    {
      "epoch": 0.12019764835578954,
      "grad_norm": 25.875,
      "learning_rate": 1.992089108693802e-05,
      "loss": 7.9117,
      "step": 42740
    },
    {
      "epoch": 0.12022577134323825,
      "grad_norm": 35.75,
      "learning_rate": 1.992085411197942e-05,
      "loss": 7.9449,
      "step": 42750
    },
    {
      "epoch": 0.12025389433068696,
      "grad_norm": 33.75,
      "learning_rate": 1.9920817128416247e-05,
      "loss": 8.2311,
      "step": 42760
    },
    {
      "epoch": 0.12028201731813568,
      "grad_norm": 39.0,
      "learning_rate": 1.9920780136248535e-05,
      "loss": 7.3514,
      "step": 42770
    },
    {
      "epoch": 0.12031014030558439,
      "grad_norm": 33.5,
      "learning_rate": 1.9920743135476318e-05,
      "loss": 8.6543,
      "step": 42780
    },
    {
      "epoch": 0.1203382632930331,
      "grad_norm": 38.5,
      "learning_rate": 1.9920706126099628e-05,
      "loss": 7.7001,
      "step": 42790
    },
    {
      "epoch": 0.1203663862804818,
      "grad_norm": 34.5,
      "learning_rate": 1.9920669108118493e-05,
      "loss": 7.618,
      "step": 42800
    },
    {
      "epoch": 0.12039450926793051,
      "grad_norm": 30.625,
      "learning_rate": 1.992063208153295e-05,
      "loss": 7.7473,
      "step": 42810
    },
    {
      "epoch": 0.12042263225537922,
      "grad_norm": 23.375,
      "learning_rate": 1.992059504634303e-05,
      "loss": 7.9674,
      "step": 42820
    },
    {
      "epoch": 0.12045075524282793,
      "grad_norm": 29.75,
      "learning_rate": 1.992055800254876e-05,
      "loss": 7.6763,
      "step": 42830
    },
    {
      "epoch": 0.12047887823027664,
      "grad_norm": 41.0,
      "learning_rate": 1.9920520950150174e-05,
      "loss": 7.8634,
      "step": 42840
    },
    {
      "epoch": 0.12050700121772535,
      "grad_norm": 19.125,
      "learning_rate": 1.9920483889147314e-05,
      "loss": 7.7972,
      "step": 42850
    },
    {
      "epoch": 0.12053512420517407,
      "grad_norm": 28.125,
      "learning_rate": 1.99204468195402e-05,
      "loss": 7.7158,
      "step": 42860
    },
    {
      "epoch": 0.12056324719262278,
      "grad_norm": 27.375,
      "learning_rate": 1.9920409741328868e-05,
      "loss": 8.3534,
      "step": 42870
    },
    {
      "epoch": 0.12059137018007149,
      "grad_norm": 50.0,
      "learning_rate": 1.9920372654513353e-05,
      "loss": 8.1732,
      "step": 42880
    },
    {
      "epoch": 0.1206194931675202,
      "grad_norm": 36.5,
      "learning_rate": 1.992033555909368e-05,
      "loss": 7.8201,
      "step": 42890
    },
    {
      "epoch": 0.12064761615496891,
      "grad_norm": 20.375,
      "learning_rate": 1.9920298455069892e-05,
      "loss": 8.4392,
      "step": 42900
    },
    {
      "epoch": 0.12067573914241762,
      "grad_norm": 22.25,
      "learning_rate": 1.9920261342442013e-05,
      "loss": 7.5271,
      "step": 42910
    },
    {
      "epoch": 0.12070386212986634,
      "grad_norm": 34.5,
      "learning_rate": 1.992022422121008e-05,
      "loss": 7.8112,
      "step": 42920
    },
    {
      "epoch": 0.12073198511731505,
      "grad_norm": 41.5,
      "learning_rate": 1.992018709137412e-05,
      "loss": 7.2164,
      "step": 42930
    },
    {
      "epoch": 0.12076010810476376,
      "grad_norm": 37.0,
      "learning_rate": 1.992014995293417e-05,
      "loss": 7.6638,
      "step": 42940
    },
    {
      "epoch": 0.12078823109221247,
      "grad_norm": 36.5,
      "learning_rate": 1.992011280589026e-05,
      "loss": 7.2362,
      "step": 42950
    },
    {
      "epoch": 0.12081635407966117,
      "grad_norm": 41.75,
      "learning_rate": 1.9920075650242417e-05,
      "loss": 8.114,
      "step": 42960
    },
    {
      "epoch": 0.12084447706710988,
      "grad_norm": 36.75,
      "learning_rate": 1.992003848599068e-05,
      "loss": 8.6004,
      "step": 42970
    },
    {
      "epoch": 0.12087260005455859,
      "grad_norm": 26.5,
      "learning_rate": 1.9920001313135082e-05,
      "loss": 8.5947,
      "step": 42980
    },
    {
      "epoch": 0.1209007230420073,
      "grad_norm": 18.25,
      "learning_rate": 1.9919964131675656e-05,
      "loss": 8.2641,
      "step": 42990
    },
    {
      "epoch": 0.12092884602945601,
      "grad_norm": 21.0,
      "learning_rate": 1.9919926941612426e-05,
      "loss": 7.8477,
      "step": 43000
    },
    {
      "epoch": 0.12095696901690473,
      "grad_norm": 21.0,
      "learning_rate": 1.991988974294543e-05,
      "loss": 7.9433,
      "step": 43010
    },
    {
      "epoch": 0.12098509200435344,
      "grad_norm": 51.5,
