{
  "best_metric": 0.3943726122379303,
  "best_model_checkpoint": "output/eminem/checkpoint-2400",
  "epoch": 5.0,
  "global_step": 2400,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 7.104344779818036e-05,
      "loss": 0.7686,
      "step": 5
    },
    {
      "epoch": 0.02,
      "learning_rate": 7.326215453099495e-05,
      "loss": 0.6054,
      "step": 10
    },
    {
      "epoch": 0.03,
      "learning_rate": 7.54759713075063e-05,
      "loss": 1.0539,
      "step": 15
    },
    {
      "epoch": 0.04,
      "learning_rate": 7.768257613962603e-05,
      "loss": 0.7369,
      "step": 20
    },
    {
      "epoch": 0.05,
      "learning_rate": 7.987965460359759e-05,
      "loss": 0.7577,
      "step": 25
    },
    {
      "epoch": 0.06,
      "learning_rate": 8.206490226751101e-05,
      "loss": 0.5693,
      "step": 30
    },
    {
      "epoch": 0.07,
      "learning_rate": 8.423602710832802e-05,
      "loss": 0.5772,
      "step": 35
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.63907519158997e-05,
      "loss": 1.6227,
      "step": 40
    },
    {
      "epoch": 0.09,
      "learning_rate": 8.852681668145864e-05,
      "loss": 0.7695,
      "step": 45
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.064198096803333e-05,
      "loss": 0.6463,
      "step": 50
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.27340262603724e-05,
      "loss": 0.7689,
      "step": 55
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.480075829185323e-05,
      "loss": 1.0531,
      "step": 60
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.684000934597923e-05,
      "loss": 0.6097,
      "step": 65
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.884964052999149e-05,
      "loss": 0.7093,
      "step": 70
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.00010082754401829355,
      "loss": 0.5035,
      "step": 75
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.00010277164526325172,
      "loss": 1.0989,
      "step": 80
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.00010467990517112986,
      "loss": 1.0782,
      "step": 85
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.00010655032224079468,
      "loss": 0.8885,
      "step": 90
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.00010838093466302798,
      "loss": 0.9381,
      "step": 95
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.00011016982237818681,
      "loss": 1.0053,
      "step": 100
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.00011191510909009365,
      "loss": 0.6927,
      "step": 105
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.00011361496423399298,
      "loss": 0.9509,
      "step": 110
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0001152676048965721,
      "loss": 0.6696,
      "step": 115
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0001168712976859924,
      "loss": 0.9941,
      "step": 120
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00011842436054997332,
      "loss": 0.5395,
      "step": 125
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00011992516454003592,
      "loss": 0.4659,
      "step": 130
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00012137213552004078,
      "loss": 0.8814,
      "step": 135
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.00012276375581724122,
      "loss": 0.7903,
      "step": 140
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00012409856581412063,
      "loss": 0.8096,
      "step": 145
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00012537516547931506,
      "loss": 0.7711,
      "step": 150
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0001265922158360627,
      "loss": 0.7044,
      "step": 155
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.00012774844036660105,
      "loss": 0.9893,
      "step": 160
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00012884262635106593,
      "loss": 0.9534,
      "step": 165
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00012987362613945392,
      "loss": 0.9657,
      "step": 170
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0001308403583553606,
      "loss": 0.8557,
      "step": 175
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.00013174180903018608,
      "loss": 0.8623,
      "step": 180
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00013257703266665794,
      "loss": 0.9682,
      "step": 185
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.0001333451532305167,
      "loss": 0.8295,
      "step": 190
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.00013404536506936147,
      "loss": 0.6588,
      "step": 195
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0001346769337576652,
      "loss": 0.7481,
      "step": 200
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0001352391968670901,
      "loss": 0.8978,
      "step": 205
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00013573156466127666,
      "loss": 1.1897,
      "step": 210
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0001361535207144,
      "loss": 1.238,
      "step": 215
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.00013650462245282926,
      "loss": 1.1457,
      "step": 220
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00013678450161932478,
      "loss": 0.8702,
      "step": 225
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.00013699286465928987,
      "loss": 1.071,
      "step": 230
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.00013712949302866842,
      "loss": 0.8298,
      "step": 235
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.00013719424342316803,
      "loss": 0.9333,
      "step": 240
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.00013718704792856656,
      "loss": 0.6711,
      "step": 245
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.000137107914091944,
      "loss": 0.786,
      "step": 250
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.00013695692491376736,
      "loss": 0.7021,
      "step": 255
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00013673423876083446,
      "loss": 1.0258,
      "step": 260
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0001364400892001687,
      "loss": 1.2098,
      "step": 265
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00013607478475403945,
      "loss": 1.0657,
      "step": 270
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00013563870857636603,
      "loss": 0.664,
      "step": 275
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00013513231805083994,
      "loss": 0.8538,
      "step": 280
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.00013455614431119578,
      "loss": 0.8278,
      "step": 285
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.00013391079168412009,
      "loss": 0.6271,
      "step": 290
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.00013319693705540213,
      "loss": 0.9119,
      "step": 295
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.00013241532915997056,
      "loss": 0.5942,
      "step": 300
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0001315667877965762,
      "loss": 0.8013,
      "step": 305
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00013065220296793026,
      "loss": 0.6856,
      "step": 310
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.00012967253394722351,
