{
  "best_metric": 2.318600654602051,
  "best_model_checkpoint": "/data4/share_nlp/data/luannd/78.52.project/weight_saving/PoetGPT_vietnamese_with_deepspeed_v0/checkpoint-7744",
  "epoch": 11.0,
  "global_step": 7744,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 1.3499999999999998e-05,
      "loss": 5.9125,
      "step": 10
    },
    {
      "epoch": 0.03,
      "learning_rate": 2.8499999999999998e-05,
      "loss": 5.6567,
      "step": 20
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.3499999999999993e-05,
      "loss": 5.3131,
      "step": 30
    },
    {
      "epoch": 0.06,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 4.8718,
      "step": 40
    },
    {
      "epoch": 0.07,
      "learning_rate": 7.199999999999999e-05,
      "loss": 4.3081,
      "step": 50
    },
    {
      "epoch": 0.09,
      "learning_rate": 8.699999999999999e-05,
      "loss": 3.6688,
      "step": 60
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.000102,
      "loss": 3.3306,
      "step": 70
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.000117,
      "loss": 3.2425,
      "step": 80
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.00013199999999999998,
      "loss": 3.1231,
      "step": 90
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.000147,
      "loss": 3.1498,
      "step": 100
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.000162,
      "loss": 3.0767,
      "step": 110
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.00017699999999999997,
      "loss": 3.0715,
      "step": 120
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00019199999999999998,
      "loss": 3.0144,
      "step": 130
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00020699999999999996,
      "loss": 3.0438,
      "step": 140
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00022199999999999998,
      "loss": 2.9902,
      "step": 150
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.000237,
      "loss": 2.9502,
      "step": 160
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.00025199999999999995,
      "loss": 2.9778,
      "step": 170
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.000267,
      "loss": 2.9055,
      "step": 180
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00028199999999999997,
      "loss": 2.8714,
      "step": 190
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00029699999999999996,
      "loss": 2.8467,
      "step": 200
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0002998135381828383,
      "loss": 2.846,
      "step": 210
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.0002990568314839864,
      "loss": 2.8177,
      "step": 220
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0002977211629518312,
      "loss": 2.8653,
      "step": 230
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00029581172054786616,
      "loss": 2.8117,
      "step": 240
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.00029333592086792107,
      "loss": 2.8291,
      "step": 250
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0002903033803348551,
      "loss": 2.797,
      "step": 260
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00028672587784675096,
      "loss": 2.7837,
      "step": 270
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00028261730902569146,
      "loss": 2.7499,
      "step": 280
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0002779936322448233,
      "loss": 2.7736,
      "step": 290
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00027287280664334875,
      "loss": 2.7529,
      "step": 300
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.00026727472237020447,
      "loss": 2.7289,
      "step": 310
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.000261221123327374,
      "loss": 2.7221,
      "step": 320
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0002547355227129109,
      "loss": 2.7534,
      "step": 330
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00024784311169171814,
      "loss": 2.7273,
      "step": 340
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0002405706615488216,
      "loss": 2.705,
      "step": 350
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0002329464197051909,
      "loss": 2.6387,
      "step": 360
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.000225,
      "loss": 2.736,
      "step": 370
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00021676226766548882,
      "loss": 2.688,
      "step": 380
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0002082652194412042,
      "loss": 2.7384,
      "step": 390
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00019954185929327506,
      "loss": 2.6948,
      "step": 400
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00019062607022145078,
      "loss": 2.7189,
      "step": 410
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00018155248265182435,
      "loss": 2.6816,
      "step": 420
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00017235633992642615,
      "loss": 2.7302,
      "step": 430
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00016307336141214873,
      "loss": 2.6641,
      "step": 440
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00015373960376071093,
      "loss": 2.7036,
      "step": 450
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00014439132085855116,
      "loss": 2.6852,
      "step": 460
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.0001350648230106275,
      "loss": 2.6867,
      "step": 470
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.000125796335905079,
      "loss": 2.7016,
      "step": 480
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00011662185990655284,
      "loss": 2.7191,
      "step": 490
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00010757703022472587,
      "loss": 2.7024,
      "step": 500
    },
    {
      "epoch": 0.72,
      "learning_rate": 9.869697850114969e-05,
      "loss": 2.6439,
      "step": 510
    },
    {
      "epoch": 0.74,
      "learning_rate": 9.001619635203888e-05,
      "loss": 2.7095,
      "step": 520
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.156840139702554e-05,
      "loss": 2.684,
      "step": 530
    },
    {
      "epoch": 0.77,
      "learning_rate": 7.33864062942472e-05,
      "loss": 2.6504,
      "step": 540
    },
    {
      "epoch": 0.78,
      "learning_rate": 6.550199129045668e-05,
      "loss": 2.6889,
      "step": 550
    },
    {
      "epoch": 0.8,
      "learning_rate": 5.794578078119291e-05,
      "loss": 2.6694,
      "step": 560
    },
    {
      "epoch": 0.81,
      "learning_rate": 5.074712436047112e-05,
      "loss": 2.669,
      "step": 570
    },
    {
      "epoch": 0.82,
      "learning_rate": 4.3933982822017876e-05,
      "loss": 2.661,
      "step": 580
    },
    {
      "epoch": 0.84,
      "learning_rate": 3.753281955483985e-05,
      "loss": 2.6576,
      "step": 590
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.15684977549647e-05,
      "loss": 2.665,
      "step": 600
    },
    {
      "epoch": 0.87,
      "learning_rate": 2.6064183852600797e-05,
      "loss": 2.6302,
      "step": 610
    },
    {
      "epoch": 0.88,
      "learning_rate": 2.1041257529821453e-05,
      "loss": 2.7063,
      "step": 620
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6519228678279718e-05,
      "loss": 2.6988,
      "step": 630
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.251566161950357e-05,
      "loss": 2.6588,
      "step": 640
    },
    {
      "epoch": 0.92,
      "learning_rate": 9.046106882113751e-06,
      "loss": 2.6728,
      "step": 650
    },
    {
      "epoch": 0.94,
      "learning_rate": 6.1240408009518346e-06,
      "loss": 2.6478,
      "step": 660
    },
    {
      "epoch": 0.95,
      "learning_rate": 3.760813172726457e-06,
      "loss": 2.6409,
      "step": 670
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.9656031714918365e-06,
      "loss": 2.73,
      "step": 680
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.453836951897885e-07,
      "loss": 2.6736,
      "step": 690
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.0489428174020875e-07,
      "loss": 2.6633,
      "step": 700
    },
    {
      "epoch": 1.0,
      "eval_loss": 2.6718015670776367,
      "eval_runtime": 11.0736,
      "eval_samples_per_second": 904.038,
      "eval_steps_per_second": 7.134,
      "step": 704
    },
    {
      "epoch": 1.01,
      "learning_rate": 4.662269987756317e-08,
      "loss": 2.6824,
      "step": 710
    },
    {
      "epoch": 1.02,
      "learning_rate": 5.707952862381681e-07,
      "loss": 2.6342,
      "step": 720
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.6753760662307215e-06,
      "loss": 2.6547,
      "step": 730
    },
    {
      "epoch": 1.05,
      "learning_rate": 3.356074662104319e-06,
      "loss": 2.6604,
      "step": 740
    },
    {
      "epoch": 1.07,
      "learning_rate": 5.606362957498179e-06,
      "loss": 2.6865,
      "step": 750
    },
    {
      "epoch": 1.08,
      "learning_rate": 8.417500453744864e-06,
      "loss": 2.6392,
      "step": 760
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.1778568219438839e-05,
      "loss": 2.6564,
      "step": 770
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.567651130140481e-05,
      "loss": 2.6074,
      "step": 780
    },
    {
      "epoch": 1.12,
      "learning_rate": 2.0096189432334208e-05,
      "loss": 2.6392,
      "step": 790
    },
    {
      "epoch": 1.14,
      "learning_rate": 2.5020435838132658e-05,
      "loss": 2.6144,
      "step": 800
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.0430123916561604e-05,
      "loss": 2.6486,
      "step": 810
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.630424152818204e-05,
      "loss": 2.6122,
      "step": 820
    },
    {
      "epoch": 1.18,
      "learning_rate": 4.2619972611042214e-05,
      "loss": 2.6399,
      "step": 830
    },
    {
      "epoch": 1.19,
      "learning_rate": 4.935278580210442e-05,
      "loss": 2.6323,
      "step": 840
    },
    {
      "epoch": 1.21,
      "learning_rate": 5.647652972118994e-05,
      "loss": 2.6635,
      "step": 850
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.396353454734303e-05,
      "loss": 2.6061,
      "step": 860
    },
    {
      "epoch": 1.24,
      "learning_rate": 7.178471949307521e-05,
      "loss": 2.5807,
      "step": 870
    },
    {
      "epoch": 1.25,
      "learning_rate": 7.990970575904072e-05,
      "loss": 2.6347,
      "step": 880
    },
    {
      "epoch": 1.26,
      "learning_rate": 8.830693453040826e-05,
      "loss": 2.6889,
      "step": 890
    },
    {
      "epoch": 1.28,
      "learning_rate": 9.694378955661275e-05,
      "loss": 2.6108,
      "step": 900
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00010578672383836428,
      "loss": 2.6338,
      "step": 910
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00011480138992984273,
      "loss": 2.596,
      "step": 920
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.00012395277334996044,
      "loss": 2.6379,
      "step": 930
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00013320532858450377,
      "loss": 2.6312,
      "step": 940
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00014252311715089535,
      "loss": 2.6165,
      "step": 950
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.00015186994718931215,
      "loss": 2.6148,
      "step": 960
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0001612095140379635,
      "loss": 2.6115,
      "step": 970
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00017050554124651096,
      "loss": 2.6476,
      "step": 980
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00017972192147990958,
      "loss": 2.6233,
      "step": 990
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00018882285676537802,
      "loss": 2.6507,
      "step": 1000
    },
