{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.026995757809487,
  "global_step": 13000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "ep_loss": 0.0,
      "epoch": 0.01,
      "learning_rate": 5e-07,
      "loss": 10.4763,
      "mlm_loss": 10.4763,
      "step": 10
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.02,
      "learning_rate": 1e-06,
      "loss": 10.3052,
      "mlm_loss": 10.3052,
      "step": 20
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.02,
      "learning_rate": 1.5e-06,
      "loss": 9.9641,
      "mlm_loss": 9.9641,
      "step": 30
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.03,
      "learning_rate": 2e-06,
      "loss": 9.5354,
      "mlm_loss": 9.5354,
      "step": 40
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.04,
      "learning_rate": 2.5e-06,
      "loss": 9.1346,
      "mlm_loss": 9.1346,
      "step": 50
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.05,
      "learning_rate": 3e-06,
      "loss": 8.8162,
      "mlm_loss": 8.8162,
      "step": 60
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.05,
      "learning_rate": 3.5e-06,
      "loss": 8.558,
      "mlm_loss": 8.558,
      "step": 70
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.06,
      "learning_rate": 4e-06,
      "loss": 8.3459,
      "mlm_loss": 8.3459,
      "step": 80
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.07,
      "learning_rate": 4.5e-06,
      "loss": 8.175,
      "mlm_loss": 8.175,
      "step": 90
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.08,
      "learning_rate": 5e-06,
      "loss": 8.0255,
      "mlm_loss": 8.0255,
      "step": 100
    },
    {
      "epoch": 0.08,
      "eval_ep_loss": -0.14438027143478394,
      "eval_loss": 7.826408386230469,
      "eval_mlm_loss": 7.826408386230469,
      "eval_runtime": 63.5165,
      "eval_samples_per_second": 1100.014,
      "eval_steps_per_second": 0.551,
      "step": 100
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.08,
      "learning_rate": 5.5e-06,
      "loss": 7.8979,
      "mlm_loss": 7.8979,
      "step": 110
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.09,
      "learning_rate": 6e-06,
      "loss": 7.7543,
      "mlm_loss": 7.7543,
      "step": 120
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.1,
      "learning_rate": 6.5e-06,
      "loss": 7.5942,
      "mlm_loss": 7.5942,
      "step": 130
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.11,
      "learning_rate": 7e-06,
      "loss": 7.4304,
      "mlm_loss": 7.4304,
      "step": 140
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.12,
      "learning_rate": 7.5e-06,
      "loss": 7.1862,
      "mlm_loss": 7.1862,
      "step": 150
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.12,
      "learning_rate": 8e-06,
      "loss": 6.924,
      "mlm_loss": 6.924,
      "step": 160
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.13,
      "learning_rate": 8.500000000000002e-06,
      "loss": 6.6521,
      "mlm_loss": 6.6521,
      "step": 170
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.14,
      "learning_rate": 9e-06,
      "loss": 6.3639,
      "mlm_loss": 6.3639,
      "step": 180
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.15,
      "learning_rate": 9.5e-06,
      "loss": 6.0915,
      "mlm_loss": 6.0915,
      "step": 190
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.15,
      "learning_rate": 1e-05,
      "loss": 5.8272,
      "mlm_loss": 5.8272,
      "step": 200
    },
    {
      "epoch": 0.15,
      "eval_ep_loss": -0.4913683831691742,
      "eval_loss": 5.531888008117676,
      "eval_mlm_loss": 5.531888008117676,
      "eval_runtime": 60.9938,
      "eval_samples_per_second": 1145.51,
      "eval_steps_per_second": 0.574,
      "step": 200
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.16,
      "learning_rate": 1.0500000000000001e-05,
      "loss": 5.5749,
      "mlm_loss": 5.5749,
      "step": 210
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.17,
      "learning_rate": 1.1e-05,
      "loss": 5.3413,
      "mlm_loss": 5.3413,
      "step": 220
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.18,
      "learning_rate": 1.15e-05,
      "loss": 5.142,
      "mlm_loss": 5.142,
      "step": 230
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.19,
      "learning_rate": 1.2e-05,
      "loss": 4.925,
      "mlm_loss": 4.925,
      "step": 240
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.19,
      "learning_rate": 1.25e-05,
      "loss": 4.7334,
      "mlm_loss": 4.7334,
      "step": 250
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.2,
      "learning_rate": 1.3e-05,
      "loss": 4.5702,
      "mlm_loss": 4.5702,
      "step": 260
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.21,
      "learning_rate": 1.35e-05,
      "loss": 4.4388,
      "mlm_loss": 4.4388,
      "step": 270
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.22,
      "learning_rate": 1.4e-05,
      "loss": 4.288,
      "mlm_loss": 4.288,
      "step": 280
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.22,
      "learning_rate": 1.4500000000000002e-05,
      "loss": 4.1403,
      "mlm_loss": 4.1403,
      "step": 290
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.23,
      "learning_rate": 1.5e-05,
      "loss": 4.0232,
      "mlm_loss": 4.0232,
      "step": 300
    },
    {
      "epoch": 0.23,
      "eval_ep_loss": -1.1010401248931885,
      "eval_loss": 3.8623688220977783,
      "eval_mlm_loss": 3.8623688220977783,
      "eval_runtime": 61.8594,
      "eval_samples_per_second": 1129.481,
      "eval_steps_per_second": 0.566,
      "step": 300
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.24,
      "learning_rate": 1.55e-05,
      "loss": 3.9135,
      "mlm_loss": 3.9135,
      "step": 310
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.25,
      "learning_rate": 1.6e-05,
      "loss": 3.8252,
      "mlm_loss": 3.8252,
      "step": 320
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.25,
      "learning_rate": 1.65e-05,
      "loss": 3.7462,
      "mlm_loss": 3.7462,
      "step": 330
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.26,
      "learning_rate": 1.7000000000000003e-05,
      "loss": 3.6491,
      "mlm_loss": 3.6491,
      "step": 340
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.27,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 3.5815,
      "mlm_loss": 3.5815,
      "step": 350
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.28,
      "learning_rate": 1.8e-05,
      "loss": 3.5211,
      "mlm_loss": 3.5211,
      "step": 360
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.29,
      "learning_rate": 1.85e-05,
      "loss": 3.452,
      "mlm_loss": 3.452,
      "step": 370
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.29,
      "learning_rate": 1.9e-05,
      "loss": 3.3747,
      "mlm_loss": 3.3747,
      "step": 380
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.3,
      "learning_rate": 1.95e-05,
      "loss": 3.3163,
      "mlm_loss": 3.3163,
      "step": 390
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.31,
      "learning_rate": 2e-05,
      "loss": 3.2679,
      "mlm_loss": 3.2679,
      "step": 400
    },
    {
      "epoch": 0.31,
      "eval_ep_loss": -1.4546582698822021,
      "eval_loss": 3.1360955238342285,
      "eval_mlm_loss": 3.1360955238342285,
      "eval_runtime": 59.2888,
      "eval_samples_per_second": 1178.451,
      "eval_steps_per_second": 0.59,
      "step": 400
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.32,
      "learning_rate": 2.05e-05,
      "loss": 3.1831,
      "mlm_loss": 3.1831,
      "step": 410
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.32,
      "learning_rate": 2.1000000000000002e-05,
      "loss": 3.1246,
      "mlm_loss": 3.1246,
      "step": 420
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.33,
      "learning_rate": 2.1499999999999997e-05,
      "loss": 3.0658,
      "mlm_loss": 3.0658,
      "step": 430
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.34,
      "learning_rate": 2.2e-05,
      "loss": 3.0032,
      "mlm_loss": 3.0032,
      "step": 440
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.35,
      "learning_rate": 2.2499999999999998e-05,
      "loss": 2.931,
      "mlm_loss": 2.931,
      "step": 450
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.35,
      "learning_rate": 2.3e-05,
      "loss": 2.904,
      "mlm_loss": 2.904,
      "step": 460
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.36,
      "learning_rate": 2.3500000000000002e-05,
      "loss": 2.8526,
      "mlm_loss": 2.8526,
      "step": 470
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.37,
      "learning_rate": 2.4e-05,
      "loss": 2.8336,
      "mlm_loss": 2.8336,
      "step": 480
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.38,
      "learning_rate": 2.4500000000000003e-05,
      "loss": 2.7953,
      "mlm_loss": 2.7953,
      "step": 490
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.39,
      "learning_rate": 2.5e-05,
      "loss": 2.7569,
      "mlm_loss": 2.7569,
      "step": 500
    },
    {
      "epoch": 0.39,
      "eval_ep_loss": -1.5984127521514893,
      "eval_loss": 2.6636624336242676,
      "eval_mlm_loss": 2.6636624336242676,
      "eval_runtime": 64.3177,
      "eval_samples_per_second": 1086.31,
      "eval_steps_per_second": 0.544,
      "step": 500
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.39,
      "learning_rate": 2.55e-05,
      "loss": 2.7157,
      "mlm_loss": 2.7157,
      "step": 510
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.4,
      "learning_rate": 2.6e-05,
      "loss": 2.6857,
      "mlm_loss": 2.6857,
      "step": 520
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.41,
      "learning_rate": 2.65e-05,
      "loss": 2.6796,
      "mlm_loss": 2.6796,
      "step": 530
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.42,
      "learning_rate": 2.7e-05,
      "loss": 2.6449,
      "mlm_loss": 2.6449,
      "step": 540
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.42,
      "learning_rate": 2.75e-05,
      "loss": 2.5932,
      "mlm_loss": 2.5932,
      "step": 550
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.43,
      "learning_rate": 2.8e-05,
      "loss": 2.5536,
      "mlm_loss": 2.5536,
      "step": 560
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.44,
      "learning_rate": 2.85e-05,
      "loss": 2.5271,
      "mlm_loss": 2.5271,
      "step": 570
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.45,
      "learning_rate": 2.9000000000000004e-05,
      "loss": 2.5057,
      "mlm_loss": 2.5057,
      "step": 580
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.46,
      "learning_rate": 2.95e-05,
      "loss": 2.476,
      "mlm_loss": 2.476,
      "step": 590
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.46,
      "learning_rate": 3e-05,
      "loss": 2.4265,
      "mlm_loss": 2.4265,
      "step": 600
    },
    {
      "epoch": 0.46,
      "eval_ep_loss": -1.7573626041412354,
      "eval_loss": 2.359508752822876,
      "eval_mlm_loss": 2.359508752822876,
      "eval_runtime": 59.8647,
      "eval_samples_per_second": 1167.114,
      "eval_steps_per_second": 0.585,
      "step": 600
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.47,
      "learning_rate": 3.05e-05,
      "loss": 2.4061,
      "mlm_loss": 2.4061,
      "step": 610
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.48,
      "learning_rate": 3.1e-05,
      "loss": 2.4071,
      "mlm_loss": 2.4071,
      "step": 620
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.49,
      "learning_rate": 3.15e-05,
      "loss": 2.3728,
      "mlm_loss": 2.3728,
      "step": 630
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.49,
      "learning_rate": 3.2e-05,
      "loss": 2.3465,
      "mlm_loss": 2.3465,
      "step": 640
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.5,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 2.3065,
      "mlm_loss": 2.3065,
      "step": 650
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.51,
      "learning_rate": 3.3e-05,
      "loss": 2.312,
      "mlm_loss": 2.312,
      "step": 660
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.52,
      "learning_rate": 3.35e-05,
      "loss": 2.29,
      "mlm_loss": 2.29,
      "step": 670
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.52,
      "learning_rate": 3.4000000000000007e-05,
      "loss": 2.2939,
      "mlm_loss": 2.2939,
      "step": 680
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.53,
      "learning_rate": 3.4500000000000005e-05,
      "loss": 2.2496,
      "mlm_loss": 2.2496,
      "step": 690
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.54,
      "learning_rate": 3.5000000000000004e-05,
      "loss": 2.2482,
      "mlm_loss": 2.2482,
      "step": 700
    },
    {
      "epoch": 0.54,
      "eval_ep_loss": -1.7850463390350342,
      "eval_loss": 2.1652746200561523,
      "eval_mlm_loss": 2.1652746200561523,
      "eval_runtime": 63.4238,
      "eval_samples_per_second": 1101.621,
      "eval_steps_per_second": 0.552,
      "step": 700
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.55,
      "learning_rate": 3.5499999999999996e-05,
      "loss": 2.2379,
      "mlm_loss": 2.2379,
      "step": 710
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.56,
      "learning_rate": 3.6e-05,
      "loss": 2.2088,
      "mlm_loss": 2.2088,
      "step": 720
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.56,
      "learning_rate": 3.65e-05,
      "loss": 2.1767,
      "mlm_loss": 2.1767,
      "step": 730
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.57,
      "learning_rate": 3.7e-05,
      "loss": 2.1589,
      "mlm_loss": 2.1589,
      "step": 740
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.58,
      "learning_rate": 3.75e-05,
      "loss": 2.1297,
      "mlm_loss": 2.1297,
      "step": 750
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.59,
      "learning_rate": 3.8e-05,
      "loss": 2.1128,
      "mlm_loss": 2.1128,
      "step": 760
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.59,
      "learning_rate": 3.85e-05,
      "loss": 2.0861,
      "mlm_loss": 2.0861,
      "step": 770
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.6,
      "learning_rate": 3.9e-05,
      "loss": 2.0557,
      "mlm_loss": 2.0557,
      "step": 780
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.61,
      "learning_rate": 3.95e-05,
      "loss": 2.0596,
      "mlm_loss": 2.0596,
      "step": 790
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.62,
      "learning_rate": 4e-05,
      "loss": 2.0349,
      "mlm_loss": 2.0349,
      "step": 800
    },
    {
      "epoch": 0.62,
      "eval_ep_loss": -1.7692768573760986,
      "eval_loss": 1.916082501411438,
      "eval_mlm_loss": 1.916082501411438,
      "eval_runtime": 62.7772,
      "eval_samples_per_second": 1112.968,
      "eval_steps_per_second": 0.558,
      "step": 800
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.62,
      "learning_rate": 4.05e-05,
      "loss": 1.9934,
      "mlm_loss": 1.9934,
      "step": 810
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.63,
      "learning_rate": 4.1e-05,
      "loss": 1.9554,
      "mlm_loss": 1.9554,
      "step": 820
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.64,
      "learning_rate": 4.1500000000000006e-05,
      "loss": 1.9348,
      "mlm_loss": 1.9348,
      "step": 830
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.65,
      "learning_rate": 4.2000000000000004e-05,
      "loss": 1.8978,
      "mlm_loss": 1.8978,
      "step": 840
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.66,
      "learning_rate": 4.25e-05,
      "loss": 1.8752,
      "mlm_loss": 1.8752,
      "step": 850
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.66,
      "learning_rate": 4.2999999999999995e-05,
      "loss": 1.8466,
      "mlm_loss": 1.8466,
      "step": 860
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.67,
      "learning_rate": 4.35e-05,
      "loss": 1.8168,
      "mlm_loss": 1.8168,
      "step": 870
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.68,
      "learning_rate": 4.4e-05,
      "loss": 1.8161,
      "mlm_loss": 1.8161,
      "step": 880
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.69,
      "learning_rate": 4.45e-05,
      "loss": 1.7637,
      "mlm_loss": 1.7637,
      "step": 890
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.69,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 1.7292,
      "mlm_loss": 1.7292,
      "step": 900
    },
    {
      "epoch": 0.69,
      "eval_ep_loss": -1.673869252204895,
      "eval_loss": 1.6480634212493896,
      "eval_mlm_loss": 1.6480634212493896,
      "eval_runtime": 61.0031,
      "eval_samples_per_second": 1145.335,
      "eval_steps_per_second": 0.574,
      "step": 900
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.7,
      "learning_rate": 4.55e-05,
      "loss": 1.7031,
      "mlm_loss": 1.7031,
      "step": 910
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.71,
      "learning_rate": 4.6e-05,
      "loss": 1.6951,
      "mlm_loss": 1.6951,
      "step": 920
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.72,
      "learning_rate": 4.65e-05,
      "loss": 1.6706,
      "mlm_loss": 1.6706,
      "step": 930
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.73,
      "learning_rate": 4.7000000000000004e-05,
      "loss": 1.6674,
      "mlm_loss": 1.6674,
      "step": 940
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.73,
      "learning_rate": 4.75e-05,
      "loss": 1.6446,
      "mlm_loss": 1.6446,
      "step": 950
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.74,
      "learning_rate": 4.8e-05,
      "loss": 1.6344,
      "mlm_loss": 1.6344,
      "step": 960
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.75,
      "learning_rate": 4.85e-05,
      "loss": 1.6182,
      "mlm_loss": 1.6182,
      "step": 970
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.76,
      "learning_rate": 4.9000000000000005e-05,
      "loss": 1.602,
      "mlm_loss": 1.602,
      "step": 980
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.76,
      "learning_rate": 4.9500000000000004e-05,
      "loss": 1.5965,
      "mlm_loss": 1.5965,
      "step": 990
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.77,
      "learning_rate": 5e-05,
      "loss": 1.6022,
      "mlm_loss": 1.6022,
      "step": 1000
    },
    {
      "epoch": 0.77,
      "eval_ep_loss": -1.7339978218078613,
      "eval_loss": 1.521234393119812,
      "eval_mlm_loss": 1.521234393119812,
      "eval_runtime": 61.977,
      "eval_samples_per_second": 1127.338,
      "eval_steps_per_second": 0.565,
      "step": 1000
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.78,
      "learning_rate": 5.05e-05,
      "loss": 1.5856,
      "mlm_loss": 1.5856,
      "step": 1010
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.79,
      "learning_rate": 5.1e-05,
      "loss": 1.5641,
      "mlm_loss": 1.5641,
      "step": 1020
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.79,
      "learning_rate": 5.15e-05,
      "loss": 1.5455,
      "mlm_loss": 1.5455,
      "step": 1030
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.8,
      "learning_rate": 5.2e-05,
      "loss": 1.5587,
      "mlm_loss": 1.5587,
      "step": 1040
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.81,
      "learning_rate": 5.25e-05,
      "loss": 1.5495,
      "mlm_loss": 1.5495,
      "step": 1050
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.82,
      "learning_rate": 5.3e-05,
      "loss": 1.5388,
      "mlm_loss": 1.5388,
      "step": 1060
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.83,
      "learning_rate": 5.35e-05,
      "loss": 1.5085,
      "mlm_loss": 1.5085,
      "step": 1070
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.83,
      "learning_rate": 5.4e-05,
      "loss": 1.5258,
      "mlm_loss": 1.5258,
      "step": 1080
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.84,
      "learning_rate": 5.45e-05,
      "loss": 1.5027,
      "mlm_loss": 1.5027,
      "step": 1090
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.85,
      "learning_rate": 5.5e-05,
      "loss": 1.4991,
      "mlm_loss": 1.4991,
      "step": 1100
    },
    {
      "epoch": 0.85,
      "eval_ep_loss": -1.8915724754333496,
      "eval_loss": 1.4428980350494385,
      "eval_mlm_loss": 1.4428980350494385,
      "eval_runtime": 60.2275,
      "eval_samples_per_second": 1160.085,
      "eval_steps_per_second": 0.581,
      "step": 1100
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.86,
      "learning_rate": 5.55e-05,
      "loss": 1.5068,
      "mlm_loss": 1.5068,
      "step": 1110
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.86,
      "learning_rate": 5.6e-05,
      "loss": 1.4812,
      "mlm_loss": 1.4812,
      "step": 1120
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.87,
      "learning_rate": 5.6500000000000005e-05,
      "loss": 1.4765,
      "mlm_loss": 1.4765,
      "step": 1130
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.88,
      "learning_rate": 5.7e-05,
      "loss": 1.4679,
      "mlm_loss": 1.4679,
      "step": 1140
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.89,
      "learning_rate": 5.75e-05,
      "loss": 1.4631,
      "mlm_loss": 1.4631,
      "step": 1150
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.89,
      "learning_rate": 5.800000000000001e-05,
      "loss": 1.4562,
      "mlm_loss": 1.4562,
      "step": 1160
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.9,
      "learning_rate": 5.8500000000000006e-05,
      "loss": 1.4542,
      "mlm_loss": 1.4542,
      "step": 1170
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.91,
      "learning_rate": 5.9e-05,
      "loss": 1.4266,
      "mlm_loss": 1.4266,
      "step": 1180
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.92,
      "learning_rate": 5.9499999999999996e-05,
      "loss": 1.4181,
      "mlm_loss": 1.4181,
      "step": 1190
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.93,
      "learning_rate": 6e-05,
      "loss": 1.4133,
      "mlm_loss": 1.4133,
      "step": 1200
    },
    {
      "epoch": 0.93,
      "eval_ep_loss": -1.9453246593475342,
      "eval_loss": 1.3670274019241333,
      "eval_mlm_loss": 1.3670274019241333,
      "eval_runtime": 61.6667,
      "eval_samples_per_second": 1133.011,
      "eval_steps_per_second": 0.568,
      "step": 1200
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.93,
      "learning_rate": 6.05e-05,
      "loss": 1.414,
      "mlm_loss": 1.414,
      "step": 1210
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.94,
      "learning_rate": 6.1e-05,
      "loss": 1.4032,
      "mlm_loss": 1.4032,
      "step": 1220
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.95,
      "learning_rate": 6.15e-05,
      "loss": 1.3958,
      "mlm_loss": 1.3958,
      "step": 1230
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.96,
      "learning_rate": 6.2e-05,
      "loss": 1.3952,
      "mlm_loss": 1.3952,
      "step": 1240
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.96,
      "learning_rate": 6.25e-05,
      "loss": 1.3829,
      "mlm_loss": 1.3829,
      "step": 1250
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.97,
      "learning_rate": 6.3e-05,
      "loss": 1.3915,
      "mlm_loss": 1.3915,
      "step": 1260
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.98,
      "learning_rate": 6.35e-05,
      "loss": 1.3881,
      "mlm_loss": 1.3881,
      "step": 1270
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.99,
      "learning_rate": 6.4e-05,
      "loss": 1.3915,
      "mlm_loss": 1.3915,
      "step": 1280
    },
    {
      "ep_loss": 0.0,
      "epoch": 0.99,
      "learning_rate": 6.450000000000001e-05,
      "loss": 1.3728,
      "mlm_loss": 1.3728,
      "step": 1290
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.0,
      "learning_rate": 6.500000000000001e-05,
      "loss": 1.3379,
      "mlm_loss": 1.3379,
      "step": 1300
    },
    {
      "epoch": 1.0,
      "eval_ep_loss": -2.041435480117798,
      "eval_loss": 1.2956568002700806,
      "eval_mlm_loss": 1.2956568002700806,
      "eval_runtime": 58.5906,
      "eval_samples_per_second": 1192.495,
      "eval_steps_per_second": 0.597,
      "step": 1300
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.01,
      "learning_rate": 6.55e-05,
      "loss": 1.3488,
      "mlm_loss": 1.3488,
      "step": 1310
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.02,
      "learning_rate": 6.6e-05,
      "loss": 1.3408,
      "mlm_loss": 1.3408,
      "step": 1320
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.03,
      "learning_rate": 6.65e-05,
      "loss": 1.3318,
      "mlm_loss": 1.3318,
      "step": 1330
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.03,
      "learning_rate": 6.7e-05,
      "loss": 1.3216,
      "mlm_loss": 1.3216,
      "step": 1340
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.04,
      "learning_rate": 6.75e-05,
      "loss": 1.3142,
      "mlm_loss": 1.3142,
      "step": 1350
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.05,
      "learning_rate": 6.800000000000001e-05,
      "loss": 1.3075,
      "mlm_loss": 1.3075,
      "step": 1360
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.06,
      "learning_rate": 6.850000000000001e-05,
      "loss": 1.312,
      "mlm_loss": 1.312,
      "step": 1370
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.06,
      "learning_rate": 6.900000000000001e-05,
      "loss": 1.2987,
      "mlm_loss": 1.2987,
      "step": 1380
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.07,
      "learning_rate": 6.950000000000001e-05,
      "loss": 1.2822,
      "mlm_loss": 1.2822,
      "step": 1390
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.08,
      "learning_rate": 7.000000000000001e-05,
      "loss": 1.2817,
      "mlm_loss": 1.2817,
      "step": 1400
    },
    {
      "epoch": 1.08,
      "eval_ep_loss": -2.044208288192749,
      "eval_loss": 1.2174968719482422,
      "eval_mlm_loss": 1.2174968719482422,
      "eval_runtime": 62.7874,
      "eval_samples_per_second": 1112.788,
      "eval_steps_per_second": 0.557,
      "step": 1400
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.09,
      "learning_rate": 7.049999999999999e-05,
      "loss": 1.2673,
      "mlm_loss": 1.2673,
      "step": 1410
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.1,
      "learning_rate": 7.099999999999999e-05,
      "loss": 1.28,
      "mlm_loss": 1.28,
      "step": 1420
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.1,
      "learning_rate": 7.149999999999999e-05,
      "loss": 1.2606,
      "mlm_loss": 1.2606,
      "step": 1430
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.11,
      "learning_rate": 7.2e-05,
      "loss": 1.2439,
      "mlm_loss": 1.2439,
      "step": 1440
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.12,
      "learning_rate": 7.25e-05,
      "loss": 1.2428,
      "mlm_loss": 1.2428,
      "step": 1450
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.13,
      "learning_rate": 7.3e-05,
      "loss": 1.2342,
      "mlm_loss": 1.2342,
      "step": 1460
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.13,
      "learning_rate": 7.35e-05,
      "loss": 1.2391,
      "mlm_loss": 1.2391,
      "step": 1470
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.14,
      "learning_rate": 7.4e-05,
      "loss": 1.2337,
      "mlm_loss": 1.2337,
      "step": 1480
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.15,
      "learning_rate": 7.45e-05,
      "loss": 1.2161,
      "mlm_loss": 1.2161,
      "step": 1490
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.16,
      "learning_rate": 7.5e-05,
      "loss": 1.2083,
      "mlm_loss": 1.2083,
      "step": 1500
    },
    {
      "epoch": 1.16,
      "eval_ep_loss": -2.247138738632202,
      "eval_loss": 1.145897388458252,
      "eval_mlm_loss": 1.145897388458252,
      "eval_runtime": 60.0275,
      "eval_samples_per_second": 1163.951,
      "eval_steps_per_second": 0.583,
      "step": 1500
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.16,
      "learning_rate": 7.55e-05,
      "loss": 1.2048,
      "mlm_loss": 1.2048,
      "step": 1510
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.17,
      "learning_rate": 7.6e-05,
      "loss": 1.1934,
      "mlm_loss": 1.1934,
      "step": 1520
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.18,
      "learning_rate": 7.65e-05,
      "loss": 1.1741,
      "mlm_loss": 1.1741,
      "step": 1530
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.19,
      "learning_rate": 7.7e-05,
      "loss": 1.1832,
      "mlm_loss": 1.1832,
      "step": 1540
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.2,
      "learning_rate": 7.75e-05,
      "loss": 1.1759,
      "mlm_loss": 1.1759,
      "step": 1550
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.2,
      "learning_rate": 7.8e-05,
      "loss": 1.1776,
      "mlm_loss": 1.1776,
      "step": 1560
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.21,
      "learning_rate": 7.85e-05,
      "loss": 1.164,
      "mlm_loss": 1.164,
      "step": 1570
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.22,
      "learning_rate": 7.9e-05,
      "loss": 1.1447,
      "mlm_loss": 1.1447,
      "step": 1580
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.23,
      "learning_rate": 7.950000000000001e-05,
      "loss": 1.1669,
      "mlm_loss": 1.1669,
      "step": 1590
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.23,
      "learning_rate": 8e-05,
      "loss": 1.1331,
      "mlm_loss": 1.1331,
      "step": 1600
    },
    {
      "epoch": 1.23,
      "eval_ep_loss": -2.236558198928833,
      "eval_loss": 1.0778173208236694,
      "eval_mlm_loss": 1.0778173208236694,
      "eval_runtime": 61.8774,
      "eval_samples_per_second": 1129.152,
      "eval_steps_per_second": 0.566,
      "step": 1600
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.24,
      "learning_rate": 8.05e-05,
      "loss": 1.1332,
      "mlm_loss": 1.1332,
      "step": 1610
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.25,
      "learning_rate": 8.1e-05,
      "loss": 1.1288,
      "mlm_loss": 1.1288,
      "step": 1620
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.26,
      "learning_rate": 8.15e-05,
      "loss": 1.1321,
      "mlm_loss": 1.1321,
      "step": 1630
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.26,
      "learning_rate": 8.2e-05,
      "loss": 1.11,
      "mlm_loss": 1.11,
      "step": 1640
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.27,