      "loss": 0.8229,
      "step": 315
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.0001286288082719761,
      "loss": 0.9531,
      "step": 320
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00012752212066629144,
      "loss": 1.3199,
      "step": 325
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.00012635363189264417,
      "loss": 1.1022,
      "step": 330
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00012512456753439896,
      "loss": 0.9544,
      "step": 335
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00012383621671034582,
      "loss": 0.9981,
      "step": 340
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00012248993072259197,
      "loss": 0.8137,
      "step": 345
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.00012108712163922375,
      "loss": 0.9984,
      "step": 350
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00011962926081325661,
      "loss": 1.0316,
      "step": 355
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0001181178773393795,
      "loss": 0.8207,
      "step": 360
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.00011655455645015076,
      "loss": 0.6663,
      "step": 365
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.00011494093785329717,
      "loss": 0.8862,
      "step": 370
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00011327871401190472,
      "loss": 0.7749,
      "step": 375
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.00011156962836924195,
      "loss": 1.0212,
      "step": 380
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0001098154735201431,
      "loss": 0.9901,
      "step": 385
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00010801808933080577,
      "loss": 0.6503,
      "step": 390
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0001061793610090449,
      "loss": 0.9526,
      "step": 395
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.00010430121712695815,
      "loss": 0.9911,
      "step": 400
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.00010238562759812645,
      "loss": 0.8466,
      "step": 405
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.00010043460161143304,
      "loss": 0.8538,
      "step": 410
    },
    {
      "epoch": 0.86,
      "learning_rate": 9.845018552372211e-05,
      "loss": 1.0898,
      "step": 415
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.643446071344141e-05,
      "loss": 0.6029,
      "step": 420
    },
    {
      "epoch": 0.88,
      "learning_rate": 9.438954139756169e-05,
      "loss": 0.6565,
      "step": 425
    },
    {
      "epoch": 0.89,
      "learning_rate": 9.231757241406026e-05,
      "loss": 0.9035,
      "step": 430
    },
    {
      "epoch": 0.9,
      "learning_rate": 9.022072697227865e-05,
      "loss": 0.6143,
      "step": 435
    },
    {
      "epoch": 0.91,
      "learning_rate": 8.810120437353156e-05,
      "loss": 0.7718,
      "step": 440
    },
    {
      "epoch": 0.92,
      "learning_rate": 8.596122770434176e-05,
      "loss": 0.9339,
      "step": 445
    },
    {
      "epoch": 0.93,
      "learning_rate": 8.380304150471618e-05,
      "loss": 0.6168,
      "step": 450
    },
    {
      "epoch": 0.94,
      "learning_rate": 8.162890941395579e-05,
      "loss": 0.917,
      "step": 455
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.944111179639565e-05,
      "loss": 0.9396,
      "step": 460
    },
    {
      "epoch": 0.96,
      "learning_rate": 7.724194334962525e-05,
      "loss": 0.8879,
      "step": 465
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.50337106976521e-05,
      "loss": 0.7428,
      "step": 470
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.281872997159678e-05,
      "loss": 0.8685,
      "step": 475
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.059932438036807e-05,
      "loss": 0.6925,
      "step": 480
    },
    {
      "epoch": 1.0,
      "learning_rate": 6.837782177395701e-05,
      "loss": 0.7737,
      "step": 485
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.8009498715400696,
      "eval_runtime": 7.9645,
      "eval_samples_per_second": 81.487,
      "eval_steps_per_second": 10.296,
      "step": 485
    },
    {
      "epoch": 1.01,
      "learning_rate": 6.615655220182037e-05,
      "loss": 0.9648,
      "step": 490
    },
    {
      "epoch": 1.02,
      "learning_rate": 6.393784546900578e-05,
      "loss": 0.7027,
      "step": 495
    },
    {
      "epoch": 1.03,
      "learning_rate": 6.172402869249251e-05,
      "loss": 0.6923,
      "step": 500
    },
    {
      "epoch": 1.04,
      "learning_rate": 5.9517423860374716e-05,
      "loss": 0.6036,
      "step": 505
    },
    {
      "epoch": 1.05,
      "learning_rate": 5.7320345396401225e-05,
      "loss": 0.5738,
      "step": 510
    },
    {
      "epoch": 1.06,
      "learning_rate": 5.5135097732487806e-05,
      "loss": 0.522,
      "step": 515
    },
    {
      "epoch": 1.07,
      "learning_rate": 5.2963972891672695e-05,
      "loss": 0.6331,
      "step": 520
    },
    {
      "epoch": 1.08,
      "learning_rate": 5.080924808409912e-05,
      "loss": 0.8264,
      "step": 525
    },
    {
      "epoch": 1.09,
      "learning_rate": 4.867318331854207e-05,
      "loss": 0.5139,
      "step": 530
    },
    {
      "epoch": 1.1,
      "learning_rate": 4.655801903196737e-05,
      "loss": 0.4191,
      "step": 535
    },
    {
      "epoch": 1.11,
      "learning_rate": 4.4465973739628306e-05,
      "loss": 0.7371,
      "step": 540
    },
    {
      "epoch": 1.12,
      "learning_rate": 4.239924170814745e-05,
      "loss": 0.7683,
      "step": 545
    },
    {
      "epoch": 1.13,
      "learning_rate": 4.0359990654021446e-05,
      "loss": 0.6885,
      "step": 550
    },
    {
      "epoch": 1.14,
      "learning_rate": 3.835035947000743e-05,
      "loss": 0.573,
      "step": 555
    },
    {
      "epoch": 1.15,
      "learning_rate": 3.6372455981705394e-05,
      "loss": 0.5655,
      "step": 560
    },
    {
      "epoch": 1.16,
      "learning_rate": 3.442835473674722e-05,
      "loss": 0.6614,
      "step": 565
    },
    {
      "epoch": 1.18,
      "learning_rate": 3.2520094828869116e-05,
      "loss": 0.553,
      "step": 570
    },
    {
      "epoch": 1.19,
      "learning_rate": 3.064967775920593e-05,
      "loss": 0.9856,
      "step": 575
    },
    {
      "epoch": 1.2,
      "learning_rate": 2.881906533697264e-05,
      "loss": 0.7987,
      "step": 580
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.7030177621813776e-05,
      "loss": 0.4053,
      "step": 585
    },
    {
      "epoch": 1.22,
      "learning_rate": 2.5284890909906917e-05,
      "loss": 0.5987,
      "step": 590
    },
    {
      "epoch": 1.23,
      "learning_rate": 2.3585035766007593e-05,
      "loss": 0.9376,
      "step": 595
    },
    {
      "epoch": 1.24,
      "learning_rate": 2.1932395103427002e-05,
      "loss": 0.6322,
      "step": 600
    },
    {
      "epoch": 1.25,
      "learning_rate": 2.032870231400813e-05,
      "loss": 0.6378,
      "step": 605
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.8775639450025844e-05,
      "loss": 0.5735,
      "step": 610
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.7274835459963273e-05,
      "loss": 1.0382,