    {
      "epoch": 1.43,
      "learning_rate": 0.00019777299753775265,
      "loss": 2.6474,
      "step": 1010
    },
    {
      "epoch": 1.45,
      "learning_rate": 0.00020653757994315076,
      "loss": 2.6162,
      "step": 1020
    },
    {
      "epoch": 1.46,
      "learning_rate": 0.00021508256086763368,
      "loss": 2.6377,
      "step": 1030
    },
    {
      "epoch": 1.48,
      "learning_rate": 0.00022337475016639342,
      "loss": 2.6132,
      "step": 1040
    },
    {
      "epoch": 1.49,
      "learning_rate": 0.00023138193957986393,
      "loss": 2.6239,
      "step": 1050
    },
    {
      "epoch": 1.51,
      "learning_rate": 0.00023907302783602514,
      "loss": 2.6212,
      "step": 1060
    },
    {
      "epoch": 1.52,
      "learning_rate": 0.0002464181414529809,
      "loss": 2.599,
      "step": 1070
    },
    {
      "epoch": 1.53,
      "learning_rate": 0.000253388750772592,
      "loss": 2.6263,
      "step": 1080
    },
    {
      "epoch": 1.55,
      "learning_rate": 0.0002599577807744739,
      "loss": 2.5971,
      "step": 1090
    },
    {
      "epoch": 1.56,
      "learning_rate": 0.0002660997162399341,
      "loss": 2.5784,
      "step": 1100
    },
    {
      "epoch": 1.58,
      "learning_rate": 0.0002717907008573785,
      "loss": 2.6068,
      "step": 1110
    },
    {
      "epoch": 1.59,
      "learning_rate": 0.0002770086298842426,
      "loss": 2.6139,
      "step": 1120
    },
    {
      "epoch": 1.61,
      "learning_rate": 0.00028173323600553423,
      "loss": 2.6228,
      "step": 1130
    },
    {
      "epoch": 1.62,
      "learning_rate": 0.0002859461680554975,
      "loss": 2.5909,
      "step": 1140
    },
    {
      "epoch": 1.63,
      "learning_rate": 0.00028963106229663063,
      "loss": 2.5944,
      "step": 1150
    },
    {
      "epoch": 1.65,
      "learning_rate": 0.0002927736059791983,
      "loss": 2.6474,
      "step": 1160
    },
    {
      "epoch": 1.66,
      "learning_rate": 0.0002953615929343616,
      "loss": 2.6134,
      "step": 1170
    },
    {
      "epoch": 1.68,
      "learning_rate": 0.0002973849709849932,
      "loss": 2.5762,
      "step": 1180
    },
    {
      "epoch": 1.69,
      "learning_rate": 0.0002988358809900258,
      "loss": 2.5956,
      "step": 1190
    },
    {
      "epoch": 1.7,
      "learning_rate": 0.0002997086873706798,
      "loss": 2.5878,
      "step": 1200
    },
    {
      "epoch": 1.72,
      "learning_rate": 0.0003,
      "loss": 2.6209,
      "step": 1210
    },
    {
      "epoch": 1.73,
      "learning_rate": 0.0002997086873706798,
      "loss": 2.6205,
      "step": 1220
    },
    {
      "epoch": 1.75,
      "learning_rate": 0.0002988358809900258,
      "loss": 2.5864,
      "step": 1230
    },
    {
      "epoch": 1.76,
      "learning_rate": 0.00029738497098499324,
      "loss": 2.5826,
      "step": 1240
    },
    {
      "epoch": 1.78,
      "learning_rate": 0.00029536159293436166,
      "loss": 2.5817,
      "step": 1250
    },
    {
      "epoch": 1.79,
      "learning_rate": 0.0002927736059791983,
      "loss": 2.5889,
      "step": 1260
    },
    {
      "epoch": 1.8,
      "learning_rate": 0.00028963106229663063,
      "loss": 2.5749,
      "step": 1270
    },
    {
      "epoch": 1.82,
      "learning_rate": 0.0002859461680554975,
      "loss": 2.6131,
      "step": 1280
    },
    {
      "epoch": 1.83,
      "learning_rate": 0.00028173323600553434,
      "loss": 2.5982,
      "step": 1290
    },
    {
      "epoch": 1.85,
      "learning_rate": 0.0002770086298842427,
      "loss": 2.5702,
      "step": 1300
    },
    {
      "epoch": 1.86,
      "learning_rate": 0.0002717907008573784,
      "loss": 2.5854,
      "step": 1310
    },
    {
      "epoch": 1.88,
      "learning_rate": 0.00026609971623993406,
      "loss": 2.554,
      "step": 1320
    },
    {
      "epoch": 1.89,
      "learning_rate": 0.0002599577807744739,
      "loss": 2.5928,
      "step": 1330
    },
    {
      "epoch": 1.9,
      "learning_rate": 0.00025338875077259205,
      "loss": 2.5768,
      "step": 1340
    },
    {
      "epoch": 1.92,
      "learning_rate": 0.00024641814145298093,
      "loss": 2.5395,
      "step": 1350
    },
    {
      "epoch": 1.93,
      "learning_rate": 0.00023907302783602525,
      "loss": 2.5265,
      "step": 1360
    },
    {
      "epoch": 1.95,
      "learning_rate": 0.00023138193957986385,
      "loss": 2.5535,
      "step": 1370
    },
    {
      "epoch": 1.96,
      "learning_rate": 0.0002233747501663934,
      "loss": 2.5308,
      "step": 1380
    },
    {
      "epoch": 1.97,
      "learning_rate": 0.00021508256086763376,
      "loss": 2.5611,
      "step": 1390
    },
    {
      "epoch": 1.99,
      "learning_rate": 0.00020653757994315084,
      "loss": 2.5567,
      "step": 1400
    },
    {
      "epoch": 2.0,
      "eval_loss": 2.5756430625915527,
      "eval_runtime": 11.0382,
      "eval_samples_per_second": 906.942,
      "eval_steps_per_second": 7.157,
      "step": 1408
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.00019777299753775273,
      "loss": 2.5977,
      "step": 1410
    },
    {
      "epoch": 2.02,
      "learning_rate": 0.00018882285676537824,
      "loss": 2.5147,
      "step": 1420
    },
    {
      "epoch": 2.03,
      "learning_rate": 0.0001797219214799098,
      "loss": 2.5313,
      "step": 1430
    },
    {
      "epoch": 2.05,
      "learning_rate": 0.0001705055412465109,
      "loss": 2.5303,
      "step": 1440
    },
    {
      "epoch": 2.06,
      "learning_rate": 0.00016120951403796358,
      "loss": 2.5039,
      "step": 1450
    },
    {
      "epoch": 2.07,
      "learning_rate": 0.00015186994718931223,
      "loss": 2.5199,
      "step": 1460
    },
    {
      "epoch": 2.09,
      "learning_rate": 0.00014252311715089543,
      "loss": 2.4787,
      "step": 1470
    },
    {
      "epoch": 2.1,
      "learning_rate": 0.00013320532858450385,
      "loss": 2.5371,
      "step": 1480
    },
    {
      "epoch": 2.12,
      "learning_rate": 0.00012395277334996052,
      "loss": 2.5247,
      "step": 1490
    },
    {
      "epoch": 2.13,
      "learning_rate": 0.00011480138992984267,
      "loss": 2.488,
      "step": 1500
    },
    {
      "epoch": 2.14,
      "learning_rate": 0.00010578672383836424,
      "loss": 2.5005,
      "step": 1510
    },
    {
      "epoch": 2.16,
      "learning_rate": 9.694378955661282e-05,
      "loss": 2.4841,
      "step": 1520
    },
    {
      "epoch": 2.17,
      "learning_rate": 8.830693453040844e-05,
      "loss": 2.5214,
      "step": 1530
    },
    {
      "epoch": 2.19,
      "learning_rate": 7.990970575904079e-05,
      "loss": 2.4842,
      "step": 1540
    },
    {
      "epoch": 2.2,
      "learning_rate": 7.178471949307551e-05,
      "loss": 2.5118,
      "step": 1550
    },
    {
      "epoch": 2.22,
      "learning_rate": 6.396353454734299e-05,
      "loss": 2.468,
      "step": 1560
    },
    {
      "epoch": 2.23,
      "learning_rate": 5.647652972119001e-05,
      "loss": 2.5586,
      "step": 1570
    },
    {
      "epoch": 2.24,
      "learning_rate": 4.935278580210444e-05,
      "loss": 2.4862,
      "step": 1580
    },
    {
      "epoch": 2.26,
      "learning_rate": 4.2619972611042316e-05,
      "loss": 2.4773,
      "step": 1590
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.630424152818206e-05,
      "loss": 2.4787,
      "step": 1600
    },
    {
      "epoch": 2.29,
      "learning_rate": 3.0430123916561723e-05,
      "loss": 2.4914,
      "step": 1610
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.5020435838132692e-05,
      "loss": 2.5184,
      "step": 1620
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.0096189432334177e-05,
      "loss": 2.5503,
      "step": 1630
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.5676511301404892e-05,
      "loss": 2.4975,
      "step": 1640
    },
    {
      "epoch": 2.34,
      "learning_rate": 1.1778568219438856e-05,
      "loss": 2.4797,
      "step": 1650
    },
    {
      "epoch": 2.36,
      "learning_rate": 8.41750045374493e-06,
      "loss": 2.5437,
      "step": 1660
    },
    {
      "epoch": 2.37,
      "learning_rate": 5.606362957498212e-06,
      "loss": 2.5227,
      "step": 1670
    },
    {
      "epoch": 2.39,
      "learning_rate": 3.356074662104369e-06,
      "loss": 2.4699,
      "step": 1680
    },
    {
      "epoch": 2.4,
      "learning_rate": 1.6753760662307048e-06,
      "loss": 2.5275,
      "step": 1690
    },
    {
      "epoch": 2.41,
      "learning_rate": 5.707952862381681e-07,
      "loss": 2.5028,
      "step": 1700
    },
    {
      "epoch": 2.43,
      "learning_rate": 4.662269987756317e-08,
      "loss": 2.4967,
      "step": 1710
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.0489428174020875e-07,
      "loss": 2.5104,
      "step": 1720
    },
    {
      "epoch": 2.46,
      "learning_rate": 7.453836951897885e-07,
      "loss": 2.4775,
      "step": 1730
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.965603171491803e-06,
      "loss": 2.4989,
      "step": 1740
    },
    {
      "epoch": 2.49,
      "learning_rate": 3.76081317272644e-06,
      "loss": 2.5113,
      "step": 1750
    },
    {
      "epoch": 2.5,
      "learning_rate": 6.1240408009518185e-06,
      "loss": 2.5008,
      "step": 1760
    },
    {
      "epoch": 2.51,
      "learning_rate": 9.046106882113702e-06,
      "loss": 2.5256,
      "step": 1770
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.251566161950357e-05,
      "loss": 2.5026,
      "step": 1780
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.6519228678279633e-05,
      "loss": 2.4909,
      "step": 1790
    },
    {
      "epoch": 2.56,
      "learning_rate": 2.104125752982142e-05,
      "loss": 2.5,
      "step": 1800
    },
    {
      "epoch": 2.57,
      "learning_rate": 2.606418385260078e-05,
      "loss": 2.5199,
      "step": 1810
    },
    {
      "epoch": 2.59,
      "learning_rate": 3.156849775496477e-05,
      "loss": 2.4495,
      "step": 1820
    },
    {
      "epoch": 2.6,
      "learning_rate": 3.753281955483985e-05,
      "loss": 2.497,
      "step": 1830
    },
    {
      "epoch": 2.61,
      "learning_rate": 4.3933982822017924e-05,
      "loss": 2.4602,
      "step": 1840
    },
    {
      "epoch": 2.63,
      "learning_rate": 5.074712436047102e-05,
      "loss": 2.4893,
      "step": 1850
    },
    {
      "epoch": 2.64,
      "learning_rate": 5.794578078119269e-05,
      "loss": 2.4704,
      "step": 1860
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.550199129045656e-05,
      "loss": 2.487,
      "step": 1870
    },
    {
      "epoch": 2.67,
      "learning_rate": 7.338640629424713e-05,
      "loss": 2.5259,
      "step": 1880
    },
    {
      "epoch": 2.68,
      "learning_rate": 8.15684013970256e-05,
      "loss": 2.5138,
      "step": 1890
    },
    {
      "epoch": 2.7,
      "learning_rate": 9.001619635203874e-05,
      "loss": 2.4926,
      "step": 1900
    },
    {
      "epoch": 2.71,
      "learning_rate": 9.869697850114967e-05,
      "loss": 2.5033,
      "step": 1910
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.00010757703022472571,
      "loss": 2.4881,
      "step": 1920
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.00011662185990655278,
      "loss": 2.5102,
      "step": 1930
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00012579633590507902,
      "loss": 2.5005,
      "step": 1940
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.00013506482301062737,
      "loss": 2.5107,
      "step": 1950
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.00014439132085855113,
      "loss": 2.5019,
      "step": 1960
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.00015373960376071074,
      "loss": 2.5012,
      "step": 1970
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.00016307336141214865,
      "loss": 2.5013,
      "step": 1980