      "learning_rate": 1.9919852535674706e-05,
      "loss": 7.5675,
      "step": 43020
    },
    {
      "epoch": 0.12101321499180215,
      "grad_norm": 61.5,
      "learning_rate": 1.9919815319800274e-05,
      "loss": 7.9803,
      "step": 43030
    },
    {
      "epoch": 0.12104133797925086,
      "grad_norm": 34.0,
      "learning_rate": 1.9919778095322172e-05,
      "loss": 8.0937,
      "step": 43040
    },
    {
      "epoch": 0.12106946096669957,
      "grad_norm": 29.75,
      "learning_rate": 1.9919740862240436e-05,
      "loss": 7.818,
      "step": 43050
    },
    {
      "epoch": 0.12109758395414828,
      "grad_norm": 29.5,
      "learning_rate": 1.991970362055509e-05,
      "loss": 8.4118,
      "step": 43060
    },
    {
      "epoch": 0.121125706941597,
      "grad_norm": 43.0,
      "learning_rate": 1.9919666370266174e-05,
      "loss": 7.9424,
      "step": 43070
    },
    {
      "epoch": 0.1211538299290457,
      "grad_norm": 21.25,
      "learning_rate": 1.991962911137372e-05,
      "loss": 7.6413,
      "step": 43080
    },
    {
      "epoch": 0.12118195291649442,
      "grad_norm": 43.25,
      "learning_rate": 1.9919591843877753e-05,
      "loss": 7.8705,
      "step": 43090
    },
    {
      "epoch": 0.12121007590394313,
      "grad_norm": 39.5,
      "learning_rate": 1.9919554567778316e-05,
      "loss": 8.3751,
      "step": 43100
    },
    {
      "epoch": 0.12123819889139184,
      "grad_norm": 23.75,
      "learning_rate": 1.991951728307543e-05,
      "loss": 8.0245,
      "step": 43110
    },
    {
      "epoch": 0.12126632187884055,
      "grad_norm": 21.375,
      "learning_rate": 1.9919479989769136e-05,
      "loss": 8.1295,
      "step": 43120
    },
    {
      "epoch": 0.12129444486628925,
      "grad_norm": 37.75,
      "learning_rate": 1.9919442687859464e-05,
      "loss": 7.4788,
      "step": 43130
    },
    {
      "epoch": 0.12132256785373796,
      "grad_norm": 31.125,
      "learning_rate": 1.991940537734644e-05,
      "loss": 8.4124,
      "step": 43140
    },
    {
      "epoch": 0.12135069084118667,
      "grad_norm": 37.75,
      "learning_rate": 1.9919368058230107e-05,
      "loss": 8.1673,
      "step": 43150
    },
    {
      "epoch": 0.12137881382863538,
      "grad_norm": 36.25,
      "learning_rate": 1.991933073051049e-05,
      "loss": 8.0086,
      "step": 43160
    },
    {
      "epoch": 0.1214069368160841,
      "grad_norm": 32.0,
      "learning_rate": 1.9919293394187624e-05,
      "loss": 8.0508,
      "step": 43170
    },
    {
      "epoch": 0.12143505980353281,
      "grad_norm": 28.75,
      "learning_rate": 1.9919256049261543e-05,
      "loss": 7.8168,
      "step": 43180
    },
    {
      "epoch": 0.12146318279098152,
      "grad_norm": 33.75,
      "learning_rate": 1.9919218695732276e-05,
      "loss": 7.3593,
      "step": 43190
    },
    {
      "epoch": 0.12149130577843023,
      "grad_norm": 34.0,
      "learning_rate": 1.9919181333599856e-05,
      "loss": 8.2071,
      "step": 43200
    },
    {
      "epoch": 0.12151942876587894,
      "grad_norm": 27.25,
      "learning_rate": 1.9919143962864318e-05,
      "loss": 8.2301,
      "step": 43210
    },
    {
      "epoch": 0.12154755175332765,
      "grad_norm": 20.875,
      "learning_rate": 1.991910658352569e-05,
      "loss": 7.572,
      "step": 43220
    },
    {
      "epoch": 0.12157567474077637,
      "grad_norm": 45.75,
      "learning_rate": 1.991906919558401e-05,
      "loss": 7.3833,
      "step": 43230
    },
    {
      "epoch": 0.12160379772822508,
      "grad_norm": 30.625,
      "learning_rate": 1.9919031799039303e-05,
      "loss": 8.8688,
      "step": 43240
    },
    {
      "epoch": 0.12163192071567379,
      "grad_norm": 26.0,
      "learning_rate": 1.991899439389161e-05,
      "loss": 8.1349,
      "step": 43250
    },
    {
      "epoch": 0.1216600437031225,
      "grad_norm": 32.0,
      "learning_rate": 1.9918956980140956e-05,
      "loss": 8.0715,
      "step": 43260
    },
    {
      "epoch": 0.12168816669057121,
      "grad_norm": 37.25,
      "learning_rate": 1.991891955778738e-05,
      "loss": 7.9527,
      "step": 43270
    },
    {
      "epoch": 0.12171628967801992,
      "grad_norm": 30.375,
      "learning_rate": 1.9918882126830913e-05,
      "loss": 7.6584,
      "step": 43280
    },
    {
      "epoch": 0.12174441266546862,
      "grad_norm": 34.75,
      "learning_rate": 1.991884468727158e-05,
      "loss": 8.1153,
      "step": 43290
    },
    {
      "epoch": 0.12177253565291733,
      "grad_norm": 49.25,
      "learning_rate": 1.991880723910942e-05,
      "loss": 8.3004,
      "step": 43300
    },
    {