      "learning_rate": 8.25e-05,
      "loss": 1.1083,
      "mlm_loss": 1.1083,
      "step": 1650
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.28,
      "learning_rate": 8.300000000000001e-05,
      "loss": 1.0999,
      "mlm_loss": 1.0999,
      "step": 1660
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.29,
      "learning_rate": 8.350000000000001e-05,
      "loss": 1.1041,
      "mlm_loss": 1.1041,
      "step": 1670
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.3,
      "learning_rate": 8.400000000000001e-05,
      "loss": 1.0823,
      "mlm_loss": 1.0823,
      "step": 1680
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.3,
      "learning_rate": 8.450000000000001e-05,
      "loss": 1.0809,
      "mlm_loss": 1.0809,
      "step": 1690
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.31,
      "learning_rate": 8.5e-05,
      "loss": 1.0859,
      "mlm_loss": 1.0859,
      "step": 1700
    },
    {
      "epoch": 1.31,
      "eval_ep_loss": -2.2500903606414795,
      "eval_loss": 1.0151546001434326,
      "eval_mlm_loss": 1.0151546001434326,
      "eval_runtime": 60.0928,
      "eval_samples_per_second": 1162.685,
      "eval_steps_per_second": 0.582,
      "step": 1700
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.32,
      "learning_rate": 8.55e-05,
      "loss": 1.0757,
      "mlm_loss": 1.0757,
      "step": 1710
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.33,
      "learning_rate": 8.599999999999999e-05,
      "loss": 1.0638,
      "mlm_loss": 1.0638,
      "step": 1720
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.33,
      "learning_rate": 8.65e-05,
      "loss": 1.0554,
      "mlm_loss": 1.0554,
      "step": 1730
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.34,
      "learning_rate": 8.7e-05,
      "loss": 1.0624,
      "mlm_loss": 1.0624,
      "step": 1740
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.35,
      "learning_rate": 8.75e-05,
      "loss": 1.0483,
      "mlm_loss": 1.0483,
      "step": 1750
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.36,
      "learning_rate": 8.8e-05,
      "loss": 1.0432,
      "mlm_loss": 1.0432,
      "step": 1760
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.37,
      "learning_rate": 8.85e-05,
      "loss": 1.0464,
      "mlm_loss": 1.0464,
      "step": 1770
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.37,
      "learning_rate": 8.9e-05,
      "loss": 1.0379,
      "mlm_loss": 1.0379,
      "step": 1780
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.38,
      "learning_rate": 8.95e-05,
      "loss": 1.0416,
      "mlm_loss": 1.0416,
      "step": 1790
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.39,
      "learning_rate": 8.999999999999999e-05,
      "loss": 1.0178,
      "mlm_loss": 1.0178,
      "step": 1800
    },
    {
      "epoch": 1.39,
      "eval_ep_loss": -2.3243699073791504,
      "eval_loss": 0.9662984013557434,
      "eval_mlm_loss": 0.9662984013557434,
      "eval_runtime": 61.639,
      "eval_samples_per_second": 1133.519,
      "eval_steps_per_second": 0.568,
      "step": 1800
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.4,
      "learning_rate": 9.05e-05,
      "loss": 1.0189,
      "mlm_loss": 1.0189,
      "step": 1810
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.4,
      "learning_rate": 9.1e-05,
      "loss": 1.0181,
      "mlm_loss": 1.0181,
      "step": 1820
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.41,
      "learning_rate": 9.15e-05,
      "loss": 1.006,
      "mlm_loss": 1.006,
      "step": 1830
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.42,
      "learning_rate": 9.2e-05,
      "loss": 1.0088,
      "mlm_loss": 1.0088,
      "step": 1840
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.43,
      "learning_rate": 9.25e-05,
      "loss": 1.008,
      "mlm_loss": 1.008,
      "step": 1850
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.43,
      "learning_rate": 9.3e-05,
      "loss": 0.9891,
      "mlm_loss": 0.9891,
      "step": 1860
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.44,
      "learning_rate": 9.35e-05,
      "loss": 0.9998,
      "mlm_loss": 0.9998,
      "step": 1870
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.45,
      "learning_rate": 9.400000000000001e-05,
      "loss": 0.9927,
      "mlm_loss": 0.9927,
      "step": 1880
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.46,
      "learning_rate": 9.45e-05,
      "loss": 0.9825,
      "mlm_loss": 0.9825,
      "step": 1890
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.47,
      "learning_rate": 9.5e-05,
      "loss": 0.979,
      "mlm_loss": 0.979,
      "step": 1900
    },
    {
      "epoch": 1.47,
      "eval_ep_loss": -2.297435760498047,
      "eval_loss": 0.9245826005935669,
      "eval_mlm_loss": 0.9245826005935669,
      "eval_runtime": 60.9274,
      "eval_samples_per_second": 1146.759,
      "eval_steps_per_second": 0.574,
      "step": 1900
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.47,
      "learning_rate": 9.55e-05,
      "loss": 0.9782,
      "mlm_loss": 0.9782,
      "step": 1910
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.48,
      "learning_rate": 9.6e-05,
      "loss": 0.9802,
      "mlm_loss": 0.9802,
      "step": 1920
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.49,
      "learning_rate": 9.65e-05,
      "loss": 0.9648,
      "mlm_loss": 0.9648,
      "step": 1930
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.5,
      "learning_rate": 9.7e-05,
      "loss": 0.9576,
      "mlm_loss": 0.9576,
      "step": 1940
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.5,
      "learning_rate": 9.750000000000001e-05,
      "loss": 0.9629,
      "mlm_loss": 0.9629,
      "step": 1950
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.51,
      "learning_rate": 9.800000000000001e-05,
      "loss": 0.9585,
      "mlm_loss": 0.9585,
      "step": 1960
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.52,
      "learning_rate": 9.850000000000001e-05,
      "loss": 0.9404,
      "mlm_loss": 0.9404,
      "step": 1970
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.53,
      "learning_rate": 9.900000000000001e-05,
      "loss": 0.9538,
      "mlm_loss": 0.9538,
      "step": 1980
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.53,
      "learning_rate": 9.95e-05,
      "loss": 0.9591,
      "mlm_loss": 0.9591,
      "step": 1990
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.54,
      "learning_rate": 0.0001,
      "loss": 0.9467,
      "mlm_loss": 0.9467,
      "step": 2000
    },
    {
      "epoch": 1.54,
      "eval_ep_loss": -2.388266086578369,
      "eval_loss": 0.8887887597084045,
      "eval_mlm_loss": 0.8887887597084045,
      "eval_runtime": 59.4206,
      "eval_samples_per_second": 1175.838,
      "eval_steps_per_second": 0.589,
      "step": 2000
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.55,
      "learning_rate": 0.0001005,
      "loss": 0.947,
      "mlm_loss": 0.947,
      "step": 2010
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.56,
      "learning_rate": 0.000101,
      "loss": 0.9407,
      "mlm_loss": 0.9407,
      "step": 2020
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.57,
      "learning_rate": 0.00010150000000000001,
      "loss": 0.9352,
      "mlm_loss": 0.9352,
      "step": 2030
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.57,
      "learning_rate": 0.000102,
      "loss": 0.9234,
      "mlm_loss": 0.9234,
      "step": 2040
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.58,
      "learning_rate": 0.0001025,
      "loss": 0.9252,
      "mlm_loss": 0.9252,
      "step": 2050
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.59,
      "learning_rate": 0.000103,
      "loss": 0.9271,
      "mlm_loss": 0.9271,
      "step": 2060
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.6,
      "learning_rate": 0.0001035,
      "loss": 0.9254,
      "mlm_loss": 0.9254,
      "step": 2070
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.6,
      "learning_rate": 0.000104,
      "loss": 0.9232,
      "mlm_loss": 0.9232,
      "step": 2080
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.61,
      "learning_rate": 0.00010449999999999999,
      "loss": 0.919,
      "mlm_loss": 0.919,
      "step": 2090
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.62,
      "learning_rate": 0.000105,
      "loss": 0.9127,
      "mlm_loss": 0.9127,
      "step": 2100
    },
    {
      "epoch": 1.62,
      "eval_ep_loss": -2.262110948562622,
      "eval_loss": 0.8549456000328064,
      "eval_mlm_loss": 0.8549456000328064,
      "eval_runtime": 59.0907,
      "eval_samples_per_second": 1182.403,
      "eval_steps_per_second": 0.592,
      "step": 2100
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.63,
      "learning_rate": 0.0001055,
      "loss": 0.9023,
      "mlm_loss": 0.9023,
      "step": 2110
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.64,
      "learning_rate": 0.000106,
      "loss": 0.9031,
      "mlm_loss": 0.9031,
      "step": 2120
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.64,
      "learning_rate": 0.0001065,
      "loss": 0.9039,
      "mlm_loss": 0.9039,
      "step": 2130
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.65,
      "learning_rate": 0.000107,
      "loss": 0.9022,
      "mlm_loss": 0.9022,
      "step": 2140
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.66,
      "learning_rate": 0.0001075,
      "loss": 0.8978,
      "mlm_loss": 0.8978,
      "step": 2150
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.67,
      "learning_rate": 0.000108,
      "loss": 0.8893,
      "mlm_loss": 0.8893,
      "step": 2160
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.67,
      "learning_rate": 0.00010850000000000001,
      "loss": 0.8796,
      "mlm_loss": 0.8796,
      "step": 2170
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.68,
      "learning_rate": 0.000109,
      "loss": 0.8925,
      "mlm_loss": 0.8925,
      "step": 2180
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.69,
      "learning_rate": 0.0001095,
      "loss": 0.8873,
      "mlm_loss": 0.8873,
      "step": 2190
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.7,
      "learning_rate": 0.00011,
      "loss": 0.8858,
      "mlm_loss": 0.8858,
      "step": 2200
    },
    {
      "epoch": 1.7,
      "eval_ep_loss": -2.3254947662353516,
      "eval_loss": 0.8307343125343323,
      "eval_mlm_loss": 0.8307343125343323,
      "eval_runtime": 62.4113,
      "eval_samples_per_second": 1119.493,
      "eval_steps_per_second": 0.561,
      "step": 2200
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.7,
      "learning_rate": 0.0001105,
      "loss": 0.888,
      "mlm_loss": 0.888,
      "step": 2210
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.71,
      "learning_rate": 0.000111,
      "loss": 0.8706,
      "mlm_loss": 0.8706,
      "step": 2220
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.72,
      "learning_rate": 0.0001115,
      "loss": 0.8854,
      "mlm_loss": 0.8854,
      "step": 2230
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.73,
      "learning_rate": 0.000112,
      "loss": 0.864,
      "mlm_loss": 0.864,
      "step": 2240
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.74,
      "learning_rate": 0.00011250000000000001,
      "loss": 0.8675,
      "mlm_loss": 0.8675,
      "step": 2250
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.74,
      "learning_rate": 0.00011300000000000001,
      "loss": 0.8654,
      "mlm_loss": 0.8654,
      "step": 2260
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.75,
      "learning_rate": 0.00011350000000000001,
      "loss": 0.8593,
      "mlm_loss": 0.8593,
      "step": 2270
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.76,
      "learning_rate": 0.000114,
      "loss": 0.8586,
      "mlm_loss": 0.8586,
      "step": 2280
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.77,
      "learning_rate": 0.0001145,
      "loss": 0.8494,
      "mlm_loss": 0.8494,
      "step": 2290
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.77,
      "learning_rate": 0.000115,
      "loss": 0.8517,
      "mlm_loss": 0.8517,
      "step": 2300
    },
    {
      "epoch": 1.77,
      "eval_ep_loss": -2.2645294666290283,
      "eval_loss": 0.803054928779602,
      "eval_mlm_loss": 0.803054928779602,
      "eval_runtime": 61.7973,
      "eval_samples_per_second": 1130.616,
      "eval_steps_per_second": 0.566,
      "step": 2300
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.78,
      "learning_rate": 0.0001155,
      "loss": 0.8396,
      "mlm_loss": 0.8396,
      "step": 2310
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.79,
      "learning_rate": 0.00011600000000000001,
      "loss": 0.8478,
      "mlm_loss": 0.8478,
      "step": 2320
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.8,
      "learning_rate": 0.00011650000000000001,
      "loss": 0.8495,
      "mlm_loss": 0.8495,
      "step": 2330
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.8,
      "learning_rate": 0.00011700000000000001,
      "loss": 0.8441,
      "mlm_loss": 0.8441,
      "step": 2340
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.81,
      "learning_rate": 0.0001175,
      "loss": 0.8441,
      "mlm_loss": 0.8441,
      "step": 2350
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.82,
      "learning_rate": 0.000118,
      "loss": 0.8406,
      "mlm_loss": 0.8406,
      "step": 2360
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.83,
      "learning_rate": 0.0001185,
      "loss": 0.8322,
      "mlm_loss": 0.8322,
      "step": 2370
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.84,
      "learning_rate": 0.00011899999999999999,
      "loss": 0.8282,
      "mlm_loss": 0.8282,
      "step": 2380
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.84,
      "learning_rate": 0.00011949999999999999,
      "loss": 0.8437,
      "mlm_loss": 0.8437,
      "step": 2390
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.85,
      "learning_rate": 0.00012,
      "loss": 0.8415,
      "mlm_loss": 0.8415,
      "step": 2400
    },
    {
      "epoch": 1.85,
      "eval_ep_loss": -2.278646945953369,
      "eval_loss": 0.7815272212028503,
      "eval_mlm_loss": 0.7815272212028503,
      "eval_runtime": 60.3734,
      "eval_samples_per_second": 1157.281,
      "eval_steps_per_second": 0.58,
      "step": 2400
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.86,
      "learning_rate": 0.0001205,
      "loss": 0.8232,
      "mlm_loss": 0.8232,
      "step": 2410
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.87,
      "learning_rate": 0.000121,
      "loss": 0.8204,
      "mlm_loss": 0.8204,
      "step": 2420
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.87,
      "learning_rate": 0.0001215,
      "loss": 0.8246,
      "mlm_loss": 0.8246,
      "step": 2430
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.88,
      "learning_rate": 0.000122,
      "loss": 0.8092,
      "mlm_loss": 0.8092,
      "step": 2440
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.89,
      "learning_rate": 0.0001225,
      "loss": 0.812,
      "mlm_loss": 0.812,
      "step": 2450
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.9,
      "learning_rate": 0.000123,
      "loss": 0.8148,
      "mlm_loss": 0.8148,
      "step": 2460
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.91,
      "learning_rate": 0.0001235,
      "loss": 0.8065,
      "mlm_loss": 0.8065,
      "step": 2470
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.91,
      "learning_rate": 0.000124,
      "loss": 0.8029,
      "mlm_loss": 0.8029,
      "step": 2480
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.92,
      "learning_rate": 0.0001245,
      "loss": 0.8148,
      "mlm_loss": 0.8148,
      "step": 2490
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.93,
      "learning_rate": 0.000125,
      "loss": 0.806,
      "mlm_loss": 0.806,
      "step": 2500
    },
    {
      "epoch": 1.93,
      "eval_ep_loss": -2.1858584880828857,
      "eval_loss": 0.7560575604438782,
      "eval_mlm_loss": 0.7560575604438782,
      "eval_runtime": 60.8551,
      "eval_samples_per_second": 1148.121,
      "eval_steps_per_second": 0.575,
      "step": 2500
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.94,
      "learning_rate": 0.00012550000000000001,
      "loss": 0.8048,
      "mlm_loss": 0.8048,
      "step": 2510
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.94,
      "learning_rate": 0.000126,
      "loss": 0.8027,
      "mlm_loss": 0.8027,
      "step": 2520
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.95,
      "learning_rate": 0.0001265,
      "loss": 0.797,
      "mlm_loss": 0.797,
      "step": 2530
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.96,
      "learning_rate": 0.000127,
      "loss": 0.8041,
      "mlm_loss": 0.8041,
      "step": 2540
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.97,
      "learning_rate": 0.0001275,
      "loss": 0.8,
      "mlm_loss": 0.8,
      "step": 2550
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.97,
      "learning_rate": 0.000128,
      "loss": 0.7955,
      "mlm_loss": 0.7955,
      "step": 2560
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.98,
      "learning_rate": 0.0001285,
      "loss": 0.7943,
      "mlm_loss": 0.7943,
      "step": 2570
    },
    {
      "ep_loss": 0.0,
      "epoch": 1.99,
      "learning_rate": 0.00012900000000000002,
      "loss": 0.7934,
      "mlm_loss": 0.7934,
      "step": 2580
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.0,
      "learning_rate": 0.0001295,
      "loss": 0.7884,
      "mlm_loss": 0.7884,
      "step": 2590
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.01,
      "learning_rate": 0.00013000000000000002,
      "loss": 0.7892,
      "mlm_loss": 0.7892,
      "step": 2600
    },
    {
      "epoch": 2.01,
      "eval_ep_loss": -2.1713080406188965,
      "eval_loss": 0.7403181195259094,
      "eval_mlm_loss": 0.7403181195259094,
      "eval_runtime": 64.561,
      "eval_samples_per_second": 1082.217,
      "eval_steps_per_second": 0.542,
      "step": 2600
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.01,
      "learning_rate": 0.0001305,
      "loss": 0.7818,
      "mlm_loss": 0.7818,
      "step": 2610
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.02,
      "learning_rate": 0.000131,
      "loss": 0.7771,
      "mlm_loss": 0.7771,
      "step": 2620
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.03,
      "learning_rate": 0.0001315,
      "loss": 0.767,
      "mlm_loss": 0.767,
      "step": 2630
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.04,
      "learning_rate": 0.000132,
      "loss": 0.7826,
      "mlm_loss": 0.7826,
      "step": 2640
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.04,
      "learning_rate": 0.00013250000000000002,
      "loss": 0.771,
      "mlm_loss": 0.771,
      "step": 2650
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.05,
      "learning_rate": 0.000133,
      "loss": 0.7774,
      "mlm_loss": 0.7774,
      "step": 2660
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.06,
      "learning_rate": 0.00013350000000000002,
      "loss": 0.7674,
      "mlm_loss": 0.7674,
      "step": 2670
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.07,
      "learning_rate": 0.000134,
      "loss": 0.7737,
      "mlm_loss": 0.7737,
      "step": 2680
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.07,
      "learning_rate": 0.00013450000000000002,
      "loss": 0.7639,
      "mlm_loss": 0.7639,
      "step": 2690
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.08,
      "learning_rate": 0.000135,
      "loss": 0.7655,
      "mlm_loss": 0.7655,
      "step": 2700
    },
    {
      "epoch": 2.08,
      "eval_ep_loss": -2.082960605621338,
      "eval_loss": 0.7183709144592285,
      "eval_mlm_loss": 0.7183709144592285,
      "eval_runtime": 61.749,
      "eval_samples_per_second": 1131.5,
      "eval_steps_per_second": 0.567,
      "step": 2700
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.09,
      "learning_rate": 0.00013550000000000001,
      "loss": 0.7611,
      "mlm_loss": 0.7611,
      "step": 2710
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.1,
      "learning_rate": 0.00013600000000000003,
      "loss": 0.7586,
      "mlm_loss": 0.7586,
      "step": 2720
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.11,
      "learning_rate": 0.0001365,
      "loss": 0.762,
      "mlm_loss": 0.762,
      "step": 2730
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.11,
      "learning_rate": 0.00013700000000000002,
      "loss": 0.7479,
      "mlm_loss": 0.7479,
      "step": 2740
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.12,
      "learning_rate": 0.0001375,
      "loss": 0.7514,
      "mlm_loss": 0.7514,
      "step": 2750
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.13,
      "learning_rate": 0.00013800000000000002,
      "loss": 0.7476,
      "mlm_loss": 0.7476,
      "step": 2760
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.14,
      "learning_rate": 0.0001385,
      "loss": 0.7629,
      "mlm_loss": 0.7629,
      "step": 2770
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.14,
      "learning_rate": 0.00013900000000000002,
      "loss": 0.7432,
      "mlm_loss": 0.7432,
      "step": 2780
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.15,
      "learning_rate": 0.0001395,
      "loss": 0.7449,
      "mlm_loss": 0.7449,
      "step": 2790
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.16,
      "learning_rate": 0.00014000000000000001,
      "loss": 0.7379,
      "mlm_loss": 0.7379,
      "step": 2800
    },
    {
      "epoch": 2.16,
      "eval_ep_loss": -2.261542558670044,
      "eval_loss": 0.7018134593963623,
      "eval_mlm_loss": 0.7018134593963623,
      "eval_runtime": 60.9379,
      "eval_samples_per_second": 1146.561,
      "eval_steps_per_second": 0.574,
      "step": 2800
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.17,
      "learning_rate": 0.00014050000000000003,
      "loss": 0.7465,
      "mlm_loss": 0.7465,
      "step": 2810
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.18,
      "learning_rate": 0.00014099999999999998,
      "loss": 0.737,
      "mlm_loss": 0.737,
      "step": 2820
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.18,
      "learning_rate": 0.0001415,
      "loss": 0.7377,
      "mlm_loss": 0.7377,
      "step": 2830
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.19,
      "learning_rate": 0.00014199999999999998,
      "loss": 0.7401,
      "mlm_loss": 0.7401,
      "step": 2840
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.2,
      "learning_rate": 0.0001425,
      "loss": 0.7266,
      "mlm_loss": 0.7266,
      "step": 2850
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.21,
      "learning_rate": 0.00014299999999999998,
      "loss": 0.7443,
      "mlm_loss": 0.7443,
      "step": 2860
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.21,
      "learning_rate": 0.0001435,
      "loss": 0.7301,
      "mlm_loss": 0.7301,
      "step": 2870
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.22,
      "learning_rate": 0.000144,
      "loss": 0.7427,
      "mlm_loss": 0.7427,
      "step": 2880
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.23,
      "learning_rate": 0.0001445,
      "loss": 0.7387,
      "mlm_loss": 0.7387,
      "step": 2890
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.24,
      "learning_rate": 0.000145,
      "loss": 0.721,
      "mlm_loss": 0.721,
      "step": 2900
    },
    {
      "epoch": 2.24,
      "eval_ep_loss": -2.377054214477539,
      "eval_loss": 0.6826241612434387,
      "eval_mlm_loss": 0.6826241612434387,
      "eval_runtime": 60.9503,
      "eval_samples_per_second": 1146.327,
      "eval_steps_per_second": 0.574,
      "step": 2900
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.24,
      "learning_rate": 0.00014549999999999999,
      "loss": 0.7352,
      "mlm_loss": 0.7352,
      "step": 2910
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.25,
      "learning_rate": 0.000146,
      "loss": 0.7219,
      "mlm_loss": 0.7219,
      "step": 2920
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.26,
      "learning_rate": 0.00014649999999999998,
      "loss": 0.729,
      "mlm_loss": 0.729,
      "step": 2930
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.27,
      "learning_rate": 0.000147,
      "loss": 0.7148,
      "mlm_loss": 0.7148,
      "step": 2940
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.28,
      "learning_rate": 0.0001475,
      "loss": 0.721,
      "mlm_loss": 0.721,
      "step": 2950
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.28,
      "learning_rate": 0.000148,
      "loss": 0.7113,
      "mlm_loss": 0.7113,
      "step": 2960
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.29,
      "learning_rate": 0.0001485,
      "loss": 0.711,
      "mlm_loss": 0.711,
      "step": 2970
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.3,
      "learning_rate": 0.000149,
      "loss": 0.7109,
      "mlm_loss": 0.7109,
      "step": 2980
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.31,
      "learning_rate": 0.0001495,
      "loss": 0.7121,
      "mlm_loss": 0.7121,
      "step": 2990
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.31,
      "learning_rate": 0.00015,
      "loss": 0.7106,
      "mlm_loss": 0.7106,
      "step": 3000
    },
    {
      "epoch": 2.31,
      "eval_ep_loss": -2.0994629859924316,
      "eval_loss": 0.6672462821006775,
      "eval_mlm_loss": 0.6672462821006775,
      "eval_runtime": 63.0197,
      "eval_samples_per_second": 1108.686,
      "eval_steps_per_second": 0.555,
      "step": 3000
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.32,
      "learning_rate": 0.0001505,
      "loss": 0.7067,
      "mlm_loss": 0.7067,
      "step": 3010
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.33,
      "learning_rate": 0.000151,
      "loss": 0.7082,
      "mlm_loss": 0.7082,
      "step": 3020
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.34,
      "learning_rate": 0.0001515,
      "loss": 0.7097,
      "mlm_loss": 0.7097,
      "step": 3030
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.34,
      "learning_rate": 0.000152,
      "loss": 0.6972,
      "mlm_loss": 0.6972,
      "step": 3040
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.35,
      "learning_rate": 0.0001525,
      "loss": 0.7051,
      "mlm_loss": 0.7051,
      "step": 3050
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.36,
      "learning_rate": 0.000153,
      "loss": 0.7029,
      "mlm_loss": 0.7029,
      "step": 3060
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.37,
      "learning_rate": 0.0001535,
      "loss": 0.7037,
      "mlm_loss": 0.7037,
      "step": 3070
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.38,
      "learning_rate": 0.000154,
      "loss": 0.6966,
      "mlm_loss": 0.6966,
      "step": 3080
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.38,
      "learning_rate": 0.00015450000000000001,
      "loss": 0.6948,
      "mlm_loss": 0.6948,
      "step": 3090
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.39,
      "learning_rate": 0.000155,
      "loss": 0.6974,
      "mlm_loss": 0.6974,
      "step": 3100
    },
    {
      "epoch": 2.39,
      "eval_ep_loss": -2.283140182495117,
      "eval_loss": 0.6495695114135742,
      "eval_mlm_loss": 0.6495695114135742,
      "eval_runtime": 59.8554,
      "eval_samples_per_second": 1167.296,
      "eval_steps_per_second": 0.585,
      "step": 3100
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.4,
      "learning_rate": 0.0001555,
      "loss": 0.6877,
      "mlm_loss": 0.6877,
      "step": 3110
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.41,
      "learning_rate": 0.000156,
      "loss": 0.6862,
      "mlm_loss": 0.6862,
      "step": 3120
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.41,
      "learning_rate": 0.0001565,
      "loss": 0.6893,
      "mlm_loss": 0.6893,
      "step": 3130
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.42,
      "learning_rate": 0.000157,
      "loss": 0.6943,
      "mlm_loss": 0.6943,
      "step": 3140
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.43,
      "learning_rate": 0.0001575,
      "loss": 0.6804,
      "mlm_loss": 0.6804,
      "step": 3150
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.44,
      "learning_rate": 0.000158,
      "loss": 0.6813,
      "mlm_loss": 0.6813,
      "step": 3160
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.45,
      "learning_rate": 0.0001585,
      "loss": 0.6828,
      "mlm_loss": 0.6828,
      "step": 3170
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.45,
      "learning_rate": 0.00015900000000000002,
      "loss": 0.6736,
      "mlm_loss": 0.6736,
      "step": 3180
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.46,
      "learning_rate": 0.0001595,
      "loss": 0.6763,
      "mlm_loss": 0.6763,
      "step": 3190
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.47,
      "learning_rate": 0.00016,
      "loss": 0.6796,
      "mlm_loss": 0.6796,
      "step": 3200
    },
    {
      "epoch": 2.47,
      "eval_ep_loss": -2.57122540473938,
      "eval_loss": 0.6341940760612488,
      "eval_mlm_loss": 0.6341940760612488,
      "eval_runtime": 63.1531,
      "eval_samples_per_second": 1106.343,
      "eval_steps_per_second": 0.554,
      "step": 3200
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.48,
      "learning_rate": 0.0001605,
      "loss": 0.6688,
      "mlm_loss": 0.6688,
      "step": 3210
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.48,
      "learning_rate": 0.000161,
      "loss": 0.6725,
      "mlm_loss": 0.6725,
      "step": 3220
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.49,
      "learning_rate": 0.0001615,
      "loss": 0.6718,
      "mlm_loss": 0.6718,
      "step": 3230
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.5,
      "learning_rate": 0.000162,
      "loss": 0.6675,
      "mlm_loss": 0.6675,
      "step": 3240
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.51,
      "learning_rate": 0.00016250000000000002,
      "loss": 0.6665,
      "mlm_loss": 0.6665,
      "step": 3250
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.51,
      "learning_rate": 0.000163,
      "loss": 0.6615,
      "mlm_loss": 0.6615,
      "step": 3260
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.52,
      "learning_rate": 0.00016350000000000002,
      "loss": 0.6784,
      "mlm_loss": 0.6784,
      "step": 3270