    },
    {
      "epoch": 2.83,
      "learning_rate": 0.0001723563399264259,
      "loss": 2.4734,
      "step": 1990
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00018155248265182446,
      "loss": 2.4939,
      "step": 2000
    },
    {
      "epoch": 2.86,
      "learning_rate": 0.00019062607022145073,
      "loss": 2.4969,
      "step": 2010
    },
    {
      "epoch": 2.87,
      "learning_rate": 0.0001995418592932751,
      "loss": 2.4625,
      "step": 2020
    },
    {
      "epoch": 2.88,
      "learning_rate": 0.0002082652194412041,
      "loss": 2.5128,
      "step": 2030
    },
    {
      "epoch": 2.9,
      "learning_rate": 0.00021676226766548885,
      "loss": 2.4877,
      "step": 2040
    },
    {
      "epoch": 2.91,
      "learning_rate": 0.00022499999999999986,
      "loss": 2.4856,
      "step": 2050
    },
    {
      "epoch": 2.93,
      "learning_rate": 0.0002329464197051909,
      "loss": 2.4225,
      "step": 2060
    },
    {
      "epoch": 2.94,
      "learning_rate": 0.00024057066154882162,
      "loss": 2.5323,
      "step": 2070
    },
    {
      "epoch": 2.95,
      "learning_rate": 0.0002478431116917181,
      "loss": 2.4751,
      "step": 2080
    },
    {
      "epoch": 2.97,
      "learning_rate": 0.0002547355227129109,
      "loss": 2.4638,
      "step": 2090
    },
    {
      "epoch": 2.98,
      "learning_rate": 0.0002612211233273739,
      "loss": 2.4553,
      "step": 2100
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0002672747223702044,
      "loss": 2.4885,
      "step": 2110
    },
    {
      "epoch": 3.0,
      "eval_loss": 2.5283169746398926,
      "eval_runtime": 11.0137,
      "eval_samples_per_second": 908.962,
      "eval_steps_per_second": 7.173,
      "step": 2112
    },
    {
      "epoch": 3.01,
      "learning_rate": 0.00027287280664334865,
      "loss": 2.4728,
      "step": 2120
    },
    {
      "epoch": 3.03,
      "learning_rate": 0.00027799363224482337,
      "loss": 2.4908,
      "step": 2130
    },
    {
      "epoch": 3.04,
      "learning_rate": 0.00028261730902569146,
      "loss": 2.4329,
      "step": 2140
    },
    {
      "epoch": 3.05,
      "learning_rate": 0.00028672587784675096,
      "loss": 2.4558,
      "step": 2150
    },
    {
      "epoch": 3.07,
      "learning_rate": 0.0002903033803348551,
      "loss": 2.4522,
      "step": 2160
    },
    {
      "epoch": 3.08,
      "learning_rate": 0.000293335920867921,
      "loss": 2.4269,
      "step": 2170
    },
    {
      "epoch": 3.1,
      "learning_rate": 0.00029581172054786616,
      "loss": 2.4584,
      "step": 2180
    },
    {
      "epoch": 3.11,
      "learning_rate": 0.0002977211629518312,
      "loss": 2.4243,
      "step": 2190
    },
    {
      "epoch": 3.12,
      "learning_rate": 0.0002990568314839864,
      "loss": 2.4672,
      "step": 2200
    },
    {
      "epoch": 3.14,
      "learning_rate": 0.0002998135381828383,
      "loss": 2.4333,
      "step": 2210
    },
    {
      "epoch": 3.15,
      "learning_rate": 0.0002999883438721462,
      "loss": 2.3994,
      "step": 2220
    },
    {
      "epoch": 3.17,
      "learning_rate": 0.000299580569577177,
      "loss": 2.4639,
      "step": 2230
    },
    {
      "epoch": 3.18,
      "learning_rate": 0.00029859179916195787,
      "loss": 2.4175,
      "step": 2240
    },
    {
      "epoch": 3.2,
      "learning_rate": 0.0002970258731772816,
      "loss": 2.4103,
      "step": 2250
    },
    {
      "epoch": 3.21,
      "learning_rate": 0.00029488887394336027,
      "loss": 2.4623,
      "step": 2260
    },
    {
      "epoch": 3.22,
      "learning_rate": 0.00029218910192506983,
      "loss": 2.4217,
      "step": 2270
    },
    {
      "epoch": 3.24,
      "learning_rate": 0.0002889370434915463,
      "loss": 2.4537,
      "step": 2280
    },
    {
      "epoch": 3.25,
      "learning_rate": 0.00028514533018536277,
      "loss": 2.4471,
      "step": 2290
    },
    {
      "epoch": 3.27,
      "learning_rate": 0.00028082868965949076,
      "loss": 2.3821,
      "step": 2300
    },
    {
      "epoch": 3.28,
      "learning_rate": 0.0002760038884726156,
      "loss": 2.406,
      "step": 2310
    },
    {
      "epoch": 3.3,
      "learning_rate": 0.0002706896669650002,
      "loss": 2.4075,
      "step": 2320
    },
    {
      "epoch": 3.31,
      "learning_rate": 0.00026490666646784665,
      "loss": 2.4283,
      "step": 2330
    },
    {
      "epoch": 3.32,
      "learning_rate": 0.00025867734912889096,
      "loss": 2.3801,
      "step": 2340
    },
    {
      "epoch": 3.34,
      "learning_rate": 0.00025202591066563786,
      "loss": 2.4049,
      "step": 2350
    },
    {
      "epoch": 3.35,
      "learning_rate": 0.00024497818638512107,
      "loss": 2.3783,
      "step": 2360
    },
    {
      "epoch": 3.37,
      "learning_rate": 0.00023756155083521851,
      "loss": 2.4167,
      "step": 2370
    },
    {
      "epoch": 3.38,
      "learning_rate": 0.00022980481147730062,
      "loss": 2.3844,
      "step": 2380
    },
    {
      "epoch": 3.39,
      "learning_rate": 0.00022173809679319783,
      "loss": 2.4114,
      "step": 2390
    },
    {
      "epoch": 3.41,
      "learning_rate": 0.00021339273926110515,
      "loss": 2.4202,
      "step": 2400
    },
    {
      "epoch": 3.42,
      "learning_rate": 0.00020480115365495915,
      "loss": 2.3722,
      "step": 2410
    },
    {
      "epoch": 3.44,
      "learning_rate": 0.00019599671113999995,
      "loss": 2.4075,
      "step": 2420
    },
    {
      "epoch": 3.45,
      "learning_rate": 0.00018701360965354402,
      "loss": 2.3886,
      "step": 2430
    },
    {
      "epoch": 3.47,
      "learning_rate": 0.00017788674107443704,
      "loss": 2.3833,
      "step": 2440
    },
    {
      "epoch": 3.48,
      "learning_rate": 0.00016865155569712278,
      "loss": 2.3775,
      "step": 2450
    },
    {
      "epoch": 3.49,
      "learning_rate": 0.00015934392453672772,
      "loss": 2.413,
      "step": 2460
    },
    {
      "epoch": 3.51,
      "learning_rate": 0.00015000000000000004,
      "loss": 2.3845,
      "step": 2470
    },
    {
      "epoch": 3.52,
      "learning_rate": 0.00014065607546327242,
      "loss": 2.3666,
      "step": 2480
    },
    {
      "epoch": 3.54,
      "learning_rate": 0.00013134844430287736,
      "loss": 2.4041,
      "step": 2490
    },
    {
      "epoch": 3.55,
      "learning_rate": 0.0001221132589255631,
      "loss": 2.4064,
      "step": 2500
    },
    {
      "epoch": 3.57,
      "learning_rate": 0.00011298639034645613,
      "loss": 2.4397,
      "step": 2510
    },
    {
      "epoch": 3.58,
      "learning_rate": 0.00010400328886000018,
      "loss": 2.4145,
      "step": 2520
    },
    {
      "epoch": 3.59,
      "learning_rate": 9.519884634504099e-05,
      "loss": 2.3853,
      "step": 2530
    },
    {
      "epoch": 3.61,
      "learning_rate": 8.660726073889497e-05,
      "loss": 2.4271,
      "step": 2540
    },
    {
      "epoch": 3.62,
      "learning_rate": 7.82619032068023e-05,
      "loss": 2.3364,
      "step": 2550
    },
    {
      "epoch": 3.64,
      "learning_rate": 7.019518852269947e-05,
      "loss": 2.3847,
      "step": 2560
    },
    {
      "epoch": 3.65,
      "learning_rate": 6.24384491647816e-05,
      "loss": 2.4106,
      "step": 2570
    },
    {
      "epoch": 3.66,
      "learning_rate": 5.502181361487904e-05,
      "loss": 2.3956,
      "step": 2580
    },
    {
      "epoch": 3.68,
      "learning_rate": 4.7974089334362206e-05,
      "loss": 2.3839,
      "step": 2590
    },
    {
      "epoch": 3.69,
      "learning_rate": 4.132265087110915e-05,
      "loss": 2.3884,
      "step": 2600
    },
    {
      "epoch": 3.71,
      "learning_rate": 3.5093333532153445e-05,
      "loss": 2.396,
      "step": 2610
    },
    {
      "epoch": 3.72,
      "learning_rate": 2.9310333034999828e-05,
      "loss": 2.3785,
      "step": 2620
    },
    {
      "epoch": 3.74,
      "learning_rate": 2.3996111527384437e-05,
      "loss": 2.4004,
      "step": 2630
    },
    {
      "epoch": 3.75,
      "learning_rate": 1.91713103405093e-05,
      "loss": 2.3918,
      "step": 2640
    },
    {
      "epoch": 3.76,
      "learning_rate": 1.4854669814637276e-05,
      "loss": 2.3066,
      "step": 2650
    },
    {
      "epoch": 3.78,
      "learning_rate": 1.1062956508453685e-05,
      "loss": 2.3492,
      "step": 2660
    },
    {
      "epoch": 3.79,
      "learning_rate": 7.810898074930194e-06,
      "loss": 2.3618,
      "step": 2670
    },
    {
      "epoch": 3.81,
      "learning_rate": 5.11112605663977e-06,
      "loss": 2.3854,
      "step": 2680
    },
    {
      "epoch": 3.82,
      "learning_rate": 2.974126822718409e-06,
      "loss": 2.3978,
      "step": 2690
    },
    {
      "epoch": 3.84,
      "learning_rate": 1.408200838042095e-06,
      "loss": 2.3669,
      "step": 2700
    },
    {
      "epoch": 3.85,
      "learning_rate": 4.194304228229806e-07,
      "loss": 2.3758,
      "step": 2710
    },
    {
      "epoch": 3.86,
      "learning_rate": 1.1656127853787445e-08,
      "loss": 2.4071,
      "step": 2720
    },
    {
      "epoch": 3.88,
      "learning_rate": 1.8646181716164831e-07,
      "loss": 2.3569,
      "step": 2730
    },
    {
      "epoch": 3.89,
      "learning_rate": 9.431685160135927e-07,
      "loss": 2.3831,
      "step": 2740
    },
    {
      "epoch": 3.91,
      "learning_rate": 2.2788370481687635e-06,
      "loss": 2.3951,
      "step": 2750
    },
    {
      "epoch": 3.92,
      "learning_rate": 4.188279452133741e-06,
      "loss": 2.4046,
      "step": 2760
    },
    {
      "epoch": 3.93,
      "learning_rate": 6.664079132078764e-06,
      "loss": 2.3911,
      "step": 2770
    },
    {
      "epoch": 3.95,
      "learning_rate": 9.696619665144767e-06,
      "loss": 2.3894,
      "step": 2780
    },
    {
      "epoch": 3.96,
      "learning_rate": 1.3274122153249145e-05,
      "loss": 2.4069,
      "step": 2790
    },
    {
      "epoch": 3.98,
      "learning_rate": 1.7382690974308465e-05,
      "loss": 2.4119,
      "step": 2800
    },
    {
      "epoch": 3.99,
      "learning_rate": 2.2006367755176625e-05,
      "loss": 2.3552,
      "step": 2810
    },
    {
      "epoch": 4.0,
      "eval_loss": 2.4395244121551514,
      "eval_runtime": 11.0583,
      "eval_samples_per_second": 905.292,
      "eval_steps_per_second": 7.144,
      "step": 2816
    },
    {
      "epoch": 4.01,
      "learning_rate": 2.712719335665126e-05,
      "loss": 2.3469,
      "step": 2820
    },
    {
      "epoch": 4.02,
      "learning_rate": 3.272527762979563e-05,
      "loss": 2.3646,
      "step": 2830
    },
    {
      "epoch": 4.03,
      "learning_rate": 3.877887667262582e-05,
      "loss": 2.3076,
      "step": 2840
    },
    {
      "epoch": 4.05,
      "learning_rate": 4.526447728708897e-05,
      "loss": 2.3542,
      "step": 2850
    },
    {
      "epoch": 4.06,
      "learning_rate": 5.2156888308281784e-05,
      "loss": 2.3176,
      "step": 2860
    },
    {
      "epoch": 4.08,
      "learning_rate": 5.9429338451178436e-05,
      "loss": 2.3324,
      "step": 2870
    },
    {
      "epoch": 4.09,
      "learning_rate": 6.705358029480876e-05,
      "loss": 2.3942,
      "step": 2880
    },
    {
      "epoch": 4.11,
      "learning_rate": 7.499999999999976e-05,
      "loss": 2.3256,
      "step": 2890
    },
    {
      "epoch": 4.12,
      "learning_rate": 8.323773233451096e-05,
      "loss": 2.3241,
      "step": 2900
    },
    {
      "epoch": 4.13,
      "learning_rate": 9.173478055879573e-05,
      "loss": 2.3697,
      "step": 2910
    },
    {
      "epoch": 4.15,
      "learning_rate": 0.00010045814070672496,
      "loss": 2.3174,
      "step": 2920
    },
    {
      "epoch": 4.16,
      "learning_rate": 0.00010937392977854936,
      "loss": 2.3787,
      "step": 2930
    },
    {
      "epoch": 4.18,