      "epoch": 0.12180065864036604,
      "grad_norm": 29.875,
      "learning_rate": 1.991876978234447e-05,
      "loss": 8.0323,
      "step": 43310
    },
    {
      "epoch": 0.12182878162781476,
      "grad_norm": 30.0,
      "learning_rate": 1.991873231697675e-05,
      "loss": 7.5858,
      "step": 43320
    },
    {
      "epoch": 0.12185690461526347,
      "grad_norm": 39.5,
      "learning_rate": 1.99186948430063e-05,
      "loss": 7.4457,
      "step": 43330
    },
    {
      "epoch": 0.12188502760271218,
      "grad_norm": 33.0,
      "learning_rate": 1.991865736043316e-05,
      "loss": 7.8812,
      "step": 43340
    },
    {
      "epoch": 0.12191315059016089,
      "grad_norm": 32.75,
      "learning_rate": 1.9918619869257348e-05,
      "loss": 7.5621,
      "step": 43350
    },
    {
      "epoch": 0.1219412735776096,
      "grad_norm": 27.375,
      "learning_rate": 1.9918582369478906e-05,
      "loss": 7.3778,
      "step": 43360
    },
    {
      "epoch": 0.12196939656505831,
      "grad_norm": 22.75,
      "learning_rate": 1.9918544861097864e-05,
      "loss": 8.5422,
      "step": 43370
    },
    {
      "epoch": 0.12199751955250702,
      "grad_norm": 27.375,
      "learning_rate": 1.991850734411425e-05,
      "loss": 7.2368,
      "step": 43380
    },
    {
      "epoch": 0.12202564253995574,
      "grad_norm": 40.25,
      "learning_rate": 1.9918469818528104e-05,
      "loss": 8.5729,
      "step": 43390
    },
    {
      "epoch": 0.12205376552740445,
      "grad_norm": 24.875,
      "learning_rate": 1.9918432284339456e-05,
      "loss": 7.8115,
      "step": 43400
    },
    {
      "epoch": 0.12208188851485316,
      "grad_norm": 53.0,
      "learning_rate": 1.991839474154834e-05,
      "loss": 8.5151,
      "step": 43410
    },
    {
      "epoch": 0.12211001150230187,
      "grad_norm": 27.125,
      "learning_rate": 1.991835719015478e-05,
      "loss": 7.097,
      "step": 43420
    },
    {
      "epoch": 0.12213813448975058,
      "grad_norm": 30.25,
      "learning_rate": 1.991831963015882e-05,
      "loss": 7.3283,
      "step": 43430
    },
    {
      "epoch": 0.1221662574771993,
      "grad_norm": 23.125,
      "learning_rate": 1.9918282061560484e-05,
      "loss": 7.653,
      "step": 43440
    },
    {
      "epoch": 0.12219438046464799,
      "grad_norm": 45.5,
      "learning_rate": 1.991824448435981e-05,
      "loss": 7.8538,
      "step": 43450
    },
    {
      "epoch": 0.1222225034520967,
      "grad_norm": 23.625,
      "learning_rate": 1.9918206898556828e-05,
      "loss": 8.9023,
      "step": 43460
    },
    {
      "epoch": 0.12225062643954542,
      "grad_norm": 29.875,
      "learning_rate": 1.9918169304151575e-05,
      "loss": 8.0306,
      "step": 43470
    },
    {
      "epoch": 0.12227874942699413,
      "grad_norm": 28.25,
      "learning_rate": 1.9918131701144075e-05,
      "loss": 7.9355,
      "step": 43480
    },
    {
      "epoch": 0.12230687241444284,
      "grad_norm": 28.625,
      "learning_rate": 1.991809408953437e-05,
      "loss": 7.4743,
      "step": 43490
    },
    {
      "epoch": 0.12233499540189155,
      "grad_norm": 46.75,
      "learning_rate": 1.9918056469322482e-05,
      "loss": 7.602,
      "step": 43500
    },
    {
      "epoch": 0.12236311838934026,
      "grad_norm": 25.125,
      "learning_rate": 1.9918018840508453e-05,
      "loss": 9.1103,
      "step": 43510
    },
    {
      "epoch": 0.12239124137678897,
      "grad_norm": 30.0,
      "learning_rate": 1.9917981203092313e-05,
      "loss": 7.5116,
      "step": 43520
    },
    {
      "epoch": 0.12241936436423768,
      "grad_norm": 24.125,
      "learning_rate": 1.9917943557074097e-05,
      "loss": 7.6431,
      "step": 43530
    },
    {
      "epoch": 0.1224474873516864,
      "grad_norm": 30.875,
      "learning_rate": 1.991790590245383e-05,
      "loss": 8.1252,
      "step": 43540
    },
    {
      "epoch": 0.12247561033913511,
      "grad_norm": 26.125,
      "learning_rate": 1.991786823923155e-05,
      "loss": 8.6345,
      "step": 43550
    },
    {
      "epoch": 0.12250373332658382,
      "grad_norm": 42.5,
      "learning_rate": 1.991783056740729e-05,
      "loss": 8.699,
      "step": 43560
    },
    {
      "epoch": 0.12253185631403253,
      "grad_norm": 34.75,
      "learning_rate": 1.9917792886981083e-05,
      "loss": 7.2963,
      "step": 43570
    },
    {
      "epoch": 0.12255997930148124,
      "grad_norm": 25.625,
      "learning_rate": 1.991775519795296e-05,
      "loss": 7.4507,
      "step": 43580
    },
    {
      "epoch": 0.12258810228892995,
      "grad_norm": 29.125,