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.53,
      "learning_rate": 0.000164,
      "loss": 0.6686,
      "mlm_loss": 0.6686,
      "step": 3280
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.54,
      "learning_rate": 0.00016450000000000001,
      "loss": 0.6697,
      "mlm_loss": 0.6697,
      "step": 3290
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.55,
      "learning_rate": 0.000165,
      "loss": 0.6621,
      "mlm_loss": 0.6621,
      "step": 3300
    },
    {
      "epoch": 2.55,
      "eval_ep_loss": -2.320197820663452,
      "eval_loss": 0.6192212104797363,
      "eval_mlm_loss": 0.6192212104797363,
      "eval_runtime": 60.3086,
      "eval_samples_per_second": 1158.524,
      "eval_steps_per_second": 0.58,
      "step": 3300
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.55,
      "learning_rate": 0.0001655,
      "loss": 0.6645,
      "mlm_loss": 0.6645,
      "step": 3310
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.56,
      "learning_rate": 0.00016600000000000002,
      "loss": 0.6594,
      "mlm_loss": 0.6594,
      "step": 3320
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.57,
      "learning_rate": 0.0001665,
      "loss": 0.6585,
      "mlm_loss": 0.6585,
      "step": 3330
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.58,
      "learning_rate": 0.00016700000000000002,
      "loss": 0.6499,
      "mlm_loss": 0.6499,
      "step": 3340
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.58,
      "learning_rate": 0.0001675,
      "loss": 0.641,
      "mlm_loss": 0.641,
      "step": 3350
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.59,
      "learning_rate": 0.00016800000000000002,
      "loss": 0.653,
      "mlm_loss": 0.653,
      "step": 3360
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.6,
      "learning_rate": 0.0001685,
      "loss": 0.6526,
      "mlm_loss": 0.6526,
      "step": 3370
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.61,
      "learning_rate": 0.00016900000000000002,
      "loss": 0.6588,
      "mlm_loss": 0.6588,
      "step": 3380
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.61,
      "learning_rate": 0.00016950000000000003,
      "loss": 0.6457,
      "mlm_loss": 0.6457,
      "step": 3390
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.62,
      "learning_rate": 0.00017,
      "loss": 0.6464,
      "mlm_loss": 0.6464,
      "step": 3400
    },
    {
      "epoch": 2.62,
      "eval_ep_loss": -2.441110610961914,
      "eval_loss": 0.6078351736068726,
      "eval_mlm_loss": 0.6078351736068726,
      "eval_runtime": 60.4806,
      "eval_samples_per_second": 1155.23,
      "eval_steps_per_second": 0.579,
      "step": 3400
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.63,
      "learning_rate": 0.00017050000000000002,
      "loss": 0.6517,
      "mlm_loss": 0.6517,
      "step": 3410
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.64,
      "learning_rate": 0.000171,
      "loss": 0.6473,
      "mlm_loss": 0.6473,
      "step": 3420
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.65,
      "learning_rate": 0.00017150000000000002,
      "loss": 0.6475,
      "mlm_loss": 0.6475,
      "step": 3430
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.65,
      "learning_rate": 0.00017199999999999998,
      "loss": 0.649,
      "mlm_loss": 0.649,
      "step": 3440
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.66,
      "learning_rate": 0.0001725,
      "loss": 0.652,
      "mlm_loss": 0.652,
      "step": 3450
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.67,
      "learning_rate": 0.000173,
      "loss": 0.6484,
      "mlm_loss": 0.6484,
      "step": 3460
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.68,
      "learning_rate": 0.0001735,
      "loss": 0.6405,
      "mlm_loss": 0.6405,
      "step": 3470
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.68,
      "learning_rate": 0.000174,
      "loss": 0.6421,
      "mlm_loss": 0.6421,
      "step": 3480
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.69,
      "learning_rate": 0.00017449999999999999,
      "loss": 0.6313,
      "mlm_loss": 0.6313,
      "step": 3490
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.7,
      "learning_rate": 0.000175,
      "loss": 0.6228,
      "mlm_loss": 0.6228,
      "step": 3500
    },
    {
      "epoch": 2.7,
      "eval_ep_loss": -2.3252084255218506,
      "eval_loss": 0.5944607257843018,
      "eval_mlm_loss": 0.5944607257843018,
      "eval_runtime": 59.8998,
      "eval_samples_per_second": 1166.431,
      "eval_steps_per_second": 0.584,
      "step": 3500
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.71,
      "learning_rate": 0.00017549999999999998,
      "loss": 0.6276,
      "mlm_loss": 0.6276,
      "step": 3510
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.72,
      "learning_rate": 0.000176,
      "loss": 0.632,
      "mlm_loss": 0.632,
      "step": 3520
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.72,
      "learning_rate": 0.00017649999999999998,
      "loss": 0.629,
      "mlm_loss": 0.629,
      "step": 3530
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.73,
      "learning_rate": 0.000177,
      "loss": 0.6327,
      "mlm_loss": 0.6327,
      "step": 3540
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.74,
      "learning_rate": 0.0001775,
      "loss": 0.6307,
      "mlm_loss": 0.6307,
      "step": 3550
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.75,
      "learning_rate": 0.000178,
      "loss": 0.6156,
      "mlm_loss": 0.6156,
      "step": 3560
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.75,
      "learning_rate": 0.0001785,
      "loss": 0.6246,
      "mlm_loss": 0.6246,
      "step": 3570
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.76,
      "learning_rate": 0.000179,
      "loss": 0.6223,
      "mlm_loss": 0.6223,
      "step": 3580
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.77,
      "learning_rate": 0.0001795,
      "loss": 0.6359,
      "mlm_loss": 0.6359,
      "step": 3590
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.78,
      "learning_rate": 0.00017999999999999998,
      "loss": 0.6192,
      "mlm_loss": 0.6192,
      "step": 3600
    },
    {
      "epoch": 2.78,
      "eval_ep_loss": -2.543795585632324,
      "eval_loss": 0.5830370783805847,
      "eval_mlm_loss": 0.5830370783805847,
      "eval_runtime": 62.3115,
      "eval_samples_per_second": 1121.285,
      "eval_steps_per_second": 0.562,
      "step": 3600
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.78,
      "learning_rate": 0.0001805,
      "loss": 0.619,
      "mlm_loss": 0.619,
      "step": 3610
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.79,
      "learning_rate": 0.000181,
      "loss": 0.6107,
      "mlm_loss": 0.6107,
      "step": 3620
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.8,
      "learning_rate": 0.0001815,
      "loss": 0.6287,
      "mlm_loss": 0.6287,
      "step": 3630
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.81,
      "learning_rate": 0.000182,
      "loss": 0.616,
      "mlm_loss": 0.616,
      "step": 3640
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.82,
      "learning_rate": 0.0001825,
      "loss": 0.6199,
      "mlm_loss": 0.6199,
      "step": 3650
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.82,
      "learning_rate": 0.000183,
      "loss": 0.6179,
      "mlm_loss": 0.6179,
      "step": 3660
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.83,
      "learning_rate": 0.0001835,
      "loss": 0.6159,
      "mlm_loss": 0.6159,
      "step": 3670
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.84,
      "learning_rate": 0.000184,
      "loss": 0.6067,
      "mlm_loss": 0.6067,
      "step": 3680
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.85,
      "learning_rate": 0.0001845,
      "loss": 0.6079,
      "mlm_loss": 0.6079,
      "step": 3690
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.85,
      "learning_rate": 0.000185,
      "loss": 0.6063,
      "mlm_loss": 0.6063,
      "step": 3700
    },
    {
      "epoch": 2.85,
      "eval_ep_loss": -2.2521820068359375,
      "eval_loss": 0.5688546299934387,
      "eval_mlm_loss": 0.5688546299934387,
      "eval_runtime": 60.0349,
      "eval_samples_per_second": 1163.807,
      "eval_steps_per_second": 0.583,
      "step": 3700
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.86,
      "learning_rate": 0.0001855,
      "loss": 0.6156,
      "mlm_loss": 0.6156,
      "step": 3710
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.87,
      "learning_rate": 0.000186,
      "loss": 0.6093,
      "mlm_loss": 0.6093,
      "step": 3720
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.88,
      "learning_rate": 0.0001865,
      "loss": 0.6073,
      "mlm_loss": 0.6073,
      "step": 3730
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.88,
      "learning_rate": 0.000187,
      "loss": 0.6039,
      "mlm_loss": 0.6039,
      "step": 3740
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.89,
      "learning_rate": 0.0001875,
      "loss": 0.5977,
      "mlm_loss": 0.5977,
      "step": 3750
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.9,
      "learning_rate": 0.00018800000000000002,
      "loss": 0.6,
      "mlm_loss": 0.6,
      "step": 3760
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.91,
      "learning_rate": 0.0001885,
      "loss": 0.6022,
      "mlm_loss": 0.6022,
      "step": 3770
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.92,
      "learning_rate": 0.000189,
      "loss": 0.6015,
      "mlm_loss": 0.6015,
      "step": 3780
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.92,
      "learning_rate": 0.0001895,
      "loss": 0.6016,
      "mlm_loss": 0.6016,
      "step": 3790
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.93,
      "learning_rate": 0.00019,
      "loss": 0.5959,
      "mlm_loss": 0.5959,
      "step": 3800
    },
    {
      "epoch": 2.93,
      "eval_ep_loss": -2.366133213043213,
      "eval_loss": 0.5573724508285522,
      "eval_mlm_loss": 0.5573724508285522,
      "eval_runtime": 61.8753,
      "eval_samples_per_second": 1129.191,
      "eval_steps_per_second": 0.566,
      "step": 3800
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.94,
      "learning_rate": 0.0001905,
      "loss": 0.6003,
      "mlm_loss": 0.6003,
      "step": 3810
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.95,
      "learning_rate": 0.000191,
      "loss": 0.5984,
      "mlm_loss": 0.5984,
      "step": 3820
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.95,
      "learning_rate": 0.00019150000000000002,
      "loss": 0.6005,
      "mlm_loss": 0.6005,
      "step": 3830
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.96,
      "learning_rate": 0.000192,
      "loss": 0.5895,
      "mlm_loss": 0.5895,
      "step": 3840
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.97,
      "learning_rate": 0.00019250000000000002,
      "loss": 0.5912,
      "mlm_loss": 0.5912,
      "step": 3850
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.98,
      "learning_rate": 0.000193,
      "loss": 0.5828,
      "mlm_loss": 0.5828,
      "step": 3860
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.98,
      "learning_rate": 0.00019350000000000001,
      "loss": 0.5865,
      "mlm_loss": 0.5865,
      "step": 3870
    },
    {
      "ep_loss": 0.0,
      "epoch": 2.99,
      "learning_rate": 0.000194,
      "loss": 0.5854,
      "mlm_loss": 0.5854,
      "step": 3880
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.0,
      "learning_rate": 0.0001945,
      "loss": 0.5883,
      "mlm_loss": 0.5883,
      "step": 3890
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.01,
      "learning_rate": 0.00019500000000000002,
      "loss": 0.5743,
      "mlm_loss": 0.5743,
      "step": 3900
    },
    {
      "epoch": 3.01,
      "eval_ep_loss": -2.332629680633545,
      "eval_loss": 0.551331102848053,
      "eval_mlm_loss": 0.551331102848053,
      "eval_runtime": 61.0219,
      "eval_samples_per_second": 1144.983,
      "eval_steps_per_second": 0.574,
      "step": 3900
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.02,
      "learning_rate": 0.0001955,
      "loss": 0.5813,
      "mlm_loss": 0.5813,
      "step": 3910
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.02,
      "learning_rate": 0.00019600000000000002,
      "loss": 0.5813,
      "mlm_loss": 0.5813,
      "step": 3920
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.03,
      "learning_rate": 0.0001965,
      "loss": 0.5899,
      "mlm_loss": 0.5899,
      "step": 3930
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.04,
      "learning_rate": 0.00019700000000000002,
      "loss": 0.5849,
      "mlm_loss": 0.5849,
      "step": 3940
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.05,
      "learning_rate": 0.0001975,
      "loss": 0.5823,
      "mlm_loss": 0.5823,
      "step": 3950
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.05,
      "learning_rate": 0.00019800000000000002,
      "loss": 0.5794,
      "mlm_loss": 0.5794,
      "step": 3960
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.06,
      "learning_rate": 0.00019850000000000003,
      "loss": 0.5837,
      "mlm_loss": 0.5837,
      "step": 3970
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.07,
      "learning_rate": 0.000199,
      "loss": 0.5783,
      "mlm_loss": 0.5783,
      "step": 3980
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.08,
      "learning_rate": 0.00019950000000000002,
      "loss": 0.5733,
      "mlm_loss": 0.5733,
      "step": 3990
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.09,
      "learning_rate": 0.0002,
      "loss": 0.5701,
      "mlm_loss": 0.5701,
      "step": 4000
    },
    {
      "epoch": 3.09,
      "eval_ep_loss": -2.3982014656066895,
      "eval_loss": 0.5421211123466492,
      "eval_mlm_loss": 0.5421211123466492,
      "eval_runtime": 62.0302,
      "eval_samples_per_second": 1126.37,
      "eval_steps_per_second": 0.564,
      "step": 4000
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.09,
      "learning_rate": 0.00020050000000000002,
      "loss": 0.5737,
      "mlm_loss": 0.5737,
      "step": 4010
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.1,
      "learning_rate": 0.000201,
      "loss": 0.5645,
      "mlm_loss": 0.5645,
      "step": 4020
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.11,
      "learning_rate": 0.00020150000000000002,
      "loss": 0.5758,
      "mlm_loss": 0.5758,
      "step": 4030
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.12,
      "learning_rate": 0.000202,
      "loss": 0.5849,
      "mlm_loss": 0.5849,
      "step": 4040
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.12,
      "learning_rate": 0.00020250000000000002,
      "loss": 0.5667,
      "mlm_loss": 0.5667,
      "step": 4050
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.13,
      "learning_rate": 0.00020300000000000003,
      "loss": 0.5726,
      "mlm_loss": 0.5726,
      "step": 4060
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.14,
      "learning_rate": 0.00020349999999999999,
      "loss": 0.5645,
      "mlm_loss": 0.5645,
      "step": 4070
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.15,
      "learning_rate": 0.000204,
      "loss": 0.5681,
      "mlm_loss": 0.5681,
      "step": 4080
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.15,
      "learning_rate": 0.00020449999999999998,
      "loss": 0.564,
      "mlm_loss": 0.564,
      "step": 4090
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.16,
      "learning_rate": 0.000205,
      "loss": 0.5655,
      "mlm_loss": 0.5655,
      "step": 4100
    },
    {
      "epoch": 3.16,
      "eval_ep_loss": -2.218219041824341,
      "eval_loss": 0.5314173698425293,
      "eval_mlm_loss": 0.5314173698425293,
      "eval_runtime": 60.8804,
      "eval_samples_per_second": 1147.644,
      "eval_steps_per_second": 0.575,
      "step": 4100
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.17,
      "learning_rate": 0.00020549999999999998,
      "loss": 0.5654,
      "mlm_loss": 0.5654,
      "step": 4110
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.18,
      "learning_rate": 0.000206,
      "loss": 0.5647,
      "mlm_loss": 0.5647,
      "step": 4120
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.19,
      "learning_rate": 0.0002065,
      "loss": 0.568,
      "mlm_loss": 0.568,
      "step": 4130
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.19,
      "learning_rate": 0.000207,
      "loss": 0.5538,
      "mlm_loss": 0.5538,
      "step": 4140
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.2,
      "learning_rate": 0.0002075,
      "loss": 0.5585,
      "mlm_loss": 0.5585,
      "step": 4150
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.21,
      "learning_rate": 0.000208,
      "loss": 0.5578,
      "mlm_loss": 0.5578,
      "step": 4160
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.22,
      "learning_rate": 0.0002085,
      "loss": 0.5508,
      "mlm_loss": 0.5508,
      "step": 4170
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.22,
      "learning_rate": 0.00020899999999999998,
      "loss": 0.5644,
      "mlm_loss": 0.5644,
      "step": 4180
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.23,
      "learning_rate": 0.0002095,
      "loss": 0.5642,
      "mlm_loss": 0.5642,
      "step": 4190
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.24,
      "learning_rate": 0.00021,
      "loss": 0.561,
      "mlm_loss": 0.561,
      "step": 4200
    },
    {
      "epoch": 3.24,
      "eval_ep_loss": -2.287240743637085,
      "eval_loss": 0.5224212408065796,
      "eval_mlm_loss": 0.5224212408065796,
      "eval_runtime": 61.3268,
      "eval_samples_per_second": 1139.291,
      "eval_steps_per_second": 0.571,
      "step": 4200
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.25,
      "learning_rate": 0.0002105,
      "loss": 0.5564,
      "mlm_loss": 0.5564,
      "step": 4210
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.25,
      "learning_rate": 0.000211,
      "loss": 0.5511,
      "mlm_loss": 0.5511,
      "step": 4220
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.26,
      "learning_rate": 0.0002115,
      "loss": 0.5578,
      "mlm_loss": 0.5578,
      "step": 4230
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.27,
      "learning_rate": 0.000212,
      "loss": 0.5531,
      "mlm_loss": 0.5531,
      "step": 4240
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.28,
      "learning_rate": 0.0002125,
      "loss": 0.5569,
      "mlm_loss": 0.5569,
      "step": 4250
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.29,
      "learning_rate": 0.000213,
      "loss": 0.5499,
      "mlm_loss": 0.5499,
      "step": 4260
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.29,
      "learning_rate": 0.0002135,
      "loss": 0.5514,
      "mlm_loss": 0.5514,
      "step": 4270
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.3,
      "learning_rate": 0.000214,
      "loss": 0.554,
      "mlm_loss": 0.554,
      "step": 4280
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.31,
      "learning_rate": 0.0002145,
      "loss": 0.5444,
      "mlm_loss": 0.5444,
      "step": 4290
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.32,
      "learning_rate": 0.000215,
      "loss": 0.5555,
      "mlm_loss": 0.5555,
      "step": 4300
    },
    {
      "epoch": 3.32,
      "eval_ep_loss": -2.407238245010376,
      "eval_loss": 0.5153276920318604,
      "eval_mlm_loss": 0.5153276920318604,
      "eval_runtime": 62.01,
      "eval_samples_per_second": 1126.738,
      "eval_steps_per_second": 0.564,
      "step": 4300
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.32,
      "learning_rate": 0.0002155,
      "loss": 0.5374,
      "mlm_loss": 0.5374,
      "step": 4310
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.33,
      "learning_rate": 0.000216,
      "loss": 0.5526,
      "mlm_loss": 0.5526,
      "step": 4320
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.34,
      "learning_rate": 0.0002165,
      "loss": 0.5451,
      "mlm_loss": 0.5451,
      "step": 4330
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.35,
      "learning_rate": 0.00021700000000000002,
      "loss": 0.5406,
      "mlm_loss": 0.5406,
      "step": 4340
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.36,
      "learning_rate": 0.0002175,
      "loss": 0.5339,
      "mlm_loss": 0.5339,
      "step": 4350
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.36,
      "learning_rate": 0.000218,
      "loss": 0.5434,
      "mlm_loss": 0.5434,
      "step": 4360
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.37,
      "learning_rate": 0.0002185,
      "loss": 0.546,
      "mlm_loss": 0.546,
      "step": 4370
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.38,
      "learning_rate": 0.000219,
      "loss": 0.5359,
      "mlm_loss": 0.5359,
      "step": 4380
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.39,
      "learning_rate": 0.0002195,
      "loss": 0.5405,
      "mlm_loss": 0.5405,
      "step": 4390
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.39,
      "learning_rate": 0.00022,
      "loss": 0.5506,
      "mlm_loss": 0.5506,
      "step": 4400
    },
    {
      "epoch": 3.39,
      "eval_ep_loss": -2.1430230140686035,
      "eval_loss": 0.5089389681816101,
      "eval_mlm_loss": 0.5089389681816101,
      "eval_runtime": 60.2829,
      "eval_samples_per_second": 1159.018,
      "eval_steps_per_second": 0.581,
      "step": 4400
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.4,
      "learning_rate": 0.0002205,
      "loss": 0.5425,
      "mlm_loss": 0.5425,
      "step": 4410
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.41,
      "learning_rate": 0.000221,
      "loss": 0.5443,
      "mlm_loss": 0.5443,
      "step": 4420
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.42,
      "learning_rate": 0.00022150000000000002,
      "loss": 0.5433,
      "mlm_loss": 0.5433,
      "step": 4430
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.42,
      "learning_rate": 0.000222,
      "loss": 0.5379,
      "mlm_loss": 0.5379,
      "step": 4440
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.43,
      "learning_rate": 0.00022250000000000001,
      "loss": 0.5349,
      "mlm_loss": 0.5349,
      "step": 4450
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.44,
      "learning_rate": 0.000223,
      "loss": 0.5385,
      "mlm_loss": 0.5385,
      "step": 4460
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.45,
      "learning_rate": 0.0002235,
      "loss": 0.5337,
      "mlm_loss": 0.5337,
      "step": 4470
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.46,
      "learning_rate": 0.000224,
      "loss": 0.5372,
      "mlm_loss": 0.5372,
      "step": 4480
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.46,
      "learning_rate": 0.0002245,
      "loss": 0.5342,
      "mlm_loss": 0.5342,
      "step": 4490
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.47,
      "learning_rate": 0.00022500000000000002,
      "loss": 0.5329,
      "mlm_loss": 0.5329,
      "step": 4500
    },
    {
      "epoch": 3.47,
      "eval_ep_loss": -2.252265691757202,
      "eval_loss": 0.501177966594696,
      "eval_mlm_loss": 0.501177966594696,
      "eval_runtime": 61.1269,
      "eval_samples_per_second": 1143.016,
      "eval_steps_per_second": 0.573,
      "step": 4500
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.48,
      "learning_rate": 0.0002255,
      "loss": 0.5321,
      "mlm_loss": 0.5321,
      "step": 4510
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.49,
      "learning_rate": 0.00022600000000000002,
      "loss": 0.5413,
      "mlm_loss": 0.5413,
      "step": 4520
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.49,
      "learning_rate": 0.0002265,
      "loss": 0.5364,
      "mlm_loss": 0.5364,
      "step": 4530
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.5,
      "learning_rate": 0.00022700000000000002,
      "loss": 0.5326,
      "mlm_loss": 0.5326,
      "step": 4540
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.51,
      "learning_rate": 0.0002275,
      "loss": 0.5327,
      "mlm_loss": 0.5327,
      "step": 4550
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.52,
      "learning_rate": 0.000228,
      "loss": 0.5334,
      "mlm_loss": 0.5334,
      "step": 4560
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.52,
      "learning_rate": 0.00022850000000000002,
      "loss": 0.5344,
      "mlm_loss": 0.5344,
      "step": 4570
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.53,
      "learning_rate": 0.000229,
      "loss": 0.5259,
      "mlm_loss": 0.5259,
      "step": 4580
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.54,
      "learning_rate": 0.00022950000000000002,
      "loss": 0.519,
      "mlm_loss": 0.519,
      "step": 4590
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.55,
      "learning_rate": 0.00023,
      "loss": 0.5243,
      "mlm_loss": 0.5243,
      "step": 4600
    },
    {
      "epoch": 3.55,
      "eval_ep_loss": -2.4623284339904785,
      "eval_loss": 0.49688172340393066,
      "eval_mlm_loss": 0.49688172340393066,
      "eval_runtime": 60.8838,
      "eval_samples_per_second": 1147.579,
      "eval_steps_per_second": 0.575,
      "step": 4600
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.56,
      "learning_rate": 0.00023050000000000002,
      "loss": 0.5237,
      "mlm_loss": 0.5237,
      "step": 4610
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.56,
      "learning_rate": 0.000231,
      "loss": 0.5249,
      "mlm_loss": 0.5249,
      "step": 4620
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.57,
      "learning_rate": 0.00023150000000000002,
      "loss": 0.5208,
      "mlm_loss": 0.5208,
      "step": 4630
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.58,
      "learning_rate": 0.00023200000000000003,
      "loss": 0.5213,
      "mlm_loss": 0.5213,
      "step": 4640
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.59,
      "learning_rate": 0.0002325,
      "loss": 0.5316,
      "mlm_loss": 0.5316,
      "step": 4650
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.59,
      "learning_rate": 0.00023300000000000003,
      "loss": 0.5231,
      "mlm_loss": 0.5231,
      "step": 4660
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.6,
      "learning_rate": 0.0002335,
      "loss": 0.524,
      "mlm_loss": 0.524,
      "step": 4670
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.61,
      "learning_rate": 0.00023400000000000002,
      "loss": 0.5274,
      "mlm_loss": 0.5274,
      "step": 4680
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.62,
      "learning_rate": 0.00023449999999999998,
      "loss": 0.5186,
      "mlm_loss": 0.5186,
      "step": 4690
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.63,
      "learning_rate": 0.000235,
      "loss": 0.5201,
      "mlm_loss": 0.5201,
      "step": 4700
    },
    {
      "epoch": 3.63,
      "eval_ep_loss": -2.275129556655884,
      "eval_loss": 0.48909062147140503,
      "eval_mlm_loss": 0.48909062147140503,
      "eval_runtime": 63.7661,
      "eval_samples_per_second": 1095.708,
      "eval_steps_per_second": 0.549,
      "step": 4700
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.63,
      "learning_rate": 0.0002355,
      "loss": 0.5167,
      "mlm_loss": 0.5167,
      "step": 4710
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.64,
      "learning_rate": 0.000236,
      "loss": 0.5152,
      "mlm_loss": 0.5152,
      "step": 4720
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.65,
      "learning_rate": 0.0002365,
      "loss": 0.5201,
      "mlm_loss": 0.5201,
      "step": 4730
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.66,
      "learning_rate": 0.000237,
      "loss": 0.5234,
      "mlm_loss": 0.5234,
      "step": 4740
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.66,
      "learning_rate": 0.0002375,
      "loss": 0.5156,
      "mlm_loss": 0.5156,
      "step": 4750
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.67,
      "learning_rate": 0.00023799999999999998,
      "loss": 0.5189,
      "mlm_loss": 0.5189,
      "step": 4760
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.68,
      "learning_rate": 0.0002385,
      "loss": 0.5189,
      "mlm_loss": 0.5189,
      "step": 4770
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.69,
      "learning_rate": 0.00023899999999999998,
      "loss": 0.51,
      "mlm_loss": 0.51,
      "step": 4780
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.69,
      "learning_rate": 0.0002395,
      "loss": 0.5158,
      "mlm_loss": 0.5158,
      "step": 4790
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.7,
      "learning_rate": 0.00024,
      "loss": 0.5121,
      "mlm_loss": 0.5121,
      "step": 4800
    },
    {
      "epoch": 3.7,
      "eval_ep_loss": -2.0928449630737305,
      "eval_loss": 0.48291251063346863,
      "eval_mlm_loss": 0.48291251063346863,
      "eval_runtime": 60.133,
      "eval_samples_per_second": 1161.908,
      "eval_steps_per_second": 0.582,
      "step": 4800
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.71,
      "learning_rate": 0.0002405,
      "loss": 0.5108,
      "mlm_loss": 0.5108,
      "step": 4810
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.72,
      "learning_rate": 0.000241,
      "loss": 0.5176,
      "mlm_loss": 0.5176,
      "step": 4820
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.73,
      "learning_rate": 0.0002415,
      "loss": 0.5102,
      "mlm_loss": 0.5102,
      "step": 4830
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.73,
      "learning_rate": 0.000242,
      "loss": 0.514,
      "mlm_loss": 0.514,
      "step": 4840
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.74,
      "learning_rate": 0.00024249999999999999,
      "loss": 0.5086,
      "mlm_loss": 0.5086,
      "step": 4850
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.75,
      "learning_rate": 0.000243,
      "loss": 0.515,
      "mlm_loss": 0.515,
      "step": 4860
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.76,
      "learning_rate": 0.0002435,
      "loss": 0.5091,
      "mlm_loss": 0.5091,
      "step": 4870
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.76,
      "learning_rate": 0.000244,
      "loss": 0.514,
      "mlm_loss": 0.514,
      "step": 4880
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.77,
      "learning_rate": 0.0002445,
      "loss": 0.5051,
      "mlm_loss": 0.5051,
      "step": 4890