      "step": 615
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.582786447995969e-05,
      "loss": 0.6316,
      "step": 620
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.4436244182758039e-05,
      "loss": 0.6119,
      "step": 625
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.3101434185879816e-05,
      "loss": 0.9124,
      "step": 630
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.1824834520685365e-05,
      "loss": 0.5674,
      "step": 635
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.060778416393771e-05,
      "loss": 0.8984,
      "step": 640
    },
    {
      "epoch": 1.33,
      "learning_rate": 9.451559633399314e-06,
      "loss": 0.4093,
      "step": 645
    },
    {
      "epoch": 1.34,
      "learning_rate": 8.35737364893442e-06,
      "loss": 1.0756,
      "step": 650
    },
    {
      "epoch": 1.35,
      "learning_rate": 7.326373860545556e-06,
      "loss": 0.7388,
      "step": 655
    },
    {
      "epoch": 1.36,
      "learning_rate": 6.359641644638905e-06,
      "loss": 0.7588,
      "step": 660
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.4581909698134475e-06,
      "loss": 0.6298,
      "step": 665
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.622967333341646e-06,
      "loss": 0.5269,
      "step": 670
    },
    {
      "epoch": 1.39,
      "learning_rate": 3.854846769483559e-06,
      "loss": 0.8377,
      "step": 675
    },
    {
      "epoch": 1.4,
      "learning_rate": 3.1546349306387612e-06,
      "loss": 0.5796,
      "step": 680
    },
    {
      "epoch": 1.41,
      "learning_rate": 2.5230662423349877e-06,
      "loss": 0.4994,
      "step": 685
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.9608031329100813e-06,
      "loss": 0.659,
      "step": 690
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.4684353387235077e-06,
      "loss": 0.5819,
      "step": 695
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.0464792855997834e-06,
      "loss": 0.5023,
      "step": 700
    },
    {
      "epoch": 1.45,
      "learning_rate": 6.953775471708487e-07,
      "loss": 0.5082,
      "step": 705
    },
    {
      "epoch": 1.46,
      "learning_rate": 4.154983806750897e-07,
      "loss": 0.6839,
      "step": 710
    },
    {
      "epoch": 1.47,
      "learning_rate": 2.071353407100288e-07,
      "loss": 0.7121,
      "step": 715
    },
    {
      "epoch": 1.48,
      "learning_rate": 7.050697133161242e-08,
      "loss": 0.4166,
      "step": 720
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.756576831951343e-09,
      "loss": 0.7094,
      "step": 725
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.295207143343513e-08,
      "loss": 0.6831,
      "step": 730
    },
    {
      "epoch": 1.52,
      "learning_rate": 9.208590805595201e-08,
      "loss": 0.6554,
      "step": 735
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.4307508623258407e-07,
      "loss": 0.3354,
      "step": 740
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.657612391656818e-07,
      "loss": 0.4874,
      "step": 745
    },
    {
      "epoch": 1.55,
      "learning_rate": 7.59910799831185e-07,
      "loss": 0.6073,
      "step": 750
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.1252152459604225e-06,
      "loss": 0.8413,
      "step": 755
    },
    {
      "epoch": 1.57,
      "learning_rate": 1.5612914236342025e-06,
      "loss": 0.3873,
      "step": 760
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.0676819491598904e-06,
      "loss": 0.7879,
      "step": 765
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.643855688804542e-06,
      "loss": 0.6646,
      "step": 770
    },
    {
      "epoch": 1.6,
      "learning_rate": 3.2892083158796966e-06,
      "loss": 0.5732,
      "step": 775
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.003062944597618e-06,
      "loss": 0.5925,
      "step": 780
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.784670840029167e-06,
      "loss": 0.5607,
      "step": 785
    },
    {
      "epoch": 1.63,
      "learning_rate": 5.63321220342353e-06,
      "loss": 0.7195,
      "step": 790
    },
    {
      "epoch": 1.64,
      "learning_rate": 6.547797032069429e-06,
      "loss": 0.4434,
      "step": 795
    },
    {
      "epoch": 1.65,
      "learning_rate": 7.527466052776141e-06,
      "loss": 0.6778,
      "step": 800
    },
    {
      "epoch": 1.66,
      "learning_rate": 8.571191728024455e-06,
      "loss": 0.7627,
      "step": 805
    },
    {
      "epoch": 1.67,
      "learning_rate": 9.677879333708184e-06,
      "loss": 0.7002,
      "step": 810
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.0846368107356494e-05,
      "loss": 0.5221,
      "step": 815
    },
    {
      "epoch": 1.69,
      "learning_rate": 1.2075432465600628e-05,
      "loss": 0.5574,
      "step": 820
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.336378328965373e-05,
      "loss": 0.7088,
      "step": 825
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.4710069277408788e-05,
      "loss": 0.7595,
      "step": 830
    },
    {
      "epoch": 1.72,
      "learning_rate": 1.6112878360775777e-05,
      "loss": 0.3203,
      "step": 835
    },
    {
      "epoch": 1.73,
      "learning_rate": 1.7570739186742892e-05,
      "loss": 0.5597,
      "step": 840
    },
    {
      "epoch": 1.74,
      "learning_rate": 1.9082122660619998e-05,
      "loss": 0.5739,
      "step": 845
    },
    {
      "epoch": 1.75,
      "learning_rate": 2.0645443549848698e-05,
      "loss": 0.531,
      "step": 850
    },
    {
      "epoch": 1.76,
      "learning_rate": 2.2259062146702276e-05,
      "loss": 0.5565,
      "step": 855
    },
    {
      "epoch": 1.77,
      "learning_rate": 2.392128598809619e-05,
      "loss": 0.7898,
      "step": 860
    },
    {
      "epoch": 1.78,
      "learning_rate": 2.5630371630759003e-05,
      "loss": 0.8282,
      "step": 865
    },
    {
      "epoch": 1.79,
      "learning_rate": 2.7384526479857852e-05,
      "loss": 0.6593,
      "step": 870
    },
    {
      "epoch": 1.8,
      "learning_rate": 2.9181910669195226e-05,
      "loss": 0.5809,
      "step": 875
    },
    {
      "epoch": 1.81,
      "learning_rate": 3.102063899095449e-05,
      "loss": 0.5236,
      "step": 880
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.289878287304121e-05,
      "loss": 0.8349,
      "step": 885
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.481437240187291e-05,
      "loss": 0.5913,
      "step": 890
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.6765398388566315e-05,
      "loss": 0.7902,
      "step": 895
    },
    {
      "epoch": 1.86,
      "learning_rate": 3.874981447627722e-05,
      "loss": 0.447,
      "step": 900
    },
    {
      "epoch": 1.87,
      "learning_rate": 4.076553928655969e-05,
      "loss": 0.5064,
      "step": 905
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.281045860243763e-05,
      "loss": 0.6135,
      "step": 910
    },
    {
      "epoch": 1.89,
      "learning_rate": 4.4882427585940885e-05,
      "loss": 0.7355,
      "step": 915
    },
    {
      "epoch": 1.9,
      "learning_rate": 4.69792730277225e-05,
      "loss": 0.6229,
      "step": 920
    },
    {
      "epoch": 1.91,
      "learning_rate": 4.909879562646775e-05,
      "loss": 0.5816,