      "learning_rate": 0.00011844751734817587,
      "loss": 2.3298,
      "step": 2940
    },
    {
      "epoch": 4.19,
      "learning_rate": 0.00012764366007357364,
      "loss": 2.2992,
      "step": 2950
    },
    {
      "epoch": 4.2,
      "learning_rate": 0.00013692663858785116,
      "loss": 2.3446,
      "step": 2960
    },
    {
      "epoch": 4.22,
      "learning_rate": 0.00014626039623928907,
      "loss": 2.3374,
      "step": 2970
    },
    {
      "epoch": 4.23,
      "learning_rate": 0.00015560867914144898,
      "loss": 2.345,
      "step": 2980
    },
    {
      "epoch": 4.25,
      "learning_rate": 0.00016493517698937217,
      "loss": 2.3703,
      "step": 2990
    },
    {
      "epoch": 4.26,
      "learning_rate": 0.0001742036640949208,
      "loss": 2.3734,
      "step": 3000
    },
    {
      "epoch": 4.28,
      "learning_rate": 0.00018337814009344703,
      "loss": 2.3125,
      "step": 3010
    },
    {
      "epoch": 4.29,
      "learning_rate": 0.00019242296977527412,
      "loss": 2.3347,
      "step": 3020
    },
    {
      "epoch": 4.3,
      "learning_rate": 0.00020130302149884988,
      "loss": 2.406,
      "step": 3030
    },
    {
      "epoch": 4.32,
      "learning_rate": 0.00020998380364796131,
      "loss": 2.3703,
      "step": 3040
    },
    {
      "epoch": 4.33,
      "learning_rate": 0.0002184315986029747,
      "loss": 2.3179,
      "step": 3050
    },
    {
      "epoch": 4.35,
      "learning_rate": 0.0002266135937057527,
      "loss": 2.3096,
      "step": 3060
    },
    {
      "epoch": 4.36,
      "learning_rate": 0.00023449800870954326,
      "loss": 2.372,
      "step": 3070
    },
    {
      "epoch": 4.38,
      "learning_rate": 0.00024205421921880715,
      "loss": 2.3321,
      "step": 3080
    },
    {
      "epoch": 4.39,
      "learning_rate": 0.00024925287563952903,
      "loss": 2.3247,
      "step": 3090
    },
    {
      "epoch": 4.4,
      "learning_rate": 0.0002560660171779819,
      "loss": 2.3729,
      "step": 3100
    },
    {
      "epoch": 4.42,
      "learning_rate": 0.00026246718044516,
      "loss": 2.3266,
      "step": 3110
    },
    {
      "epoch": 4.43,
      "learning_rate": 0.0002684315022450353,
      "loss": 2.3244,
      "step": 3120
    },
    {
      "epoch": 4.45,
      "learning_rate": 0.00027393581614739896,
      "loss": 2.3222,
      "step": 3130
    },
    {
      "epoch": 4.46,
      "learning_rate": 0.0002789587424701784,
      "loss": 2.3556,
      "step": 3140
    },
    {
      "epoch": 4.47,
      "learning_rate": 0.00028348077132172016,
      "loss": 2.3062,
      "step": 3150
    },
    {
      "epoch": 4.49,
      "learning_rate": 0.0002874843383804963,
      "loss": 2.3373,
      "step": 3160
    },
    {
      "epoch": 4.5,
      "learning_rate": 0.0002909538931178862,
      "loss": 2.3501,
      "step": 3170
    },
    {
      "epoch": 4.52,
      "learning_rate": 0.0002938759591990482,
      "loss": 2.3569,
      "step": 3180
    },
    {
      "epoch": 4.53,
      "learning_rate": 0.00029623918682727355,
      "loss": 2.3147,
      "step": 3190
    },
    {
      "epoch": 4.55,
      "learning_rate": 0.0002980343968285081,
      "loss": 2.2977,
      "step": 3200
    },
    {
      "epoch": 4.56,
      "learning_rate": 0.0002992546163048102,
      "loss": 2.3241,
      "step": 3210
    },
    {
      "epoch": 4.57,
      "learning_rate": 0.0002998951057182598,
      "loss": 2.3418,
      "step": 3220
    },
    {
      "epoch": 4.59,
      "learning_rate": 0.0002999533773001224,
      "loss": 2.3403,
      "step": 3230
    },
    {
      "epoch": 4.6,
      "learning_rate": 0.0002994292047137618,
      "loss": 2.306,
      "step": 3240
    },
    {
      "epoch": 4.62,
      "learning_rate": 0.00029832462393376933,
      "loss": 2.2761,
      "step": 3250
    },
    {
      "epoch": 4.63,
      "learning_rate": 0.0002966439253378957,
      "loss": 2.3357,
      "step": 3260
    },
    {
      "epoch": 4.64,
      "learning_rate": 0.0002943936370425018,
      "loss": 2.3251,
      "step": 3270
    },
    {
      "epoch": 4.66,
      "learning_rate": 0.0002915824995462553,
      "loss": 2.3676,
      "step": 3280
    },
    {
      "epoch": 4.67,
      "learning_rate": 0.00028822143178056103,
      "loss": 2.3547,
      "step": 3290
    },
    {
      "epoch": 4.69,
      "learning_rate": 0.00028432348869859505,
      "loss": 2.3155,
      "step": 3300
    },
    {
      "epoch": 4.7,
      "learning_rate": 0.00027990381056766585,
      "loss": 2.3275,
      "step": 3310
    },
    {
      "epoch": 4.72,
      "learning_rate": 0.00027497956416186735,
      "loss": 2.3019,
      "step": 3320
    },
    {
      "epoch": 4.73,
      "learning_rate": 0.0002695698760834384,
      "loss": 2.3236,
      "step": 3330
    },
    {
      "epoch": 4.74,
      "learning_rate": 0.00026369575847181784,
      "loss": 2.2911,
      "step": 3340
    },
    {
      "epoch": 4.76,
      "learning_rate": 0.000257380027388958,
      "loss": 2.3297,
      "step": 3350
    },
    {
      "epoch": 4.77,
      "learning_rate": 0.0002506472141978957,
      "loss": 2.3119,
      "step": 3360
    },
    {
      "epoch": 4.79,
      "learning_rate": 0.0002435234702788101,
      "loss": 2.3075,
      "step": 3370
    },
    {
      "epoch": 4.8,
      "learning_rate": 0.00023603646545265687,
      "loss": 2.3113,
      "step": 3380
    },
    {
      "epoch": 4.82,
      "learning_rate": 0.00022821528050692507,
      "loss": 2.3155,
      "step": 3390
    },
    {
      "epoch": 4.83,
      "learning_rate": 0.00022009029424095958,
      "loss": 2.315,
      "step": 3400
    },
    {
      "epoch": 4.84,
      "learning_rate": 0.00021169306546959193,
      "loss": 2.281,
      "step": 3410
    },
    {
      "epoch": 4.86,
      "learning_rate": 0.00020305621044338731,
      "loss": 2.2756,
      "step": 3420
    },
    {
      "epoch": 4.87,
      "learning_rate": 0.00019421327616163563,
      "loss": 2.2932,
      "step": 3430
    },
    {
      "epoch": 4.89,
      "learning_rate": 0.0001851986100701572,
      "loss": 2.2784,
      "step": 3440
    },
    {
      "epoch": 4.9,
      "learning_rate": 0.00017604722665003937,
      "loss": 2.3281,
      "step": 3450
    },
    {
      "epoch": 4.91,
      "learning_rate": 0.00016679467141549642,
      "loss": 2.3306,
      "step": 3460
    },
    {
      "epoch": 4.93,
      "learning_rate": 0.00015747688284910473,
      "loss": 2.311,
      "step": 3470
    },
    {
      "epoch": 4.94,
      "learning_rate": 0.0001481300528106878,
      "loss": 2.3097,
      "step": 3480
    },
    {
      "epoch": 4.96,
      "learning_rate": 0.00013879048596203628,
      "loss": 2.3104,
      "step": 3490
    },
    {
      "epoch": 4.97,
      "learning_rate": 0.00012949445875348934,
      "loss": 2.2894,
      "step": 3500
    },
    {
      "epoch": 4.99,
      "learning_rate": 0.00012027807852009062,
      "loss": 2.3139,
      "step": 3510
    },
    {
      "epoch": 5.0,
      "learning_rate": 0.00011117714323462205,
      "loss": 2.3084,
      "step": 3520
    },
    {
      "epoch": 5.0,
      "eval_loss": 2.3810958862304688,
      "eval_runtime": 11.048,
      "eval_samples_per_second": 906.136,
      "eval_steps_per_second": 7.151,
      "step": 3520
    },
    {
      "epoch": 5.01,
      "learning_rate": 0.00010222700246224737,
      "loss": 2.2791,
      "step": 3530
    },
    {
      "epoch": 5.03,
      "learning_rate": 9.346242005684964e-05,
      "loss": 2.2723,
      "step": 3540
    },
    {
      "epoch": 5.04,
      "learning_rate": 8.491743913236614e-05,
      "loss": 2.2202,
      "step": 3550
    },
    {
      "epoch": 5.06,
      "learning_rate": 7.662524983360638e-05,
      "loss": 2.2527,
      "step": 3560
    },
    {
      "epoch": 5.07,
      "learning_rate": 6.861806042013623e-05,
      "loss": 2.2603,
      "step": 3570
    },
    {
      "epoch": 5.09,
      "learning_rate": 6.092697216397482e-05,
      "loss": 2.2643,
      "step": 3580
    },
    {
      "epoch": 5.1,
      "learning_rate": 5.3581858547019076e-05,
      "loss": 2.3002,
      "step": 3590
    },
    {
      "epoch": 5.11,
      "learning_rate": 4.661124922740784e-05,
      "loss": 2.2654,
      "step": 3600
    },
    {
      "epoch": 5.13,
      "learning_rate": 4.004221922552624e-05,
      "loss": 2.2413,
      "step": 3610
    },
    {
      "epoch": 5.14,
      "learning_rate": 3.3900283760066006e-05,
      "loss": 2.2112,
      "step": 3620
    },
    {
      "epoch": 5.16,
      "learning_rate": 2.8209299142621573e-05,
      "loss": 2.2311,
      "step": 3630
    },
    {
      "epoch": 5.17,
      "learning_rate": 2.2991370115757362e-05,
      "loss": 2.258,
      "step": 3640
    },
    {
      "epoch": 5.18,
      "learning_rate": 1.8266763994465914e-05,
      "loss": 2.2304,
      "step": 3650
    },
    {
      "epoch": 5.2,
      "learning_rate": 1.4053831944502642e-05,
      "loss": 2.2356,
      "step": 3660
    },
    {
      "epoch": 5.21,
      "learning_rate": 1.0368937703369245e-05,
      "loss": 2.2375,
      "step": 3670
    },
    {
      "epoch": 5.23,
      "learning_rate": 7.226394020801679e-06,
      "loss": 2.2039,
      "step": 3680
    },
    {
      "epoch": 5.24,
      "learning_rate": 4.6384070656383054e-06,
      "loss": 2.232,
      "step": 3690
    },
    {
      "epoch": 5.26,
      "learning_rate": 2.6150290150067422e-06,
      "loss": 2.2633,
      "step": 3700
    },
    {
      "epoch": 5.27,
      "learning_rate": 1.1641190099741572e-06,
      "loss": 2.2459,
      "step": 3710
    },
    {
      "epoch": 5.28,
      "learning_rate": 2.913126293202395e-07,
      "loss": 2.2498,
      "step": 3720
    },
    {
      "epoch": 5.3,
      "learning_rate": 0.0,
      "loss": 2.228,
      "step": 3730
    },
    {
      "epoch": 5.31,
      "learning_rate": 2.913126293202228e-07,
      "loss": 2.2358,
      "step": 3740
    },
    {
      "epoch": 5.33,
      "learning_rate": 1.1641190099741237e-06,
      "loss": 2.2618,
      "step": 3750
    },
    {
      "epoch": 5.34,
      "learning_rate": 2.6150290150066923e-06,
      "loss": 2.2322,
      "step": 3760
    },
    {
      "epoch": 5.36,
      "learning_rate": 4.6384070656382385e-06,
      "loss": 2.2604,
      "step": 3770
    },
    {
      "epoch": 5.37,
      "learning_rate": 7.2263940208015954e-06,
      "loss": 2.2499,
      "step": 3780
    },
    {
      "epoch": 5.38,
      "learning_rate": 1.0368937703369145e-05,
      "loss": 2.2434,
      "step": 3790
    },
    {
      "epoch": 5.4,
      "learning_rate": 1.4053831944502525e-05,
      "loss": 2.2428,
      "step": 3800
    },
    {
      "epoch": 5.41,
      "learning_rate": 1.82667639944658e-05,
      "loss": 2.2561,
      "step": 3810
    },
    {
      "epoch": 5.43,
      "learning_rate": 2.299137011575723e-05,
      "loss": 2.2548,
      "step": 3820
    },
    {
      "epoch": 5.44,
      "learning_rate": 2.820929914262142e-05,
      "loss": 2.2404,
      "step": 3830
    },
    {
      "epoch": 5.45,
      "learning_rate": 3.3900283760065837e-05,
      "loss": 2.2243,
      "step": 3840
    },
    {
      "epoch": 5.47,
      "learning_rate": 4.004221922552608e-05,
      "loss": 2.2692,
      "step": 3850
    },
    {
      "epoch": 5.48,
      "learning_rate": 4.6611249227407644e-05,
      "loss": 2.2215,
      "step": 3860
    },
    {
      "epoch": 5.5,
      "learning_rate": 5.358185854701887e-05,
      "loss": 2.2461,
      "step": 3870
    },
    {
      "epoch": 5.51,
      "learning_rate": 6.0926972163974606e-05,
      "loss": 2.2281,
      "step": 3880
    },
    {
      "epoch": 5.53,
      "learning_rate": 6.861806042013602e-05,
      "loss": 2.2252,
      "step": 3890
    },
    {
      "epoch": 5.54,
      "learning_rate": 7.662524983360616e-05,
      "loss": 2.2318,
      "step": 3900
    },
    {
      "epoch": 5.55,
      "learning_rate": 8.491743913236591e-05,