      "learning_rate": 1.9917717500322952e-05,
      "loss": 7.2891,
      "step": 43590
    },
    {
      "epoch": 0.12261622527637867,
      "grad_norm": 24.625,
      "learning_rate": 1.9917679794091094e-05,
      "loss": 7.9408,
      "step": 43600
    },
    {
      "epoch": 0.12264434826382736,
      "grad_norm": 151.0,
      "learning_rate": 1.991764207925742e-05,
      "loss": 7.6819,
      "step": 43610
    },
    {
      "epoch": 0.12267247125127607,
      "grad_norm": 23.375,
      "learning_rate": 1.991760435582196e-05,
      "loss": 6.8868,
      "step": 43620
    },
    {
      "epoch": 0.12270059423872479,
      "grad_norm": 26.25,
      "learning_rate": 1.991756662378475e-05,
      "loss": 7.0039,
      "step": 43630
    },
    {
      "epoch": 0.1227287172261735,
      "grad_norm": 27.875,
      "learning_rate": 1.991752888314582e-05,
      "loss": 8.2015,
      "step": 43640
    },
    {
      "epoch": 0.12275684021362221,
      "grad_norm": 44.25,
      "learning_rate": 1.99174911339052e-05,
      "loss": 7.7878,
      "step": 43650
    },
    {
      "epoch": 0.12278496320107092,
      "grad_norm": 26.25,
      "learning_rate": 1.991745337606293e-05,
      "loss": 7.4298,
      "step": 43660
    },
    {
      "epoch": 0.12281308618851963,
      "grad_norm": 27.75,
      "learning_rate": 1.991741560961904e-05,
      "loss": 8.3192,
      "step": 43670
    },
    {
      "epoch": 0.12284120917596834,
      "grad_norm": 26.25,
      "learning_rate": 1.9917377834573557e-05,
      "loss": 8.4248,
      "step": 43680
    },
    {
      "epoch": 0.12286933216341706,
      "grad_norm": 25.625,
      "learning_rate": 1.991734005092652e-05,
      "loss": 8.1707,
      "step": 43690
    },
    {
      "epoch": 0.12289745515086577,
      "grad_norm": 22.0,
      "learning_rate": 1.9917302258677962e-05,
      "loss": 7.7223,
      "step": 43700
    },
    {
      "epoch": 0.12292557813831448,
      "grad_norm": 28.625,
      "learning_rate": 1.9917264457827913e-05,
      "loss": 8.0313,
      "step": 43710
    },
    {
      "epoch": 0.12295370112576319,
      "grad_norm": 21.875,
      "learning_rate": 1.9917226648376407e-05,
      "loss": 7.4001,
      "step": 43720
    },
    {
      "epoch": 0.1229818241132119,
      "grad_norm": 29.25,
      "learning_rate": 1.9917188830323474e-05,
      "loss": 8.0921,
      "step": 43730
    },
    {
      "epoch": 0.12300994710066061,
      "grad_norm": 22.875,
      "learning_rate": 1.9917151003669153e-05,
      "loss": 7.9055,
      "step": 43740
    },
    {
      "epoch": 0.12303807008810932,
      "grad_norm": 33.5,
      "learning_rate": 1.9917113168413472e-05,
      "loss": 7.8247,
      "step": 43750
    },
    {
      "epoch": 0.12306619307555804,
      "grad_norm": 38.5,
      "learning_rate": 1.9917075324556464e-05,
      "loss": 7.8692,
      "step": 43760
    },
    {
      "epoch": 0.12309431606300675,
      "grad_norm": 29.625,
      "learning_rate": 1.991703747209816e-05,
      "loss": 7.8179,
      "step": 43770
    },
    {
      "epoch": 0.12312243905045545,
      "grad_norm": 33.5,
      "learning_rate": 1.99169996110386e-05,
      "loss": 8.3988,
      "step": 43780
    },
    {
      "epoch": 0.12315056203790416,
      "grad_norm": 35.5,
      "learning_rate": 1.9916961741377814e-05,
      "loss": 7.8793,
      "step": 43790
    },
    {
      "epoch": 0.12317868502535287,
      "grad_norm": 29.75,
      "learning_rate": 1.9916923863115828e-05,
      "loss": 8.2983,
      "step": 43800
    },
    {
      "epoch": 0.12320680801280158,
      "grad_norm": 29.875,
      "learning_rate": 1.9916885976252683e-05,
      "loss": 8.2648,
      "step": 43810
    },
    {
      "epoch": 0.12323493100025029,
      "grad_norm": 28.625,
      "learning_rate": 1.991684808078841e-05,
      "loss": 8.0012,
      "step": 43820
    },
    {
      "epoch": 0.123263053987699,
      "grad_norm": 23.625,
      "learning_rate": 1.991681017672304e-05,
      "loss": 8.2324,
      "step": 43830
    },
    {
      "epoch": 0.12329117697514771,
      "grad_norm": 32.5,
      "learning_rate": 1.9916772264056604e-05,
      "loss": 8.0498,
      "step": 43840
    },
    {
      "epoch": 0.12331929996259643,
      "grad_norm": 43.0,
      "learning_rate": 1.991673434278914e-05,
      "loss": 8.2379,
      "step": 43850
    },
    {
      "epoch": 0.12334742295004514,
      "grad_norm": 27.25,
      "learning_rate": 1.991669641292068e-05,
      "loss": 8.0651,
      "step": 43860
    },
    {
      "epoch": 0.12337554593749385,
      "grad_norm": 27.125,
      "learning_rate": 1.9916658474451252e-05,
      "loss": 7.9731,
      "step": 43870
    },
    {