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.78,
      "learning_rate": 0.000245,
      "loss": 0.526,
      "mlm_loss": 0.526,
      "step": 4900
    },
    {
      "epoch": 3.78,
      "eval_ep_loss": -2.452838182449341,
      "eval_loss": 0.4754766523838043,
      "eval_mlm_loss": 0.4754766523838043,
      "eval_runtime": 60.9895,
      "eval_samples_per_second": 1145.591,
      "eval_steps_per_second": 0.574,
      "step": 4900
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.79,
      "learning_rate": 0.0002455,
      "loss": 0.5104,
      "mlm_loss": 0.5104,
      "step": 4910
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.79,
      "learning_rate": 0.000246,
      "loss": 0.4975,
      "mlm_loss": 0.4975,
      "step": 4920
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.8,
      "learning_rate": 0.00024650000000000003,
      "loss": 0.511,
      "mlm_loss": 0.511,
      "step": 4930
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.81,
      "learning_rate": 0.000247,
      "loss": 0.5081,
      "mlm_loss": 0.5081,
      "step": 4940
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.82,
      "learning_rate": 0.0002475,
      "loss": 0.5067,
      "mlm_loss": 0.5067,
      "step": 4950
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.83,
      "learning_rate": 0.000248,
      "loss": 0.5062,
      "mlm_loss": 0.5062,
      "step": 4960
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.83,
      "learning_rate": 0.0002485,
      "loss": 0.5054,
      "mlm_loss": 0.5054,
      "step": 4970
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.84,
      "learning_rate": 0.000249,
      "loss": 0.5032,
      "mlm_loss": 0.5032,
      "step": 4980
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.85,
      "learning_rate": 0.0002495,
      "loss": 0.5022,
      "mlm_loss": 0.5022,
      "step": 4990
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.86,
      "learning_rate": 0.00025,
      "loss": 0.5077,
      "mlm_loss": 0.5077,
      "step": 5000
    },
    {
      "epoch": 3.86,
      "eval_ep_loss": -2.5484063625335693,
      "eval_loss": 0.47381114959716797,
      "eval_mlm_loss": 0.47381114959716797,
      "eval_runtime": 59.7879,
      "eval_samples_per_second": 1168.615,
      "eval_steps_per_second": 0.585,
      "step": 5000
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.86,
      "learning_rate": 0.0002505,
      "loss": 0.5129,
      "mlm_loss": 0.5129,
      "step": 5010
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.87,
      "learning_rate": 0.00025100000000000003,
      "loss": 0.5031,
      "mlm_loss": 0.5031,
      "step": 5020
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.88,
      "learning_rate": 0.0002515,
      "loss": 0.4941,
      "mlm_loss": 0.4941,
      "step": 5030
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.89,
      "learning_rate": 0.000252,
      "loss": 0.4976,
      "mlm_loss": 0.4976,
      "step": 5040
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.9,
      "learning_rate": 0.0002525,
      "loss": 0.5053,
      "mlm_loss": 0.5053,
      "step": 5050
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.9,
      "learning_rate": 0.000253,
      "loss": 0.495,
      "mlm_loss": 0.495,
      "step": 5060
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.91,
      "learning_rate": 0.0002535,
      "loss": 0.4972,
      "mlm_loss": 0.4972,
      "step": 5070
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.92,
      "learning_rate": 0.000254,
      "loss": 0.4934,
      "mlm_loss": 0.4934,
      "step": 5080
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.93,
      "learning_rate": 0.0002545,
      "loss": 0.4907,
      "mlm_loss": 0.4907,
      "step": 5090
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.93,
      "learning_rate": 0.000255,
      "loss": 0.4936,
      "mlm_loss": 0.4936,
      "step": 5100
    },
    {
      "epoch": 3.93,
      "eval_ep_loss": -2.5289254188537598,
      "eval_loss": 0.46655842661857605,
      "eval_mlm_loss": 0.46655842661857605,
      "eval_runtime": 62.2403,
      "eval_samples_per_second": 1122.569,
      "eval_steps_per_second": 0.562,
      "step": 5100
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.94,
      "learning_rate": 0.00025550000000000003,
      "loss": 0.4996,
      "mlm_loss": 0.4996,
      "step": 5110
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.95,
      "learning_rate": 0.000256,
      "loss": 0.4911,
      "mlm_loss": 0.4911,
      "step": 5120
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.96,
      "learning_rate": 0.0002565,
      "loss": 0.4994,
      "mlm_loss": 0.4994,
      "step": 5130
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.96,
      "learning_rate": 0.000257,
      "loss": 0.4899,
      "mlm_loss": 0.4899,
      "step": 5140
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.97,
      "learning_rate": 0.0002575,
      "loss": 0.5057,
      "mlm_loss": 0.5057,
      "step": 5150
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.98,
      "learning_rate": 0.00025800000000000004,
      "loss": 0.4845,
      "mlm_loss": 0.4845,
      "step": 5160
    },
    {
      "ep_loss": 0.0,
      "epoch": 3.99,
      "learning_rate": 0.0002585,
      "loss": 0.4925,
      "mlm_loss": 0.4925,
      "step": 5170
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.0,
      "learning_rate": 0.000259,
      "loss": 0.4976,
      "mlm_loss": 0.4976,
      "step": 5180
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.0,
      "learning_rate": 0.0002595,
      "loss": 0.495,
      "mlm_loss": 0.495,
      "step": 5190
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.01,
      "learning_rate": 0.00026000000000000003,
      "loss": 0.4902,
      "mlm_loss": 0.4902,
      "step": 5200
    },
    {
      "epoch": 4.01,
      "eval_ep_loss": -2.2716147899627686,
      "eval_loss": 0.46275386214256287,
      "eval_mlm_loss": 0.46275386214256287,
      "eval_runtime": 59.0432,
      "eval_samples_per_second": 1183.353,
      "eval_steps_per_second": 0.593,
      "step": 5200
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.02,
      "learning_rate": 0.0002605,
      "loss": 0.4882,
      "mlm_loss": 0.4882,
      "step": 5210
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.03,
      "learning_rate": 0.000261,
      "loss": 0.4889,
      "mlm_loss": 0.4889,
      "step": 5220
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.03,
      "learning_rate": 0.0002615,
      "loss": 0.4967,
      "mlm_loss": 0.4967,
      "step": 5230
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.04,
      "learning_rate": 0.000262,
      "loss": 0.4841,
      "mlm_loss": 0.4841,
      "step": 5240
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.05,
      "learning_rate": 0.00026250000000000004,
      "loss": 0.4883,
      "mlm_loss": 0.4883,
      "step": 5250
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.06,
      "learning_rate": 0.000263,
      "loss": 0.4877,
      "mlm_loss": 0.4877,
      "step": 5260
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.06,
      "learning_rate": 0.0002635,
      "loss": 0.4785,
      "mlm_loss": 0.4785,
      "step": 5270
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.07,
      "learning_rate": 0.000264,
      "loss": 0.4896,
      "mlm_loss": 0.4896,
      "step": 5280
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.08,
      "learning_rate": 0.00026450000000000003,
      "loss": 0.4866,
      "mlm_loss": 0.4866,
      "step": 5290
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.09,
      "learning_rate": 0.00026500000000000004,
      "loss": 0.4871,
      "mlm_loss": 0.4871,
      "step": 5300
    },
    {
      "epoch": 4.09,
      "eval_ep_loss": -2.3992135524749756,
      "eval_loss": 0.4574643075466156,
      "eval_mlm_loss": 0.4574643075466156,
      "eval_runtime": 62.4777,
      "eval_samples_per_second": 1118.303,
      "eval_steps_per_second": 0.56,
      "step": 5300
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.1,
      "learning_rate": 0.0002655,
      "loss": 0.4836,
      "mlm_loss": 0.4836,
      "step": 5310
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.1,
      "learning_rate": 0.000266,
      "loss": 0.487,
      "mlm_loss": 0.487,
      "step": 5320
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.11,
      "learning_rate": 0.0002665,
      "loss": 0.4922,
      "mlm_loss": 0.4922,
      "step": 5330
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.12,
      "learning_rate": 0.00026700000000000004,
      "loss": 0.4886,
      "mlm_loss": 0.4886,
      "step": 5340
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.13,
      "learning_rate": 0.0002675,
      "loss": 0.4774,
      "mlm_loss": 0.4774,
      "step": 5350
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.13,
      "learning_rate": 0.000268,
      "loss": 0.4833,
      "mlm_loss": 0.4833,
      "step": 5360
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.14,
      "learning_rate": 0.0002685,
      "loss": 0.4878,
      "mlm_loss": 0.4878,
      "step": 5370
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.15,
      "learning_rate": 0.00026900000000000003,
      "loss": 0.4778,
      "mlm_loss": 0.4778,
      "step": 5380
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.16,
      "learning_rate": 0.00026950000000000005,
      "loss": 0.4833,
      "mlm_loss": 0.4833,
      "step": 5390
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.17,
      "learning_rate": 0.00027,
      "loss": 0.4849,
      "mlm_loss": 0.4849,
      "step": 5400
    },
    {
      "epoch": 4.17,
      "eval_ep_loss": -2.320817708969116,
      "eval_loss": 0.4534033536911011,
      "eval_mlm_loss": 0.4534033536911011,
      "eval_runtime": 61.6917,
      "eval_samples_per_second": 1132.551,
      "eval_steps_per_second": 0.567,
      "step": 5400
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.17,
      "learning_rate": 0.0002705,
      "loss": 0.4889,
      "mlm_loss": 0.4889,
      "step": 5410
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.18,
      "learning_rate": 0.00027100000000000003,
      "loss": 0.4793,
      "mlm_loss": 0.4793,
      "step": 5420
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.19,
      "learning_rate": 0.00027150000000000004,
      "loss": 0.476,
      "mlm_loss": 0.476,
      "step": 5430
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.2,
      "learning_rate": 0.00027200000000000005,
      "loss": 0.4808,
      "mlm_loss": 0.4808,
      "step": 5440
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.2,
      "learning_rate": 0.0002725,
      "loss": 0.475,
      "mlm_loss": 0.475,
      "step": 5450
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.21,
      "learning_rate": 0.000273,
      "loss": 0.4763,
      "mlm_loss": 0.4763,
      "step": 5460
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.22,
      "learning_rate": 0.00027350000000000003,
      "loss": 0.4796,
      "mlm_loss": 0.4796,
      "step": 5470
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.23,
      "learning_rate": 0.00027400000000000005,
      "loss": 0.4828,
      "mlm_loss": 0.4828,
      "step": 5480
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.23,
      "learning_rate": 0.0002745,
      "loss": 0.4799,
      "mlm_loss": 0.4799,
      "step": 5490
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.24,
      "learning_rate": 0.000275,
      "loss": 0.4769,
      "mlm_loss": 0.4769,
      "step": 5500
    },
    {
      "epoch": 4.24,
      "eval_ep_loss": -2.215350389480591,
      "eval_loss": 0.4512145221233368,
      "eval_mlm_loss": 0.4512145221233368,
      "eval_runtime": 60.8844,
      "eval_samples_per_second": 1147.569,
      "eval_steps_per_second": 0.575,
      "step": 5500
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.25,
      "learning_rate": 0.00027550000000000003,
      "loss": 0.4799,
      "mlm_loss": 0.4799,
      "step": 5510
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.26,
      "learning_rate": 0.00027600000000000004,
      "loss": 0.4732,
      "mlm_loss": 0.4732,
      "step": 5520
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.27,
      "learning_rate": 0.00027650000000000005,
      "loss": 0.4756,
      "mlm_loss": 0.4756,
      "step": 5530
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.27,
      "learning_rate": 0.000277,
      "loss": 0.4749,
      "mlm_loss": 0.4749,
      "step": 5540
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.28,
      "learning_rate": 0.0002775,
      "loss": 0.4656,
      "mlm_loss": 0.4656,
      "step": 5550
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.29,
      "learning_rate": 0.00027800000000000004,
      "loss": 0.473,
      "mlm_loss": 0.473,
      "step": 5560
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.3,
      "learning_rate": 0.00027850000000000005,
      "loss": 0.4819,
      "mlm_loss": 0.4819,
      "step": 5570
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.3,
      "learning_rate": 0.000279,
      "loss": 0.4825,
      "mlm_loss": 0.4825,
      "step": 5580
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.31,
      "learning_rate": 0.0002795,
      "loss": 0.4799,
      "mlm_loss": 0.4799,
      "step": 5590
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.32,
      "learning_rate": 0.00028000000000000003,
      "loss": 0.4684,
      "mlm_loss": 0.4684,
      "step": 5600
    },
    {
      "epoch": 4.32,
      "eval_ep_loss": -2.7064967155456543,
      "eval_loss": 0.44732606410980225,
      "eval_mlm_loss": 0.44732606410980225,
      "eval_runtime": 60.0339,
      "eval_samples_per_second": 1163.826,
      "eval_steps_per_second": 0.583,
      "step": 5600
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.33,
      "learning_rate": 0.00028050000000000004,
      "loss": 0.47,
      "mlm_loss": 0.47,
      "step": 5610
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.33,
      "learning_rate": 0.00028100000000000005,
      "loss": 0.4698,
      "mlm_loss": 0.4698,
      "step": 5620
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.34,
      "learning_rate": 0.00028149999999999996,
      "loss": 0.4758,
      "mlm_loss": 0.4758,
      "step": 5630
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.35,
      "learning_rate": 0.00028199999999999997,
      "loss": 0.475,
      "mlm_loss": 0.475,
      "step": 5640
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.36,
      "learning_rate": 0.0002825,
      "loss": 0.4805,
      "mlm_loss": 0.4805,
      "step": 5650
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.37,
      "learning_rate": 0.000283,
      "loss": 0.4731,
      "mlm_loss": 0.4731,
      "step": 5660
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.37,
      "learning_rate": 0.0002835,
      "loss": 0.4738,
      "mlm_loss": 0.4738,
      "step": 5670
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.38,
      "learning_rate": 0.00028399999999999996,
      "loss": 0.4672,
      "mlm_loss": 0.4672,
      "step": 5680
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.39,
      "learning_rate": 0.0002845,
      "loss": 0.4614,
      "mlm_loss": 0.4614,
      "step": 5690
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.4,
      "learning_rate": 0.000285,
      "loss": 0.4717,
      "mlm_loss": 0.4717,
      "step": 5700
    },
    {
      "epoch": 4.4,
      "eval_ep_loss": -2.254852294921875,
      "eval_loss": 0.44317254424095154,
      "eval_mlm_loss": 0.44317254424095154,
      "eval_runtime": 59.6572,
      "eval_samples_per_second": 1171.174,
      "eval_steps_per_second": 0.587,
      "step": 5700
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.4,
      "learning_rate": 0.0002855,
      "loss": 0.4665,
      "mlm_loss": 0.4665,
      "step": 5710
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.41,
      "learning_rate": 0.00028599999999999996,
      "loss": 0.4725,
      "mlm_loss": 0.4725,
      "step": 5720
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.42,
      "learning_rate": 0.00028649999999999997,
      "loss": 0.4688,
      "mlm_loss": 0.4688,
      "step": 5730
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.43,
      "learning_rate": 0.000287,
      "loss": 0.4742,
      "mlm_loss": 0.4742,
      "step": 5740
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.44,
      "learning_rate": 0.0002875,
      "loss": 0.4656,
      "mlm_loss": 0.4656,
      "step": 5750
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.44,
      "learning_rate": 0.000288,
      "loss": 0.4758,
      "mlm_loss": 0.4758,
      "step": 5760
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.45,
      "learning_rate": 0.00028849999999999997,
      "loss": 0.4705,
      "mlm_loss": 0.4705,
      "step": 5770
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.46,
      "learning_rate": 0.000289,
      "loss": 0.4655,
      "mlm_loss": 0.4655,
      "step": 5780
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.47,
      "learning_rate": 0.0002895,
      "loss": 0.4636,
      "mlm_loss": 0.4636,
      "step": 5790
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.47,
      "learning_rate": 0.00029,
      "loss": 0.4646,
      "mlm_loss": 0.4646,
      "step": 5800
    },
    {
      "epoch": 4.47,
      "eval_ep_loss": -2.2536895275115967,
      "eval_loss": 0.43992048501968384,
      "eval_mlm_loss": 0.43992048501968384,
      "eval_runtime": 62.7548,
      "eval_samples_per_second": 1113.364,
      "eval_steps_per_second": 0.558,
      "step": 5800
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.48,
      "learning_rate": 0.00029049999999999996,
      "loss": 0.4691,
      "mlm_loss": 0.4691,
      "step": 5810
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.49,
      "learning_rate": 0.00029099999999999997,
      "loss": 0.4634,
      "mlm_loss": 0.4634,
      "step": 5820
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.5,
      "learning_rate": 0.0002915,
      "loss": 0.46,
      "mlm_loss": 0.46,
      "step": 5830
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.5,
      "learning_rate": 0.000292,
      "loss": 0.4648,
      "mlm_loss": 0.4648,
      "step": 5840
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.51,
      "learning_rate": 0.0002925,
      "loss": 0.4624,
      "mlm_loss": 0.4624,
      "step": 5850
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.52,
      "learning_rate": 0.00029299999999999997,
      "loss": 0.4649,
      "mlm_loss": 0.4649,
      "step": 5860
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.53,
      "learning_rate": 0.0002935,
      "loss": 0.4626,
      "mlm_loss": 0.4626,
      "step": 5870
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.54,
      "learning_rate": 0.000294,
      "loss": 0.4648,
      "mlm_loss": 0.4648,
      "step": 5880
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.54,
      "learning_rate": 0.0002945,
      "loss": 0.4694,
      "mlm_loss": 0.4694,
      "step": 5890
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.55,
      "learning_rate": 0.000295,
      "loss": 0.4699,
      "mlm_loss": 0.4699,
      "step": 5900
    },
    {
      "epoch": 4.55,
      "eval_ep_loss": -2.540950298309326,
      "eval_loss": 0.4340103566646576,
      "eval_mlm_loss": 0.4340103566646576,
      "eval_runtime": 60.4098,
      "eval_samples_per_second": 1156.584,
      "eval_steps_per_second": 0.579,
      "step": 5900
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.56,
      "learning_rate": 0.00029549999999999997,
      "loss": 0.4615,
      "mlm_loss": 0.4615,
      "step": 5910
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.57,
      "learning_rate": 0.000296,
      "loss": 0.4628,
      "mlm_loss": 0.4628,
      "step": 5920
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.57,
      "learning_rate": 0.0002965,
      "loss": 0.4614,
      "mlm_loss": 0.4614,
      "step": 5930
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.58,
      "learning_rate": 0.000297,
      "loss": 0.4584,
      "mlm_loss": 0.4584,
      "step": 5940
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.59,
      "learning_rate": 0.00029749999999999997,
      "loss": 0.4624,
      "mlm_loss": 0.4624,
      "step": 5950
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.6,
      "learning_rate": 0.000298,
      "loss": 0.4559,
      "mlm_loss": 0.4559,
      "step": 5960
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.6,
      "learning_rate": 0.0002985,
      "loss": 0.4583,
      "mlm_loss": 0.4583,
      "step": 5970
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.61,
      "learning_rate": 0.000299,
      "loss": 0.4641,
      "mlm_loss": 0.4641,
      "step": 5980
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.62,
      "learning_rate": 0.0002995,
      "loss": 0.4639,
      "mlm_loss": 0.4639,
      "step": 5990
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.63,
      "learning_rate": 0.0003,
      "loss": 0.4632,
      "mlm_loss": 0.4632,
      "step": 6000
    },
    {
      "epoch": 4.63,
      "eval_ep_loss": -2.32681941986084,
      "eval_loss": 0.4312271177768707,
      "eval_mlm_loss": 0.4312271177768707,
      "eval_runtime": 60.5376,
      "eval_samples_per_second": 1154.142,
      "eval_steps_per_second": 0.578,
      "step": 6000
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.64,
      "learning_rate": 0.0003005,
      "loss": 0.4618,
      "mlm_loss": 0.4618,
      "step": 6010
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.64,
      "learning_rate": 0.000301,
      "loss": 0.4553,
      "mlm_loss": 0.4553,
      "step": 6020
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.65,
      "learning_rate": 0.0003015,
      "loss": 0.4649,
      "mlm_loss": 0.4649,
      "step": 6030
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.66,
      "learning_rate": 0.000302,
      "loss": 0.4569,
      "mlm_loss": 0.4569,
      "step": 6040
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.67,
      "learning_rate": 0.0003025,
      "loss": 0.4658,
      "mlm_loss": 0.4658,
      "step": 6050
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.67,
      "learning_rate": 0.000303,
      "loss": 0.4571,
      "mlm_loss": 0.4571,
      "step": 6060
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.68,
      "learning_rate": 0.0003035,
      "loss": 0.4574,
      "mlm_loss": 0.4574,
      "step": 6070
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.69,
      "learning_rate": 0.000304,
      "loss": 0.4539,
      "mlm_loss": 0.4539,
      "step": 6080
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.7,
      "learning_rate": 0.0003045,
      "loss": 0.4544,
      "mlm_loss": 0.4544,
      "step": 6090
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.7,
      "learning_rate": 0.000305,
      "loss": 0.4532,
      "mlm_loss": 0.4532,
      "step": 6100
    },
    {
      "epoch": 4.7,
      "eval_ep_loss": -2.5871927738189697,
      "eval_loss": 0.4311540722846985,
      "eval_mlm_loss": 0.4311540722846985,
      "eval_runtime": 62.3642,
      "eval_samples_per_second": 1120.337,
      "eval_steps_per_second": 0.561,
      "step": 6100
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.71,
      "learning_rate": 0.0003055,
      "loss": 0.4553,
      "mlm_loss": 0.4553,
      "step": 6110
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.72,
      "learning_rate": 0.000306,
      "loss": 0.4557,
      "mlm_loss": 0.4557,
      "step": 6120
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.73,
      "learning_rate": 0.0003065,
      "loss": 0.4568,
      "mlm_loss": 0.4568,
      "step": 6130
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.74,
      "learning_rate": 0.000307,
      "loss": 0.4633,
      "mlm_loss": 0.4633,
      "step": 6140
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.74,
      "learning_rate": 0.0003075,
      "loss": 0.4517,
      "mlm_loss": 0.4517,
      "step": 6150
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.75,
      "learning_rate": 0.000308,
      "loss": 0.4539,
      "mlm_loss": 0.4539,
      "step": 6160
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.76,
      "learning_rate": 0.0003085,
      "loss": 0.4606,
      "mlm_loss": 0.4606,
      "step": 6170
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.77,
      "learning_rate": 0.00030900000000000003,
      "loss": 0.4502,
      "mlm_loss": 0.4502,
      "step": 6180
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.77,
      "learning_rate": 0.0003095,
      "loss": 0.453,
      "mlm_loss": 0.453,
      "step": 6190
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.78,
      "learning_rate": 0.00031,
      "loss": 0.4499,
      "mlm_loss": 0.4499,
      "step": 6200
    },
    {
      "epoch": 4.78,
      "eval_ep_loss": -2.451572895050049,
      "eval_loss": 0.42733505368232727,
      "eval_mlm_loss": 0.42733505368232727,
      "eval_runtime": 59.9548,
      "eval_samples_per_second": 1165.361,
      "eval_steps_per_second": 0.584,
      "step": 6200
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.79,
      "learning_rate": 0.0003105,
      "loss": 0.4582,
      "mlm_loss": 0.4582,
      "step": 6210
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.8,
      "learning_rate": 0.000311,
      "loss": 0.4587,
      "mlm_loss": 0.4587,
      "step": 6220
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.81,
      "learning_rate": 0.0003115,
      "loss": 0.4557,
      "mlm_loss": 0.4557,
      "step": 6230
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.81,
      "learning_rate": 0.000312,
      "loss": 0.4503,
      "mlm_loss": 0.4503,
      "step": 6240
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.82,
      "learning_rate": 0.0003125,
      "loss": 0.4513,
      "mlm_loss": 0.4513,
      "step": 6250
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.83,
      "learning_rate": 0.000313,
      "loss": 0.4532,
      "mlm_loss": 0.4532,
      "step": 6260
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.84,
      "learning_rate": 0.00031350000000000003,
      "loss": 0.4535,
      "mlm_loss": 0.4535,
      "step": 6270
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.84,
      "learning_rate": 0.000314,
      "loss": 0.453,
      "mlm_loss": 0.453,
      "step": 6280
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.85,
      "learning_rate": 0.0003145,
      "loss": 0.4541,
      "mlm_loss": 0.4541,
      "step": 6290
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.86,
      "learning_rate": 0.000315,
      "loss": 0.4532,
      "mlm_loss": 0.4532,
      "step": 6300
    },
    {
      "epoch": 4.86,
      "eval_ep_loss": -2.6052770614624023,
      "eval_loss": 0.4238816797733307,
      "eval_mlm_loss": 0.4238816797733307,
      "eval_runtime": 60.9957,
      "eval_samples_per_second": 1145.474,
      "eval_steps_per_second": 0.574,
      "step": 6300
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.87,
      "learning_rate": 0.0003155,
      "loss": 0.4505,
      "mlm_loss": 0.4505,
      "step": 6310
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.87,
      "learning_rate": 0.000316,
      "loss": 0.4476,
      "mlm_loss": 0.4476,
      "step": 6320
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.88,
      "learning_rate": 0.0003165,
      "loss": 0.4495,
      "mlm_loss": 0.4495,
      "step": 6330
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.89,
      "learning_rate": 0.000317,
      "loss": 0.4483,
      "mlm_loss": 0.4483,
      "step": 6340
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.9,
      "learning_rate": 0.0003175,
      "loss": 0.4498,
      "mlm_loss": 0.4498,
      "step": 6350
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.91,
      "learning_rate": 0.00031800000000000003,
      "loss": 0.4505,
      "mlm_loss": 0.4505,
      "step": 6360
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.91,
      "learning_rate": 0.0003185,
      "loss": 0.4362,
      "mlm_loss": 0.4362,
      "step": 6370
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.92,
      "learning_rate": 0.000319,
      "loss": 0.449,
      "mlm_loss": 0.449,
      "step": 6380
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.93,
      "learning_rate": 0.0003195,
      "loss": 0.4558,
      "mlm_loss": 0.4558,
      "step": 6390
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.94,
      "learning_rate": 0.00032,
      "loss": 0.4484,
      "mlm_loss": 0.4484,
      "step": 6400
    },
    {
      "epoch": 4.94,
      "eval_ep_loss": -2.464764356613159,
      "eval_loss": 0.42165419459342957,
      "eval_mlm_loss": 0.42165419459342957,
      "eval_runtime": 60.6893,
      "eval_samples_per_second": 1151.257,
      "eval_steps_per_second": 0.577,
      "step": 6400
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.94,
      "learning_rate": 0.00032050000000000004,
      "loss": 0.4442,
      "mlm_loss": 0.4442,
      "step": 6410
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.95,
      "learning_rate": 0.000321,
      "loss": 0.4494,
      "mlm_loss": 0.4494,
      "step": 6420
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.96,
      "learning_rate": 0.0003215,
      "loss": 0.4492,
      "mlm_loss": 0.4492,
      "step": 6430
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.97,
      "learning_rate": 0.000322,
      "loss": 0.4481,
      "mlm_loss": 0.4481,
      "step": 6440
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.97,
      "learning_rate": 0.00032250000000000003,
      "loss": 0.4499,
      "mlm_loss": 0.4499,
      "step": 6450
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.98,
      "learning_rate": 0.000323,
      "loss": 0.454,
      "mlm_loss": 0.454,
      "step": 6460
    },
    {
      "ep_loss": 0.0,
      "epoch": 4.99,
      "learning_rate": 0.0003235,
      "loss": 0.4466,
      "mlm_loss": 0.4466,
      "step": 6470
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.0,
      "learning_rate": 0.000324,
      "loss": 0.4473,
      "mlm_loss": 0.4473,
      "step": 6480
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.01,
      "learning_rate": 0.00032450000000000003,
      "loss": 0.445,
      "mlm_loss": 0.445,
      "step": 6490
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.01,
      "learning_rate": 0.00032500000000000004,
      "loss": 0.439,
      "mlm_loss": 0.439,
      "step": 6500
    },
    {
      "epoch": 5.01,
      "eval_ep_loss": -2.375345230102539,
      "eval_loss": 0.4176154136657715,
      "eval_mlm_loss": 0.4176154136657715,
      "eval_runtime": 61.6206,
      "eval_samples_per_second": 1133.858,
      "eval_steps_per_second": 0.568,