      "loss": 2.2259,
      "step": 3910
    },
    {
      "epoch": 5.57,
      "learning_rate": 9.34624200568494e-05,
      "loss": 2.195,
      "step": 3920
    },
    {
      "epoch": 5.58,
      "learning_rate": 0.00010222700246224714,
      "loss": 2.297,
      "step": 3930
    },
    {
      "epoch": 5.6,
      "learning_rate": 0.00011117714323462178,
      "loss": 2.245,
      "step": 3940
    },
    {
      "epoch": 5.61,
      "learning_rate": 0.00012027807852009036,
      "loss": 2.2401,
      "step": 3950
    },
    {
      "epoch": 5.62,
      "learning_rate": 0.0001294944587534891,
      "loss": 2.2179,
      "step": 3960
    },
    {
      "epoch": 5.64,
      "learning_rate": 0.00013879048596203604,
      "loss": 2.2755,
      "step": 3970
    },
    {
      "epoch": 5.65,
      "learning_rate": 0.00014813005281068752,
      "loss": 2.2392,
      "step": 3980
    },
    {
      "epoch": 5.67,
      "learning_rate": 0.00015747688284910446,
      "loss": 2.2611,
      "step": 3990
    },
    {
      "epoch": 5.68,
      "learning_rate": 0.00016679467141549617,
      "loss": 2.2239,
      "step": 4000
    },
    {
      "epoch": 5.7,
      "learning_rate": 0.0001760472266500391,
      "loss": 2.2209,
      "step": 4010
    },
    {
      "epoch": 5.71,
      "learning_rate": 0.00018519861007015696,
      "loss": 2.2581,
      "step": 4020
    },
    {
      "epoch": 5.72,
      "learning_rate": 0.00019421327616163538,
      "loss": 2.2747,
      "step": 4030
    },
    {
      "epoch": 5.74,
      "learning_rate": 0.0002030562104433871,
      "loss": 2.2458,
      "step": 4040
    },
    {
      "epoch": 5.75,
      "learning_rate": 0.00021169306546959168,
      "loss": 2.2509,
      "step": 4050
    },
    {
      "epoch": 5.77,
      "learning_rate": 0.00022009029424095936,
      "loss": 2.2503,
      "step": 4060
    },
    {
      "epoch": 5.78,
      "learning_rate": 0.00022821528050692485,
      "loss": 2.2346,
      "step": 4070
    },
    {
      "epoch": 5.8,
      "learning_rate": 0.00023603646545265668,
      "loss": 2.2528,
      "step": 4080
    },
    {
      "epoch": 5.81,
      "learning_rate": 0.00024352347027880992,
      "loss": 2.2423,
      "step": 4090
    },
    {
      "epoch": 5.82,
      "learning_rate": 0.0002506472141978955,
      "loss": 2.2623,
      "step": 4100
    },
    {
      "epoch": 5.84,
      "learning_rate": 0.0002573800273889578,
      "loss": 2.2391,
      "step": 4110
    },
    {
      "epoch": 5.85,
      "learning_rate": 0.0002636957584718177,
      "loss": 2.252,
      "step": 4120
    },
    {
      "epoch": 5.87,
      "learning_rate": 0.0002695698760834382,
      "loss": 2.2544,
      "step": 4130
    },
    {
      "epoch": 5.88,
      "learning_rate": 0.0002749795641618672,
      "loss": 2.2513,
      "step": 4140
    },
    {
      "epoch": 5.89,
      "learning_rate": 0.00027990381056766574,
      "loss": 2.2127,
      "step": 4150
    },
    {
      "epoch": 5.91,
      "learning_rate": 0.0002843234886985949,
      "loss": 2.2516,
      "step": 4160
    },
    {
      "epoch": 5.92,
      "learning_rate": 0.000288221431780561,
      "loss": 2.2518,
      "step": 4170
    },
    {
      "epoch": 5.94,
      "learning_rate": 0.0002915824995462552,
      "loss": 2.2243,
      "step": 4180
    },
    {
      "epoch": 5.95,
      "learning_rate": 0.00029439363704250176,
      "loss": 2.2356,
      "step": 4190
    },
    {
      "epoch": 5.97,
      "learning_rate": 0.00029664392533789563,
      "loss": 2.2513,
      "step": 4200
    },
    {
      "epoch": 5.98,
      "learning_rate": 0.0002983246239337692,
      "loss": 2.2249,
      "step": 4210
    },
    {
      "epoch": 5.99,
      "learning_rate": 0.0002994292047137618,
      "loss": 2.2587,
      "step": 4220
    },
    {
      "epoch": 6.0,
      "eval_loss": 2.3698606491088867,
      "eval_runtime": 11.0095,
      "eval_samples_per_second": 909.309,
      "eval_steps_per_second": 7.176,
      "step": 4224
    },
    {
      "epoch": 6.01,
      "learning_rate": 0.0002999533773001224,
      "loss": 2.2167,
      "step": 4230
    },
    {
      "epoch": 6.02,
      "learning_rate": 0.0002998951057182598,
      "loss": 2.1821,
      "step": 4240
    },
    {
      "epoch": 6.04,
      "learning_rate": 0.0002992546163048102,
      "loss": 2.2571,
      "step": 4250
    },
    {
      "epoch": 6.05,
      "learning_rate": 0.0002980343968285081,
      "loss": 2.213,
      "step": 4260
    },
    {
      "epoch": 6.07,
      "learning_rate": 0.0002962391868272736,
      "loss": 2.1956,
      "step": 4270
    },
    {
      "epoch": 6.08,
      "learning_rate": 0.0002938759591990481,
      "loss": 2.2039,
      "step": 4280
    },
    {
      "epoch": 6.09,
      "learning_rate": 0.0002909538931178863,
      "loss": 2.2088,
      "step": 4290
    },
    {
      "epoch": 6.11,
      "learning_rate": 0.00028748433838049643,
      "loss": 2.1783,
      "step": 4300
    },
    {
      "epoch": 6.12,
      "learning_rate": 0.0002834807713217205,
      "loss": 2.2402,
      "step": 4310
    },
    {
      "epoch": 6.14,
      "learning_rate": 0.0002789587424701785,
      "loss": 2.1845,
      "step": 4320
    },
    {
      "epoch": 6.15,
      "learning_rate": 0.0002739358161473994,
      "loss": 2.2124,
      "step": 4330
    },
    {
      "epoch": 6.16,
      "learning_rate": 0.0002684315022450354,
      "loss": 2.1913,
      "step": 4340
    },
    {
      "epoch": 6.18,
      "learning_rate": 0.00026246718044516056,
      "loss": 2.2663,
      "step": 4350
    },
    {
      "epoch": 6.19,
      "learning_rate": 0.00025606601717798207,
      "loss": 2.2365,
      "step": 4360
    },
    {
      "epoch": 6.21,
      "learning_rate": 0.0002492528756395288,
      "loss": 2.2329,
      "step": 4370
    },
    {
      "epoch": 6.22,
      "learning_rate": 0.00024205421921880737,
      "loss": 2.1998,
      "step": 4380
    },
    {
      "epoch": 6.24,
      "learning_rate": 0.00023449800870954305,
      "loss": 2.2371,
      "step": 4390
    },
    {
      "epoch": 6.25,
      "learning_rate": 0.00022661359370575293,
      "loss": 2.2358,
      "step": 4400
    },
    {
      "epoch": 6.26,
      "learning_rate": 0.00021843159860297448,
      "loss": 2.1672,
      "step": 4410
    },
    {
      "epoch": 6.28,
      "learning_rate": 0.00020998380364796156,
      "loss": 2.1938,
      "step": 4420
    },
    {
      "epoch": 6.29,
      "learning_rate": 0.00020130302149885012,
      "loss": 2.1829,
      "step": 4430
    },
    {
      "epoch": 6.31,
      "learning_rate": 0.00019242296977527433,
      "loss": 2.2059,
      "step": 4440
    },
    {
      "epoch": 6.32,
      "learning_rate": 0.0001833781400934473,
      "loss": 2.2004,
      "step": 4450
    },
    {
      "epoch": 6.34,
      "learning_rate": 0.00017420366409492155,
      "loss": 2.2182,
      "step": 4460
    },
    {
      "epoch": 6.35,
      "learning_rate": 0.00016493517698937244,
      "loss": 2.2017,
      "step": 4470
    },
    {
      "epoch": 6.36,
      "learning_rate": 0.0001556086791414492,
      "loss": 2.173,
      "step": 4480
    },
    {
      "epoch": 6.38,
      "learning_rate": 0.00014626039623928932,
      "loss": 2.1953,
      "step": 4490
    },
    {
      "epoch": 6.39,
      "learning_rate": 0.0001369266385878509,
      "loss": 2.2158,
      "step": 4500
    },
    {
      "epoch": 6.41,
      "learning_rate": 0.00012764366007357388,
      "loss": 2.1664,
      "step": 4510
    },
    {
      "epoch": 6.42,
      "learning_rate": 0.0001184475173481756,
      "loss": 2.1875,
      "step": 4520
    },
    {
      "epoch": 6.43,
      "learning_rate": 0.0001093739297785496,
      "loss": 2.2206,
      "step": 4530
    },
    {
      "epoch": 6.45,
      "learning_rate": 0.0001004581407067247,
      "loss": 2.2095,
      "step": 4540
    },
    {
      "epoch": 6.46,
      "learning_rate": 9.173478055879596e-05,
      "loss": 2.218,
      "step": 4550
    },
    {
      "epoch": 6.48,
      "learning_rate": 8.323773233451119e-05,
      "loss": 2.1932,
      "step": 4560
    },
    {
      "epoch": 6.49,
      "learning_rate": 7.500000000000044e-05,
      "loss": 2.2209,
      "step": 4570
    },
    {
      "epoch": 6.51,
      "learning_rate": 6.705358029480897e-05,
      "loss": 2.1806,
      "step": 4580
    },
    {
      "epoch": 6.52,
      "learning_rate": 5.942933845117864e-05,
      "loss": 2.2136,
      "step": 4590
    },
    {
      "epoch": 6.53,
      "learning_rate": 5.215688830828199e-05,
      "loss": 2.2195,
      "step": 4600
    },
    {
      "epoch": 6.55,
      "learning_rate": 4.5264477287089516e-05,
      "loss": 2.1989,
      "step": 4610
    },
    {
      "epoch": 6.56,
      "learning_rate": 3.877887667262599e-05,
      "loss": 2.1715,
      "step": 4620
    },
    {
      "epoch": 6.58,
      "learning_rate": 3.272527762979546e-05,
      "loss": 2.1788,
      "step": 4630
    },
    {
      "epoch": 6.59,
      "learning_rate": 2.7127193356651412e-05,
      "loss": 2.1751,
      "step": 4640
    },
    {
      "epoch": 6.61,
      "learning_rate": 2.200636775517649e-05,
      "loss": 2.2413,
      "step": 4650
    },
    {
      "epoch": 6.62,
      "learning_rate": 1.7382690974308584e-05,
      "loss": 2.1537,
      "step": 4660
    },
    {
      "epoch": 6.63,
      "learning_rate": 1.3274122153249028e-05,
      "loss": 2.2157,
      "step": 4670
    },
    {
      "epoch": 6.65,
      "learning_rate": 9.696619665145034e-06,
      "loss": 2.2003,
      "step": 4680
    },
    {
      "epoch": 6.66,
      "learning_rate": 6.664079132078831e-06,
      "loss": 2.1929,
      "step": 4690
    },
    {
      "epoch": 6.68,
      "learning_rate": 4.188279452133875e-06,
      "loss": 2.2119,
      "step": 4700
    },
    {
      "epoch": 6.69,
      "learning_rate": 2.2788370481688135e-06,
      "loss": 2.1549,
      "step": 4710
    },
    {
      "epoch": 6.7,
      "learning_rate": 9.431685160136759e-07,
      "loss": 2.1555,
      "step": 4720
    },
    {
      "epoch": 6.72,
      "learning_rate": 1.8646181716164831e-07,
      "loss": 2.1977,
      "step": 4730
    },
    {
      "epoch": 6.73,
      "learning_rate": 1.1656127853787445e-08,
      "loss": 2.1932,
      "step": 4740
    },
    {
      "epoch": 6.75,
      "learning_rate": 4.194304228229639e-07,
      "loss": 2.1683,
      "step": 4750
    },
    {
      "epoch": 6.76,
      "learning_rate": 1.408200838042145e-06,
      "loss": 2.1984,
      "step": 4760
    },
    {
      "epoch": 6.78,
      "learning_rate": 2.974126822718409e-06,
      "loss": 2.1759,
      "step": 4770
    },
    {
      "epoch": 6.79,
      "learning_rate": 5.11112605663977e-06,
      "loss": 2.1944,
      "step": 4780
    },
    {
      "epoch": 6.8,
      "learning_rate": 7.810898074930111e-06,
      "loss": 2.1653,
      "step": 4790
    },
    {
      "epoch": 6.82,
      "learning_rate": 1.1062956508453785e-05,
      "loss": 2.1588,
      "step": 4800
    },
    {
      "epoch": 6.83,
      "learning_rate": 1.485466981463706e-05,
      "loss": 2.2177,
      "step": 4810
    },
    {
      "epoch": 6.85,
      "learning_rate": 1.9171310340509167e-05,
      "loss": 2.1766,
      "step": 4820
    },
    {
      "epoch": 6.86,
      "learning_rate": 2.399611152738402e-05,
      "loss": 2.2032,
      "step": 4830
    },
    {
      "epoch": 6.88,
      "learning_rate": 2.9310333034999828e-05,
      "loss": 2.1885,
      "step": 4840
    },
    {
      "epoch": 6.89,
      "learning_rate": 3.509333353215311e-05,
      "loss": 2.1923,
      "step": 4850
    },
    {
      "epoch": 6.9,
      "learning_rate": 4.1322650871108964e-05,
      "loss": 2.2324,
      "step": 4860
    },
    {
      "epoch": 6.92,
      "learning_rate": 4.797408933436242e-05,
      "loss": 2.2365,
      "step": 4870
    },
    {
      "epoch": 6.93,
      "learning_rate": 5.5021813614879056e-05,