      "step": 925
    },
    {
      "epoch": 1.92,
      "learning_rate": 5.123877229565941e-05,
      "loss": 0.558,
      "step": 930
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.3396958495283104e-05,
      "loss": 0.5434,
      "step": 935
    },
    {
      "epoch": 1.94,
      "learning_rate": 5.557109058604349e-05,
      "loss": 0.7625,
      "step": 940
    },
    {
      "epoch": 1.95,
      "learning_rate": 5.7758888203603614e-05,
      "loss": 0.6406,
      "step": 945
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.9958056650374005e-05,
      "loss": 0.6691,
      "step": 950
    },
    {
      "epoch": 1.97,
      "learning_rate": 6.216628930234716e-05,
      "loss": 0.6314,
      "step": 955
    },
    {
      "epoch": 1.98,
      "learning_rate": 6.438127002840443e-05,
      "loss": 0.7209,
      "step": 960
    },
    {
      "epoch": 1.99,
      "learning_rate": 6.660067561963313e-05,
      "loss": 0.5188,
      "step": 965
    },
    {
      "epoch": 2.0,
      "learning_rate": 6.882217822604419e-05,
      "loss": 0.5676,
      "step": 970
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.8053014278411865,
      "eval_runtime": 7.9895,
      "eval_samples_per_second": 81.231,
      "eval_steps_per_second": 10.263,
      "step": 970
    },
    {
      "epoch": 2.01,
      "learning_rate": 7.104344779818084e-05,
      "loss": 0.8119,
      "step": 975
    },
    {
      "epoch": 2.02,
      "learning_rate": 7.326215453099347e-05,
      "loss": 0.8241,
      "step": 980
    },
    {
      "epoch": 2.03,
      "learning_rate": 7.547597130750675e-05,
      "loss": 0.4897,
      "step": 985
    },
    {
      "epoch": 2.04,
      "learning_rate": 7.768257613962455e-05,
      "loss": 0.7487,
      "step": 990
    },
    {
      "epoch": 2.05,
      "learning_rate": 7.987965460359805e-05,
      "loss": 0.7651,
      "step": 995
    },
    {
      "epoch": 2.06,
      "learning_rate": 8.206490226751148e-05,
      "loss": 0.6692,
      "step": 1000
    },
    {
      "epoch": 2.07,
      "learning_rate": 8.423602710832848e-05,
      "loss": 0.7985,
      "step": 1005
    },
    {
      "epoch": 2.08,
      "learning_rate": 8.639075191590017e-05,
      "loss": 0.5855,
      "step": 1010
    },
    {
      "epoch": 2.09,
      "learning_rate": 8.852681668145908e-05,
      "loss": 0.6353,
      "step": 1015
    },
    {
      "epoch": 2.1,
      "learning_rate": 9.064198096803377e-05,
      "loss": 0.552,
      "step": 1020
    },
    {
      "epoch": 2.11,
      "learning_rate": 9.273402626037102e-05,
      "loss": 0.4755,
      "step": 1025
    },
    {
      "epoch": 2.12,
      "learning_rate": 9.480075829185367e-05,
      "loss": 0.6665,
      "step": 1030
    },
    {
      "epoch": 2.13,
      "learning_rate": 9.684000934597788e-05,
      "loss": 0.3769,
      "step": 1035
    },
    {
      "epoch": 2.14,
      "learning_rate": 9.884964052999192e-05,
      "loss": 0.6127,
      "step": 1040
    },
    {
      "epoch": 2.15,
      "learning_rate": 0.00010082754401829396,
      "loss": 0.7927,
      "step": 1045
    },
    {
      "epoch": 2.16,
      "learning_rate": 0.00010277164526325214,
      "loss": 0.5635,
      "step": 1050
    },
    {
      "epoch": 2.18,
      "learning_rate": 0.00010467990517113026,
      "loss": 0.694,
      "step": 1055
    },
    {
      "epoch": 2.19,
      "learning_rate": 0.00010655032224079507,
      "loss": 0.75,
      "step": 1060
    },
    {
      "epoch": 2.2,
      "learning_rate": 0.00010838093466302835,
      "loss": 0.4615,
      "step": 1065
    },
    {
      "epoch": 2.21,
      "learning_rate": 0.00011016982237818718,
      "loss": 0.4177,
      "step": 1070
    },
    {
      "epoch": 2.22,
      "learning_rate": 0.00011191510909009402,
      "loss": 0.7636,
      "step": 1075
    },
    {
      "epoch": 2.23,
      "learning_rate": 0.00011361496423399185,
      "loss": 0.5608,
      "step": 1080
    },
    {
      "epoch": 2.24,
      "learning_rate": 0.00011526760489657245,
      "loss": 0.4228,
      "step": 1085
    },
    {
      "epoch": 2.25,
      "learning_rate": 0.00011687129768599134,
      "loss": 0.6887,
      "step": 1090
    },
    {
      "epoch": 2.26,
      "learning_rate": 0.00011842436054997366,
      "loss": 0.4605,
      "step": 1095
    },
    {
      "epoch": 2.27,
      "learning_rate": 0.00011992516454003623,
      "loss": 0.5089,
      "step": 1100
    },
    {
      "epoch": 2.28,
      "learning_rate": 0.00012137213552004109,
      "loss": 0.7078,
      "step": 1105
    },
    {
      "epoch": 2.29,
      "learning_rate": 0.00012276375581724152,
      "loss": 0.5178,
      "step": 1110
    },
    {
      "epoch": 2.3,
      "learning_rate": 0.0001240985658141209,
      "loss": 0.6848,
      "step": 1115
    },
    {
      "epoch": 2.31,
      "learning_rate": 0.0001253751654793153,
      "loss": 0.5177,
      "step": 1120
    },
    {
      "epoch": 2.32,
      "learning_rate": 0.00012659221583606188,
      "loss": 0.8266,
      "step": 1125
    },
    {
      "epoch": 2.33,
      "learning_rate": 0.0001277484403666013,
      "loss": 0.5183,
      "step": 1130
    },
    {
      "epoch": 2.34,
      "learning_rate": 0.00012884262635106522,
      "loss": 0.8944,
      "step": 1135
    },
    {
      "epoch": 2.35,
      "learning_rate": 0.00012987362613945414,
      "loss": 0.8195,
      "step": 1140
    },
    {
      "epoch": 2.36,
      "learning_rate": 0.00013084035835536078,
      "loss": 0.8404,
      "step": 1145
    },
    {
      "epoch": 2.37,
      "learning_rate": 0.00013174180903018627,
      "loss": 0.6088,
      "step": 1150
    },
    {
      "epoch": 2.38,
      "learning_rate": 0.00013257703266665808,
      "loss": 0.6745,
      "step": 1155
    },
    {
      "epoch": 2.39,
      "learning_rate": 0.00013334515323051684,
      "loss": 0.6451,
      "step": 1160
    },
    {
      "epoch": 2.4,
      "learning_rate": 0.0001340453650693616,
      "loss": 0.7023,
      "step": 1165
    },
    {
      "epoch": 2.41,
      "learning_rate": 0.00013467693375766531,
      "loss": 0.8257,
      "step": 1170
    },
    {
      "epoch": 2.42,
      "learning_rate": 0.0001352391968670902,
      "loss": 0.5857,
      "step": 1175
    },
    {
      "epoch": 2.43,
      "learning_rate": 0.00013573156466127633,
      "loss": 0.3617,
      "step": 1180
    },
    {
      "epoch": 2.44,
      "learning_rate": 0.0001361535207144001,
      "loss": 0.7152,
      "step": 1185
    },
    {
      "epoch": 2.45,
      "learning_rate": 0.00013650462245282905,
      "loss": 0.7494,
      "step": 1190
    },
    {
      "epoch": 2.46,
      "learning_rate": 0.0001367845016193248,
      "loss": 0.5441,
      "step": 1195
    },
    {
      "epoch": 2.47,
      "learning_rate": 0.00013699286465928992,
      "loss": 0.916,
      "step": 1200
    },
    {
      "epoch": 2.48,
      "learning_rate": 0.00013712949302866845,
      "loss": 0.6266,
      "step": 1205
    },
    {
      "epoch": 2.49,
      "learning_rate": 0.00013719424342316803,
      "loss": 0.8704,
      "step": 1210
    },
    {
      "epoch": 2.51,
      "learning_rate": 0.00013718704792856654,
      "loss": 0.6092,
      "step": 1215
    },
    {
      "epoch": 2.52,
      "learning_rate": 0.00013710791409194398,
      "loss": 0.6756,