      "epoch": 0.12340366892494256,
      "grad_norm": 30.5,
      "learning_rate": 1.9916620527380893e-05,
      "loss": 7.6134,
      "step": 43880
    },
    {
      "epoch": 0.12343179191239127,
      "grad_norm": 22.5,
      "learning_rate": 1.991658257170964e-05,
      "loss": 8.1534,
      "step": 43890
    },
    {
      "epoch": 0.12345991489983998,
      "grad_norm": 21.875,
      "learning_rate": 1.9916544607437517e-05,
      "loss": 7.5378,
      "step": 43900
    },
    {
      "epoch": 0.1234880378872887,
      "grad_norm": 31.75,
      "learning_rate": 1.991650663456456e-05,
      "loss": 7.6263,
      "step": 43910
    },
    {
      "epoch": 0.12351616087473741,
      "grad_norm": 36.75,
      "learning_rate": 1.9916468653090802e-05,
      "loss": 7.3201,
      "step": 43920
    },
    {
      "epoch": 0.12354428386218612,
      "grad_norm": 47.0,
      "learning_rate": 1.9916430663016283e-05,
      "loss": 7.5229,
      "step": 43930
    },
    {
      "epoch": 0.12357240684963482,
      "grad_norm": 33.75,
      "learning_rate": 1.9916392664341028e-05,
      "loss": 7.271,
      "step": 43940
    },
    {
      "epoch": 0.12360052983708353,
      "grad_norm": 31.0,
      "learning_rate": 1.991635465706507e-05,
      "loss": 7.6172,
      "step": 43950
    },
    {
      "epoch": 0.12362865282453224,
      "grad_norm": 37.0,
      "learning_rate": 1.991631664118844e-05,
      "loss": 8.4038,
      "step": 43960
    },
    {
      "epoch": 0.12365677581198095,
      "grad_norm": 34.0,
      "learning_rate": 1.991627861671118e-05,
      "loss": 8.38,
      "step": 43970
    },
    {
      "epoch": 0.12368489879942966,
      "grad_norm": 25.75,
      "learning_rate": 1.9916240583633318e-05,
      "loss": 8.4333,
      "step": 43980
    },
    {
      "epoch": 0.12371302178687837,
      "grad_norm": 31.625,
      "learning_rate": 1.9916202541954885e-05,
      "loss": 7.6844,
      "step": 43990
    },
    {
      "epoch": 0.12374114477432709,
      "grad_norm": 26.25,
      "learning_rate": 1.9916164491675916e-05,
      "loss": 8.0323,
      "step": 44000
    },
    {
      "epoch": 0.1237692677617758,
      "grad_norm": 35.25,
      "learning_rate": 1.9916126432796443e-05,
      "loss": 8.7711,
      "step": 44010
    },
    {
      "epoch": 0.12379739074922451,
      "grad_norm": 22.0,
      "learning_rate": 1.9916088365316502e-05,
      "loss": 7.7043,
      "step": 44020
    },
    {
      "epoch": 0.12382551373667322,
      "grad_norm": 28.875,
      "learning_rate": 1.991605028923612e-05,
      "loss": 7.6165,
      "step": 44030
    },
    {
      "epoch": 0.12385363672412193,
      "grad_norm": 28.375,
      "learning_rate": 1.9916012204555336e-05,
      "loss": 8.4116,
      "step": 44040
    },
    {
      "epoch": 0.12388175971157064,
      "grad_norm": 35.0,
      "learning_rate": 1.9915974111274182e-05,
      "loss": 8.0699,
      "step": 44050
    },
    {
      "epoch": 0.12390988269901936,
      "grad_norm": 23.25,
      "learning_rate": 1.991593600939269e-05,
      "loss": 7.9621,
      "step": 44060
    },
    {
      "epoch": 0.12393800568646807,
      "grad_norm": 41.25,
      "learning_rate": 1.9915897898910888e-05,
      "loss": 7.5328,
      "step": 44070
    },
    {
      "epoch": 0.12396612867391678,
      "grad_norm": 35.0,
      "learning_rate": 1.9915859779828823e-05,
      "loss": 7.6328,
      "step": 44080
    },
    {
      "epoch": 0.12399425166136549,
      "grad_norm": 22.25,
      "learning_rate": 1.9915821652146513e-05,
      "loss": 7.4859,
      "step": 44090
    },
    {
      "epoch": 0.12402237464881419,
      "grad_norm": 35.75,
      "learning_rate": 1.9915783515863997e-05,
      "loss": 7.3922,
      "step": 44100
    },
    {
      "epoch": 0.1240504976362629,
      "grad_norm": 30.875,
      "learning_rate": 1.9915745370981308e-05,
      "loss": 7.8724,
      "step": 44110
    },
    {
      "epoch": 0.12407862062371161,
      "grad_norm": 24.25,
      "learning_rate": 1.991570721749848e-05,
      "loss": 7.9552,
      "step": 44120
    },
    {
      "epoch": 0.12410674361116032,
      "grad_norm": 33.75,
      "learning_rate": 1.9915669055415547e-05,
      "loss": 9.177,
      "step": 44130
    },
    {
      "epoch": 0.12413486659860903,
      "grad_norm": 67.0,
      "learning_rate": 1.991563088473254e-05,
      "loss": 8.2073,
      "step": 44140
    },
    {
      "epoch": 0.12416298958605775,
      "grad_norm": 22.625,
      "learning_rate": 1.991559270544949e-05,
      "loss": 7.975,
      "step": 44150
    },
    {
      "epoch": 0.12419111257350646,
      "grad_norm": 26.0,