      "step": 6500
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.02,
      "learning_rate": 0.0003255,
      "loss": 0.4483,
      "mlm_loss": 0.4483,
      "step": 6510
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.03,
      "learning_rate": 0.000326,
      "loss": 0.4435,
      "mlm_loss": 0.4435,
      "step": 6520
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.04,
      "learning_rate": 0.0003265,
      "loss": 0.4455,
      "mlm_loss": 0.4455,
      "step": 6530
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.04,
      "learning_rate": 0.00032700000000000003,
      "loss": 0.4464,
      "mlm_loss": 0.4464,
      "step": 6540
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.05,
      "learning_rate": 0.00032750000000000005,
      "loss": 0.4434,
      "mlm_loss": 0.4434,
      "step": 6550
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.06,
      "learning_rate": 0.000328,
      "loss": 0.4363,
      "mlm_loss": 0.4363,
      "step": 6560
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.07,
      "learning_rate": 0.0003285,
      "loss": 0.4354,
      "mlm_loss": 0.4354,
      "step": 6570
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.08,
      "learning_rate": 0.00032900000000000003,
      "loss": 0.4485,
      "mlm_loss": 0.4485,
      "step": 6580
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.08,
      "learning_rate": 0.00032950000000000004,
      "loss": 0.445,
      "mlm_loss": 0.445,
      "step": 6590
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.09,
      "learning_rate": 0.00033,
      "loss": 0.4461,
      "mlm_loss": 0.4461,
      "step": 6600
    },
    {
      "epoch": 5.09,
      "eval_ep_loss": -2.5544066429138184,
      "eval_loss": 0.41740256547927856,
      "eval_mlm_loss": 0.41740256547927856,
      "eval_runtime": 59.4616,
      "eval_samples_per_second": 1175.028,
      "eval_steps_per_second": 0.589,
      "step": 6600
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.1,
      "learning_rate": 0.0003305,
      "loss": 0.4421,
      "mlm_loss": 0.4421,
      "step": 6610
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.11,
      "learning_rate": 0.000331,
      "loss": 0.4406,
      "mlm_loss": 0.4406,
      "step": 6620
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.11,
      "learning_rate": 0.00033150000000000003,
      "loss": 0.4488,
      "mlm_loss": 0.4488,
      "step": 6630
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.12,
      "learning_rate": 0.00033200000000000005,
      "loss": 0.436,
      "mlm_loss": 0.436,
      "step": 6640
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.13,
      "learning_rate": 0.0003325,
      "loss": 0.4398,
      "mlm_loss": 0.4398,
      "step": 6650
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.14,
      "learning_rate": 0.000333,
      "loss": 0.4381,
      "mlm_loss": 0.4381,
      "step": 6660
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.14,
      "learning_rate": 0.00033350000000000003,
      "loss": 0.4479,
      "mlm_loss": 0.4479,
      "step": 6670
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.15,
      "learning_rate": 0.00033400000000000004,
      "loss": 0.4409,
      "mlm_loss": 0.4409,
      "step": 6680
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.16,
      "learning_rate": 0.00033450000000000005,
      "loss": 0.4395,
      "mlm_loss": 0.4395,
      "step": 6690
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.17,
      "learning_rate": 0.000335,
      "loss": 0.4415,
      "mlm_loss": 0.4415,
      "step": 6700
    },
    {
      "epoch": 5.17,
      "eval_ep_loss": -2.5074503421783447,
      "eval_loss": 0.41386091709136963,
      "eval_mlm_loss": 0.41386091709136963,
      "eval_runtime": 61.5941,
      "eval_samples_per_second": 1134.346,
      "eval_steps_per_second": 0.568,
      "step": 6700
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.18,
      "learning_rate": 0.0003355,
      "loss": 0.443,
      "mlm_loss": 0.443,
      "step": 6710
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.18,
      "learning_rate": 0.00033600000000000004,
      "loss": 0.4355,
      "mlm_loss": 0.4355,
      "step": 6720
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.19,
      "learning_rate": 0.00033650000000000005,
      "loss": 0.4301,
      "mlm_loss": 0.4301,
      "step": 6730
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.2,
      "learning_rate": 0.000337,
      "loss": 0.4414,
      "mlm_loss": 0.4414,
      "step": 6740
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.21,
      "learning_rate": 0.0003375,
      "loss": 0.4368,
      "mlm_loss": 0.4368,
      "step": 6750
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.21,
      "learning_rate": 0.00033800000000000003,
      "loss": 0.4378,
      "mlm_loss": 0.4378,
      "step": 6760
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.22,
      "learning_rate": 0.00033850000000000004,
      "loss": 0.4402,
      "mlm_loss": 0.4402,
      "step": 6770
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.23,
      "learning_rate": 0.00033900000000000005,
      "loss": 0.439,
      "mlm_loss": 0.439,
      "step": 6780
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.24,
      "learning_rate": 0.0003395,
      "loss": 0.4454,
      "mlm_loss": 0.4454,
      "step": 6790
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.24,
      "learning_rate": 0.00034,
      "loss": 0.4379,
      "mlm_loss": 0.4379,
      "step": 6800
    },
    {
      "epoch": 5.24,
      "eval_ep_loss": -2.6114962100982666,
      "eval_loss": 0.4116796851158142,
      "eval_mlm_loss": 0.4116796851158142,
      "eval_runtime": 61.6861,
      "eval_samples_per_second": 1132.654,
      "eval_steps_per_second": 0.567,
      "step": 6800
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.25,
      "learning_rate": 0.00034050000000000004,
      "loss": 0.4407,
      "mlm_loss": 0.4407,
      "step": 6810
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.26,
      "learning_rate": 0.00034100000000000005,
      "loss": 0.4312,
      "mlm_loss": 0.4312,
      "step": 6820
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.27,
      "learning_rate": 0.0003415,
      "loss": 0.4327,
      "mlm_loss": 0.4327,
      "step": 6830
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.28,
      "learning_rate": 0.000342,
      "loss": 0.4405,
      "mlm_loss": 0.4405,
      "step": 6840
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.28,
      "learning_rate": 0.00034250000000000003,
      "loss": 0.4395,
      "mlm_loss": 0.4395,
      "step": 6850
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.29,
      "learning_rate": 0.00034300000000000004,
      "loss": 0.4398,
      "mlm_loss": 0.4398,
      "step": 6860
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.3,
      "learning_rate": 0.00034350000000000006,
      "loss": 0.4372,
      "mlm_loss": 0.4372,
      "step": 6870
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.31,
      "learning_rate": 0.00034399999999999996,
      "loss": 0.4438,
      "mlm_loss": 0.4438,
      "step": 6880
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.31,
      "learning_rate": 0.00034449999999999997,
      "loss": 0.4356,
      "mlm_loss": 0.4356,
      "step": 6890
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.32,
      "learning_rate": 0.000345,
      "loss": 0.4357,
      "mlm_loss": 0.4357,
      "step": 6900
    },
    {
      "epoch": 5.32,
      "eval_ep_loss": -2.2833127975463867,
      "eval_loss": 0.40905898809432983,
      "eval_mlm_loss": 0.40905898809432983,
      "eval_runtime": 61.29,
      "eval_samples_per_second": 1139.974,
      "eval_steps_per_second": 0.571,
      "step": 6900
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.33,
      "learning_rate": 0.0003455,
      "loss": 0.4357,
      "mlm_loss": 0.4357,
      "step": 6910
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.34,
      "learning_rate": 0.000346,
      "loss": 0.4243,
      "mlm_loss": 0.4243,
      "step": 6920
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.35,
      "learning_rate": 0.00034649999999999997,
      "loss": 0.4404,
      "mlm_loss": 0.4404,
      "step": 6930
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.35,
      "learning_rate": 0.000347,
      "loss": 0.4319,
      "mlm_loss": 0.4319,
      "step": 6940
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.36,
      "learning_rate": 0.0003475,
      "loss": 0.4362,
      "mlm_loss": 0.4362,
      "step": 6950
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.37,
      "learning_rate": 0.000348,
      "loss": 0.4355,
      "mlm_loss": 0.4355,
      "step": 6960
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.38,
      "learning_rate": 0.00034849999999999996,
      "loss": 0.43,
      "mlm_loss": 0.43,
      "step": 6970
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.38,
      "learning_rate": 0.00034899999999999997,
      "loss": 0.4342,
      "mlm_loss": 0.4342,
      "step": 6980
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.39,
      "learning_rate": 0.0003495,
      "loss": 0.4345,
      "mlm_loss": 0.4345,
      "step": 6990
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.4,
      "learning_rate": 0.00035,
      "loss": 0.4324,
      "mlm_loss": 0.4324,
      "step": 7000
    },
    {
      "epoch": 5.4,
      "eval_ep_loss": -2.64093279838562,
      "eval_loss": 0.4105168282985687,
      "eval_mlm_loss": 0.4105168282985687,
      "eval_runtime": 60.7916,
      "eval_samples_per_second": 1149.32,
      "eval_steps_per_second": 0.576,
      "step": 7000
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.41,
      "learning_rate": 0.0003505,
      "loss": 0.4319,
      "mlm_loss": 0.4319,
      "step": 7010
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.41,
      "learning_rate": 0.00035099999999999997,
      "loss": 0.4372,
      "mlm_loss": 0.4372,
      "step": 7020
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.42,
      "learning_rate": 0.0003515,
      "loss": 0.4342,
      "mlm_loss": 0.4342,
      "step": 7030
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.43,
      "learning_rate": 0.000352,
      "loss": 0.4308,
      "mlm_loss": 0.4308,
      "step": 7040
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.44,
      "learning_rate": 0.0003525,
      "loss": 0.4373,
      "mlm_loss": 0.4373,
      "step": 7050
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.45,
      "learning_rate": 0.00035299999999999996,
      "loss": 0.4271,
      "mlm_loss": 0.4271,
      "step": 7060
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.45,
      "learning_rate": 0.0003535,
      "loss": 0.4355,
      "mlm_loss": 0.4355,
      "step": 7070
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.46,
      "learning_rate": 0.000354,
      "loss": 0.4359,
      "mlm_loss": 0.4359,
      "step": 7080
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.47,
      "learning_rate": 0.0003545,
      "loss": 0.436,
      "mlm_loss": 0.436,
      "step": 7090
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.48,
      "learning_rate": 0.000355,
      "loss": 0.4362,
      "mlm_loss": 0.4362,
      "step": 7100
    },
    {
      "epoch": 5.48,
      "eval_ep_loss": -2.3400020599365234,
      "eval_loss": 0.4073105752468109,
      "eval_mlm_loss": 0.4073105752468109,
      "eval_runtime": 63.2925,
      "eval_samples_per_second": 1103.907,
      "eval_steps_per_second": 0.553,
      "step": 7100
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.48,
      "learning_rate": 0.00035549999999999997,
      "loss": 0.4321,
      "mlm_loss": 0.4321,
      "step": 7110
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.49,
      "learning_rate": 0.000356,
      "loss": 0.4343,
      "mlm_loss": 0.4343,
      "step": 7120
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.5,
      "learning_rate": 0.0003565,
      "loss": 0.4307,
      "mlm_loss": 0.4307,
      "step": 7130
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.51,
      "learning_rate": 0.000357,
      "loss": 0.4284,
      "mlm_loss": 0.4284,
      "step": 7140
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.51,
      "learning_rate": 0.0003575,
      "loss": 0.4262,
      "mlm_loss": 0.4262,
      "step": 7150
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.52,
      "learning_rate": 0.000358,
      "loss": 0.4248,
      "mlm_loss": 0.4248,
      "step": 7160
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.53,
      "learning_rate": 0.0003585,
      "loss": 0.4419,
      "mlm_loss": 0.4419,
      "step": 7170
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.54,
      "learning_rate": 0.000359,
      "loss": 0.4267,
      "mlm_loss": 0.4267,
      "step": 7180
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.55,
      "learning_rate": 0.0003595,
      "loss": 0.4337,
      "mlm_loss": 0.4337,
      "step": 7190
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.55,
      "learning_rate": 0.00035999999999999997,
      "loss": 0.4264,
      "mlm_loss": 0.4264,
      "step": 7200
    },
    {
      "epoch": 5.55,
      "eval_ep_loss": -2.597167491912842,
      "eval_loss": 0.40466931462287903,
      "eval_mlm_loss": 0.40466931462287903,
      "eval_runtime": 61.8389,
      "eval_samples_per_second": 1129.855,
      "eval_steps_per_second": 0.566,
      "step": 7200
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.56,
      "learning_rate": 0.0003605,
      "loss": 0.4232,
      "mlm_loss": 0.4232,
      "step": 7210
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.57,
      "learning_rate": 0.000361,
      "loss": 0.4253,
      "mlm_loss": 0.4253,
      "step": 7220
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.58,
      "learning_rate": 0.0003615,
      "loss": 0.437,
      "mlm_loss": 0.437,
      "step": 7230
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.58,
      "learning_rate": 0.000362,
      "loss": 0.4319,
      "mlm_loss": 0.4319,
      "step": 7240
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.59,
      "learning_rate": 0.0003625,
      "loss": 0.4257,
      "mlm_loss": 0.4257,
      "step": 7250
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.6,
      "learning_rate": 0.000363,
      "loss": 0.4318,
      "mlm_loss": 0.4318,
      "step": 7260
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.61,
      "learning_rate": 0.0003635,
      "loss": 0.4308,
      "mlm_loss": 0.4308,
      "step": 7270
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.62,
      "learning_rate": 0.000364,
      "loss": 0.4304,
      "mlm_loss": 0.4304,
      "step": 7280
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.62,
      "learning_rate": 0.0003645,
      "loss": 0.424,
      "mlm_loss": 0.424,
      "step": 7290
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.63,
      "learning_rate": 0.000365,
      "loss": 0.4315,
      "mlm_loss": 0.4315,
      "step": 7300
    },
    {
      "epoch": 5.63,
      "eval_ep_loss": -2.466102361679077,
      "eval_loss": 0.40299102663993835,
      "eval_mlm_loss": 0.40299102663993835,
      "eval_runtime": 59.5652,
      "eval_samples_per_second": 1172.984,
      "eval_steps_per_second": 0.588,
      "step": 7300
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.64,
      "learning_rate": 0.0003655,
      "loss": 0.4305,
      "mlm_loss": 0.4305,
      "step": 7310
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.65,
      "learning_rate": 0.000366,
      "loss": 0.4215,
      "mlm_loss": 0.4215,
      "step": 7320
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.65,
      "learning_rate": 0.0003665,
      "loss": 0.4314,
      "mlm_loss": 0.4314,
      "step": 7330
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.66,
      "learning_rate": 0.000367,
      "loss": 0.4334,
      "mlm_loss": 0.4334,
      "step": 7340
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.67,
      "learning_rate": 0.0003675,
      "loss": 0.4294,
      "mlm_loss": 0.4294,
      "step": 7350
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.68,
      "learning_rate": 0.000368,
      "loss": 0.4212,
      "mlm_loss": 0.4212,
      "step": 7360
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.68,
      "learning_rate": 0.0003685,
      "loss": 0.4282,
      "mlm_loss": 0.4282,
      "step": 7370
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.69,
      "learning_rate": 0.000369,
      "loss": 0.4249,
      "mlm_loss": 0.4249,
      "step": 7380
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.7,
      "learning_rate": 0.0003695,
      "loss": 0.4236,
      "mlm_loss": 0.4236,
      "step": 7390
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.71,
      "learning_rate": 0.00037,
      "loss": 0.4287,
      "mlm_loss": 0.4287,
      "step": 7400
    },
    {
      "epoch": 5.71,
      "eval_ep_loss": -2.350245475769043,
      "eval_loss": 0.401913046836853,
      "eval_mlm_loss": 0.401913046836853,
      "eval_runtime": 60.065,
      "eval_samples_per_second": 1163.223,
      "eval_steps_per_second": 0.583,
      "step": 7400
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.72,
      "learning_rate": 0.0003705,
      "loss": 0.4203,
      "mlm_loss": 0.4203,
      "step": 7410
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.72,
      "learning_rate": 0.000371,
      "loss": 0.4252,
      "mlm_loss": 0.4252,
      "step": 7420
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.73,
      "learning_rate": 0.00037150000000000003,
      "loss": 0.4264,
      "mlm_loss": 0.4264,
      "step": 7430
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.74,
      "learning_rate": 0.000372,
      "loss": 0.4225,
      "mlm_loss": 0.4225,
      "step": 7440
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.75,
      "learning_rate": 0.0003725,
      "loss": 0.4262,
      "mlm_loss": 0.4262,
      "step": 7450
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.75,
      "learning_rate": 0.000373,
      "loss": 0.4231,
      "mlm_loss": 0.4231,
      "step": 7460
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.76,
      "learning_rate": 0.0003735,
      "loss": 0.4292,
      "mlm_loss": 0.4292,
      "step": 7470
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.77,
      "learning_rate": 0.000374,
      "loss": 0.4239,
      "mlm_loss": 0.4239,
      "step": 7480
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.78,
      "learning_rate": 0.0003745,
      "loss": 0.4209,
      "mlm_loss": 0.4209,
      "step": 7490
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.78,
      "learning_rate": 0.000375,
      "loss": 0.4198,
      "mlm_loss": 0.4198,
      "step": 7500
    },
    {
      "epoch": 5.78,
      "eval_ep_loss": -2.3534927368164062,
      "eval_loss": 0.3982246518135071,
      "eval_mlm_loss": 0.3982246518135071,
      "eval_runtime": 63.0218,
      "eval_samples_per_second": 1108.649,
      "eval_steps_per_second": 0.555,
      "step": 7500
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.79,
      "learning_rate": 0.0003755,
      "loss": 0.4212,
      "mlm_loss": 0.4212,
      "step": 7510
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.8,
      "learning_rate": 0.00037600000000000003,
      "loss": 0.4141,
      "mlm_loss": 0.4141,
      "step": 7520
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.81,
      "learning_rate": 0.0003765,
      "loss": 0.4266,
      "mlm_loss": 0.4266,
      "step": 7530
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.82,
      "learning_rate": 0.000377,
      "loss": 0.4255,
      "mlm_loss": 0.4255,
      "step": 7540
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.82,
      "learning_rate": 0.0003775,
      "loss": 0.4194,
      "mlm_loss": 0.4194,
      "step": 7550
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.83,
      "learning_rate": 0.000378,
      "loss": 0.428,
      "mlm_loss": 0.428,
      "step": 7560
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.84,
      "learning_rate": 0.0003785,
      "loss": 0.4221,
      "mlm_loss": 0.4221,
      "step": 7570
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.85,
      "learning_rate": 0.000379,
      "loss": 0.4253,
      "mlm_loss": 0.4253,
      "step": 7580
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.85,
      "learning_rate": 0.0003795,
      "loss": 0.4174,
      "mlm_loss": 0.4174,
      "step": 7590
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.86,
      "learning_rate": 0.00038,
      "loss": 0.4204,
      "mlm_loss": 0.4204,
      "step": 7600
    },
    {
      "epoch": 5.86,
      "eval_ep_loss": -2.3331265449523926,
      "eval_loss": 0.3971656858921051,
      "eval_mlm_loss": 0.3971656858921051,
      "eval_runtime": 61.175,
      "eval_samples_per_second": 1142.117,
      "eval_steps_per_second": 0.572,
      "step": 7600
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.87,
      "learning_rate": 0.00038050000000000003,
      "loss": 0.4127,
      "mlm_loss": 0.4127,
      "step": 7610
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.88,
      "learning_rate": 0.000381,
      "loss": 0.4226,
      "mlm_loss": 0.4226,
      "step": 7620
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.89,
      "learning_rate": 0.0003815,
      "loss": 0.418,
      "mlm_loss": 0.418,
      "step": 7630
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.89,
      "learning_rate": 0.000382,
      "loss": 0.4291,
      "mlm_loss": 0.4291,
      "step": 7640
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.9,
      "learning_rate": 0.00038250000000000003,
      "loss": 0.4155,
      "mlm_loss": 0.4155,
      "step": 7650
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.91,
      "learning_rate": 0.00038300000000000004,
      "loss": 0.4213,
      "mlm_loss": 0.4213,
      "step": 7660
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.92,
      "learning_rate": 0.0003835,
      "loss": 0.4273,
      "mlm_loss": 0.4273,
      "step": 7670
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.92,
      "learning_rate": 0.000384,
      "loss": 0.4252,
      "mlm_loss": 0.4252,
      "step": 7680
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.93,
      "learning_rate": 0.0003845,
      "loss": 0.4201,
      "mlm_loss": 0.4201,
      "step": 7690
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.94,
      "learning_rate": 0.00038500000000000003,
      "loss": 0.4198,
      "mlm_loss": 0.4198,
      "step": 7700
    },
    {
      "epoch": 5.94,
      "eval_ep_loss": -2.514362335205078,
      "eval_loss": 0.39368629455566406,
      "eval_mlm_loss": 0.39368629455566406,
      "eval_runtime": 60.7647,
      "eval_samples_per_second": 1149.828,
      "eval_steps_per_second": 0.576,
      "step": 7700
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.95,
      "learning_rate": 0.0003855,
      "loss": 0.4167,
      "mlm_loss": 0.4167,
      "step": 7710
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.95,
      "learning_rate": 0.000386,
      "loss": 0.4235,
      "mlm_loss": 0.4235,
      "step": 7720
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.96,
      "learning_rate": 0.0003865,
      "loss": 0.4233,
      "mlm_loss": 0.4233,
      "step": 7730
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.97,
      "learning_rate": 0.00038700000000000003,
      "loss": 0.4227,
      "mlm_loss": 0.4227,
      "step": 7740
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.98,
      "learning_rate": 0.00038750000000000004,
      "loss": 0.4194,
      "mlm_loss": 0.4194,
      "step": 7750
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.99,
      "learning_rate": 0.000388,
      "loss": 0.4194,
      "mlm_loss": 0.4194,
      "step": 7760
    },
    {
      "ep_loss": 0.0,
      "epoch": 5.99,
      "learning_rate": 0.0003885,
      "loss": 0.4171,
      "mlm_loss": 0.4171,
      "step": 7770
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.0,
      "learning_rate": 0.000389,
      "loss": 0.4197,
      "mlm_loss": 0.4197,
      "step": 7780
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.01,
      "learning_rate": 0.00038950000000000003,
      "loss": 0.4138,
      "mlm_loss": 0.4138,
      "step": 7790
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.02,
      "learning_rate": 0.00039000000000000005,
      "loss": 0.4149,
      "mlm_loss": 0.4149,
      "step": 7800
    },
    {
      "epoch": 6.02,
      "eval_ep_loss": -2.336341619491577,
      "eval_loss": 0.3958059251308441,
      "eval_mlm_loss": 0.3958059251308441,
      "eval_runtime": 59.0112,
      "eval_samples_per_second": 1183.995,
      "eval_steps_per_second": 0.593,
      "step": 7800
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.02,
      "learning_rate": 0.0003905,
      "loss": 0.4197,
      "mlm_loss": 0.4197,
      "step": 7810
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.03,
      "learning_rate": 0.000391,
      "loss": 0.4135,
      "mlm_loss": 0.4135,
      "step": 7820
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.04,
      "learning_rate": 0.00039150000000000003,
      "loss": 0.4207,
      "mlm_loss": 0.4207,
      "step": 7830
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.05,
      "learning_rate": 0.00039200000000000004,
      "loss": 0.419,
      "mlm_loss": 0.419,
      "step": 7840
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.05,
      "learning_rate": 0.0003925,
      "loss": 0.4169,
      "mlm_loss": 0.4169,
      "step": 7850
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.06,
      "learning_rate": 0.000393,
      "loss": 0.4183,
      "mlm_loss": 0.4183,
      "step": 7860
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.07,
      "learning_rate": 0.0003935,
      "loss": 0.4127,
      "mlm_loss": 0.4127,
      "step": 7870
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.08,
      "learning_rate": 0.00039400000000000004,
      "loss": 0.421,
      "mlm_loss": 0.421,
      "step": 7880
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.09,
      "learning_rate": 0.00039450000000000005,
      "loss": 0.4135,
      "mlm_loss": 0.4135,
      "step": 7890
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.09,
      "learning_rate": 0.000395,
      "loss": 0.4174,
      "mlm_loss": 0.4174,
      "step": 7900
    },
    {
      "epoch": 6.09,
      "eval_ep_loss": -2.3622541427612305,
      "eval_loss": 0.3937676250934601,
      "eval_mlm_loss": 0.3937676250934601,
      "eval_runtime": 60.7463,
      "eval_samples_per_second": 1150.176,
      "eval_steps_per_second": 0.576,
      "step": 7900
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.1,
      "learning_rate": 0.0003955,
      "loss": 0.4256,
      "mlm_loss": 0.4256,
      "step": 7910
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.11,
      "learning_rate": 0.00039600000000000003,
      "loss": 0.4217,
      "mlm_loss": 0.4217,
      "step": 7920
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.12,
      "learning_rate": 0.00039650000000000004,
      "loss": 0.4239,
      "mlm_loss": 0.4239,
      "step": 7930
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.12,
      "learning_rate": 0.00039700000000000005,
      "loss": 0.4183,
      "mlm_loss": 0.4183,
      "step": 7940
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.13,
      "learning_rate": 0.0003975,
      "loss": 0.4172,
      "mlm_loss": 0.4172,
      "step": 7950
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.14,
      "learning_rate": 0.000398,
      "loss": 0.42,
      "mlm_loss": 0.42,
      "step": 7960
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.15,
      "learning_rate": 0.00039850000000000004,
      "loss": 0.4149,
      "mlm_loss": 0.4149,
      "step": 7970
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.16,
      "learning_rate": 0.00039900000000000005,
      "loss": 0.4128,
      "mlm_loss": 0.4128,
      "step": 7980
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.16,
      "learning_rate": 0.0003995,
      "loss": 0.4153,
      "mlm_loss": 0.4153,
      "step": 7990
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.17,
      "learning_rate": 0.0004,
      "loss": 0.418,
      "mlm_loss": 0.418,
      "step": 8000
    },
    {
      "epoch": 6.17,
      "eval_ep_loss": -2.42641282081604,
      "eval_loss": 0.3915087580680847,
      "eval_mlm_loss": 0.3915087580680847,
      "eval_runtime": 60.1518,
      "eval_samples_per_second": 1161.545,
      "eval_steps_per_second": 0.582,
      "step": 8000
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.18,
      "learning_rate": 0.00040050000000000003,
      "loss": 0.4161,
      "mlm_loss": 0.4161,
      "step": 8010
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.19,
      "learning_rate": 0.00040100000000000004,
      "loss": 0.4158,
      "mlm_loss": 0.4158,
      "step": 8020
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.19,
      "learning_rate": 0.00040150000000000006,
      "loss": 0.4224,
      "mlm_loss": 0.4224,
      "step": 8030
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.2,
      "learning_rate": 0.000402,
      "loss": 0.4203,
      "mlm_loss": 0.4203,
      "step": 8040
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.21,
      "learning_rate": 0.0004025,
      "loss": 0.4134,
      "mlm_loss": 0.4134,
      "step": 8050
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.22,
      "learning_rate": 0.00040300000000000004,
      "loss": 0.4153,
      "mlm_loss": 0.4153,
      "step": 8060
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.22,
      "learning_rate": 0.00040350000000000005,
      "loss": 0.4126,
      "mlm_loss": 0.4126,
      "step": 8070
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.23,
      "learning_rate": 0.000404,
      "loss": 0.4142,
      "mlm_loss": 0.4142,
      "step": 8080
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.24,
      "learning_rate": 0.0004045,
      "loss": 0.4141,
      "mlm_loss": 0.4141,
      "step": 8090
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.25,
      "learning_rate": 0.00040500000000000003,
      "loss": 0.4129,
      "mlm_loss": 0.4129,
      "step": 8100
    },
    {
      "epoch": 6.25,
      "eval_ep_loss": -2.3123908042907715,
      "eval_loss": 0.39225736260414124,
      "eval_mlm_loss": 0.39225736260414124,
      "eval_runtime": 61.9154,
      "eval_samples_per_second": 1128.459,
      "eval_steps_per_second": 0.565,
      "step": 8100
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.26,
      "learning_rate": 0.00040550000000000004,
      "loss": 0.4155,
      "mlm_loss": 0.4155,
      "step": 8110
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.26,