      "step": 1220
    },
    {
      "epoch": 2.53,
      "learning_rate": 0.00013695692491376746,
      "loss": 0.6538,
      "step": 1225
    },
    {
      "epoch": 2.54,
      "learning_rate": 0.0001367342387608344,
      "loss": 0.5702,
      "step": 1230
    },
    {
      "epoch": 2.55,
      "learning_rate": 0.0001364400892001689,
      "loss": 0.8016,
      "step": 1235
    },
    {
      "epoch": 2.56,
      "learning_rate": 0.00013607478475403972,
      "loss": 0.6645,
      "step": 1240
    },
    {
      "epoch": 2.57,
      "learning_rate": 0.00013563870857636595,
      "loss": 0.9163,
      "step": 1245
    },
    {
      "epoch": 2.58,
      "learning_rate": 0.0001351323180508398,
      "loss": 0.6785,
      "step": 1250
    },
    {
      "epoch": 2.59,
      "learning_rate": 0.00013455614431119567,
      "loss": 0.7625,
      "step": 1255
    },
    {
      "epoch": 2.6,
      "learning_rate": 0.00013391079168411995,
      "loss": 0.5317,
      "step": 1260
    },
    {
      "epoch": 2.61,
      "learning_rate": 0.00013319693705540197,
      "loss": 0.6176,
      "step": 1265
    },
    {
      "epoch": 2.62,
      "learning_rate": 0.0001324153291599711,
      "loss": 0.4044,
      "step": 1270
    },
    {
      "epoch": 2.63,
      "learning_rate": 0.000131566787796576,
      "loss": 0.6815,
      "step": 1275
    },
    {
      "epoch": 2.64,
      "learning_rate": 0.0001306522029679309,
      "loss": 0.7421,
      "step": 1280
    },
    {
      "epoch": 2.65,
      "learning_rate": 0.0001296725339472242,
      "loss": 0.9262,
      "step": 1285
    },
    {
      "epoch": 2.66,
      "learning_rate": 0.00012862880827197592,
      "loss": 0.5789,
      "step": 1290
    },
    {
      "epoch": 2.67,
      "learning_rate": 0.0001275221206662922,
      "loss": 0.6176,
      "step": 1295
    },
    {
      "epoch": 2.68,
      "learning_rate": 0.00012635363189264392,
      "loss": 0.5932,
      "step": 1300
    },
    {
      "epoch": 2.69,
      "learning_rate": 0.00012512456753439868,
      "loss": 0.5771,
      "step": 1305
    },
    {
      "epoch": 2.7,
      "learning_rate": 0.00012383621671034555,
      "loss": 0.5197,
      "step": 1310
    },
    {
      "epoch": 2.71,
      "learning_rate": 0.00012248993072259167,
      "loss": 0.5347,
      "step": 1315
    },
    {
      "epoch": 2.72,
      "learning_rate": 0.00012108712163922345,
      "loss": 0.7577,
      "step": 1320
    },
    {
      "epoch": 2.73,
      "learning_rate": 0.0001196292608132576,
      "loss": 0.6237,
      "step": 1325
    },
    {
      "epoch": 2.74,
      "learning_rate": 0.00011811787733938051,
      "loss": 0.9522,
      "step": 1330
    },
    {
      "epoch": 2.75,
      "learning_rate": 0.00011655455645015182,
      "loss": 0.6511,
      "step": 1335
    },
    {
      "epoch": 2.76,
      "learning_rate": 0.00011494093785329826,
      "loss": 0.7683,
      "step": 1340
    },
    {
      "epoch": 2.77,
      "learning_rate": 0.00011327871401190438,
      "loss": 0.8408,
      "step": 1345
    },
    {
      "epoch": 2.78,
      "learning_rate": 0.00011156962836924156,
      "loss": 0.7132,
      "step": 1350
    },
    {
      "epoch": 2.79,
      "learning_rate": 0.00010981547352014274,
      "loss": 0.7741,
      "step": 1355
    },
    {
      "epoch": 2.8,
      "learning_rate": 0.00010801808933080536,
      "loss": 0.6007,
      "step": 1360
    },
    {
      "epoch": 2.81,
      "learning_rate": 0.0001061793610090445,
      "loss": 0.7367,
      "step": 1365
    },
    {
      "epoch": 2.82,
      "learning_rate": 0.00010430121712695943,
      "loss": 0.7996,
      "step": 1370
    },
    {
      "epoch": 2.84,
      "learning_rate": 0.00010238562759812602,
      "loss": 0.5623,
      "step": 1375
    },
    {
      "epoch": 2.85,
      "learning_rate": 0.00010043460161143261,
      "loss": 0.5343,
      "step": 1380
    },
    {
      "epoch": 2.86,
      "learning_rate": 9.845018552372344e-05,
      "loss": 1.0444,
      "step": 1385
    },
    {
      "epoch": 2.87,
      "learning_rate": 9.643446071344098e-05,
      "loss": 0.6887,
      "step": 1390
    },
    {
      "epoch": 2.88,
      "learning_rate": 9.438954139756306e-05,
      "loss": 0.8487,
      "step": 1395
    },
    {
      "epoch": 2.89,
      "learning_rate": 9.231757241405981e-05,
      "loss": 0.7323,
      "step": 1400
    },
    {
      "epoch": 2.9,
      "learning_rate": 9.02207269722782e-05,
      "loss": 0.6256,
      "step": 1405
    },
    {
      "epoch": 2.91,
      "learning_rate": 8.81012043735311e-05,
      "loss": 0.832,
      "step": 1410
    },
    {
      "epoch": 2.92,
      "learning_rate": 8.59612277043413e-05,
      "loss": 0.539,
      "step": 1415
    },
    {
      "epoch": 2.93,
      "learning_rate": 8.380304150471572e-05,
      "loss": 0.6064,
      "step": 1420
    },
    {
      "epoch": 2.94,
      "learning_rate": 8.162890941395533e-05,
      "loss": 0.6207,
      "step": 1425
    },
    {
      "epoch": 2.95,
      "learning_rate": 7.944111179639712e-05,
      "loss": 0.5802,
      "step": 1430
    },
    {
      "epoch": 2.96,
      "learning_rate": 7.72419433496248e-05,
      "loss": 0.6887,
      "step": 1435
    },
    {
      "epoch": 2.97,
      "learning_rate": 7.503371069765357e-05,
      "loss": 0.8306,
      "step": 1440
    },
    {
      "epoch": 2.98,
      "learning_rate": 7.28187299715963e-05,
      "loss": 0.4574,
      "step": 1445
    },
    {
      "epoch": 2.99,
      "learning_rate": 7.059932438036761e-05,
      "loss": 0.6537,
      "step": 1450
    },
    {
      "epoch": 3.0,
      "learning_rate": 6.837782177395655e-05,
      "loss": 0.8464,
      "step": 1455
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.7975038886070251,
      "eval_runtime": 7.9678,
      "eval_samples_per_second": 81.452,
      "eval_steps_per_second": 10.291,
      "step": 1455
    },
    {
      "epoch": 3.02,
      "learning_rate": 3.533340602884326e-05,
      "loss": 0.6584,
      "step": 1460
    },
    {
      "epoch": 3.03,
      "learning_rate": 3.340418885898272e-05,
      "loss": 0.9188,
      "step": 1465
    },
    {
      "epoch": 3.04,
      "learning_rate": 3.1512039922545005e-05,
      "loss": 0.7885,
      "step": 1470
    },
    {
      "epoch": 3.05,
      "learning_rate": 2.9658952031065493e-05,
      "loss": 0.6785,
      "step": 1475
    },
    {
      "epoch": 3.06,
      "learning_rate": 2.784687685697823e-05,
      "loss": 0.9316,
      "step": 1480
    },
    {
      "epoch": 3.07,
      "learning_rate": 2.6077722878123704e-05,
      "loss": 0.7074,
      "step": 1485
    },
    {
      "epoch": 3.08,
      "learning_rate": 2.4353353367727083e-05,
      "loss": 0.5083,
      "step": 1490
    },
    {
      "epoch": 3.09,
      "learning_rate": 2.2675584432009507e-05,
      "loss": 0.7031,
      "step": 1495
    },
    {
      "epoch": 3.1,
      "learning_rate": 2.104618309745989e-05,
      "loss": 0.5955,
      "step": 1500
    },
    {
      "epoch": 3.11,
      "learning_rate": 1.94668654497983e-05,
      "loss": 0.6515,
      "step": 1505
    },
    {
      "epoch": 3.12,
      "learning_rate": 1.7939294826601462e-05,
      "loss": 0.6982,
      "step": 1510
    },
    {
      "epoch": 3.13,
      "learning_rate": 1.6465080065474947e-05,