      "loss": 2.2018,
      "step": 4880
    },
    {
      "epoch": 6.95,
      "learning_rate": 6.243844916478162e-05,
      "loss": 2.2164,
      "step": 4890
    },
    {
      "epoch": 6.96,
      "learning_rate": 7.019518852269926e-05,
      "loss": 2.1815,
      "step": 4900
    },
    {
      "epoch": 6.97,
      "learning_rate": 7.826190320680255e-05,
      "loss": 2.1761,
      "step": 4910
    },
    {
      "epoch": 6.99,
      "learning_rate": 8.660726073889499e-05,
      "loss": 2.1938,
      "step": 4920
    },
    {
      "epoch": 7.0,
      "eval_loss": 2.346975326538086,
      "eval_runtime": 11.0292,
      "eval_samples_per_second": 907.68,
      "eval_steps_per_second": 7.163,
      "step": 4928
    },
    {
      "epoch": 7.0,
      "learning_rate": 9.519884634504074e-05,
      "loss": 2.157,
      "step": 4930
    },
    {
      "epoch": 7.02,
      "learning_rate": 0.00010400328885999944,
      "loss": 2.1342,
      "step": 4940
    },
    {
      "epoch": 7.03,
      "learning_rate": 0.00011298639034645615,
      "loss": 2.169,
      "step": 4950
    },
    {
      "epoch": 7.05,
      "learning_rate": 0.00012211325892556206,
      "loss": 2.1329,
      "step": 4960
    },
    {
      "epoch": 7.06,
      "learning_rate": 0.0001313484443028771,
      "loss": 2.1549,
      "step": 4970
    },
    {
      "epoch": 7.07,
      "learning_rate": 0.00014065607546327163,
      "loss": 2.1629,
      "step": 4980
    },
    {
      "epoch": 7.09,
      "learning_rate": 0.00014999999999999955,
      "loss": 2.1504,
      "step": 4990
    },
    {
      "epoch": 7.1,
      "learning_rate": 0.000159343924536728,
      "loss": 2.1788,
      "step": 5000
    },
    {
      "epoch": 7.12,
      "learning_rate": 0.0001686515556971225,
      "loss": 2.1536,
      "step": 5010
    },
    {
      "epoch": 7.13,
      "learning_rate": 0.00017788674107443756,
      "loss": 2.1729,
      "step": 5020
    },
    {
      "epoch": 7.14,
      "learning_rate": 0.00018701360965354348,
      "loss": 2.1063,
      "step": 5030
    },
    {
      "epoch": 7.16,
      "learning_rate": 0.0001959967111400002,
      "loss": 2.1522,
      "step": 5040
    },
    {
      "epoch": 7.17,
      "learning_rate": 0.0002048011536549589,
      "loss": 2.1394,
      "step": 5050
    },
    {
      "epoch": 7.19,
      "learning_rate": 0.00021339273926110466,
      "loss": 2.1487,
      "step": 5060
    },
    {
      "epoch": 7.2,
      "learning_rate": 0.00022173809679319713,
      "loss": 2.1357,
      "step": 5070
    },
    {
      "epoch": 7.22,
      "learning_rate": 0.00022980481147730043,
      "loss": 2.1608,
      "step": 5080
    },
    {
      "epoch": 7.23,
      "learning_rate": 0.00023756155083521808,
      "loss": 2.1628,
      "step": 5090
    },
    {
      "epoch": 7.24,
      "learning_rate": 0.00024497818638512063,
      "loss": 2.1579,
      "step": 5100
    },
    {
      "epoch": 7.26,
      "learning_rate": 0.0002520259106656373,
      "loss": 2.1437,
      "step": 5110
    },
    {
      "epoch": 7.27,
      "learning_rate": 0.0002586773491288908,
      "loss": 2.173,
      "step": 5120
    },
    {
      "epoch": 7.29,
      "learning_rate": 0.00026490666646784665,
      "loss": 2.1681,
      "step": 5130
    },
    {
      "epoch": 7.3,
      "learning_rate": 0.00027068966696499995,
      "loss": 2.168,
      "step": 5140
    },
    {
      "epoch": 7.32,
      "learning_rate": 0.00027600388847261577,
      "loss": 2.1713,
      "step": 5150
    },
    {
      "epoch": 7.33,
      "learning_rate": 0.00028082868965949065,
      "loss": 2.1775,
      "step": 5160
    },
    {
      "epoch": 7.34,
      "learning_rate": 0.00028514533018536277,
      "loss": 2.1691,
      "step": 5170
    },
    {
      "epoch": 7.36,
      "learning_rate": 0.00028893704349154605,
      "loss": 2.201,
      "step": 5180
    },
    {
      "epoch": 7.37,
      "learning_rate": 0.0002921891019250697,
      "loss": 2.1381,
      "step": 5190
    },
    {
      "epoch": 7.39,
      "learning_rate": 0.0002948888739433601,
      "loss": 2.1945,
      "step": 5200
    },
    {
      "epoch": 7.4,
      "learning_rate": 0.0002970258731772815,
      "loss": 2.212,
      "step": 5210
    },
    {
      "epoch": 7.41,
      "learning_rate": 0.00029859179916195776,
      "loss": 2.1935,
      "step": 5220
    },
    {
      "epoch": 7.43,
      "learning_rate": 0.00029958056957717696,
      "loss": 2.1762,
      "step": 5230
    },
    {
      "epoch": 7.44,
      "learning_rate": 0.0002999883438721462,
      "loss": 2.1784,
      "step": 5240
    },
    {
      "epoch": 7.46,
      "learning_rate": 0.00029981353818283837,
      "loss": 2.156,
      "step": 5250
    },
    {
      "epoch": 7.47,
      "learning_rate": 0.00029905683148398634,
      "loss": 2.1444,
      "step": 5260
    },
    {
      "epoch": 7.49,
      "learning_rate": 0.0002977211629518312,
      "loss": 2.1508,
      "step": 5270
    },
    {
      "epoch": 7.5,
      "learning_rate": 0.00029581172054786616,
      "loss": 2.1578,
      "step": 5280
    },
    {
      "epoch": 7.51,
      "learning_rate": 0.00029333592086792123,
      "loss": 2.1284,
      "step": 5290
    },
    {
      "epoch": 7.53,
      "learning_rate": 0.0002903033803348551,
      "loss": 2.1465,
      "step": 5300
    },
    {
      "epoch": 7.54,
      "learning_rate": 0.0002867258778467511,
      "loss": 2.1643,
      "step": 5310
    },
    {
      "epoch": 7.56,
      "learning_rate": 0.0002826173090256916,
      "loss": 2.2208,
      "step": 5320
    },
    {
      "epoch": 7.57,
      "learning_rate": 0.00027799363224482364,
      "loss": 2.1873,
      "step": 5330
    },
    {
      "epoch": 7.59,
      "learning_rate": 0.00027287280664334875,
      "loss": 2.1881,
      "step": 5340
    },
    {
      "epoch": 7.6,
      "learning_rate": 0.00026727472237020506,
      "loss": 2.1576,
      "step": 5350
    },
    {
      "epoch": 7.61,
      "learning_rate": 0.0002612211233273738,
      "loss": 2.1402,
      "step": 5360
    },
    {
      "epoch": 7.63,
      "learning_rate": 0.0002547355227129107,
      "loss": 2.1339,
      "step": 5370
    },
    {
      "epoch": 7.64,
      "learning_rate": 0.0002478431116917187,
      "loss": 2.1533,
      "step": 5380
    },
    {
      "epoch": 7.66,
      "learning_rate": 0.00024057066154882119,
      "loss": 2.185,
      "step": 5390
    },
    {
      "epoch": 7.67,
      "learning_rate": 0.00023294641970519127,
      "loss": 2.1806,
      "step": 5400
    },
    {
      "epoch": 7.68,
      "learning_rate": 0.00022500000000000032,
      "loss": 2.134,
      "step": 5410
    },
    {
      "epoch": 7.7,
      "learning_rate": 0.00021676226766548907,
      "loss": 2.1749,
      "step": 5420
    },
    {
      "epoch": 7.71,
      "learning_rate": 0.00020826521944120434,
      "loss": 2.1698,
      "step": 5430
    },
    {
      "epoch": 7.73,
      "learning_rate": 0.0001995418592932751,
      "loss": 2.1687,
      "step": 5440
    },
    {
      "epoch": 7.74,
      "learning_rate": 0.0001906260702214507,
      "loss": 2.1888,
      "step": 5450
    },
    {
      "epoch": 7.76,
      "learning_rate": 0.00018155248265182522,
      "loss": 2.145,
      "step": 5460
    },
    {
      "epoch": 7.77,
      "learning_rate": 0.0001723563399264259,
      "loss": 2.1814,
      "step": 5470
    },
    {
      "epoch": 7.78,
      "learning_rate": 0.0001630733614121494,
      "loss": 2.1604,
      "step": 5480
    },
    {
      "epoch": 7.8,
      "learning_rate": 0.00015373960376071152,
      "loss": 2.1584,
      "step": 5490
    },
    {
      "epoch": 7.81,
      "learning_rate": 0.00014439132085855056,
      "loss": 2.146,
      "step": 5500
    },
    {
      "epoch": 7.83,
      "learning_rate": 0.0001350648230106279,
      "loss": 2.1995,
      "step": 5510
    },
    {
      "epoch": 7.84,
      "learning_rate": 0.0001257963359050793,
      "loss": 2.1522,
      "step": 5520
    },
    {
      "epoch": 7.86,
      "learning_rate": 0.00011662185990655302,
      "loss": 2.1457,
      "step": 5530
    },
    {
      "epoch": 7.87,
      "learning_rate": 0.00010757703022472596,
      "loss": 2.1699,
      "step": 5540
    },
    {
      "epoch": 7.88,
      "learning_rate": 9.869697850114966e-05,
      "loss": 2.1149,
      "step": 5550
    },
    {
      "epoch": 7.9,
      "learning_rate": 9.001619635203874e-05,
      "loss": 2.1681,
      "step": 5560
    },
    {
      "epoch": 7.91,
      "learning_rate": 8.156840139702631e-05,
      "loss": 2.1431,
      "step": 5570
    },
    {
      "epoch": 7.93,
      "learning_rate": 7.33864062942469e-05,
      "loss": 2.1279,
      "step": 5580
    },
    {
      "epoch": 7.94,
      "learning_rate": 6.550199129045721e-05,
      "loss": 2.1389,
      "step": 5590
    },
    {
      "epoch": 7.95,
      "learning_rate": 5.794578078119331e-05,
      "loss": 2.1316,
      "step": 5600
    },
    {
      "epoch": 7.97,
      "learning_rate": 5.0747124360471404e-05,
      "loss": 2.0931,
      "step": 5610
    },
    {
      "epoch": 7.98,
      "learning_rate": 4.393398282201809e-05,
      "loss": 2.1771,
      "step": 5620
    },
    {
      "epoch": 8.0,
      "learning_rate": 3.7532819554840014e-05,
      "loss": 2.1491,
      "step": 5630
    },
    {
      "epoch": 8.0,
      "eval_loss": 2.3224949836730957,
      "eval_runtime": 11.0019,
      "eval_samples_per_second": 909.934,
      "eval_steps_per_second": 7.181,
      "step": 5632
    },
    {
      "epoch": 8.01,
      "learning_rate": 3.156849775496477e-05,
      "loss": 2.0926,
      "step": 5640
    },
    {
      "epoch": 8.03,
      "learning_rate": 2.606418385260078e-05,
      "loss": 2.077,
      "step": 5650
    },
    {
      "epoch": 8.04,
      "learning_rate": 2.104125752982142e-05,
      "loss": 2.0516,
      "step": 5660
    },
    {
      "epoch": 8.05,
      "learning_rate": 1.6519228678279616e-05,
      "loss": 2.0906,
      "step": 5670
    },
    {
      "epoch": 8.07,
      "learning_rate": 1.2515661619503886e-05,
      "loss": 2.095,
      "step": 5680
    },
    {
      "epoch": 8.08,
      "learning_rate": 9.046106882113602e-06,
      "loss": 2.0801,
      "step": 5690
    },
    {
      "epoch": 8.1,
      "learning_rate": 6.124040800951968e-06,
      "loss": 2.0714,
      "step": 5700
    },
    {
      "epoch": 8.11,
      "learning_rate": 3.760813172726557e-06,
      "loss": 2.1109,
      "step": 5710
    },
    {
      "epoch": 8.12,
      "learning_rate": 1.9656031714918865e-06,
      "loss": 2.12,
      "step": 5720
    },
    {
      "epoch": 8.14,
      "learning_rate": 7.453836951898218e-07,
      "loss": 2.0788,
      "step": 5730
    },
    {
      "epoch": 8.15,
      "learning_rate": 1.048942817402254e-07,
      "loss": 2.0993,
      "step": 5740
    },
    {
      "epoch": 8.17,
      "learning_rate": 4.662269987756317e-08,
      "loss": 2.0666,
      "step": 5750
    },
    {
      "epoch": 8.18,
      "learning_rate": 5.707952862381681e-07,
      "loss": 2.1035,
      "step": 5760
    },
    {
      "epoch": 8.2,
      "learning_rate": 1.675376066230738e-06,
      "loss": 2.0755,
      "step": 5770
    },
    {
      "epoch": 8.21,
      "learning_rate": 3.356074662104369e-06,
      "loss": 2.0395,
      "step": 5780
    },
    {
      "epoch": 8.22,
      "learning_rate": 5.606362957497995e-06,
      "loss": 2.049,
      "step": 5790
    },
    {
      "epoch": 8.24,
      "learning_rate": 8.417500453745013e-06,
      "loss": 2.1038,
      "step": 5800
    },
    {
      "epoch": 8.25,
      "learning_rate": 1.1778568219438656e-05,
      "loss": 2.0996,
      "step": 5810
    },
    {
      "epoch": 8.27,
      "learning_rate": 1.5676511301404658e-05,
      "loss": 2.055,
      "step": 5820
    },
    {
      "epoch": 8.28,
      "learning_rate": 2.009618943233406e-05,
      "loss": 2.0607,
      "step": 5830
    },
    {
      "epoch": 8.3,