      "learning_rate": 1.9915554517566432e-05,
      "loss": 8.4547,
      "step": 44160
    },
    {
      "epoch": 0.12421923556095517,
      "grad_norm": 32.0,
      "learning_rate": 1.9915516321083403e-05,
      "loss": 9.1469,
      "step": 44170
    },
    {
      "epoch": 0.12424735854840388,
      "grad_norm": 30.125,
      "learning_rate": 1.9915478116000433e-05,
      "loss": 7.6905,
      "step": 44180
    },
    {
      "epoch": 0.12427548153585259,
      "grad_norm": 31.0,
      "learning_rate": 1.9915439902317554e-05,
      "loss": 7.8882,
      "step": 44190
    },
    {
      "epoch": 0.1243036045233013,
      "grad_norm": 26.75,
      "learning_rate": 1.99154016800348e-05,
      "loss": 7.2601,
      "step": 44200
    },
    {
      "epoch": 0.12433172751075001,
      "grad_norm": 29.75,
      "learning_rate": 1.9915363449152205e-05,
      "loss": 7.9805,
      "step": 44210
    },
    {
      "epoch": 0.12435985049819873,
      "grad_norm": 21.75,
      "learning_rate": 1.9915325209669803e-05,
      "loss": 8.4179,
      "step": 44220
    },
    {
      "epoch": 0.12438797348564744,
      "grad_norm": 43.0,
      "learning_rate": 1.991528696158762e-05,
      "loss": 7.4565,
      "step": 44230
    },
    {
      "epoch": 0.12441609647309615,
      "grad_norm": 25.875,
      "learning_rate": 1.99152487049057e-05,
      "loss": 7.4887,
      "step": 44240
    },
    {
      "epoch": 0.12444421946054486,
      "grad_norm": 19.375,
      "learning_rate": 1.991521043962407e-05,
      "loss": 8.0544,
      "step": 44250
    },
    {
      "epoch": 0.12447234244799356,
      "grad_norm": 40.5,
      "learning_rate": 1.9915172165742764e-05,
      "loss": 7.8951,
      "step": 44260
    },
    {
      "epoch": 0.12450046543544227,
      "grad_norm": 46.5,
      "learning_rate": 1.9915133883261816e-05,
      "loss": 7.781,
      "step": 44270
    },
    {
      "epoch": 0.12452858842289098,
      "grad_norm": 31.75,
      "learning_rate": 1.991509559218126e-05,
      "loss": 8.1637,
      "step": 44280
    },
    {
      "epoch": 0.1245567114103397,
      "grad_norm": 25.0,
      "learning_rate": 1.9915057292501125e-05,
      "loss": 7.8485,
      "step": 44290
    },
    {
      "epoch": 0.1245848343977884,
      "grad_norm": 39.0,
      "learning_rate": 1.9915018984221447e-05,
      "loss": 8.0817,
      "step": 44300
    },
    {
      "epoch": 0.12461295738523712,
      "grad_norm": 36.25,
      "learning_rate": 1.991498066734226e-05,
      "loss": 8.2061,
      "step": 44310
    },
    {
      "epoch": 0.12464108037268583,
      "grad_norm": 37.25,
      "learning_rate": 1.9914942341863598e-05,
      "loss": 7.939,
      "step": 44320
    },
    {
      "epoch": 0.12466920336013454,
      "grad_norm": 31.25,
      "learning_rate": 1.991490400778549e-05,
      "loss": 8.3724,
      "step": 44330
    },
    {
      "epoch": 0.12469732634758325,
      "grad_norm": 28.875,
      "learning_rate": 1.9914865665107972e-05,
      "loss": 8.0691,
      "step": 44340
    },
    {
      "epoch": 0.12472544933503196,
      "grad_norm": 31.375,
      "learning_rate": 1.991482731383108e-05,
      "loss": 7.817,
      "step": 44350
    },
    {
      "epoch": 0.12475357232248067,
      "grad_norm": 31.25,
      "learning_rate": 1.9914788953954843e-05,
      "loss": 8.3676,
      "step": 44360
    },
    {
      "epoch": 0.12478169530992939,
      "grad_norm": 32.75,
      "learning_rate": 1.9914750585479298e-05,
      "loss": 8.0087,
      "step": 44370
    },
    {
      "epoch": 0.1248098182973781,
      "grad_norm": 26.75,
      "learning_rate": 1.991471220840447e-05,
      "loss": 7.7846,
      "step": 44380
    },
    {
      "epoch": 0.12483794128482681,
      "grad_norm": 26.625,
      "learning_rate": 1.9914673822730404e-05,
      "loss": 8.355,
      "step": 44390
    },
    {
      "epoch": 0.12486606427227552,
      "grad_norm": 23.25,
      "learning_rate": 1.9914635428457125e-05,
      "loss": 8.1914,
      "step": 44400
    },
    {
      "epoch": 0.12489418725972423,
      "grad_norm": 26.125,
      "learning_rate": 1.991459702558467e-05,
      "loss": 7.5201,
      "step": 44410
    },
    {
      "epoch": 0.12492231024717294,
      "grad_norm": 22.5,
      "learning_rate": 1.9914558614113072e-05,
      "loss": 7.3034,
      "step": 44420
    },
    {
      "epoch": 0.12495043323462164,
      "grad_norm": 23.625,
      "learning_rate": 1.9914520194042365e-05,
      "loss": 7.4072,
      "step": 44430
    },
    {
      "epoch": 0.12497855622207035,
      "grad_norm": 31.875,
      "learning_rate": 1.9914481765372576e-05,
      "loss": 7.3925,
      "step": 44440
    },
    {