      "learning_rate": 0.00040600000000000006,
      "loss": 0.4174,
      "mlm_loss": 0.4174,
      "step": 8120
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.27,
      "learning_rate": 0.00040649999999999996,
      "loss": 0.4099,
      "mlm_loss": 0.4099,
      "step": 8130
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.28,
      "learning_rate": 0.00040699999999999997,
      "loss": 0.4123,
      "mlm_loss": 0.4123,
      "step": 8140
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.29,
      "learning_rate": 0.0004075,
      "loss": 0.4216,
      "mlm_loss": 0.4216,
      "step": 8150
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.29,
      "learning_rate": 0.000408,
      "loss": 0.4095,
      "mlm_loss": 0.4095,
      "step": 8160
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.3,
      "learning_rate": 0.0004085,
      "loss": 0.413,
      "mlm_loss": 0.413,
      "step": 8170
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.31,
      "learning_rate": 0.00040899999999999997,
      "loss": 0.4165,
      "mlm_loss": 0.4165,
      "step": 8180
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.32,
      "learning_rate": 0.0004095,
      "loss": 0.4191,
      "mlm_loss": 0.4191,
      "step": 8190
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.32,
      "learning_rate": 0.00041,
      "loss": 0.4161,
      "mlm_loss": 0.4161,
      "step": 8200
    },
    {
      "epoch": 6.32,
      "eval_ep_loss": -2.4482126235961914,
      "eval_loss": 0.388824462890625,
      "eval_mlm_loss": 0.388824462890625,
      "eval_runtime": 59.1336,
      "eval_samples_per_second": 1181.545,
      "eval_steps_per_second": 0.592,
      "step": 8200
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.33,
      "learning_rate": 0.0004105,
      "loss": 0.4218,
      "mlm_loss": 0.4218,
      "step": 8210
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.34,
      "learning_rate": 0.00041099999999999996,
      "loss": 0.4085,
      "mlm_loss": 0.4085,
      "step": 8220
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.35,
      "learning_rate": 0.0004115,
      "loss": 0.4114,
      "mlm_loss": 0.4114,
      "step": 8230
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.36,
      "learning_rate": 0.000412,
      "loss": 0.4129,
      "mlm_loss": 0.4129,
      "step": 8240
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.36,
      "learning_rate": 0.0004125,
      "loss": 0.422,
      "mlm_loss": 0.422,
      "step": 8250
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.37,
      "learning_rate": 0.000413,
      "loss": 0.4184,
      "mlm_loss": 0.4184,
      "step": 8260
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.38,
      "learning_rate": 0.00041349999999999997,
      "loss": 0.4192,
      "mlm_loss": 0.4192,
      "step": 8270
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.39,
      "learning_rate": 0.000414,
      "loss": 0.4168,
      "mlm_loss": 0.4168,
      "step": 8280
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.39,
      "learning_rate": 0.0004145,
      "loss": 0.4171,
      "mlm_loss": 0.4171,
      "step": 8290
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.4,
      "learning_rate": 0.000415,
      "loss": 0.4131,
      "mlm_loss": 0.4131,
      "step": 8300
    },
    {
      "epoch": 6.4,
      "eval_ep_loss": -2.358128547668457,
      "eval_loss": 0.38723644614219666,
      "eval_mlm_loss": 0.38723644614219666,
      "eval_runtime": 60.0999,
      "eval_samples_per_second": 1162.548,
      "eval_steps_per_second": 0.582,
      "step": 8300
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.41,
      "learning_rate": 0.00041549999999999996,
      "loss": 0.4149,
      "mlm_loss": 0.4149,
      "step": 8310
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.42,
      "learning_rate": 0.000416,
      "loss": 0.4068,
      "mlm_loss": 0.4068,
      "step": 8320
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.42,
      "learning_rate": 0.0004165,
      "loss": 0.4102,
      "mlm_loss": 0.4102,
      "step": 8330
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.43,
      "learning_rate": 0.000417,
      "loss": 0.4129,
      "mlm_loss": 0.4129,
      "step": 8340
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.44,
      "learning_rate": 0.0004175,
      "loss": 0.413,
      "mlm_loss": 0.413,
      "step": 8350
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.45,
      "learning_rate": 0.00041799999999999997,
      "loss": 0.4143,
      "mlm_loss": 0.4143,
      "step": 8360
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.46,
      "learning_rate": 0.0004185,
      "loss": 0.4134,
      "mlm_loss": 0.4134,
      "step": 8370
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.46,
      "learning_rate": 0.000419,
      "loss": 0.406,
      "mlm_loss": 0.406,
      "step": 8380
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.47,
      "learning_rate": 0.0004195,
      "loss": 0.4144,
      "mlm_loss": 0.4144,
      "step": 8390
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.48,
      "learning_rate": 0.00042,
      "loss": 0.4131,
      "mlm_loss": 0.4131,
      "step": 8400
    },
    {
      "epoch": 6.48,
      "eval_ep_loss": -2.0901758670806885,
      "eval_loss": 0.3867802917957306,
      "eval_mlm_loss": 0.3867802917957306,
      "eval_runtime": 63.0955,
      "eval_samples_per_second": 1107.354,
      "eval_steps_per_second": 0.555,
      "step": 8400
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.49,
      "learning_rate": 0.0004205,
      "loss": 0.4011,
      "mlm_loss": 0.4011,
      "step": 8410
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.49,
      "learning_rate": 0.000421,
      "loss": 0.4117,
      "mlm_loss": 0.4117,
      "step": 8420
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.5,
      "learning_rate": 0.0004215,
      "loss": 0.414,
      "mlm_loss": 0.414,
      "step": 8430
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.51,
      "learning_rate": 0.000422,
      "loss": 0.4186,
      "mlm_loss": 0.4186,
      "step": 8440
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.52,
      "learning_rate": 0.00042249999999999997,
      "loss": 0.4149,
      "mlm_loss": 0.4149,
      "step": 8450
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.53,
      "learning_rate": 0.000423,
      "loss": 0.4154,
      "mlm_loss": 0.4154,
      "step": 8460
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.53,
      "learning_rate": 0.0004235,
      "loss": 0.409,
      "mlm_loss": 0.409,
      "step": 8470
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.54,
      "learning_rate": 0.000424,
      "loss": 0.4098,
      "mlm_loss": 0.4098,
      "step": 8480
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.55,
      "learning_rate": 0.0004245,
      "loss": 0.4179,
      "mlm_loss": 0.4179,
      "step": 8490
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.56,
      "learning_rate": 0.000425,
      "loss": 0.4147,
      "mlm_loss": 0.4147,
      "step": 8500
    },
    {
      "epoch": 6.56,
      "eval_ep_loss": -2.5853233337402344,
      "eval_loss": 0.3858720660209656,
      "eval_mlm_loss": 0.3858720660209656,
      "eval_runtime": 60.8689,
      "eval_samples_per_second": 1147.86,
      "eval_steps_per_second": 0.575,
      "step": 8500
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.56,
      "learning_rate": 0.0004255,
      "loss": 0.4094,
      "mlm_loss": 0.4094,
      "step": 8510
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.57,
      "learning_rate": 0.000426,
      "loss": 0.4136,
      "mlm_loss": 0.4136,
      "step": 8520
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.58,
      "learning_rate": 0.0004265,
      "loss": 0.4077,
      "mlm_loss": 0.4077,
      "step": 8530
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.59,
      "learning_rate": 0.000427,
      "loss": 0.4077,
      "mlm_loss": 0.4077,
      "step": 8540
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.59,
      "learning_rate": 0.0004275,
      "loss": 0.4115,
      "mlm_loss": 0.4115,
      "step": 8550
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.6,
      "learning_rate": 0.000428,
      "loss": 0.4068,
      "mlm_loss": 0.4068,
      "step": 8560
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.61,
      "learning_rate": 0.0004285,
      "loss": 0.4096,
      "mlm_loss": 0.4096,
      "step": 8570
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.62,
      "learning_rate": 0.000429,
      "loss": 0.4146,
      "mlm_loss": 0.4146,
      "step": 8580
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.63,
      "learning_rate": 0.0004295,
      "loss": 0.4095,
      "mlm_loss": 0.4095,
      "step": 8590
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.63,
      "learning_rate": 0.00043,
      "loss": 0.4105,
      "mlm_loss": 0.4105,
      "step": 8600
    },
    {
      "epoch": 6.63,
      "eval_ep_loss": -2.559201955795288,
      "eval_loss": 0.3874005973339081,
      "eval_mlm_loss": 0.3874005973339081,
      "eval_runtime": 62.3828,
      "eval_samples_per_second": 1120.003,
      "eval_steps_per_second": 0.561,
      "step": 8600
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.64,
      "learning_rate": 0.0004305,
      "loss": 0.4121,
      "mlm_loss": 0.4121,
      "step": 8610
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.65,
      "learning_rate": 0.000431,
      "loss": 0.4115,
      "mlm_loss": 0.4115,
      "step": 8620
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.66,
      "learning_rate": 0.0004315,
      "loss": 0.4124,
      "mlm_loss": 0.4124,
      "step": 8630
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.66,
      "learning_rate": 0.000432,
      "loss": 0.4158,
      "mlm_loss": 0.4158,
      "step": 8640
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.67,
      "learning_rate": 0.0004325,
      "loss": 0.4099,
      "mlm_loss": 0.4099,
      "step": 8650
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.68,
      "learning_rate": 0.000433,
      "loss": 0.4157,
      "mlm_loss": 0.4157,
      "step": 8660
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.69,
      "learning_rate": 0.0004335,
      "loss": 0.4032,
      "mlm_loss": 0.4032,
      "step": 8670
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.69,
      "learning_rate": 0.00043400000000000003,
      "loss": 0.4094,
      "mlm_loss": 0.4094,
      "step": 8680
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.7,
      "learning_rate": 0.0004345,
      "loss": 0.4057,
      "mlm_loss": 0.4057,
      "step": 8690
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.71,
      "learning_rate": 0.000435,
      "loss": 0.4144,
      "mlm_loss": 0.4144,
      "step": 8700
    },
    {
      "epoch": 6.71,
      "eval_ep_loss": -2.2817635536193848,
      "eval_loss": 0.3839435577392578,
      "eval_mlm_loss": 0.3839435577392578,
      "eval_runtime": 61.6764,
      "eval_samples_per_second": 1132.831,
      "eval_steps_per_second": 0.567,
      "step": 8700
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.72,
      "learning_rate": 0.0004355,
      "loss": 0.4114,
      "mlm_loss": 0.4114,
      "step": 8710
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.73,
      "learning_rate": 0.000436,
      "loss": 0.4091,
      "mlm_loss": 0.4091,
      "step": 8720
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.73,
      "learning_rate": 0.0004365,
      "loss": 0.4058,
      "mlm_loss": 0.4058,
      "step": 8730
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.74,
      "learning_rate": 0.000437,
      "loss": 0.4009,
      "mlm_loss": 0.4009,
      "step": 8740
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.75,
      "learning_rate": 0.0004375,
      "loss": 0.405,
      "mlm_loss": 0.405,
      "step": 8750
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.76,
      "learning_rate": 0.000438,
      "loss": 0.4084,
      "mlm_loss": 0.4084,
      "step": 8760
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.76,
      "learning_rate": 0.00043850000000000003,
      "loss": 0.409,
      "mlm_loss": 0.409,
      "step": 8770
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.77,
      "learning_rate": 0.000439,
      "loss": 0.407,
      "mlm_loss": 0.407,
      "step": 8780
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.78,
      "learning_rate": 0.0004395,
      "loss": 0.4003,
      "mlm_loss": 0.4003,
      "step": 8790
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.79,
      "learning_rate": 0.00044,
      "loss": 0.4021,
      "mlm_loss": 0.4021,
      "step": 8800
    },
    {
      "epoch": 6.79,
      "eval_ep_loss": -2.1474719047546387,
      "eval_loss": 0.38189372420310974,
      "eval_mlm_loss": 0.38189372420310974,
      "eval_runtime": 59.6801,
      "eval_samples_per_second": 1170.726,
      "eval_steps_per_second": 0.586,
      "step": 8800
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.8,
      "learning_rate": 0.00044050000000000003,
      "loss": 0.4166,
      "mlm_loss": 0.4166,
      "step": 8810
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.8,
      "learning_rate": 0.000441,
      "loss": 0.4055,
      "mlm_loss": 0.4055,
      "step": 8820
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.81,
      "learning_rate": 0.0004415,
      "loss": 0.4095,
      "mlm_loss": 0.4095,
      "step": 8830
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.82,
      "learning_rate": 0.000442,
      "loss": 0.4063,
      "mlm_loss": 0.4063,
      "step": 8840
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.83,
      "learning_rate": 0.0004425,
      "loss": 0.4124,
      "mlm_loss": 0.4124,
      "step": 8850
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.83,
      "learning_rate": 0.00044300000000000003,
      "loss": 0.4052,
      "mlm_loss": 0.4052,
      "step": 8860
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.84,
      "learning_rate": 0.0004435,
      "loss": 0.4106,
      "mlm_loss": 0.4106,
      "step": 8870
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.85,
      "learning_rate": 0.000444,
      "loss": 0.4042,
      "mlm_loss": 0.4042,
      "step": 8880
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.86,
      "learning_rate": 0.0004445,
      "loss": 0.4126,
      "mlm_loss": 0.4126,
      "step": 8890
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.86,
      "learning_rate": 0.00044500000000000003,
      "loss": 0.4019,
      "mlm_loss": 0.4019,
      "step": 8900
    },
    {
      "epoch": 6.86,
      "eval_ep_loss": -2.1015372276306152,
      "eval_loss": 0.3846256732940674,
      "eval_mlm_loss": 0.3846256732940674,
      "eval_runtime": 59.4967,
      "eval_samples_per_second": 1174.335,
      "eval_steps_per_second": 0.588,
      "step": 8900
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.87,
      "learning_rate": 0.00044550000000000004,
      "loss": 0.4092,
      "mlm_loss": 0.4092,
      "step": 8910
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.88,
      "learning_rate": 0.000446,
      "loss": 0.4053,
      "mlm_loss": 0.4053,
      "step": 8920
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.89,
      "learning_rate": 0.0004465,
      "loss": 0.4071,
      "mlm_loss": 0.4071,
      "step": 8930
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.9,
      "learning_rate": 0.000447,
      "loss": 0.4087,
      "mlm_loss": 0.4087,
      "step": 8940
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.9,
      "learning_rate": 0.00044750000000000004,
      "loss": 0.4075,
      "mlm_loss": 0.4075,
      "step": 8950
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.91,
      "learning_rate": 0.000448,
      "loss": 0.4134,
      "mlm_loss": 0.4134,
      "step": 8960
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.92,
      "learning_rate": 0.0004485,
      "loss": 0.407,
      "mlm_loss": 0.407,
      "step": 8970
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.93,
      "learning_rate": 0.000449,
      "loss": 0.4081,
      "mlm_loss": 0.4081,
      "step": 8980
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.93,
      "learning_rate": 0.00044950000000000003,
      "loss": 0.4047,
      "mlm_loss": 0.4047,
      "step": 8990
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.94,
      "learning_rate": 0.00045000000000000004,
      "loss": 0.4046,
      "mlm_loss": 0.4046,
      "step": 9000
    },
    {
      "epoch": 6.94,
      "eval_ep_loss": -2.1676366329193115,
      "eval_loss": 0.3808509409427643,
      "eval_mlm_loss": 0.3808509409427643,
      "eval_runtime": 60.7859,
      "eval_samples_per_second": 1149.428,
      "eval_steps_per_second": 0.576,
      "step": 9000
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.95,
      "learning_rate": 0.0004505,
      "loss": 0.4108,
      "mlm_loss": 0.4108,
      "step": 9010
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.96,
      "learning_rate": 0.000451,
      "loss": 0.4095,
      "mlm_loss": 0.4095,
      "step": 9020
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.96,
      "learning_rate": 0.0004515,
      "loss": 0.4069,
      "mlm_loss": 0.4069,
      "step": 9030
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.97,
      "learning_rate": 0.00045200000000000004,
      "loss": 0.407,
      "mlm_loss": 0.407,
      "step": 9040
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.98,
      "learning_rate": 0.00045250000000000005,
      "loss": 0.4056,
      "mlm_loss": 0.4056,
      "step": 9050
    },
    {
      "ep_loss": 0.0,
      "epoch": 6.99,
      "learning_rate": 0.000453,
      "loss": 0.4107,
      "mlm_loss": 0.4107,
      "step": 9060
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.0,
      "learning_rate": 0.0004535,
      "loss": 0.4085,
      "mlm_loss": 0.4085,
      "step": 9070
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.0,
      "learning_rate": 0.00045400000000000003,
      "loss": 0.4146,
      "mlm_loss": 0.4146,
      "step": 9080
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.01,
      "learning_rate": 0.00045450000000000004,
      "loss": 0.4041,
      "mlm_loss": 0.4041,
      "step": 9090
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.02,
      "learning_rate": 0.000455,
      "loss": 0.414,
      "mlm_loss": 0.414,
      "step": 9100
    },
    {
      "epoch": 7.02,
      "eval_ep_loss": -2.345536231994629,
      "eval_loss": 0.3819408118724823,
      "eval_mlm_loss": 0.3819408118724823,
      "eval_runtime": 61.289,
      "eval_samples_per_second": 1139.992,
      "eval_steps_per_second": 0.571,
      "step": 9100
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.03,
      "learning_rate": 0.0004555,
      "loss": 0.4014,
      "mlm_loss": 0.4014,
      "step": 9110
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.03,
      "learning_rate": 0.000456,
      "loss": 0.4039,
      "mlm_loss": 0.4039,
      "step": 9120
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.04,
      "learning_rate": 0.00045650000000000004,
      "loss": 0.4066,
      "mlm_loss": 0.4066,
      "step": 9130
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.05,
      "learning_rate": 0.00045700000000000005,
      "loss": 0.4107,
      "mlm_loss": 0.4107,
      "step": 9140
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.06,
      "learning_rate": 0.0004575,
      "loss": 0.4035,
      "mlm_loss": 0.4035,
      "step": 9150
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.07,
      "learning_rate": 0.000458,
      "loss": 0.4037,
      "mlm_loss": 0.4037,
      "step": 9160
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.07,
      "learning_rate": 0.00045850000000000003,
      "loss": 0.4035,
      "mlm_loss": 0.4035,
      "step": 9170
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.08,
      "learning_rate": 0.00045900000000000004,
      "loss": 0.4054,
      "mlm_loss": 0.4054,
      "step": 9180
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.09,
      "learning_rate": 0.00045950000000000006,
      "loss": 0.4078,
      "mlm_loss": 0.4078,
      "step": 9190
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.1,
      "learning_rate": 0.00046,
      "loss": 0.4066,
      "mlm_loss": 0.4066,
      "step": 9200
    },
    {
      "epoch": 7.1,
      "eval_ep_loss": -2.1998836994171143,
      "eval_loss": 0.37953051924705505,
      "eval_mlm_loss": 0.37953051924705505,
      "eval_runtime": 59.882,
      "eval_samples_per_second": 1166.779,
      "eval_steps_per_second": 0.584,
      "step": 9200
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.1,
      "learning_rate": 0.0004605,
      "loss": 0.3965,
      "mlm_loss": 0.3965,
      "step": 9210
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.11,
      "learning_rate": 0.00046100000000000004,
      "loss": 0.3973,
      "mlm_loss": 0.3973,
      "step": 9220
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.12,
      "learning_rate": 0.00046150000000000005,
      "loss": 0.3996,
      "mlm_loss": 0.3996,
      "step": 9230
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.13,
      "learning_rate": 0.000462,
      "loss": 0.4075,
      "mlm_loss": 0.4075,
      "step": 9240
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.13,
      "learning_rate": 0.0004625,
      "loss": 0.4042,
      "mlm_loss": 0.4042,
      "step": 9250
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.14,
      "learning_rate": 0.00046300000000000003,
      "loss": 0.3976,
      "mlm_loss": 0.3976,
      "step": 9260
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.15,
      "learning_rate": 0.00046350000000000004,
      "loss": 0.4026,
      "mlm_loss": 0.4026,
      "step": 9270
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.16,
      "learning_rate": 0.00046400000000000006,
      "loss": 0.4047,
      "mlm_loss": 0.4047,
      "step": 9280
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.17,
      "learning_rate": 0.0004645,
      "loss": 0.403,
      "mlm_loss": 0.403,
      "step": 9290
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.17,
      "learning_rate": 0.000465,
      "loss": 0.4064,
      "mlm_loss": 0.4064,
      "step": 9300
    },
    {
      "epoch": 7.17,
      "eval_ep_loss": -2.131833553314209,
      "eval_loss": 0.38073843717575073,
      "eval_mlm_loss": 0.38073843717575073,
      "eval_runtime": 61.1101,
      "eval_samples_per_second": 1143.331,
      "eval_steps_per_second": 0.573,
      "step": 9300
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.18,
      "learning_rate": 0.00046550000000000004,
      "loss": 0.4023,
      "mlm_loss": 0.4023,
      "step": 9310
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.19,
      "learning_rate": 0.00046600000000000005,
      "loss": 0.4037,
      "mlm_loss": 0.4037,
      "step": 9320
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.2,
      "learning_rate": 0.0004665,
      "loss": 0.4024,
      "mlm_loss": 0.4024,
      "step": 9330
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.2,
      "learning_rate": 0.000467,
      "loss": 0.4031,
      "mlm_loss": 0.4031,
      "step": 9340
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.21,
      "learning_rate": 0.00046750000000000003,
      "loss": 0.4069,
      "mlm_loss": 0.4069,
      "step": 9350
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.22,
      "learning_rate": 0.00046800000000000005,
      "loss": 0.4134,
      "mlm_loss": 0.4134,
      "step": 9360
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.23,
      "learning_rate": 0.00046850000000000006,
      "loss": 0.3998,
      "mlm_loss": 0.3998,
      "step": 9370
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.23,
      "learning_rate": 0.00046899999999999996,
      "loss": 0.403,
      "mlm_loss": 0.403,
      "step": 9380
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.24,
      "learning_rate": 0.0004695,
      "loss": 0.4067,
      "mlm_loss": 0.4067,
      "step": 9390
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.25,
      "learning_rate": 0.00047,
      "loss": 0.4022,
      "mlm_loss": 0.4022,
      "step": 9400
    },
    {
      "epoch": 7.25,
      "eval_ep_loss": -2.311575412750244,
      "eval_loss": 0.3808383643627167,
      "eval_mlm_loss": 0.3808383643627167,
      "eval_runtime": 61.1652,
      "eval_samples_per_second": 1142.301,
      "eval_steps_per_second": 0.572,
      "step": 9400
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.26,
      "learning_rate": 0.0004705,
      "loss": 0.405,
      "mlm_loss": 0.405,
      "step": 9410
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.27,
      "learning_rate": 0.000471,
      "loss": 0.3949,
      "mlm_loss": 0.3949,
      "step": 9420
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.27,
      "learning_rate": 0.00047149999999999997,
      "loss": 0.405,
      "mlm_loss": 0.405,
      "step": 9430
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.28,
      "learning_rate": 0.000472,
      "loss": 0.4092,
      "mlm_loss": 0.4092,
      "step": 9440
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.29,
      "learning_rate": 0.0004725,
      "loss": 0.4102,
      "mlm_loss": 0.4102,
      "step": 9450
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.3,
      "learning_rate": 0.000473,
      "loss": 0.4066,
      "mlm_loss": 0.4066,
      "step": 9460
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.3,
      "learning_rate": 0.00047349999999999996,
      "loss": 0.3987,
      "mlm_loss": 0.3987,
      "step": 9470
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.31,
      "learning_rate": 0.000474,
      "loss": 0.399,
      "mlm_loss": 0.399,
      "step": 9480
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.32,
      "learning_rate": 0.0004745,
      "loss": 0.4033,
      "mlm_loss": 0.4033,
      "step": 9490
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.33,
      "learning_rate": 0.000475,
      "loss": 0.3982,
      "mlm_loss": 0.3982,
      "step": 9500
    },
    {
      "epoch": 7.33,
      "eval_ep_loss": -2.2588560581207275,
      "eval_loss": 0.37661808729171753,
      "eval_mlm_loss": 0.37661808729171753,
      "eval_runtime": 61.5272,
      "eval_samples_per_second": 1135.579,
      "eval_steps_per_second": 0.569,
      "step": 9500
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.34,
      "learning_rate": 0.0004755,
      "loss": 0.4048,
      "mlm_loss": 0.4048,
      "step": 9510
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.34,
      "learning_rate": 0.00047599999999999997,
      "loss": 0.4034,
      "mlm_loss": 0.4034,
      "step": 9520
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.35,
      "learning_rate": 0.0004765,
      "loss": 0.4001,
      "mlm_loss": 0.4001,
      "step": 9530
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.36,
      "learning_rate": 0.000477,
      "loss": 0.406,
      "mlm_loss": 0.406,
      "step": 9540
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.37,
      "learning_rate": 0.0004775,
      "loss": 0.4007,
      "mlm_loss": 0.4007,
      "step": 9550
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.37,
      "learning_rate": 0.00047799999999999996,
      "loss": 0.416,
      "mlm_loss": 0.416,
      "step": 9560
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.38,
      "learning_rate": 0.0004785,
      "loss": 0.398,
      "mlm_loss": 0.398,
      "step": 9570
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.39,
      "learning_rate": 0.000479,
      "loss": 0.4045,
      "mlm_loss": 0.4045,
      "step": 9580
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.4,
      "learning_rate": 0.0004795,
      "loss": 0.4011,
      "mlm_loss": 0.4011,
      "step": 9590
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.4,
      "learning_rate": 0.00048,
      "loss": 0.3991,
      "mlm_loss": 0.3991,
      "step": 9600
    },
    {
      "epoch": 7.4,
      "eval_ep_loss": -2.4242005348205566,
      "eval_loss": 0.37998148798942566,
      "eval_mlm_loss": 0.37998148798942566,
      "eval_runtime": 59.2852,
      "eval_samples_per_second": 1178.524,
      "eval_steps_per_second": 0.59,
      "step": 9600
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.41,
      "learning_rate": 0.00048049999999999997,
      "loss": 0.3983,
      "mlm_loss": 0.3983,
      "step": 9610
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.42,
      "learning_rate": 0.000481,
      "loss": 0.4056,
      "mlm_loss": 0.4056,
      "step": 9620
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.43,
      "learning_rate": 0.0004815,
      "loss": 0.4034,
      "mlm_loss": 0.4034,
      "step": 9630
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.44,
      "learning_rate": 0.000482,
      "loss": 0.4071,
      "mlm_loss": 0.4071,
      "step": 9640
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.44,
      "learning_rate": 0.0004825,
      "loss": 0.3994,
      "mlm_loss": 0.3994,
      "step": 9650
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.45,
      "learning_rate": 0.000483,
      "loss": 0.398,
      "mlm_loss": 0.398,
      "step": 9660
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.46,
      "learning_rate": 0.0004835,
      "loss": 0.4038,
      "mlm_loss": 0.4038,
      "step": 9670
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.47,
      "learning_rate": 0.000484,
      "loss": 0.401,
      "mlm_loss": 0.401,
      "step": 9680
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.47,
      "learning_rate": 0.0004845,
      "loss": 0.4049,
      "mlm_loss": 0.4049,
      "step": 9690
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.48,
      "learning_rate": 0.00048499999999999997,
      "loss": 0.4003,
      "mlm_loss": 0.4003,
      "step": 9700
    },
    {
      "epoch": 7.48,
      "eval_ep_loss": -2.468691825866699,
      "eval_loss": 0.37805142998695374,
      "eval_mlm_loss": 0.37805142998695374,
      "eval_runtime": 60.3023,
      "eval_samples_per_second": 1158.646,
      "eval_steps_per_second": 0.58,
      "step": 9700
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.49,
      "learning_rate": 0.0004855,
      "loss": 0.3978,
      "mlm_loss": 0.3978,
      "step": 9710
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.5,
      "learning_rate": 0.000486,
      "loss": 0.3994,
      "mlm_loss": 0.3994,
      "step": 9720
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.5,
      "learning_rate": 0.0004865,
      "loss": 0.4029,
      "mlm_loss": 0.4029,
      "step": 9730
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.51,
      "learning_rate": 0.000487,
      "loss": 0.3991,
      "mlm_loss": 0.3991,