      "loss": 0.7575,
      "step": 1515
    },
    {
      "epoch": 3.14,
      "learning_rate": 1.504577380961979e-05,
      "loss": 0.7294,
      "step": 1520
    },
    {
      "epoch": 3.15,
      "learning_rate": 1.3682870872588951e-05,
      "loss": 0.566,
      "step": 1525
    },
    {
      "epoch": 3.16,
      "learning_rate": 1.2377806663955526e-05,
      "loss": 0.4467,
      "step": 1530
    },
    {
      "epoch": 3.17,
      "learning_rate": 1.1131955677535227e-05,
      "loss": 0.8649,
      "step": 1535
    },
    {
      "epoch": 3.18,
      "learning_rate": 9.946630043766268e-06,
      "loss": 0.6318,
      "step": 1540
    },
    {
      "epoch": 3.19,
      "learning_rate": 8.8230781477809e-06,
      "loss": 0.5197,
      "step": 1545
    },
    {
      "epoch": 3.2,
      "learning_rate": 7.762483314597675e-06,
      "loss": 0.7195,
      "step": 1550
    },
    {
      "epoch": 3.21,
      "learning_rate": 6.765962562849582e-06,
      "loss": 0.6544,
      "step": 1555
    },
    {
      "epoch": 3.22,
      "learning_rate": 5.834565428339295e-06,
      "loss": 0.6373,
      "step": 1560
    },
    {
      "epoch": 3.23,
      "learning_rate": 4.969272858664371e-06,
      "loss": 0.7969,
      "step": 1565
    },
    {
      "epoch": 3.24,
      "learning_rate": 4.170996180083756e-06,
      "loss": 0.5398,
      "step": 1570
    },
    {
      "epoch": 3.25,
      "learning_rate": 3.440576137712397e-06,
      "loss": 0.6392,
      "step": 1575
    },
    {
      "epoch": 3.26,
      "learning_rate": 2.778782010045892e-06,
      "loss": 0.8183,
      "step": 1580
    },
    {
      "epoch": 3.27,
      "learning_rate": 2.186310798754732e-06,
      "loss": 0.4459,
      "step": 1585
    },
    {
      "epoch": 3.29,
      "learning_rate": 1.6637864946060192e-06,
      "loss": 0.6059,
      "step": 1590
    },
    {
      "epoch": 3.3,
      "learning_rate": 1.2117594202713427e-06,
      "loss": 0.9818,
      "step": 1595
    },
    {
      "epoch": 3.31,
      "learning_rate": 8.307056507336243e-07,
      "loss": 0.3983,
      "step": 1600
    },
    {
      "epoch": 3.32,
      "learning_rate": 5.21026511876941e-07,
      "loss": 0.9094,
      "step": 1605
    },
    {
      "epoch": 3.33,
      "learning_rate": 2.8304815781598153e-07,
      "loss": 0.5006,
      "step": 1610
    },
    {
      "epoch": 3.34,
      "learning_rate": 1.1702122738749632e-07,
      "loss": 0.5685,
      "step": 1615
    },
    {
      "epoch": 3.35,
      "learning_rate": 2.312058017904961e-08,
      "loss": 0.7558,
      "step": 1620
    },
    {
      "epoch": 3.36,
      "learning_rate": 1.445112365939849e-09,
      "loss": 0.5613,
      "step": 1625
    },
    {
      "epoch": 3.37,
      "learning_rate": 5.2017652553981476e-08,
      "loss": 0.7474,
      "step": 1630
    },
    {
      "epoch": 3.38,
      "learning_rate": 1.7478493773641084e-07,
      "loss": 0.7388,
      "step": 1635
    },
    {
      "epoch": 3.39,
      "learning_rate": 3.69617669391192e-07,
      "loss": 0.5788,
      "step": 1640
    },
    {
      "epoch": 3.4,
      "learning_rate": 6.363106496559428e-07,
      "loss": 0.6158,
      "step": 1645
    },
    {
      "epoch": 3.41,
      "learning_rate": 9.745829974457439e-07,
      "loss": 0.6697,
      "step": 1650
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.3840784442740293e-06,
      "loss": 0.5538,
      "step": 1655
    },
    {
      "epoch": 3.43,
      "learning_rate": 1.8643657094772689e-06,
      "loss": 0.7529,
      "step": 1660
    },
    {
      "epoch": 3.44,
      "learning_rate": 2.4149389544374383e-06,
      "loss": 0.5164,
      "step": 1665
    },
    {
      "epoch": 3.45,
      "learning_rate": 3.035218315332177e-06,
      "loss": 0.5686,
      "step": 1670
    },
    {
      "epoch": 3.46,
      "learning_rate": 3.7245505138493062e-06,
      "loss": 0.7307,
      "step": 1675
    },
    {
      "epoch": 3.47,
      "learning_rate": 4.482209545215727e-06,
      "loss": 0.4993,
      "step": 1680
    },
    {
      "epoch": 3.48,
      "learning_rate": 5.307397442832051e-06,
      "loss": 0.5905,
      "step": 1685
    },
    {
      "epoch": 3.49,
      "learning_rate": 6.199245118679951e-06,
      "loss": 0.7372,
      "step": 1690
    },
    {
      "epoch": 3.5,
      "learning_rate": 7.15681327866027e-06,
      "loss": 0.7588,
      "step": 1695
    },
    {
      "epoch": 3.51,
      "learning_rate": 8.179093411845718e-06,
      "loss": 0.7609,
      "step": 1700
    },
    {
      "epoch": 3.52,
      "learning_rate": 9.265008852652901e-06,
      "loss": 0.4208,
      "step": 1705
    },
    {
      "epoch": 3.53,
      "learning_rate": 1.0413415914780494e-05,
      "loss": 0.5304,
      "step": 1710
    },
    {
      "epoch": 3.54,
      "learning_rate": 1.1623105095742268e-05,
      "loss": 0.47,
      "step": 1715
    },
    {
      "epoch": 3.55,
      "learning_rate": 1.2892802350718737e-05,
      "loss": 0.4479,
      "step": 1720
    },
    {
      "epoch": 3.56,
      "learning_rate": 1.4221170434376129e-05,
      "loss": 0.3907,
      "step": 1725
    },
    {
      "epoch": 3.57,
      "learning_rate": 1.5606810309252903e-05,
      "loss": 0.8871,
      "step": 1730
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.7048262619230822e-05,
      "loss": 0.5595,
      "step": 1735
    },
    {
      "epoch": 3.6,
      "learning_rate": 1.8544009226529548e-05,
      "loss": 0.6169,
      "step": 1740
    },
    {
      "epoch": 3.61,
      "learning_rate": 2.0092474810601755e-05,
      "loss": 0.557,
      "step": 1745
    },
    {
      "epoch": 3.62,
      "learning_rate": 2.169202852727143e-05,
      "loss": 0.8663,
      "step": 1750
    },
    {
      "epoch": 3.63,
      "learning_rate": 2.334098572632982e-05,
      "loss": 0.754,
      "step": 1755
    },
    {
      "epoch": 3.64,
      "learning_rate": 2.5037609725813137e-05,
      "loss": 0.7106,
      "step": 1760
    },
    {
      "epoch": 3.65,
      "learning_rate": 2.6780113641080935e-05,
      "loss": 0.8902,
      "step": 1765
    },
    {
      "epoch": 3.66,
      "learning_rate": 2.8566662266755925e-05,
      "loss": 0.5382,
      "step": 1770
    },
    {
      "epoch": 3.67,
      "learning_rate": 3.039537400956214e-05,
      "loss": 0.5907,
      "step": 1775
    },
    {
      "epoch": 3.68,
      "learning_rate": 3.226432287002555e-05,
      "loss": 1.0195,
      "step": 1780
    },
    {
      "epoch": 3.69,
      "learning_rate": 3.417154047093731e-05,
      "loss": 0.5186,
      "step": 1785
    },
    {
      "epoch": 3.7,
      "learning_rate": 3.611501813044067e-05,
      "loss": 0.4889,
      "step": 1790
    },
    {
      "epoch": 3.71,
      "learning_rate": 3.8092708977579776e-05,
      "loss": 0.6288,
      "step": 1795
    },
    {
      "epoch": 3.72,
      "learning_rate": 4.010253010806977e-05,
      "loss": 0.7543,
      "step": 1800
    },
    {
      "epoch": 3.73,
      "learning_rate": 4.214236477799971e-05,
      "loss": 0.5513,
      "step": 1805
    },
    {
      "epoch": 3.74,
      "learning_rate": 4.421006463320525e-05,
      "loss": 0.6435,
      "step": 1810
    },
    {
      "epoch": 3.75,
      "learning_rate": 4.630345197188684e-05,
      "loss": 0.4894,
      "step": 1815
    },
    {
      "epoch": 3.76,