      "learning_rate": 2.5020435838132556e-05,
      "loss": 2.0802,
      "step": 5840
    },
    {
      "epoch": 8.31,
      "learning_rate": 3.043012391656157e-05,
      "loss": 2.0957,
      "step": 5850
    },
    {
      "epoch": 8.32,
      "learning_rate": 3.630424152818206e-05,
      "loss": 2.0716,
      "step": 5860
    },
    {
      "epoch": 8.34,
      "learning_rate": 4.2619972611041564e-05,
      "loss": 2.1055,
      "step": 5870
    },
    {
      "epoch": 8.35,
      "learning_rate": 4.935278580210464e-05,
      "loss": 2.0589,
      "step": 5880
    },
    {
      "epoch": 8.37,
      "learning_rate": 5.6476529721190224e-05,
      "loss": 2.0697,
      "step": 5890
    },
    {
      "epoch": 8.38,
      "learning_rate": 6.396353454734256e-05,
      "loss": 2.0682,
      "step": 5900
    },
    {
      "epoch": 8.39,
      "learning_rate": 7.178471949307574e-05,
      "loss": 2.0672,
      "step": 5910
    },
    {
      "epoch": 8.41,
      "learning_rate": 7.990970575904032e-05,
      "loss": 2.0863,
      "step": 5920
    },
    {
      "epoch": 8.42,
      "learning_rate": 8.830693453040795e-05,
      "loss": 2.089,
      "step": 5930
    },
    {
      "epoch": 8.44,
      "learning_rate": 9.694378955661256e-05,
      "loss": 2.0583,
      "step": 5940
    },
    {
      "epoch": 8.45,
      "learning_rate": 0.00010578672383836425,
      "loss": 2.0616,
      "step": 5950
    },
    {
      "epoch": 8.47,
      "learning_rate": 0.00011480138992984269,
      "loss": 2.0936,
      "step": 5960
    },
    {
      "epoch": 8.48,
      "learning_rate": 0.00012395277334996052,
      "loss": 2.0863,
      "step": 5970
    },
    {
      "epoch": 8.49,
      "learning_rate": 0.00013320532858450296,
      "loss": 2.1113,
      "step": 5980
    },
    {
      "epoch": 8.51,
      "learning_rate": 0.0001425231171508957,
      "loss": 2.04,
      "step": 5990
    },
    {
      "epoch": 8.52,
      "learning_rate": 0.0001518699471893126,
      "loss": 2.0697,
      "step": 6000
    },
    {
      "epoch": 8.54,
      "learning_rate": 0.00016120951403796307,
      "loss": 2.0468,
      "step": 6010
    },
    {
      "epoch": 8.55,
      "learning_rate": 0.0001705055412465116,
      "loss": 2.0831,
      "step": 6020
    },
    {
      "epoch": 8.57,
      "learning_rate": 0.00017972192147990926,
      "loss": 2.1102,
      "step": 6030
    },
    {
      "epoch": 8.58,
      "learning_rate": 0.00018882285676537786,
      "loss": 2.0944,
      "step": 6040
    },
    {
      "epoch": 8.59,
      "learning_rate": 0.0001977729975377525,
      "loss": 2.108,
      "step": 6050
    },
    {
      "epoch": 8.61,
      "learning_rate": 0.00020653757994315073,
      "loss": 2.0661,
      "step": 6060
    },
    {
      "epoch": 8.62,
      "learning_rate": 0.00021508256086763376,
      "loss": 2.078,
      "step": 6070
    },
    {
      "epoch": 8.64,
      "learning_rate": 0.00022337475016639353,
      "loss": 2.1152,
      "step": 6080
    },
    {
      "epoch": 8.65,
      "learning_rate": 0.0002313819395798632,
      "loss": 2.1097,
      "step": 6090
    },
    {
      "epoch": 8.66,
      "learning_rate": 0.00023907302783602552,
      "loss": 2.0812,
      "step": 6100
    },
    {
      "epoch": 8.68,
      "learning_rate": 0.0002464181414529804,
      "loss": 2.1115,
      "step": 6110
    },
    {
      "epoch": 8.69,
      "learning_rate": 0.00025338875077259167,
      "loss": 2.0616,
      "step": 6120
    },
    {
      "epoch": 8.71,
      "learning_rate": 0.0002599577807744744,
      "loss": 2.1264,
      "step": 6130
    },
    {
      "epoch": 8.72,
      "learning_rate": 0.0002655054352858896,
      "loss": 2.0978,
      "step": 6140
    },
    {
      "epoch": 8.74,
      "learning_rate": 0.0002712425392983004,
      "loss": 2.1099,
      "step": 6150
    },
    {
      "epoch": 8.75,
      "learning_rate": 0.00027650871687193255,
      "loss": 2.0902,
      "step": 6160
    },
    {
      "epoch": 8.76,
      "learning_rate": 0.0002812835132863128,
      "loss": 2.102,
      "step": 6170
    },
    {
      "epoch": 8.78,
      "learning_rate": 0.0002855483824281353,
      "loss": 2.082,
      "step": 6180
    },
    {
      "epoch": 8.79,
      "learning_rate": 0.0002892867588274879,
      "loss": 2.1016,
      "step": 6190
    },
    {
      "epoch": 8.81,
      "learning_rate": 0.00029248412200092686,
      "loss": 2.0823,
      "step": 6200
    },
    {
      "epoch": 8.82,
      "learning_rate": 0.0002951280528514794,
      "loss": 2.1198,
      "step": 6210
    },
    {
      "epoch": 8.84,
      "learning_rate": 0.00029720828190650815,
      "loss": 2.0912,
      "step": 6220
    },
    {
      "epoch": 8.85,
      "learning_rate": 0.00029871672920607153,
      "loss": 2.0547,
      "step": 6230
    },
    {
      "epoch": 8.86,
      "learning_rate": 0.00029964753568684926,
      "loss": 2.1026,
      "step": 6240
    },
    {
      "epoch": 8.88,
      "learning_rate": 0.0002999970859397307,
      "loss": 2.106,
      "step": 6250
    },
    {
      "epoch": 8.89,
      "learning_rate": 0.00029976402225267247,
      "loss": 2.1424,
      "step": 6260
    },
    {
      "epoch": 8.91,
      "learning_rate": 0.0002989492498842809,
      "loss": 2.1021,
      "step": 6270
    },
    {
      "epoch": 8.92,
      "learning_rate": 0.00029755593354763527,
      "loss": 2.1026,
      "step": 6280
    },
    {
      "epoch": 8.93,
      "learning_rate": 0.00029558948511800866,
      "loss": 2.0828,
      "step": 6290
    },
    {
      "epoch": 8.95,
      "learning_rate": 0.00029305754261223406,
      "loss": 2.1417,
      "step": 6300
    },
    {
      "epoch": 8.96,
      "learning_rate": 0.00028996994052135996,
      "loss": 2.0579,
      "step": 6310
    },
    {
      "epoch": 8.98,
      "learning_rate": 0.00028633867161183155,
      "loss": 2.1051,
      "step": 6320
    },
    {
      "epoch": 8.99,
      "learning_rate": 0.00028217784034356626,
      "loss": 2.0623,
      "step": 6330
    },
    {
      "epoch": 9.0,
      "eval_loss": 2.327604293823242,
      "eval_runtime": 11.0081,
      "eval_samples_per_second": 909.418,
      "eval_steps_per_second": 7.177,
      "step": 6336
    },
    {
      "epoch": 9.01,
      "learning_rate": 0.0002775036080858562,
      "loss": 2.091,
      "step": 6340
    },
    {
      "epoch": 9.02,
      "learning_rate": 0.0002723341303438894,
      "loss": 2.0147,
      "step": 6350
    },
    {
      "epoch": 9.03,
      "learning_rate": 0.00026668948623970694,
      "loss": 2.0584,
      "step": 6360
    },
    {
      "epoch": 9.05,
      "learning_rate": 0.0002605916005215189,
      "loss": 2.0441,
      "step": 6370
    },
    {
      "epoch": 9.06,
      "learning_rate": 0.00025406415840428147,
      "loss": 2.0037,
      "step": 6380
    },
    {
      "epoch": 9.08,
      "learning_rate": 0.00024713251357234075,
      "loss": 2.0342,
      "step": 6390
    },
    {
      "epoch": 9.09,
      "learning_rate": 0.00023982358970145017,
      "loss": 2.0517,
      "step": 6400
    },
    {
      "epoch": 9.11,
      "learning_rate": 0.00023216577588268072,
      "loss": 2.0496,
      "step": 6410
    },
    {
      "epoch": 9.12,
      "learning_rate": 0.00022418881635441105,
      "loss": 2.0683,
      "step": 6420
    },
    {
      "epoch": 9.13,
      "learning_rate": 0.00021592369497069755,
      "loss": 2.0416,
      "step": 6430
    },
    {
      "epoch": 9.15,
      "learning_rate": 0.00020740251485476324,
      "loss": 2.0329,
      "step": 6440
    },
    {
      "epoch": 9.16,
      "learning_rate": 0.00019865837370507073,
      "loss": 2.0409,
      "step": 6450
    },
    {
      "epoch": 9.18,
      "learning_rate": 0.00018972523523827966,
      "loss": 2.0436,
      "step": 6460
    },
    {
      "epoch": 9.19,
      "learning_rate": 0.00018063779726845152,
      "loss": 2.0631,
      "step": 6470
    },
    {
      "epoch": 9.2,
      "learning_rate": 0.0001714313569349074,
      "loss": 2.0563,
      "step": 6480
    },
    {
      "epoch": 9.22,
      "learning_rate": 0.0001621416736021808,
      "loss": 2.0395,
      "step": 6490
    },
    {
      "epoch": 9.23,
      "learning_rate": 0.00015280482996463552,
      "loss": 2.0422,
      "step": 6500
    },
    {
      "epoch": 9.25,
      "learning_rate": 0.0001434570918951997,
      "loss": 2.038,
      "step": 6510
    },
    {
      "epoch": 9.26,
      "learning_rate": 0.00013413476758260932,
      "loss": 2.0441,
      "step": 6520
    },
    {
      "epoch": 9.28,
      "learning_rate": 0.00012487406650428943,
      "loss": 2.036,
      "step": 6530
    },
    {
      "epoch": 9.29,
      "learning_rate": 0.00011571095878264738,
      "loss": 2.074,
      "step": 6540
    },
    {
      "epoch": 9.3,
      "learning_rate": 0.00010668103547105523,
      "loss": 2.063,
      "step": 6550
    },
    {
      "epoch": 9.32,
      "learning_rate": 9.781937031221653e-05,
      "loss": 2.0583,
      "step": 6560
    },
    {
      "epoch": 9.33,
      "learning_rate": 8.916038350582922e-05,
      "loss": 2.0469,
      "step": 6570
    },
    {
      "epoch": 9.35,
      "learning_rate": 8.073770801474436e-05,
      "loss": 2.0674,
      "step": 6580
    },
    {
      "epoch": 9.36,
      "learning_rate": 7.25840589288743e-05,
      "loss": 2.0586,
      "step": 6590
    },
    {
      "epoch": 9.38,
      "learning_rate": 6.473110639426635e-05,
      "loss": 2.0402,
      "step": 6600
    },
    {
      "epoch": 9.39,
      "learning_rate": 5.72093526009319e-05,
      "loss": 2.0753,
      "step": 6610
    },
    {
      "epoch": 9.4,
      "learning_rate": 5.0048013307199424e-05,
      "loss": 1.9955,
      "step": 6620
    },
    {
      "epoch": 9.42,
      "learning_rate": 4.3274904360790423e-05,
      "loss": 2.0728,
      "step": 6630
    },
    {
      "epoch": 9.43,
      "learning_rate": 3.691633365738289e-05,
      "loss": 2.0395,
      "step": 6640
    },
    {
      "epoch": 9.45,
      "learning_rate": 3.09969989563152e-05,
      "loss": 2.0321,
      "step": 6650
    },
    {
      "epoch": 9.46,
      "learning_rate": 2.5539891950326658e-05,
      "loss": 2.0234,
      "step": 6660
    },
    {
      "epoch": 9.47,
      "learning_rate": 2.056620896195836e-05,
      "loss": 2.0023,
      "step": 6670
    },
    {
      "epoch": 9.49,
      "learning_rate": 1.6095268613458483e-05,
      "loss": 2.0256,
      "step": 6680
    },
    {
      "epoch": 9.5,
      "learning_rate": 1.2144436790007034e-05,
      "loss": 2.005,
      "step": 6690
    },
    {
      "epoch": 9.52,
      "learning_rate": 8.729059187690579e-06,
      "loss": 2.0098,
      "step": 6700
    },
    {
      "epoch": 9.53,
      "learning_rate": 5.862401708235109e-06,
      "loss": 2.0488,
      "step": 6710
    },
    {
      "epoch": 9.55,
      "learning_rate": 3.555598932010012e-06,
      "loss": 2.0012,
      "step": 6720
    },
    {
      "epoch": 9.56,
      "learning_rate": 1.8176108694427927e-06,
      "loss": 2.0404,
      "step": 6730
    },
    {
      "epoch": 9.57,
      "learning_rate": 6.551881588300112e-07,
      "loss": 2.0141,
      "step": 6740
    },
    {
      "epoch": 9.59,
      "learning_rate": 7.284584572083696e-08,
      "loss": 2.0593,
      "step": 6750
    },
    {
      "epoch": 9.6,
      "learning_rate": 7.284584572082031e-08,
      "loss": 2.0341,
      "step": 6760
    },
    {
      "epoch": 9.62,
      "learning_rate": 6.551881588299612e-07,
      "loss": 2.037,
      "step": 6770
    },
    {
      "epoch": 9.63,
      "learning_rate": 1.817610869442726e-06,
      "loss": 2.0146,
      "step": 6780
    },
    {
      "epoch": 9.64,
      "learning_rate": 3.5555989320099122e-06,
      "loss": 2.0295,
      "step": 6790
    },
    {
      "epoch": 9.66,
      "learning_rate": 5.862401708234976e-06,
      "loss": 2.0359,
      "step": 6800
    },
    {
      "epoch": 9.67,