      "step": 9740
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.52,
      "learning_rate": 0.0004875,
      "loss": 0.4061,
      "mlm_loss": 0.4061,
      "step": 9750
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.53,
      "learning_rate": 0.000488,
      "loss": 0.4047,
      "mlm_loss": 0.4047,
      "step": 9760
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.54,
      "learning_rate": 0.0004885,
      "loss": 0.4071,
      "mlm_loss": 0.4071,
      "step": 9770
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.54,
      "learning_rate": 0.000489,
      "loss": 0.4014,
      "mlm_loss": 0.4014,
      "step": 9780
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.55,
      "learning_rate": 0.0004895,
      "loss": 0.4023,
      "mlm_loss": 0.4023,
      "step": 9790
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.56,
      "learning_rate": 0.00049,
      "loss": 0.4052,
      "mlm_loss": 0.4052,
      "step": 9800
    },
    {
      "epoch": 7.56,
      "eval_ep_loss": -2.403273344039917,
      "eval_loss": 0.3772997260093689,
      "eval_mlm_loss": 0.3772997260093689,
      "eval_runtime": 59.2376,
      "eval_samples_per_second": 1179.471,
      "eval_steps_per_second": 0.591,
      "step": 9800
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.57,
      "learning_rate": 0.0004905,
      "loss": 0.394,
      "mlm_loss": 0.394,
      "step": 9810
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.57,
      "learning_rate": 0.000491,
      "loss": 0.4071,
      "mlm_loss": 0.4071,
      "step": 9820
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.58,
      "learning_rate": 0.0004915,
      "loss": 0.3992,
      "mlm_loss": 0.3992,
      "step": 9830
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.59,
      "learning_rate": 0.000492,
      "loss": 0.4024,
      "mlm_loss": 0.4024,
      "step": 9840
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.6,
      "learning_rate": 0.0004925,
      "loss": 0.4042,
      "mlm_loss": 0.4042,
      "step": 9850
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.61,
      "learning_rate": 0.0004930000000000001,
      "loss": 0.4067,
      "mlm_loss": 0.4067,
      "step": 9860
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.61,
      "learning_rate": 0.0004935,
      "loss": 0.3995,
      "mlm_loss": 0.3995,
      "step": 9870
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.62,
      "learning_rate": 0.000494,
      "loss": 0.3998,
      "mlm_loss": 0.3998,
      "step": 9880
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.63,
      "learning_rate": 0.0004945,
      "loss": 0.4011,
      "mlm_loss": 0.4011,
      "step": 9890
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.64,
      "learning_rate": 0.000495,
      "loss": 0.4003,
      "mlm_loss": 0.4003,
      "step": 9900
    },
    {
      "epoch": 7.64,
      "eval_ep_loss": -2.388214111328125,
      "eval_loss": 0.3742392659187317,
      "eval_mlm_loss": 0.3742392659187317,
      "eval_runtime": 60.995,
      "eval_samples_per_second": 1145.488,
      "eval_steps_per_second": 0.574,
      "step": 9900
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.64,
      "learning_rate": 0.0004955,
      "loss": 0.3885,
      "mlm_loss": 0.3885,
      "step": 9910
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.65,
      "learning_rate": 0.000496,
      "loss": 0.3988,
      "mlm_loss": 0.3988,
      "step": 9920
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.66,
      "learning_rate": 0.0004965,
      "loss": 0.4044,
      "mlm_loss": 0.4044,
      "step": 9930
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.67,
      "learning_rate": 0.000497,
      "loss": 0.4002,
      "mlm_loss": 0.4002,
      "step": 9940
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.67,
      "learning_rate": 0.0004975,
      "loss": 0.4078,
      "mlm_loss": 0.4078,
      "step": 9950
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.68,
      "learning_rate": 0.000498,
      "loss": 0.3993,
      "mlm_loss": 0.3993,
      "step": 9960
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.69,
      "learning_rate": 0.0004985,
      "loss": 0.3985,
      "mlm_loss": 0.3985,
      "step": 9970
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.7,
      "learning_rate": 0.0004989500000000001,
      "loss": 0.4052,
      "mlm_loss": 0.4052,
      "step": 9980
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.71,
      "learning_rate": 0.00049945,
      "loss": 0.4045,
      "mlm_loss": 0.4045,
      "step": 9990
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.71,
      "learning_rate": 0.00049995,
      "loss": 0.3959,
      "mlm_loss": 0.3959,
      "step": 10000
    },
    {
      "epoch": 7.71,
      "eval_ep_loss": -2.4986319541931152,
      "eval_loss": 0.3774937689304352,
      "eval_mlm_loss": 0.3774937689304352,
      "eval_runtime": 60.247,
      "eval_samples_per_second": 1159.709,
      "eval_steps_per_second": 0.581,
      "step": 10000
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.72,
      "learning_rate": 0.0004997173366834171,
      "loss": 0.4023,
      "mlm_loss": 0.4023,
      "step": 10010
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.73,
      "learning_rate": 0.0004994032663316583,
      "loss": 0.407,
      "mlm_loss": 0.407,
      "step": 10020
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.74,
      "learning_rate": 0.0004990891959798995,
      "loss": 0.4024,
      "mlm_loss": 0.4024,
      "step": 10030
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.74,
      "learning_rate": 0.0004987751256281407,
      "loss": 0.3991,
      "mlm_loss": 0.3991,
      "step": 10040
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.75,
      "learning_rate": 0.000498461055276382,
      "loss": 0.4058,
      "mlm_loss": 0.4058,
      "step": 10050
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.76,
      "learning_rate": 0.0004981469849246231,
      "loss": 0.4046,
      "mlm_loss": 0.4046,
      "step": 10060
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.77,
      "learning_rate": 0.0004978329145728643,
      "loss": 0.397,
      "mlm_loss": 0.397,
      "step": 10070
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.77,
      "learning_rate": 0.0004975188442211055,
      "loss": 0.3967,
      "mlm_loss": 0.3967,
      "step": 10080
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.78,
      "learning_rate": 0.0004972047738693468,
      "loss": 0.3986,
      "mlm_loss": 0.3986,
      "step": 10090
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.79,
      "learning_rate": 0.000496890703517588,
      "loss": 0.4018,
      "mlm_loss": 0.4018,
      "step": 10100
    },
    {
      "epoch": 7.79,
      "eval_ep_loss": -2.651991605758667,
      "eval_loss": 0.374184250831604,
      "eval_mlm_loss": 0.374184250831604,
      "eval_runtime": 61.0227,
      "eval_samples_per_second": 1144.966,
      "eval_steps_per_second": 0.574,
      "step": 10100
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.8,
      "learning_rate": 0.0004965766331658292,
      "loss": 0.4029,
      "mlm_loss": 0.4029,
      "step": 10110
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.81,
      "learning_rate": 0.0004962625628140703,
      "loss": 0.4006,
      "mlm_loss": 0.4006,
      "step": 10120
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.81,
      "learning_rate": 0.0004959484924623116,
      "loss": 0.4003,
      "mlm_loss": 0.4003,
      "step": 10130
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.82,
      "learning_rate": 0.0004956344221105528,
      "loss": 0.3982,
      "mlm_loss": 0.3982,
      "step": 10140
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.83,
      "learning_rate": 0.000495320351758794,
      "loss": 0.4044,
      "mlm_loss": 0.4044,
      "step": 10150
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.84,
      "learning_rate": 0.0004950062814070351,
      "loss": 0.4023,
      "mlm_loss": 0.4023,
      "step": 10160
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.84,
      "learning_rate": 0.0004946922110552764,
      "loss": 0.4024,
      "mlm_loss": 0.4024,
      "step": 10170
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.85,
      "learning_rate": 0.0004943781407035176,
      "loss": 0.4036,
      "mlm_loss": 0.4036,
      "step": 10180
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.86,
      "learning_rate": 0.0004940640703517588,
      "loss": 0.3962,
      "mlm_loss": 0.3962,
      "step": 10190
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.87,
      "learning_rate": 0.00049375,
      "loss": 0.4061,
      "mlm_loss": 0.4061,
      "step": 10200
    },
    {
      "epoch": 7.87,
      "eval_ep_loss": -2.436077356338501,
      "eval_loss": 0.3732386827468872,
      "eval_mlm_loss": 0.3732386827468872,
      "eval_runtime": 59.8162,
      "eval_samples_per_second": 1168.062,
      "eval_steps_per_second": 0.585,
      "step": 10200
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.88,
      "learning_rate": 0.0004934359296482412,
      "loss": 0.3959,
      "mlm_loss": 0.3959,
      "step": 10210
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.88,
      "learning_rate": 0.0004931218592964824,
      "loss": 0.3997,
      "mlm_loss": 0.3997,
      "step": 10220
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.89,
      "learning_rate": 0.0004928077889447236,
      "loss": 0.4013,
      "mlm_loss": 0.4013,
      "step": 10230
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.9,
      "learning_rate": 0.0004924937185929649,
      "loss": 0.3956,
      "mlm_loss": 0.3956,
      "step": 10240
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.91,
      "learning_rate": 0.0004921796482412061,
      "loss": 0.3967,
      "mlm_loss": 0.3967,
      "step": 10250
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.91,
      "learning_rate": 0.0004918655778894472,
      "loss": 0.3967,
      "mlm_loss": 0.3967,
      "step": 10260
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.92,
      "learning_rate": 0.0004915515075376884,
      "loss": 0.3954,
      "mlm_loss": 0.3954,
      "step": 10270
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.93,
      "learning_rate": 0.0004912374371859297,
      "loss": 0.3976,
      "mlm_loss": 0.3976,
      "step": 10280
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.94,
      "learning_rate": 0.0004909233668341709,
      "loss": 0.3945,
      "mlm_loss": 0.3945,
      "step": 10290
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.94,
      "learning_rate": 0.0004906092964824121,
      "loss": 0.3979,
      "mlm_loss": 0.3979,
      "step": 10300
    },
    {
      "epoch": 7.94,
      "eval_ep_loss": -2.4026572704315186,
      "eval_loss": 0.37325599789619446,
      "eval_mlm_loss": 0.37325599789619446,
      "eval_runtime": 61.5586,
      "eval_samples_per_second": 1135.0,
      "eval_steps_per_second": 0.569,
      "step": 10300
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.95,
      "learning_rate": 0.0004902952261306532,
      "loss": 0.3971,
      "mlm_loss": 0.3971,
      "step": 10310
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.96,
      "learning_rate": 0.0004899811557788945,
      "loss": 0.4018,
      "mlm_loss": 0.4018,
      "step": 10320
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.97,
      "learning_rate": 0.0004896670854271357,
      "loss": 0.3993,
      "mlm_loss": 0.3993,
      "step": 10330
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.98,
      "learning_rate": 0.0004893530150753769,
      "loss": 0.3995,
      "mlm_loss": 0.3995,
      "step": 10340
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.98,
      "learning_rate": 0.0004890389447236181,
      "loss": 0.3948,
      "mlm_loss": 0.3948,
      "step": 10350
    },
    {
      "ep_loss": 0.0,
      "epoch": 7.99,
      "learning_rate": 0.0004887248743718593,
      "loss": 0.4028,
      "mlm_loss": 0.4028,
      "step": 10360
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.0,
      "learning_rate": 0.0004884108040201005,
      "loss": 0.3972,
      "mlm_loss": 0.3972,
      "step": 10370
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.01,
      "learning_rate": 0.0004880967336683417,
      "loss": 0.3899,
      "mlm_loss": 0.3899,
      "step": 10380
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.01,
      "learning_rate": 0.00048778266331658295,
      "loss": 0.3993,
      "mlm_loss": 0.3993,
      "step": 10390
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.02,
      "learning_rate": 0.0004874685929648242,
      "loss": 0.3945,
      "mlm_loss": 0.3945,
      "step": 10400
    },
    {
      "epoch": 8.02,
      "eval_ep_loss": -2.2991716861724854,
      "eval_loss": 0.3698738217353821,
      "eval_mlm_loss": 0.3698738217353821,
      "eval_runtime": 62.0944,
      "eval_samples_per_second": 1125.206,
      "eval_steps_per_second": 0.564,
      "step": 10400
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.03,
      "learning_rate": 0.0004871545226130653,
      "loss": 0.3855,
      "mlm_loss": 0.3855,
      "step": 10410
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.04,
      "learning_rate": 0.0004868404522613065,
      "loss": 0.3925,
      "mlm_loss": 0.3925,
      "step": 10420
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.04,
      "learning_rate": 0.00048652638190954775,
      "loss": 0.3901,
      "mlm_loss": 0.3901,
      "step": 10430
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.05,
      "learning_rate": 0.0004862123115577889,
      "loss": 0.39,
      "mlm_loss": 0.39,
      "step": 10440
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.06,
      "learning_rate": 0.00048589824120603015,
      "loss": 0.3904,
      "mlm_loss": 0.3904,
      "step": 10450
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.07,
      "learning_rate": 0.0004855841708542714,
      "loss": 0.3908,
      "mlm_loss": 0.3908,
      "step": 10460
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.08,
      "learning_rate": 0.00048527010050251256,
      "loss": 0.3931,
      "mlm_loss": 0.3931,
      "step": 10470
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.08,
      "learning_rate": 0.0004849560301507538,
      "loss": 0.3899,
      "mlm_loss": 0.3899,
      "step": 10480
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.09,
      "learning_rate": 0.000484641959798995,
      "loss": 0.3908,
      "mlm_loss": 0.3908,
      "step": 10490
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.1,
      "learning_rate": 0.00048432788944723624,
      "loss": 0.3909,
      "mlm_loss": 0.3909,
      "step": 10500
    },
    {
      "epoch": 8.1,
      "eval_ep_loss": -2.246208429336548,
      "eval_loss": 0.36925286054611206,
      "eval_mlm_loss": 0.36925286054611206,
      "eval_runtime": 61.3519,
      "eval_samples_per_second": 1138.823,
      "eval_steps_per_second": 0.57,
      "step": 10500
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.11,
      "learning_rate": 0.00048401381909547736,
      "loss": 0.3922,
      "mlm_loss": 0.3922,
      "step": 10510
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.11,
      "learning_rate": 0.0004836997487437186,
      "loss": 0.3878,
      "mlm_loss": 0.3878,
      "step": 10520
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.12,
      "learning_rate": 0.0004833856783919598,
      "loss": 0.3901,
      "mlm_loss": 0.3901,
      "step": 10530
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.13,
      "learning_rate": 0.000483071608040201,
      "loss": 0.3872,
      "mlm_loss": 0.3872,
      "step": 10540
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.14,
      "learning_rate": 0.0004827575376884422,
      "loss": 0.3931,
      "mlm_loss": 0.3931,
      "step": 10550
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.15,
      "learning_rate": 0.00048244346733668345,
      "loss": 0.3928,
      "mlm_loss": 0.3928,
      "step": 10560
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.15,
      "learning_rate": 0.0004821293969849246,
      "loss": 0.3975,
      "mlm_loss": 0.3975,
      "step": 10570
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.16,
      "learning_rate": 0.00048181532663316585,
      "loss": 0.4002,
      "mlm_loss": 0.4002,
      "step": 10580
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.17,
      "learning_rate": 0.0004815012562814071,
      "loss": 0.3961,
      "mlm_loss": 0.3961,
      "step": 10590
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.18,
      "learning_rate": 0.00048118718592964825,
      "loss": 0.392,
      "mlm_loss": 0.392,
      "step": 10600
    },
    {
      "epoch": 8.18,
      "eval_ep_loss": -2.7679216861724854,
      "eval_loss": 0.36933663487434387,
      "eval_mlm_loss": 0.36933663487434387,
      "eval_runtime": 61.879,
      "eval_samples_per_second": 1129.123,
      "eval_steps_per_second": 0.566,
      "step": 10600
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.18,
      "learning_rate": 0.0004808731155778894,
      "loss": 0.3957,
      "mlm_loss": 0.3957,
      "step": 10610
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.19,
      "learning_rate": 0.00048055904522613065,
      "loss": 0.4001,
      "mlm_loss": 0.4001,
      "step": 10620
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.2,
      "learning_rate": 0.0004802449748743719,
      "loss": 0.398,
      "mlm_loss": 0.398,
      "step": 10630
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.21,
      "learning_rate": 0.00047993090452261305,
      "loss": 0.3897,
      "mlm_loss": 0.3897,
      "step": 10640
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.21,
      "learning_rate": 0.0004796168341708543,
      "loss": 0.3949,
      "mlm_loss": 0.3949,
      "step": 10650
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.22,
      "learning_rate": 0.0004793027638190955,
      "loss": 0.3844,
      "mlm_loss": 0.3844,
      "step": 10660
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.23,
      "learning_rate": 0.0004789886934673367,
      "loss": 0.3931,
      "mlm_loss": 0.3931,
      "step": 10670
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.24,
      "learning_rate": 0.0004786746231155779,
      "loss": 0.3892,
      "mlm_loss": 0.3892,
      "step": 10680
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.25,
      "learning_rate": 0.00047836055276381914,
      "loss": 0.3946,
      "mlm_loss": 0.3946,
      "step": 10690
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.25,
      "learning_rate": 0.0004780464824120603,
      "loss": 0.393,
      "mlm_loss": 0.393,
      "step": 10700
    },
    {
      "epoch": 8.25,
      "eval_ep_loss": -2.653468608856201,
      "eval_loss": 0.36575964093208313,
      "eval_mlm_loss": 0.36575964093208313,
      "eval_runtime": 61.0152,
      "eval_samples_per_second": 1145.108,
      "eval_steps_per_second": 0.574,
      "step": 10700
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.26,
      "learning_rate": 0.0004777324120603015,
      "loss": 0.3865,
      "mlm_loss": 0.3865,
      "step": 10710
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.27,
      "learning_rate": 0.0004774183417085427,
      "loss": 0.3936,
      "mlm_loss": 0.3936,
      "step": 10720
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.28,
      "learning_rate": 0.0004771042713567839,
      "loss": 0.3874,
      "mlm_loss": 0.3874,
      "step": 10730
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.28,
      "learning_rate": 0.0004767902010050251,
      "loss": 0.3839,
      "mlm_loss": 0.3839,
      "step": 10740
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.29,
      "learning_rate": 0.00047647613065326635,
      "loss": 0.3819,
      "mlm_loss": 0.3819,
      "step": 10750
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.3,
      "learning_rate": 0.0004761620603015076,
      "loss": 0.3841,
      "mlm_loss": 0.3841,
      "step": 10760
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.31,
      "learning_rate": 0.00047584798994974875,
      "loss": 0.3896,
      "mlm_loss": 0.3896,
      "step": 10770
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.31,
      "learning_rate": 0.00047553391959799,
      "loss": 0.388,
      "mlm_loss": 0.388,
      "step": 10780
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.32,
      "learning_rate": 0.0004752198492462312,
      "loss": 0.3864,
      "mlm_loss": 0.3864,
      "step": 10790
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.33,
      "learning_rate": 0.0004749057788944724,
      "loss": 0.3887,
      "mlm_loss": 0.3887,
      "step": 10800
    },
    {
      "epoch": 8.33,
      "eval_ep_loss": -2.6650192737579346,
      "eval_loss": 0.362870454788208,
      "eval_mlm_loss": 0.362870454788208,
      "eval_runtime": 59.7372,
      "eval_samples_per_second": 1169.606,
      "eval_steps_per_second": 0.586,
      "step": 10800
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.34,
      "learning_rate": 0.00047459170854271355,
      "loss": 0.3898,
      "mlm_loss": 0.3898,
      "step": 10810
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.35,
      "learning_rate": 0.0004742776381909548,
      "loss": 0.3881,
      "mlm_loss": 0.3881,
      "step": 10820
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.35,
      "learning_rate": 0.00047396356783919595,
      "loss": 0.3868,
      "mlm_loss": 0.3868,
      "step": 10830
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.36,
      "learning_rate": 0.0004736494974874372,
      "loss": 0.3924,
      "mlm_loss": 0.3924,
      "step": 10840
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.37,
      "learning_rate": 0.0004733354271356784,
      "loss": 0.3849,
      "mlm_loss": 0.3849,
      "step": 10850
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.38,
      "learning_rate": 0.0004730213567839196,
      "loss": 0.387,
      "mlm_loss": 0.387,
      "step": 10860
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.38,
      "learning_rate": 0.0004727072864321608,
      "loss": 0.3829,
      "mlm_loss": 0.3829,
      "step": 10870
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.39,
      "learning_rate": 0.00047239321608040204,
      "loss": 0.3881,
      "mlm_loss": 0.3881,
      "step": 10880
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.4,
      "learning_rate": 0.00047207914572864327,
      "loss": 0.3957,
      "mlm_loss": 0.3957,
      "step": 10890
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.41,
      "learning_rate": 0.00047176507537688444,
      "loss": 0.3856,
      "mlm_loss": 0.3856,
      "step": 10900
    },
    {
      "epoch": 8.41,
      "eval_ep_loss": -2.556627035140991,
      "eval_loss": 0.3644386827945709,
      "eval_mlm_loss": 0.3644386827945709,
      "eval_runtime": 60.0106,
      "eval_samples_per_second": 1164.279,
      "eval_steps_per_second": 0.583,
      "step": 10900
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.41,
      "learning_rate": 0.0004714510050251256,
      "loss": 0.3918,
      "mlm_loss": 0.3918,
      "step": 10910
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.42,
      "learning_rate": 0.00047113693467336684,
      "loss": 0.3888,
      "mlm_loss": 0.3888,
      "step": 10920
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.43,
      "learning_rate": 0.000470822864321608,
      "loss": 0.3874,
      "mlm_loss": 0.3874,
      "step": 10930
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.44,
      "learning_rate": 0.00047050879396984925,
      "loss": 0.3826,
      "mlm_loss": 0.3826,
      "step": 10940
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.45,
      "learning_rate": 0.0004701947236180905,
      "loss": 0.3817,
      "mlm_loss": 0.3817,
      "step": 10950
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.45,
      "learning_rate": 0.00046988065326633165,
      "loss": 0.3877,
      "mlm_loss": 0.3877,
      "step": 10960
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.46,
      "learning_rate": 0.0004695665829145729,
      "loss": 0.3841,
      "mlm_loss": 0.3841,
      "step": 10970
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.47,
      "learning_rate": 0.0004692525125628141,
      "loss": 0.3854,
      "mlm_loss": 0.3854,
      "step": 10980
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.48,
      "learning_rate": 0.0004689384422110553,
      "loss": 0.3818,
      "mlm_loss": 0.3818,
      "step": 10990
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.48,
      "learning_rate": 0.0004686243718592965,
      "loss": 0.3844,
      "mlm_loss": 0.3844,
      "step": 11000
    },
    {
      "epoch": 8.48,
      "eval_ep_loss": -2.669065475463867,
      "eval_loss": 0.36087533831596375,
      "eval_mlm_loss": 0.36087533831596375,
      "eval_runtime": 59.8288,
      "eval_samples_per_second": 1167.815,
      "eval_steps_per_second": 0.585,
      "step": 11000
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.49,
      "learning_rate": 0.0004683103015075377,
      "loss": 0.3867,
      "mlm_loss": 0.3867,
      "step": 11010
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.5,
      "learning_rate": 0.0004679962311557789,
      "loss": 0.383,
      "mlm_loss": 0.383,
      "step": 11020
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.51,
      "learning_rate": 0.0004676821608040201,
      "loss": 0.3897,
      "mlm_loss": 0.3897,
      "step": 11030
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.52,
      "learning_rate": 0.0004673680904522613,
      "loss": 0.3908,
      "mlm_loss": 0.3908,
      "step": 11040
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.52,
      "learning_rate": 0.00046705402010050254,
      "loss": 0.3854,
      "mlm_loss": 0.3854,
      "step": 11050
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.53,
      "learning_rate": 0.0004667399497487437,
      "loss": 0.3886,
      "mlm_loss": 0.3886,
      "step": 11060
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.54,
      "learning_rate": 0.00046642587939698494,
      "loss": 0.3814,
      "mlm_loss": 0.3814,
      "step": 11070
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.55,
      "learning_rate": 0.00046611180904522617,
      "loss": 0.3888,
      "mlm_loss": 0.3888,
      "step": 11080
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.55,
      "learning_rate": 0.00046579773869346734,
      "loss": 0.3884,
      "mlm_loss": 0.3884,
      "step": 11090
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.56,
      "learning_rate": 0.00046548366834170857,
      "loss": 0.39,
      "mlm_loss": 0.39,
      "step": 11100
    },
    {
      "epoch": 8.56,
      "eval_ep_loss": -2.3106160163879395,
      "eval_loss": 0.3573386073112488,
      "eval_mlm_loss": 0.3573386073112488,
      "eval_runtime": 61.2527,
      "eval_samples_per_second": 1140.668,
      "eval_steps_per_second": 0.571,
      "step": 11100
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.57,
      "learning_rate": 0.00046516959798994974,
      "loss": 0.3767,
      "mlm_loss": 0.3767,
      "step": 11110
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.58,
      "learning_rate": 0.0004648555276381909,
      "loss": 0.3712,
      "mlm_loss": 0.3712,
      "step": 11120
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.58,
      "learning_rate": 0.00046454145728643215,
      "loss": 0.3787,
      "mlm_loss": 0.3787,
      "step": 11130
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.59,
      "learning_rate": 0.0004642273869346734,
      "loss": 0.3815,
      "mlm_loss": 0.3815,
      "step": 11140
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.6,
      "learning_rate": 0.0004639133165829146,
      "loss": 0.3837,
      "mlm_loss": 0.3837,
      "step": 11150
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.61,
      "learning_rate": 0.0004635992462311558,
      "loss": 0.3851,
      "mlm_loss": 0.3851,
      "step": 11160
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.62,
      "learning_rate": 0.000463285175879397,
      "loss": 0.3818,
      "mlm_loss": 0.3818,
      "step": 11170
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.62,
      "learning_rate": 0.00046297110552763823,
      "loss": 0.3841,
      "mlm_loss": 0.3841,
      "step": 11180
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.63,
      "learning_rate": 0.0004626570351758794,
      "loss": 0.3803,
      "mlm_loss": 0.3803,
      "step": 11190
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.64,
      "learning_rate": 0.00046234296482412063,
      "loss": 0.3799,
      "mlm_loss": 0.3799,
      "step": 11200
    },
    {
      "epoch": 8.64,
      "eval_ep_loss": -2.7373206615448,
      "eval_loss": 0.35678714513778687,
      "eval_mlm_loss": 0.35678714513778687,
      "eval_runtime": 63.4192,
      "eval_samples_per_second": 1101.701,
      "eval_steps_per_second": 0.552,
      "step": 11200
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.65,
      "learning_rate": 0.0004620288944723618,
      "loss": 0.3788,
      "mlm_loss": 0.3788,
      "step": 11210
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.65,
      "learning_rate": 0.000461714824120603,
      "loss": 0.385,
      "mlm_loss": 0.385,
      "step": 11220
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.66,
      "learning_rate": 0.0004614007537688442,
      "loss": 0.3886,
      "mlm_loss": 0.3886,
      "step": 11230
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.67,
      "learning_rate": 0.00046108668341708544,
      "loss": 0.3816,
      "mlm_loss": 0.3816,
      "step": 11240
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.68,
      "learning_rate": 0.00046077261306532667,
      "loss": 0.3913,
      "mlm_loss": 0.3913,
      "step": 11250
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.68,
      "learning_rate": 0.00046045854271356784,
      "loss": 0.3816,
      "mlm_loss": 0.3816,
      "step": 11260
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.69,
      "learning_rate": 0.00046014447236180907,
      "loss": 0.3809,
      "mlm_loss": 0.3809,
      "step": 11270
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.7,
      "learning_rate": 0.0004598304020100503,
      "loss": 0.3901,
      "mlm_loss": 0.3901,
      "step": 11280
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.71,
      "learning_rate": 0.00045951633165829147,
      "loss": 0.3791,
      "mlm_loss": 0.3791,
      "step": 11290
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.72,
      "learning_rate": 0.0004592022613065327,
      "loss": 0.3778,
      "mlm_loss": 0.3778,
      "step": 11300
    },
    {
      "epoch": 8.72,