      "epoch": 0.12500667920951908,
      "grad_norm": 36.0,
      "learning_rate": 1.9914443328103746e-05,
      "loss": 8.5892,
      "step": 44450
    },
    {
      "epoch": 0.12503480219696778,
      "grad_norm": 43.25,
      "learning_rate": 1.9914404882235905e-05,
      "loss": 8.1262,
      "step": 44460
    },
    {
      "epoch": 0.1250629251844165,
      "grad_norm": 27.875,
      "learning_rate": 1.991436642776909e-05,
      "loss": 8.3774,
      "step": 44470
    },
    {
      "epoch": 0.1250910481718652,
      "grad_norm": 22.625,
      "learning_rate": 1.9914327964703327e-05,
      "loss": 8.1674,
      "step": 44480
    },
    {
      "epoch": 0.12511917115931392,
      "grad_norm": 29.0,
      "learning_rate": 1.9914289493038654e-05,
      "loss": 8.1085,
      "step": 44490
    },
    {
      "epoch": 0.12514729414676262,
      "grad_norm": 67.0,
      "learning_rate": 1.9914251012775106e-05,
      "loss": 8.2124,
      "step": 44500
    },
    {
      "epoch": 0.12517541713421132,
      "grad_norm": 29.375,
      "learning_rate": 1.9914212523912713e-05,
      "loss": 7.7332,
      "step": 44510
    },
    {
      "epoch": 0.12520354012166005,
      "grad_norm": 44.75,
      "learning_rate": 1.991417402645151e-05,
      "loss": 7.2597,
      "step": 44520
    },
    {
      "epoch": 0.12523166310910874,
      "grad_norm": 30.125,
      "learning_rate": 1.9914135520391532e-05,
      "loss": 8.8132,
      "step": 44530
    },
    {
      "epoch": 0.12525978609655747,
      "grad_norm": 32.5,
      "learning_rate": 1.991409700573281e-05,
      "loss": 8.173,
      "step": 44540
    },
    {
      "epoch": 0.12528790908400617,
      "grad_norm": 34.75,
      "learning_rate": 1.9914058482475376e-05,
      "loss": 7.7879,
      "step": 44550
    },
    {
      "epoch": 0.1253160320714549,
      "grad_norm": 30.25,
      "learning_rate": 1.9914019950619264e-05,
      "loss": 8.2806,
      "step": 44560
    },
    {
      "epoch": 0.1253441550589036,
      "grad_norm": 37.0,
      "learning_rate": 1.991398141016451e-05,
      "loss": 8.5648,
      "step": 44570
    },
    {
      "epoch": 0.12537227804635231,
      "grad_norm": 31.875,
      "learning_rate": 1.9913942861111147e-05,
      "loss": 8.1016,
      "step": 44580
    },
    {
      "epoch": 0.125400401033801,
      "grad_norm": 36.75,
      "learning_rate": 1.991390430345921e-05,
      "loss": 7.4242,
      "step": 44590
    },
    {
      "epoch": 0.12542852402124974,
      "grad_norm": 24.625,
      "learning_rate": 1.9913865737208726e-05,
      "loss": 7.926,
      "step": 44600
    },
    {
      "epoch": 0.12545664700869844,
      "grad_norm": 24.75,
      "learning_rate": 1.9913827162359736e-05,
      "loss": 8.4242,
      "step": 44610
    },
    {
      "epoch": 0.12548476999614716,
      "grad_norm": 35.0,
      "learning_rate": 1.9913788578912267e-05,
      "loss": 7.8015,
      "step": 44620
    },
    {
      "epoch": 0.12551289298359586,
      "grad_norm": 23.875,
      "learning_rate": 1.9913749986866357e-05,
      "loss": 8.2369,
      "step": 44630
    },
    {
      "epoch": 0.12554101597104458,
      "grad_norm": 34.5,
      "learning_rate": 1.9913711386222035e-05,
      "loss": 8.1834,
      "step": 44640
    },
    {
      "epoch": 0.12556913895849328,
      "grad_norm": 21.875,
      "learning_rate": 1.9913672776979344e-05,
      "loss": 8.241,
      "step": 44650
    },
    {
      "epoch": 0.125597261945942,
      "grad_norm": 22.375,
      "learning_rate": 1.9913634159138305e-05,
      "loss": 7.342,
      "step": 44660
    },
    {
      "epoch": 0.1256253849333907,
      "grad_norm": 28.25,
      "learning_rate": 1.9913595532698958e-05,
      "loss": 7.6523,
      "step": 44670
    },
    {
      "epoch": 0.1256535079208394,
      "grad_norm": 32.25,
      "learning_rate": 1.9913556897661338e-05,
      "loss": 8.2159,
      "step": 44680
    },
    {
      "epoch": 0.12568163090828813,
      "grad_norm": 33.25,
      "learning_rate": 1.9913518254025477e-05,
      "loss": 7.8621,
      "step": 44690
    },
    {
      "epoch": 0.12570975389573683,
      "grad_norm": 26.75,
      "learning_rate": 1.9913479601791404e-05,
      "loss": 7.1099,
      "step": 44700
    },
    {
      "epoch": 0.12573787688318555,
      "grad_norm": 28.25,
      "learning_rate": 1.9913440940959158e-05,
      "loss": 7.7682,
      "step": 44710
    },
    {
      "epoch": 0.12576599987063425,
      "grad_norm": 30.125,
      "learning_rate": 1.991340227152877e-05,
      "loss": 8.1713,
      "step": 44720
    },
    {
      "epoch": 0.12579412285808297,
      "grad_norm": 37.25,