      "learning_rate": 4.8420322038201655e-05,
      "loss": 0.4861,
      "step": 1820
    },
    {
      "epoch": 3.77,
      "learning_rate": 5.0558445344283775e-05,
      "loss": 0.634,
      "step": 1825
    },
    {
      "epoch": 3.78,
      "learning_rate": 5.2715570018359465e-05,
      "loss": 0.5037,
      "step": 1830
    },
    {
      "epoch": 3.79,
      "learning_rate": 5.48894241764059e-05,
      "loss": 0.6078,
      "step": 1835
    },
    {
      "epoch": 3.8,
      "learning_rate": 5.70777183149052e-05,
      "loss": 0.7133,
      "step": 1840
    },
    {
      "epoch": 3.81,
      "learning_rate": 5.927814772215574e-05,
      "loss": 0.6985,
      "step": 1845
    },
    {
      "epoch": 3.82,
      "learning_rate": 6.148839490558139e-05,
      "loss": 0.6508,
      "step": 1850
    },
    {
      "epoch": 3.83,
      "learning_rate": 6.370613203253432e-05,
      "loss": 0.4598,
      "step": 1855
    },
    {
      "epoch": 3.84,
      "learning_rate": 6.592902338193773e-05,
      "loss": 0.428,
      "step": 1860
    },
    {
      "epoch": 3.85,
      "learning_rate": 6.815472780430632e-05,
      "loss": 0.5706,
      "step": 1865
    },
    {
      "epoch": 3.86,
      "learning_rate": 7.038090118741434e-05,
      "loss": 0.9235,
      "step": 1870
    },
    {
      "epoch": 3.87,
      "learning_rate": 7.260519892513427e-05,
      "loss": 0.8443,
      "step": 1875
    },
    {
      "epoch": 3.88,
      "learning_rate": 7.482527838676278e-05,
      "loss": 0.5691,
      "step": 1880
    },
    {
      "epoch": 3.89,
      "learning_rate": 7.703880138428551e-05,
      "loss": 0.4075,
      "step": 1885
    },
    {
      "epoch": 3.9,
      "learning_rate": 7.924343663496286e-05,
      "loss": 0.4282,
      "step": 1890
    },
    {
      "epoch": 3.92,
      "learning_rate": 8.14368622166238e-05,
      "loss": 0.6324,
      "step": 1895
    },
    {
      "epoch": 3.93,
      "learning_rate": 8.361676801313385e-05,
      "loss": 0.7816,
      "step": 1900
    },
    {
      "epoch": 3.94,
      "learning_rate": 8.578085814737905e-05,
      "loss": 0.6076,
      "step": 1905
    },
    {
      "epoch": 3.95,
      "learning_rate": 8.792685339932241e-05,
      "loss": 0.5257,
      "step": 1910
    },
    {
      "epoch": 3.96,
      "learning_rate": 9.005249360644818e-05,
      "loss": 0.6216,
      "step": 1915
    },
    {
      "epoch": 3.97,
      "learning_rate": 9.215554004418427e-05,
      "loss": 0.7805,
      "step": 1920
    },
    {
      "epoch": 3.98,
      "learning_rate": 9.423377778371423e-05,
      "loss": 0.7339,
      "step": 1925
    },
    {
      "epoch": 3.99,
      "learning_rate": 9.628501802474621e-05,
      "loss": 0.6319,
      "step": 1930
    },
    {
      "epoch": 4.0,
      "learning_rate": 9.830710040076315e-05,
      "loss": 0.5267,
      "step": 1935
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.4826279878616333,
      "eval_runtime": 14.7153,
      "eval_samples_per_second": 44.716,
      "eval_steps_per_second": 5.64,
      "step": 1936
    },
    {
      "epoch": 4.04,
      "learning_rate": 0.00011450235959621898,
      "loss": 0.3751,
      "step": 1940
    },
    {
      "epoch": 4.05,
      "learning_rate": 0.00011280977272542054,
      "loss": 0.7741,
      "step": 1945
    },
    {
      "epoch": 4.06,
      "learning_rate": 0.00011106984492265664,
      "loss": 0.6279,
      "step": 1950
    },
    {
      "epoch": 4.07,
      "learning_rate": 0.00010928443934619291,
      "loss": 0.4172,
      "step": 1955
    },
    {
      "epoch": 4.08,
      "learning_rate": 0.00010745546785304314,
      "loss": 0.8202,
      "step": 1960
    },
    {
      "epoch": 4.09,
      "learning_rate": 0.0001055848889516962,
      "loss": 0.7253,
      "step": 1965
    },
    {
      "epoch": 4.1,
      "learning_rate": 0.00010367470570488607,
      "loss": 0.6737,
      "step": 1970
    },
    {
      "epoch": 4.11,
      "learning_rate": 0.00010172696358468322,
      "loss": 0.602,
      "step": 1975
    },
    {
      "epoch": 4.12,
      "learning_rate": 9.974374828213395e-05,
      "loss": 0.7079,
      "step": 1980
    },
    {
      "epoch": 4.14,
      "learning_rate": 9.772718347385836e-05,
      "loss": 0.8964,
      "step": 1985
    },
    {
      "epoch": 4.15,
      "learning_rate": 9.567942854796078e-05,
      "loss": 0.4818,
      "step": 1990
    },
    {
      "epoch": 4.16,
      "learning_rate": 9.360267629171191e-05,
      "loss": 0.7278,
      "step": 1995
    },
    {
      "epoch": 4.17,
      "learning_rate": 9.149915054343752e-05,
      "loss": 0.6072,
      "step": 2000
    },
    {
      "epoch": 4.18,
      "learning_rate": 8.937110381119999e-05,
      "loss": 0.8248,
      "step": 2005
    },
    {
      "epoch": 4.19,
      "learning_rate": 8.722081486074574e-05,
      "loss": 0.8012,
      "step": 2010
    },
    {
      "epoch": 4.2,
      "learning_rate": 8.50505862753343e-05,
      "loss": 0.542,
      "step": 2015
    },
    {
      "epoch": 4.21,
      "learning_rate": 8.286274199009886e-05,
      "loss": 0.5911,
      "step": 2020
    },
    {
      "epoch": 4.22,
      "learning_rate": 8.06596248034977e-05,
      "loss": 0.6034,
      "step": 2025
    },
    {
      "epoch": 4.23,
      "learning_rate": 7.844359386859629e-05,
      "loss": 0.6177,
      "step": 2030
    },
    {
      "epoch": 4.24,
      "learning_rate": 7.621702216682333e-05,
      "loss": 0.7298,
      "step": 2035
    },
    {
      "epoch": 4.25,
      "learning_rate": 7.398229396693239e-05,
      "loss": 0.51,
      "step": 2040
    },
    {
      "epoch": 4.26,
      "learning_rate": 7.174180227184603e-05,
      "loss": 0.8855,
      "step": 2045
    },
    {
      "epoch": 4.27,
      "learning_rate": 6.94979462561934e-05,
      "loss": 0.6728,
      "step": 2050
    },
    {
      "epoch": 4.28,
      "learning_rate": 6.725312869720243e-05,
      "loss": 0.4899,
      "step": 2055
    },
    {
      "epoch": 4.29,
      "learning_rate": 6.500975340173405e-05,
      "loss": 0.4827,
      "step": 2060
    },
    {
      "epoch": 4.3,
      "learning_rate": 6.27702226322509e-05,
      "loss": 0.6521,
      "step": 2065
    },
    {
      "epoch": 4.31,
      "learning_rate": 6.05369345343951e-05,
      "loss": 0.5115,
      "step": 2070
    },
    {
      "epoch": 4.32,
      "learning_rate": 5.831228056899374e-05,
      "loss": 0.3792,
      "step": 2075
    },
    {
      "epoch": 4.33,
      "learning_rate": 5.609864295123839e-05,
      "loss": 0.5564,
      "step": 2080
    },
    {
      "epoch": 4.34,
      "learning_rate": 5.389839209973954e-05,
      "loss": 0.8714,
      "step": 2085
    },
    {
      "epoch": 4.35,
      "learning_rate": 5.171388409821159e-05,
      "loss": 0.4843,
      "step": 2090
    },
    {
      "epoch": 4.36,
      "learning_rate": 4.9547458172542385e-05,
      "loss": 0.7155,
      "step": 2095
    },
    {
      "epoch": 4.38,
      "learning_rate": 4.74014341858799e-05,
      "loss": 0.4362,
      "step": 2100
    },
    {
      "epoch": 4.39,
      "learning_rate": 4.5278110154455465e-05,
      "loss": 0.4663,
      "step": 2105
    },
    {
      "epoch": 4.4,
      "learning_rate": 4.317975978682545e-05,
      "loss": 0.5994,
      "step": 2110
    },
    {
      "epoch": 4.41,