      "learning_rate": 8.729059187690413e-06,
      "loss": 2.0138,
      "step": 6810
    },
    {
      "epoch": 9.69,
      "learning_rate": 1.214443679000685e-05,
      "loss": 2.028,
      "step": 6820
    },
    {
      "epoch": 9.7,
      "learning_rate": 1.6095268613458266e-05,
      "loss": 2.0097,
      "step": 6830
    },
    {
      "epoch": 9.72,
      "learning_rate": 2.0566208961958125e-05,
      "loss": 2.0522,
      "step": 6840
    },
    {
      "epoch": 9.73,
      "learning_rate": 2.553989195032639e-05,
      "loss": 2.0516,
      "step": 6850
    },
    {
      "epoch": 9.74,
      "learning_rate": 3.0996998956314905e-05,
      "loss": 2.022,
      "step": 6860
    },
    {
      "epoch": 9.76,
      "learning_rate": 3.691633365738259e-05,
      "loss": 2.0524,
      "step": 6870
    },
    {
      "epoch": 9.77,
      "learning_rate": 4.327490436079009e-05,
      "loss": 2.0464,
      "step": 6880
    },
    {
      "epoch": 9.79,
      "learning_rate": 5.004801330719908e-05,
      "loss": 2.0572,
      "step": 6890
    },
    {
      "epoch": 9.8,
      "learning_rate": 5.720935260093152e-05,
      "loss": 2.0273,
      "step": 6900
    },
    {
      "epoch": 9.82,
      "learning_rate": 6.473110639426594e-05,
      "loss": 2.03,
      "step": 6910
    },
    {
      "epoch": 9.83,
      "learning_rate": 7.258405892887389e-05,
      "loss": 2.027,
      "step": 6920
    },
    {
      "epoch": 9.84,
      "learning_rate": 8.073770801474394e-05,
      "loss": 2.0345,
      "step": 6930
    },
    {
      "epoch": 9.86,
      "learning_rate": 8.916038350582879e-05,
      "loss": 2.0404,
      "step": 6940
    },
    {
      "epoch": 9.87,
      "learning_rate": 9.781937031221607e-05,
      "loss": 2.0137,
      "step": 6950
    },
    {
      "epoch": 9.89,
      "learning_rate": 0.00010668103547105477,
      "loss": 2.0181,
      "step": 6960
    },
    {
      "epoch": 9.9,
      "learning_rate": 0.0001157109587826469,
      "loss": 2.0438,
      "step": 6970
    },
    {
      "epoch": 9.91,
      "learning_rate": 0.00012487406650428897,
      "loss": 2.0424,
      "step": 6980
    },
    {
      "epoch": 9.93,
      "learning_rate": 0.00013413476758260883,
      "loss": 2.0435,
      "step": 6990
    },
    {
      "epoch": 9.94,
      "learning_rate": 0.0001434570918951992,
      "loss": 2.0117,
      "step": 7000
    },
    {
      "epoch": 9.96,
      "learning_rate": 0.00015280482996463506,
      "loss": 2.0098,
      "step": 7010
    },
    {
      "epoch": 9.97,
      "learning_rate": 0.00016214167360218033,
      "loss": 2.0263,
      "step": 7020
    },
    {
      "epoch": 9.99,
      "learning_rate": 0.00017143135693490692,
      "loss": 2.0697,
      "step": 7030
    },
    {
      "epoch": 10.0,
      "learning_rate": 0.00018063779726845106,
      "loss": 2.0672,
      "step": 7040
    },
    {
      "epoch": 10.0,
      "eval_loss": 2.330075263977051,
      "eval_runtime": 11.0756,
      "eval_samples_per_second": 903.883,
      "eval_steps_per_second": 7.133,
      "step": 7040
    },
    {
      "epoch": 10.01,
      "learning_rate": 0.0001897252352382792,
      "loss": 2.0053,
      "step": 7050
    },
    {
      "epoch": 10.03,
      "learning_rate": 0.0001986583737050703,
      "loss": 2.0154,
      "step": 7060
    },
    {
      "epoch": 10.04,
      "learning_rate": 0.0002074025148547628,
      "loss": 1.998,
      "step": 7070
    },
    {
      "epoch": 10.06,
      "learning_rate": 0.00021592369497069712,
      "loss": 2.0056,
      "step": 7080
    },
    {
      "epoch": 10.07,
      "learning_rate": 0.00022418881635441067,
      "loss": 1.9888,
      "step": 7090
    },
    {
      "epoch": 10.09,
      "learning_rate": 0.00023216577588268034,
      "loss": 2.0266,
      "step": 7100
    },
    {
      "epoch": 10.1,
      "learning_rate": 0.00023982358970144976,
      "loss": 1.9969,
      "step": 7110
    },
    {
      "epoch": 10.11,
      "learning_rate": 0.0002471325135723404,
      "loss": 2.0036,
      "step": 7120
    },
    {
      "epoch": 10.13,
      "learning_rate": 0.00025406415840428115,
      "loss": 2.0173,
      "step": 7130
    },
    {
      "epoch": 10.14,
      "learning_rate": 0.0002605916005215186,
      "loss": 2.0406,
      "step": 7140
    },
    {
      "epoch": 10.16,
      "learning_rate": 0.0002666894862397066,
      "loss": 2.0007,
      "step": 7150
    },
    {
      "epoch": 10.17,
      "learning_rate": 0.0002723341303438891,
      "loss": 2.0359,
      "step": 7160
    },
    {
      "epoch": 10.18,
      "learning_rate": 0.000277503608085856,
      "loss": 2.0217,
      "step": 7170
    },
    {
      "epoch": 10.2,
      "learning_rate": 0.00028217784034356605,
      "loss": 2.0029,
      "step": 7180
    },
    {
      "epoch": 10.21,
      "learning_rate": 0.0002863386716118314,
      "loss": 2.0112,
      "step": 7190
    },
    {
      "epoch": 10.23,
      "learning_rate": 0.0002899699405213598,
      "loss": 2.0221,
      "step": 7200
    },
    {
      "epoch": 10.24,
      "learning_rate": 0.00029305754261223395,
      "loss": 2.0064,
      "step": 7210
    },
    {
      "epoch": 10.26,
      "learning_rate": 0.00029558948511800855,
      "loss": 2.019,
      "step": 7220
    },
    {
      "epoch": 10.27,
      "learning_rate": 0.00029755593354763516,
      "loss": 2.0286,
      "step": 7230
    },
    {
      "epoch": 10.28,
      "learning_rate": 0.00029894924988428087,
      "loss": 2.0046,
      "step": 7240
    },
    {
      "epoch": 10.3,
      "learning_rate": 0.00029976402225267247,
      "loss": 2.0264,
      "step": 7250
    },
    {
      "epoch": 10.31,
      "learning_rate": 0.0002999970859397307,
      "loss": 2.0141,
      "step": 7260
    },
    {
      "epoch": 10.33,
      "learning_rate": 0.00029964753568684926,
      "loss": 2.0015,
      "step": 7270
    },
    {
      "epoch": 10.34,
      "learning_rate": 0.0002987167292060716,
      "loss": 2.0057,
      "step": 7280
    },
    {
      "epoch": 10.36,
      "learning_rate": 0.00029720828190650826,
      "loss": 2.0256,
      "step": 7290
    },
    {
      "epoch": 10.37,
      "learning_rate": 0.00029512805285147956,
      "loss": 2.0477,
      "step": 7300
    },
    {
      "epoch": 10.38,
      "learning_rate": 0.00029248412200092697,
      "loss": 2.0311,
      "step": 7310
    },
    {
      "epoch": 10.4,
      "learning_rate": 0.00028928675882748813,
      "loss": 1.9869,
      "step": 7320
    },
    {
      "epoch": 10.41,
      "learning_rate": 0.00028554838242813554,
      "loss": 2.0306,
      "step": 7330
    },
    {
      "epoch": 10.43,
      "learning_rate": 0.00028128351328631304,
      "loss": 2.0188,
      "step": 7340
    },
    {
      "epoch": 10.44,
      "learning_rate": 0.0002765087168719328,
      "loss": 2.0359,
      "step": 7350
    },
    {
      "epoch": 10.45,
      "learning_rate": 0.00027124253929830067,
      "loss": 2.008,
      "step": 7360
    },
    {
      "epoch": 10.47,
      "learning_rate": 0.00026550543528588993,
      "loss": 1.9957,
      "step": 7370
    },
    {
      "epoch": 10.48,
      "learning_rate": 0.000259319688712759,
      "loss": 2.0123,
      "step": 7380
    },
    {
      "epoch": 10.5,
      "learning_rate": 0.0002527093260602245,
      "loss": 2.0441,
      "step": 7390
    },
    {
      "epoch": 10.51,
      "learning_rate": 0.00024570002308995167,
      "loss": 2.0213,
      "step": 7400
    },
    {
      "epoch": 10.53,
      "learning_rate": 0.00023831900511498102,
      "loss": 2.0308,
      "step": 7410
    },
    {
      "epoch": 10.54,
      "learning_rate": 0.00023059494125202381,
      "loss": 2.0259,
      "step": 7420
    },
    {
      "epoch": 10.55,
      "learning_rate": 0.00022255783306578613,
      "loss": 2.0105,
      "step": 7430
    },
    {
      "epoch": 10.57,
      "learning_rate": 0.0002142388980378395,
      "loss": 2.0307,
      "step": 7440
    },
    {
      "epoch": 10.58,
      "learning_rate": 0.00020567044831266566,
      "loss": 2.0057,
      "step": 7450
    },
    {
      "epoch": 10.6,
      "learning_rate": 0.00019688576519184654,
      "loss": 1.9967,
      "step": 7460
    },
    {
      "epoch": 10.61,
      "learning_rate": 0.0001879189698638844,
      "loss": 2.0092,
      "step": 7470
    },
    {
      "epoch": 10.62,
      "learning_rate": 0.00017880489087176112,
      "loss": 1.9987,
      "step": 7480
    },
    {
      "epoch": 10.64,
      "learning_rate": 0.00016957892883300732,
      "loss": 2.0128,
      "step": 7490
    },
    {
      "epoch": 10.65,
      "learning_rate": 0.000160276918937754,
      "loss": 2.0124,
      "step": 7500
    },
    {
      "epoch": 10.67,
      "learning_rate": 0.00015093499175880545,
      "loss": 2.0014,
      "step": 7510
    },
    {
      "epoch": 10.68,
      "learning_rate": 0.00014158943291442156,
      "loss": 2.0083,
      "step": 7520
    },
    {
      "epoch": 10.7,
      "learning_rate": 0.0001322765421288613,
      "loss": 2.0226,
      "step": 7530
    },
    {
      "epoch": 10.71,
      "learning_rate": 0.0001230324922381422,
      "loss": 2.0203,
      "step": 7540
    },
    {
      "epoch": 10.72,
      "learning_rate": 0.00011389318868865406,
      "loss": 2.0343,
      "step": 7550
    },
    {
      "epoch": 10.74,
      "learning_rate": 0.00010489413007435998,
      "loss": 2.0033,
      "step": 7560
    },
    {
      "epoch": 10.75,
      "learning_rate": 9.60702702542747e-05,
      "loss": 2.0476,
      "step": 7570
    },
    {
      "epoch": 10.77,
      "learning_rate": 8.745588258580053e-05,
      "loss": 1.9919,
      "step": 7580
    },
    {
      "epoch": 10.78,
      "learning_rate": 7.908442680122653e-05,
      "loss": 1.999,
      "step": 7590
    },
    {
      "epoch": 10.8,
      "learning_rate": 7.098841904449448e-05,
      "loss": 2.0013,
      "step": 7600
    },
    {
      "epoch": 10.81,
      "learning_rate": 6.319930557302952e-05,
      "loss": 1.9996,
      "step": 7610
    },
    {
      "epoch": 10.82,
      "learning_rate": 5.57473406151682e-05,
      "loss": 1.9714,
      "step": 7620
    },
    {
      "epoch": 10.84,
      "learning_rate": 4.8661468857651156e-05,
      "loss": 2.0055,
      "step": 7630
    },
    {
      "epoch": 10.85,
      "learning_rate": 4.196921301958112e-05,
      "loss": 2.0256,
      "step": 7640
    },
    {
      "epoch": 10.87,
      "learning_rate": 3.569656694954841e-05,
      "loss": 2.0128,
      "step": 7650
    },
    {
      "epoch": 10.88,
      "learning_rate": 2.9867894661145786e-05,
      "loss": 1.9936,
      "step": 7660
    },
    {
      "epoch": 10.89,
      "learning_rate": 2.4505835699037535e-05,
      "loss": 1.9781,
      "step": 7670
    },
    {
      "epoch": 10.91,
      "learning_rate": 1.9631217203152903e-05,
      "loss": 2.0075,
      "step": 7680
    },
    {
      "epoch": 10.92,
      "learning_rate": 1.5262973012573674e-05,
      "loss": 1.9988,
      "step": 7690
    },
    {
      "epoch": 10.94,
      "learning_rate": 1.141807012330722e-05,
      "loss": 2.0045,
      "step": 7700
    },
    {
      "epoch": 10.95,
      "learning_rate": 8.111442785622413e-06,
      "loss": 1.9608,
      "step": 7710
    },
    {
      "epoch": 10.97,
      "learning_rate": 5.3559344969059505e-06,
      "loss": 1.9888,
      "step": 7720
    },
    {
      "epoch": 10.98,
      "learning_rate": 3.1622481153527944e-06,
      "loss": 1.9767,
      "step": 7730
    },
    {
      "epoch": 10.99,
      "learning_rate": 1.5389042882661184e-06,
      "loss": 2.0293,
      "step": 7740
    },
    {
      "epoch": 11.0,
      "eval_loss": 2.318600654602051,
      "eval_runtime": 11.0328,
      "eval_samples_per_second": 907.383,
      "eval_steps_per_second": 7.16,
      "step": 7744
    }
  ],
  "max_steps": 21120,
  "num_train_epochs": 30,
  "total_flos": 7.58623645990912e+16,
  "trial_name": null,
  "trial_params": null
}