      "learning_rate": 1.991336359350028e-05,
      "loss": 7.5024,
      "step": 44730
    },
    {
      "epoch": 0.12582224584553167,
      "grad_norm": 70.5,
      "learning_rate": 1.991332490687371e-05,
      "loss": 7.9639,
      "step": 44740
    },
    {
      "epoch": 0.1258503688329804,
      "grad_norm": 31.625,
      "learning_rate": 1.9913286211649105e-05,
      "loss": 7.2957,
      "step": 44750
    },
    {
      "epoch": 0.1258784918204291,
      "grad_norm": 24.125,
      "learning_rate": 1.9913247507826487e-05,
      "loss": 7.6057,
      "step": 44760
    },
    {
      "epoch": 0.12590661480787782,
      "grad_norm": 23.25,
      "learning_rate": 1.99132087954059e-05,
      "loss": 7.565,
      "step": 44770
    },
    {
      "epoch": 0.12593473779532652,
      "grad_norm": 36.75,
      "learning_rate": 1.991317007438737e-05,
      "loss": 7.8619,
      "step": 44780
    },
    {
      "epoch": 0.12596286078277524,
      "grad_norm": 32.25,
      "learning_rate": 1.9913131344770937e-05,
      "loss": 7.9224,
      "step": 44790
    },
    {
      "epoch": 0.12599098377022394,
      "grad_norm": 33.25,
      "learning_rate": 1.991309260655663e-05,
      "loss": 7.7658,
      "step": 44800
    },
    {
      "epoch": 0.12601910675767267,
      "grad_norm": 29.5,
      "learning_rate": 1.9913053859744485e-05,
      "loss": 7.4781,
      "step": 44810
    },
    {
      "epoch": 0.12604722974512136,
      "grad_norm": 25.125,
      "learning_rate": 1.9913015104334535e-05,
      "loss": 7.9938,
      "step": 44820
    },
    {
      "epoch": 0.12607535273257006,
      "grad_norm": 26.25,
      "learning_rate": 1.991297634032681e-05,
      "loss": 7.9185,
      "step": 44830
    },
    {
      "epoch": 0.1261034757200188,
      "grad_norm": 23.75,
      "learning_rate": 1.9912937567721348e-05,
      "loss": 7.5742,
      "step": 44840
    },
    {
      "epoch": 0.12613159870746748,
      "grad_norm": 27.375,
      "learning_rate": 1.9912898786518184e-05,
      "loss": 8.0946,
      "step": 44850
    },
    {
      "epoch": 0.1261597216949162,
      "grad_norm": 29.25,
      "learning_rate": 1.991285999671735e-05,
      "loss": 8.0335,
      "step": 44860
    },
    {
      "epoch": 0.1261878446823649,
      "grad_norm": 60.0,
      "learning_rate": 1.9912821198318874e-05,
      "loss": 8.1619,
      "step": 44870
    },
    {
      "epoch": 0.12621596766981363,
      "grad_norm": 23.875,
      "learning_rate": 1.99127823913228e-05,
      "loss": 7.6644,
      "step": 44880
    },
    {
      "epoch": 0.12624409065726233,
      "grad_norm": 32.5,
      "learning_rate": 1.9912743575729152e-05,
      "loss": 8.5606,
      "step": 44890
    },
    {
      "epoch": 0.12627221364471106,
      "grad_norm": 27.875,
      "learning_rate": 1.991270475153797e-05,
      "loss": 7.9702,
      "step": 44900
    },
    {
      "epoch": 0.12630033663215975,
      "grad_norm": 29.5,
      "learning_rate": 1.9912665918749283e-05,
      "loss": 7.9345,
      "step": 44910
    },
    {
      "epoch": 0.12632845961960848,
      "grad_norm": 42.0,
      "learning_rate": 1.9912627077363128e-05,
      "loss": 7.3068,
      "step": 44920
    },
    {
      "epoch": 0.12635658260705718,
      "grad_norm": 39.25,
      "learning_rate": 1.9912588227379537e-05,
      "loss": 7.3179,
      "step": 44930
    },
    {
      "epoch": 0.1263847055945059,
      "grad_norm": 41.25,
      "learning_rate": 1.991254936879855e-05,
      "loss": 8.3695,
      "step": 44940
    },
    {
      "epoch": 0.1264128285819546,
      "grad_norm": 24.375,
      "learning_rate": 1.991251050162019e-05,
      "loss": 7.7571,
      "step": 44950
    },
    {
      "epoch": 0.12644095156940333,
      "grad_norm": 28.75,
      "learning_rate": 1.9912471625844496e-05,
      "loss": 7.386,
      "step": 44960
    },
    {
      "epoch": 0.12646907455685202,
      "grad_norm": 27.5,
      "learning_rate": 1.99124327414715e-05,
      "loss": 8.2161,
      "step": 44970
    },
    {
      "epoch": 0.12649719754430075,
      "grad_norm": 25.75,
      "learning_rate": 1.9912393848501242e-05,
      "loss": 8.0966,
      "step": 44980
    },
    {
      "epoch": 0.12652532053174945,
      "grad_norm": 33.0,
      "learning_rate": 1.9912354946933748e-05,
      "loss": 7.6225,
      "step": 44990
    },
    {
      "epoch": 0.12655344351919814,
      "grad_norm": 32.5,
      "learning_rate": 1.9912316036769056e-05,
      "loss": 8.0792,
      "step": 45000
    }
  ],
  "logging_steps": 10,
  "max_steps": 1066743,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 5.300677993447987e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}