      "learning_rate": 4.110863004912692e-05,
      "loss": 0.4463,
      "step": 2115
    },
    {
      "epoch": 4.42,
      "learning_rate": 3.9066938758951916e-05,
      "loss": 0.6226,
      "step": 2120
    },
    {
      "epoch": 4.43,
      "learning_rate": 3.7056872210476396e-05,
      "loss": 0.5947,
      "step": 2125
    },
    {
      "epoch": 4.44,
      "learning_rate": 3.508058283331054e-05,
      "loss": 0.5276,
      "step": 2130
    },
    {
      "epoch": 4.45,
      "learning_rate": 3.314018688761322e-05,
      "loss": 0.5118,
      "step": 2135
    },
    {
      "epoch": 4.46,
      "learning_rate": 3.123776219796979e-05,
      "loss": 0.6581,
      "step": 2140
    },
    {
      "epoch": 4.47,
      "learning_rate": 2.937534592838677e-05,
      "loss": 0.6129,
      "step": 2145
    },
    {
      "epoch": 4.48,
      "learning_rate": 2.7554932400854067e-05,
      "loss": 0.6305,
      "step": 2150
    },
    {
      "epoch": 4.49,
      "learning_rate": 2.577847095977183e-05,
      "loss": 0.6368,
      "step": 2155
    },
    {
      "epoch": 4.5,
      "learning_rate": 2.4047863884551244e-05,
      "loss": 0.6684,
      "step": 2160
    },
    {
      "epoch": 4.51,
      "learning_rate": 2.2364964352589423e-05,
      "loss": 0.7136,
      "step": 2165
    },
    {
      "epoch": 4.52,
      "learning_rate": 2.0731574454860283e-05,
      "loss": 0.5243,
      "step": 2170
    },
    {
      "epoch": 4.53,
      "learning_rate": 1.914944326618096e-05,
      "loss": 0.665,
      "step": 2175
    },
    {
      "epoch": 4.54,
      "learning_rate": 1.7620264972250878e-05,
      "loss": 0.5977,
      "step": 2180
    },
    {
      "epoch": 4.55,
      "learning_rate": 1.6145677055492958e-05,
      "loss": 0.7612,
      "step": 2185
    },
    {
      "epoch": 4.56,
      "learning_rate": 1.4727258541581522e-05,
      "loss": 0.7172,
      "step": 2190
    },
    {
      "epoch": 4.57,
      "learning_rate": 1.3366528308588766e-05,
      "loss": 0.6571,
      "step": 2195
    },
    {
      "epoch": 4.58,
      "learning_rate": 1.2064943460529662e-05,
      "loss": 0.664,
      "step": 2200
    },
    {
      "epoch": 4.59,
      "learning_rate": 1.0823897767063955e-05,
      "loss": 1.1318,
      "step": 2205
    },
    {
      "epoch": 4.6,
      "learning_rate": 9.644720171000711e-06,
      "loss": 0.5895,
      "step": 2210
    },
    {
      "epoch": 4.61,
      "learning_rate": 8.52867336524718e-06,
      "loss": 0.5683,
      "step": 2215
    },
    {
      "epoch": 4.62,
      "learning_rate": 7.476952440678663e-06,
      "loss": 0.8862,
      "step": 2220
    },
    {
      "epoch": 4.64,
      "learning_rate": 6.490683606400273e-06,
      "loss": 0.5336,
      "step": 2225
    },
    {
      "epoch": 4.65,
      "learning_rate": 5.570922983785744e-06,
      "loss": 0.7185,
      "step": 2230
    },
    {
      "epoch": 4.66,
      "learning_rate": 4.718655475547088e-06,
      "loss": 0.6121,
      "step": 2235
    },
    {
      "epoch": 4.67,
      "learning_rate": 3.934793711076979e-06,
      "loss": 0.4995,
      "step": 2240
    },
    {
      "epoch": 4.68,
      "learning_rate": 3.2201770691879365e-06,
      "loss": 0.639,
      "step": 2245
    },
    {
      "epoch": 4.69,
      "learning_rate": 2.5755707792803602e-06,
      "loss": 0.6703,
      "step": 2250
    },
    {
      "epoch": 4.7,
      "learning_rate": 2.001665101912862e-06,
      "loss": 0.5835,
      "step": 2255
    },
    {
      "epoch": 4.71,
      "learning_rate": 1.499074589660724e-06,
      "loss": 0.8827,
      "step": 2260
    },
    {
      "epoch": 4.72,
      "learning_rate": 1.0683374290336727e-06,
      "loss": 0.521,
      "step": 2265
    },
    {
      "epoch": 4.73,
      "learning_rate": 7.099148641708803e-07,
      "loss": 0.5759,
      "step": 2270
    },
    {
      "epoch": 4.74,
      "learning_rate": 4.2419070293249417e-07,
      "loss": 0.4576,
      "step": 2275
    },
    {
      "epoch": 4.75,
      "learning_rate": 2.1147090590762733e-07,
      "loss": 0.9183,
      "step": 2280
    },
    {
      "epoch": 4.76,
      "learning_rate": 7.19832587822494e-08,
      "loss": 0.6811,
      "step": 2285
    },
    {
      "epoch": 4.77,
      "learning_rate": 5.877128423112166e-09,
      "loss": 0.6599,
      "step": 2290
    },
    {
      "epoch": 4.78,
      "learning_rate": 1.3223302930315373e-08,
      "loss": 0.8066,
      "step": 2295
    },
    {
      "epoch": 4.79,
      "learning_rate": 9.401391583621219e-08,
      "loss": 0.93,
      "step": 2300
    },
    {
      "epoch": 4.8,
      "learning_rate": 2.4816245452881085e-07,
      "loss": 0.7526,
      "step": 2305
    },
    {
      "epoch": 4.81,
      "learning_rate": 4.755038528919192e-07,
      "loss": 0.6209,
      "step": 2310
    },
    {
      "epoch": 4.82,
      "learning_rate": 7.757946680618806e-07,
      "loss": 0.5472,
      "step": 2315
    },
    {
      "epoch": 4.83,
      "learning_rate": 1.1487133411127057e-06,
      "loss": 0.5128,
      "step": 2320
    },
    {
      "epoch": 4.84,
      "learning_rate": 1.5938605413876288e-06,
      "loss": 0.5971,
      "step": 2325
    },
    {
      "epoch": 4.85,
      "learning_rate": 2.1107595941164267e-06,
      "loss": 0.9001,
      "step": 2330
    },
    {
      "epoch": 4.86,
      "learning_rate": 2.698856990845814e-06,
      "loss": 0.624,
      "step": 2335
    },
    {
      "epoch": 4.88,
      "learning_rate": 3.3575229821519346e-06,
      "loss": 0.469,
      "step": 2340
    },
    {
      "epoch": 4.89,
      "learning_rate": 4.086052251995796e-06,
      "loss": 0.7627,
      "step": 2345
    },
    {
      "epoch": 4.9,
      "learning_rate": 4.883664672985776e-06,
      "loss": 0.5373,
      "step": 2350
    },
    {
      "epoch": 4.91,
      "learning_rate": 5.749506141764069e-06,
      "loss": 0.684,
      "step": 2355
    },
    {
      "epoch": 4.92,
      "learning_rate": 6.682649493598944e-06,
      "loss": 0.6701,
      "step": 2360
    },
    {
      "epoch": 4.93,
      "learning_rate": 7.682095495217462e-06,
      "loss": 0.5721,
      "step": 2365
    },
    {
      "epoch": 4.94,
      "learning_rate": 8.746773914805009e-06,
      "loss": 0.5432,
      "step": 2370
    },
    {
      "epoch": 4.95,
      "learning_rate": 9.875544668046303e-06,
      "loss": 0.6582,
      "step": 2375
    },
    {
      "epoch": 4.96,
      "learning_rate": 1.1067199038944353e-05,
      "loss": 0.5341,
      "step": 2380
    },
    {
      "epoch": 4.97,
      "learning_rate": 1.2320460974148876e-05,
      "loss": 0.677,
      "step": 2385
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.3633988449393231e-05,
      "loss": 0.5661,
      "step": 2390
    },
    {
      "epoch": 4.99,
      "learning_rate": 1.5006374906553291e-05,
      "loss": 0.4745,
      "step": 2395
    },
    {
      "epoch": 5.0,
      "learning_rate": 1.6436150759836066e-05,
      "loss": 0.746,
      "step": 2400
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.3943726122379303,
      "eval_runtime": 15.4496,
      "eval_samples_per_second": 44.661,
      "eval_steps_per_second": 5.631,
      "step": 2400
    }
  ],
  "max_steps": 2400,
  "num_train_epochs": 5,
  "total_flos": 2505790586880000.0,
  "trial_name": null,
  "trial_params": null
}