      "eval_ep_loss": -2.907904624938965,
      "eval_loss": 0.3548813462257385,
      "eval_mlm_loss": 0.3548813462257385,
      "eval_runtime": 63.2595,
      "eval_samples_per_second": 1104.482,
      "eval_steps_per_second": 0.553,
      "step": 11300
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.72,
      "learning_rate": 0.00045888819095477387,
      "loss": 0.3802,
      "mlm_loss": 0.3802,
      "step": 11310
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.73,
      "learning_rate": 0.00045857412060301505,
      "loss": 0.3811,
      "mlm_loss": 0.3811,
      "step": 11320
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.74,
      "learning_rate": 0.0004582600502512563,
      "loss": 0.385,
      "mlm_loss": 0.385,
      "step": 11330
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.75,
      "learning_rate": 0.0004579459798994975,
      "loss": 0.3828,
      "mlm_loss": 0.3828,
      "step": 11340
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.75,
      "learning_rate": 0.0004576319095477387,
      "loss": 0.3789,
      "mlm_loss": 0.3789,
      "step": 11350
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.76,
      "learning_rate": 0.0004573178391959799,
      "loss": 0.3802,
      "mlm_loss": 0.3802,
      "step": 11360
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.77,
      "learning_rate": 0.00045700376884422113,
      "loss": 0.38,
      "mlm_loss": 0.38,
      "step": 11370
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.78,
      "learning_rate": 0.00045668969849246236,
      "loss": 0.3842,
      "mlm_loss": 0.3842,
      "step": 11380
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.79,
      "learning_rate": 0.00045637562814070353,
      "loss": 0.3843,
      "mlm_loss": 0.3843,
      "step": 11390
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.79,
      "learning_rate": 0.00045606155778894476,
      "loss": 0.3811,
      "mlm_loss": 0.3811,
      "step": 11400
    },
    {
      "epoch": 8.79,
      "eval_ep_loss": -2.4214723110198975,
      "eval_loss": 0.35386431217193604,
      "eval_mlm_loss": 0.35386431217193604,
      "eval_runtime": 59.4503,
      "eval_samples_per_second": 1175.25,
      "eval_steps_per_second": 0.589,
      "step": 11400
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.8,
      "learning_rate": 0.00045574748743718594,
      "loss": 0.3777,
      "mlm_loss": 0.3777,
      "step": 11410
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.81,
      "learning_rate": 0.0004554334170854271,
      "loss": 0.3857,
      "mlm_loss": 0.3857,
      "step": 11420
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.82,
      "learning_rate": 0.00045511934673366834,
      "loss": 0.3818,
      "mlm_loss": 0.3818,
      "step": 11430
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.82,
      "learning_rate": 0.00045480527638190957,
      "loss": 0.3794,
      "mlm_loss": 0.3794,
      "step": 11440
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.83,
      "learning_rate": 0.00045449120603015074,
      "loss": 0.3783,
      "mlm_loss": 0.3783,
      "step": 11450
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.84,
      "learning_rate": 0.00045417713567839197,
      "loss": 0.3826,
      "mlm_loss": 0.3826,
      "step": 11460
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.85,
      "learning_rate": 0.0004538630653266332,
      "loss": 0.3786,
      "mlm_loss": 0.3786,
      "step": 11470
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.85,
      "learning_rate": 0.00045354899497487437,
      "loss": 0.3731,
      "mlm_loss": 0.3731,
      "step": 11480
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.86,
      "learning_rate": 0.0004532349246231156,
      "loss": 0.3766,
      "mlm_loss": 0.3766,
      "step": 11490
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.87,
      "learning_rate": 0.0004529208542713568,
      "loss": 0.386,
      "mlm_loss": 0.386,
      "step": 11500
    },
    {
      "epoch": 8.87,
      "eval_ep_loss": -2.3580639362335205,
      "eval_loss": 0.3554854094982147,
      "eval_mlm_loss": 0.3554854094982147,
      "eval_runtime": 59.7228,
      "eval_samples_per_second": 1169.889,
      "eval_steps_per_second": 0.586,
      "step": 11500
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.88,
      "learning_rate": 0.000452606783919598,
      "loss": 0.3765,
      "mlm_loss": 0.3765,
      "step": 11510
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.89,
      "learning_rate": 0.0004522927135678392,
      "loss": 0.375,
      "mlm_loss": 0.375,
      "step": 11520
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.89,
      "learning_rate": 0.0004519786432160804,
      "loss": 0.3831,
      "mlm_loss": 0.3831,
      "step": 11530
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.9,
      "learning_rate": 0.00045166457286432163,
      "loss": 0.3797,
      "mlm_loss": 0.3797,
      "step": 11540
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.91,
      "learning_rate": 0.0004513505025125628,
      "loss": 0.3729,
      "mlm_loss": 0.3729,
      "step": 11550
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.92,
      "learning_rate": 0.00045103643216080403,
      "loss": 0.3736,
      "mlm_loss": 0.3736,
      "step": 11560
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.92,
      "learning_rate": 0.00045072236180904526,
      "loss": 0.3756,
      "mlm_loss": 0.3756,
      "step": 11570
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.93,
      "learning_rate": 0.00045040829145728643,
      "loss": 0.3791,
      "mlm_loss": 0.3791,
      "step": 11580
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.94,
      "learning_rate": 0.00045009422110552766,
      "loss": 0.3701,
      "mlm_loss": 0.3701,
      "step": 11590
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.95,
      "learning_rate": 0.0004497801507537689,
      "loss": 0.374,
      "mlm_loss": 0.374,
      "step": 11600
    },
    {
      "epoch": 8.95,
      "eval_ep_loss": -2.2494945526123047,
      "eval_loss": 0.35051652789115906,
      "eval_mlm_loss": 0.35051652789115906,
      "eval_runtime": 60.6153,
      "eval_samples_per_second": 1152.663,
      "eval_steps_per_second": 0.577,
      "step": 11600
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.95,
      "learning_rate": 0.00044946608040201,
      "loss": 0.3771,
      "mlm_loss": 0.3771,
      "step": 11610
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.96,
      "learning_rate": 0.00044915201005025124,
      "loss": 0.3787,
      "mlm_loss": 0.3787,
      "step": 11620
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.97,
      "learning_rate": 0.00044883793969849247,
      "loss": 0.3783,
      "mlm_loss": 0.3783,
      "step": 11630
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.98,
      "learning_rate": 0.0004485238693467337,
      "loss": 0.3683,
      "mlm_loss": 0.3683,
      "step": 11640
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.99,
      "learning_rate": 0.00044820979899497487,
      "loss": 0.3755,
      "mlm_loss": 0.3755,
      "step": 11650
    },
    {
      "ep_loss": 0.0,
      "epoch": 8.99,
      "learning_rate": 0.0004478957286432161,
      "loss": 0.3751,
      "mlm_loss": 0.3751,
      "step": 11660
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.0,
      "learning_rate": 0.0004475816582914573,
      "loss": 0.3806,
      "mlm_loss": 0.3806,
      "step": 11670
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.01,
      "learning_rate": 0.0004472675879396985,
      "loss": 0.3723,
      "mlm_loss": 0.3723,
      "step": 11680
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.02,
      "learning_rate": 0.0004469535175879397,
      "loss": 0.3727,
      "mlm_loss": 0.3727,
      "step": 11690
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.02,
      "learning_rate": 0.00044663944723618095,
      "loss": 0.3717,
      "mlm_loss": 0.3717,
      "step": 11700
    },
    {
      "epoch": 9.02,
      "eval_ep_loss": -2.543911933898926,
      "eval_loss": 0.34973713755607605,
      "eval_mlm_loss": 0.34973713755607605,
      "eval_runtime": 62.1216,
      "eval_samples_per_second": 1124.713,
      "eval_steps_per_second": 0.563,
      "step": 11700
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.03,
      "learning_rate": 0.0004463253768844221,
      "loss": 0.3694,
      "mlm_loss": 0.3694,
      "step": 11710
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.04,
      "learning_rate": 0.0004460113065326633,
      "loss": 0.3673,
      "mlm_loss": 0.3673,
      "step": 11720
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.05,
      "learning_rate": 0.00044569723618090453,
      "loss": 0.3679,
      "mlm_loss": 0.3679,
      "step": 11730
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.06,
      "learning_rate": 0.0004453831658291457,
      "loss": 0.3757,
      "mlm_loss": 0.3757,
      "step": 11740
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.06,
      "learning_rate": 0.00044506909547738693,
      "loss": 0.3702,
      "mlm_loss": 0.3702,
      "step": 11750
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.07,
      "learning_rate": 0.00044475502512562816,
      "loss": 0.3697,
      "mlm_loss": 0.3697,
      "step": 11760
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.08,
      "learning_rate": 0.0004444409547738694,
      "loss": 0.3721,
      "mlm_loss": 0.3721,
      "step": 11770
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.09,
      "learning_rate": 0.00044412688442211056,
      "loss": 0.3713,
      "mlm_loss": 0.3713,
      "step": 11780
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.09,
      "learning_rate": 0.0004438128140703518,
      "loss": 0.3701,
      "mlm_loss": 0.3701,
      "step": 11790
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.1,
      "learning_rate": 0.000443498743718593,
      "loss": 0.368,
      "mlm_loss": 0.368,
      "step": 11800
    },
    {
      "epoch": 9.1,
      "eval_ep_loss": -2.5209760665893555,
      "eval_loss": 0.349372535943985,
      "eval_mlm_loss": 0.349372535943985,
      "eval_runtime": 58.6125,
      "eval_samples_per_second": 1192.049,
      "eval_steps_per_second": 0.597,
      "step": 11800
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.11,
      "learning_rate": 0.00044318467336683414,
      "loss": 0.3664,
      "mlm_loss": 0.3664,
      "step": 11810
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.12,
      "learning_rate": 0.00044287060301507537,
      "loss": 0.3766,
      "mlm_loss": 0.3766,
      "step": 11820
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.12,
      "learning_rate": 0.0004425565326633166,
      "loss": 0.3677,
      "mlm_loss": 0.3677,
      "step": 11830
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.13,
      "learning_rate": 0.00044224246231155777,
      "loss": 0.3701,
      "mlm_loss": 0.3701,
      "step": 11840
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.14,
      "learning_rate": 0.000441928391959799,
      "loss": 0.3789,
      "mlm_loss": 0.3789,
      "step": 11850
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.15,
      "learning_rate": 0.0004416143216080402,
      "loss": 0.3716,
      "mlm_loss": 0.3716,
      "step": 11860
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.16,
      "learning_rate": 0.00044130025125628145,
      "loss": 0.3672,
      "mlm_loss": 0.3672,
      "step": 11870
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.16,
      "learning_rate": 0.0004409861809045226,
      "loss": 0.369,
      "mlm_loss": 0.369,
      "step": 11880
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.17,
      "learning_rate": 0.00044067211055276385,
      "loss": 0.377,
      "mlm_loss": 0.377,
      "step": 11890
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.18,
      "learning_rate": 0.0004403580402010051,
      "loss": 0.3692,
      "mlm_loss": 0.3692,
      "step": 11900
    },
    {
      "epoch": 9.18,
      "eval_ep_loss": -2.5095667839050293,
      "eval_loss": 0.3459388017654419,
      "eval_mlm_loss": 0.3459388017654419,
      "eval_runtime": 59.1505,
      "eval_samples_per_second": 1181.208,
      "eval_steps_per_second": 0.592,
      "step": 11900
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.19,
      "learning_rate": 0.0004400439698492462,
      "loss": 0.3717,
      "mlm_loss": 0.3717,
      "step": 11910
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.19,
      "learning_rate": 0.00043972989949748743,
      "loss": 0.3706,
      "mlm_loss": 0.3706,
      "step": 11920
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.2,
      "learning_rate": 0.00043941582914572866,
      "loss": 0.3766,
      "mlm_loss": 0.3766,
      "step": 11930
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.21,
      "learning_rate": 0.00043910175879396983,
      "loss": 0.3675,
      "mlm_loss": 0.3675,
      "step": 11940
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.22,
      "learning_rate": 0.00043878768844221106,
      "loss": 0.3653,
      "mlm_loss": 0.3653,
      "step": 11950
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.22,
      "learning_rate": 0.0004384736180904523,
      "loss": 0.3683,
      "mlm_loss": 0.3683,
      "step": 11960
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.23,
      "learning_rate": 0.00043815954773869346,
      "loss": 0.3695,
      "mlm_loss": 0.3695,
      "step": 11970
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.24,
      "learning_rate": 0.0004378454773869347,
      "loss": 0.3674,
      "mlm_loss": 0.3674,
      "step": 11980
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.25,
      "learning_rate": 0.0004375314070351759,
      "loss": 0.3621,
      "mlm_loss": 0.3621,
      "step": 11990
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.26,
      "learning_rate": 0.00043721733668341715,
      "loss": 0.3714,
      "mlm_loss": 0.3714,
      "step": 12000
    },
    {
      "epoch": 9.26,
      "eval_ep_loss": -2.6775925159454346,
      "eval_loss": 0.3442252576351166,
      "eval_mlm_loss": 0.3442252576351166,
      "eval_runtime": 60.7149,
      "eval_samples_per_second": 1150.773,
      "eval_steps_per_second": 0.576,
      "step": 12000
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.26,
      "learning_rate": 0.00043690326633165827,
      "loss": 0.3706,
      "mlm_loss": 0.3706,
      "step": 12010
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.27,
      "learning_rate": 0.0004365891959798995,
      "loss": 0.3642,
      "mlm_loss": 0.3642,
      "step": 12020
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.28,
      "learning_rate": 0.0004362751256281407,
      "loss": 0.3663,
      "mlm_loss": 0.3663,
      "step": 12030
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.29,
      "learning_rate": 0.0004359610552763819,
      "loss": 0.3743,
      "mlm_loss": 0.3743,
      "step": 12040
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.29,
      "learning_rate": 0.0004356469849246231,
      "loss": 0.3701,
      "mlm_loss": 0.3701,
      "step": 12050
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.3,
      "learning_rate": 0.00043533291457286435,
      "loss": 0.3713,
      "mlm_loss": 0.3713,
      "step": 12060
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.31,
      "learning_rate": 0.0004350188442211055,
      "loss": 0.3684,
      "mlm_loss": 0.3684,
      "step": 12070
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.32,
      "learning_rate": 0.00043470477386934675,
      "loss": 0.3603,
      "mlm_loss": 0.3603,
      "step": 12080
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.33,
      "learning_rate": 0.000434390703517588,
      "loss": 0.3599,
      "mlm_loss": 0.3599,
      "step": 12090
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.33,
      "learning_rate": 0.00043407663316582916,
      "loss": 0.3653,
      "mlm_loss": 0.3653,
      "step": 12100
    },
    {
      "epoch": 9.33,
      "eval_ep_loss": -2.689305543899536,
      "eval_loss": 0.34572362899780273,
      "eval_mlm_loss": 0.34572362899780273,
      "eval_runtime": 60.5622,
      "eval_samples_per_second": 1153.673,
      "eval_steps_per_second": 0.578,
      "step": 12100
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.34,
      "learning_rate": 0.00043376256281407033,
      "loss": 0.3746,
      "mlm_loss": 0.3746,
      "step": 12110
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.35,
      "learning_rate": 0.00043344849246231156,
      "loss": 0.3707,
      "mlm_loss": 0.3707,
      "step": 12120
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.36,
      "learning_rate": 0.0004331344221105528,
      "loss": 0.375,
      "mlm_loss": 0.375,
      "step": 12130
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.36,
      "learning_rate": 0.00043282035175879396,
      "loss": 0.3675,
      "mlm_loss": 0.3675,
      "step": 12140
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.37,
      "learning_rate": 0.0004325062814070352,
      "loss": 0.3683,
      "mlm_loss": 0.3683,
      "step": 12150
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.38,
      "learning_rate": 0.0004321922110552764,
      "loss": 0.3646,
      "mlm_loss": 0.3646,
      "step": 12160
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.39,
      "learning_rate": 0.0004318781407035176,
      "loss": 0.37,
      "mlm_loss": 0.37,
      "step": 12170
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.39,
      "learning_rate": 0.0004315640703517588,
      "loss": 0.3701,
      "mlm_loss": 0.3701,
      "step": 12180
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.4,
      "learning_rate": 0.00043125000000000005,
      "loss": 0.364,
      "mlm_loss": 0.364,
      "step": 12190
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.41,
      "learning_rate": 0.00043093592964824117,
      "loss": 0.3639,
      "mlm_loss": 0.3639,
      "step": 12200
    },
    {
      "epoch": 9.41,
      "eval_ep_loss": -2.2502214908599854,
      "eval_loss": 0.3424089848995209,
      "eval_mlm_loss": 0.3424089848995209,
      "eval_runtime": 60.2107,
      "eval_samples_per_second": 1160.408,
      "eval_steps_per_second": 0.581,
      "step": 12200
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.42,
      "learning_rate": 0.0004306218592964824,
      "loss": 0.3668,
      "mlm_loss": 0.3668,
      "step": 12210
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.43,
      "learning_rate": 0.0004303077889447236,
      "loss": 0.3643,
      "mlm_loss": 0.3643,
      "step": 12220
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.43,
      "learning_rate": 0.0004299937185929648,
      "loss": 0.3682,
      "mlm_loss": 0.3682,
      "step": 12230
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.44,
      "learning_rate": 0.00042971105527638193,
      "loss": 0.3682,
      "mlm_loss": 0.3682,
      "step": 12240
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.45,
      "learning_rate": 0.00042939698492462316,
      "loss": 0.3663,
      "mlm_loss": 0.3663,
      "step": 12250
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.46,
      "learning_rate": 0.0004291143216080402,
      "loss": 0.3633,
      "mlm_loss": 0.3633,
      "step": 12260
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.46,
      "learning_rate": 0.0004288002512562814,
      "loss": 0.3657,
      "mlm_loss": 0.3657,
      "step": 12270
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.47,
      "learning_rate": 0.00042848618090452265,
      "loss": 0.3608,
      "mlm_loss": 0.3608,
      "step": 12280
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.48,
      "learning_rate": 0.0004281721105527638,
      "loss": 0.3674,
      "mlm_loss": 0.3674,
      "step": 12290
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.49,
      "learning_rate": 0.00042785804020100505,
      "loss": 0.3701,
      "mlm_loss": 0.3701,
      "step": 12300
    },
    {
      "epoch": 9.49,
      "eval_ep_loss": -2.4860949516296387,
      "eval_loss": 0.34168559312820435,
      "eval_mlm_loss": 0.34168559312820435,
      "eval_runtime": 60.0999,
      "eval_samples_per_second": 1162.548,
      "eval_steps_per_second": 0.582,
      "step": 12300
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.49,
      "learning_rate": 0.0004275439698492463,
      "loss": 0.3626,
      "mlm_loss": 0.3626,
      "step": 12310
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.5,
      "learning_rate": 0.00042722989949748745,
      "loss": 0.3667,
      "mlm_loss": 0.3667,
      "step": 12320
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.51,
      "learning_rate": 0.0004269158291457286,
      "loss": 0.3564,
      "mlm_loss": 0.3564,
      "step": 12330
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.52,
      "learning_rate": 0.00042660175879396985,
      "loss": 0.3653,
      "mlm_loss": 0.3653,
      "step": 12340
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.53,
      "learning_rate": 0.00042628768844221103,
      "loss": 0.3696,
      "mlm_loss": 0.3696,
      "step": 12350
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.53,
      "learning_rate": 0.00042597361809045226,
      "loss": 0.3653,
      "mlm_loss": 0.3653,
      "step": 12360
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.54,
      "learning_rate": 0.0004256595477386935,
      "loss": 0.3645,
      "mlm_loss": 0.3645,
      "step": 12370
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.55,
      "learning_rate": 0.0004253454773869347,
      "loss": 0.368,
      "mlm_loss": 0.368,
      "step": 12380
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.56,
      "learning_rate": 0.0004250314070351759,
      "loss": 0.3637,
      "mlm_loss": 0.3637,
      "step": 12390
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.56,
      "learning_rate": 0.0004247173366834171,
      "loss": 0.3694,
      "mlm_loss": 0.3694,
      "step": 12400
    },
    {
      "epoch": 9.56,
      "eval_ep_loss": -2.249842405319214,
      "eval_loss": 0.3413134813308716,
      "eval_mlm_loss": 0.3413134813308716,
      "eval_runtime": 60.5795,
      "eval_samples_per_second": 1153.344,
      "eval_steps_per_second": 0.578,
      "step": 12400
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.57,
      "learning_rate": 0.00042440326633165834,
      "loss": 0.3612,
      "mlm_loss": 0.3612,
      "step": 12410
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.58,
      "learning_rate": 0.0004240891959798995,
      "loss": 0.3594,
      "mlm_loss": 0.3594,
      "step": 12420
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.59,
      "learning_rate": 0.0004237751256281407,
      "loss": 0.3705,
      "mlm_loss": 0.3705,
      "step": 12430
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.6,
      "learning_rate": 0.0004234610552763819,
      "loss": 0.3608,
      "mlm_loss": 0.3608,
      "step": 12440
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.6,
      "learning_rate": 0.0004231469849246231,
      "loss": 0.3611,
      "mlm_loss": 0.3611,
      "step": 12450
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.61,
      "learning_rate": 0.0004228329145728643,
      "loss": 0.3615,
      "mlm_loss": 0.3615,
      "step": 12460
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.62,
      "learning_rate": 0.00042251884422110555,
      "loss": 0.3641,
      "mlm_loss": 0.3641,
      "step": 12470
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.63,
      "learning_rate": 0.0004222047738693467,
      "loss": 0.3664,
      "mlm_loss": 0.3664,
      "step": 12480
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.63,
      "learning_rate": 0.00042189070351758795,
      "loss": 0.3612,
      "mlm_loss": 0.3612,
      "step": 12490
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.64,
      "learning_rate": 0.0004215766331658292,
      "loss": 0.37,
      "mlm_loss": 0.37,
      "step": 12500
    },
    {
      "epoch": 9.64,
      "eval_ep_loss": -2.4926164150238037,
      "eval_loss": 0.3400452435016632,
      "eval_mlm_loss": 0.3400452435016632,
      "eval_runtime": 62.1559,
      "eval_samples_per_second": 1124.093,
      "eval_steps_per_second": 0.563,
      "step": 12500
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.65,
      "learning_rate": 0.0004212625628140704,
      "loss": 0.3693,
      "mlm_loss": 0.3693,
      "step": 12510
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.66,
      "learning_rate": 0.0004209484924623116,
      "loss": 0.3623,
      "mlm_loss": 0.3623,
      "step": 12520
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.66,
      "learning_rate": 0.00042063442211055275,
      "loss": 0.3637,
      "mlm_loss": 0.3637,
      "step": 12530
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.67,
      "learning_rate": 0.000420320351758794,
      "loss": 0.363,
      "mlm_loss": 0.363,
      "step": 12540
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.68,
      "learning_rate": 0.00042000628140703516,
      "loss": 0.355,
      "mlm_loss": 0.355,
      "step": 12550
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.69,
      "learning_rate": 0.0004196922110552764,
      "loss": 0.3651,
      "mlm_loss": 0.3651,
      "step": 12560
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.7,
      "learning_rate": 0.0004193781407035176,
      "loss": 0.3593,
      "mlm_loss": 0.3593,
      "step": 12570
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.7,
      "learning_rate": 0.0004190640703517588,
      "loss": 0.3591,
      "mlm_loss": 0.3591,
      "step": 12580
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.71,
      "learning_rate": 0.00041875,
      "loss": 0.3585,
      "mlm_loss": 0.3585,
      "step": 12590
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.72,
      "learning_rate": 0.00041843592964824124,
      "loss": 0.3618,
      "mlm_loss": 0.3618,
      "step": 12600
    },
    {
      "epoch": 9.72,
      "eval_ep_loss": -2.574556827545166,
      "eval_loss": 0.33879169821739197,
      "eval_mlm_loss": 0.33879169821739197,
      "eval_runtime": 60.5312,
      "eval_samples_per_second": 1154.265,
      "eval_steps_per_second": 0.578,
      "step": 12600
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.73,
      "learning_rate": 0.0004181218592964824,
      "loss": 0.3587,
      "mlm_loss": 0.3587,
      "step": 12610
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.73,
      "learning_rate": 0.0004178077889447236,
      "loss": 0.3668,
      "mlm_loss": 0.3668,
      "step": 12620
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.74,
      "learning_rate": 0.0004174937185929648,
      "loss": 0.3616,
      "mlm_loss": 0.3616,
      "step": 12630
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.75,
      "learning_rate": 0.00041717964824120605,
      "loss": 0.3626,
      "mlm_loss": 0.3626,
      "step": 12640
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.76,
      "learning_rate": 0.0004168655778894472,
      "loss": 0.3514,
      "mlm_loss": 0.3514,
      "step": 12650
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.76,
      "learning_rate": 0.00041655150753768845,
      "loss": 0.3605,
      "mlm_loss": 0.3605,
      "step": 12660
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.77,
      "learning_rate": 0.0004162374371859297,
      "loss": 0.3591,
      "mlm_loss": 0.3591,
      "step": 12670
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.78,
      "learning_rate": 0.00041592336683417085,
      "loss": 0.3642,
      "mlm_loss": 0.3642,
      "step": 12680
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.79,
      "learning_rate": 0.0004156092964824121,
      "loss": 0.3683,
      "mlm_loss": 0.3683,
      "step": 12690
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.8,
      "learning_rate": 0.0004152952261306533,
      "loss": 0.3614,
      "mlm_loss": 0.3614,
      "step": 12700
    },
    {
      "epoch": 9.8,
      "eval_ep_loss": -2.610844135284424,
      "eval_loss": 0.33695146441459656,
      "eval_mlm_loss": 0.33695146441459656,
      "eval_runtime": 59.2376,
      "eval_samples_per_second": 1179.471,
      "eval_steps_per_second": 0.591,
      "step": 12700
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.8,
      "learning_rate": 0.0004149811557788945,
      "loss": 0.3551,
      "mlm_loss": 0.3551,
      "step": 12710
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.81,
      "learning_rate": 0.00041466708542713565,
      "loss": 0.357,
      "mlm_loss": 0.357,
      "step": 12720
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.82,
      "learning_rate": 0.0004143530150753769,
      "loss": 0.3631,
      "mlm_loss": 0.3631,
      "step": 12730
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.83,
      "learning_rate": 0.00041403894472361806,
      "loss": 0.3603,
      "mlm_loss": 0.3603,
      "step": 12740
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.83,
      "learning_rate": 0.0004137248743718593,
      "loss": 0.3624,
      "mlm_loss": 0.3624,
      "step": 12750
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.84,
      "learning_rate": 0.0004134108040201005,
      "loss": 0.3619,
      "mlm_loss": 0.3619,
      "step": 12760
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.85,
      "learning_rate": 0.00041309673366834174,
      "loss": 0.3623,
      "mlm_loss": 0.3623,
      "step": 12770
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.86,
      "learning_rate": 0.0004127826633165829,
      "loss": 0.3615,
      "mlm_loss": 0.3615,
      "step": 12780
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.87,
      "learning_rate": 0.00041246859296482414,
      "loss": 0.3569,
      "mlm_loss": 0.3569,
      "step": 12790
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.87,
      "learning_rate": 0.00041215452261306537,
      "loss": 0.3561,
      "mlm_loss": 0.3561,
      "step": 12800
    },
    {
      "epoch": 9.87,
      "eval_ep_loss": -2.4833860397338867,
      "eval_loss": 0.3353259861469269,
      "eval_mlm_loss": 0.3353259861469269,
      "eval_runtime": 59.1455,
      "eval_samples_per_second": 1181.307,
      "eval_steps_per_second": 0.592,
      "step": 12800
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.88,
      "learning_rate": 0.00041184045226130654,
      "loss": 0.362,
      "mlm_loss": 0.362,
      "step": 12810
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.89,
      "learning_rate": 0.0004115263819095477,
      "loss": 0.3589,
      "mlm_loss": 0.3589,
      "step": 12820
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.9,
      "learning_rate": 0.00041121231155778895,
      "loss": 0.3614,
      "mlm_loss": 0.3614,
      "step": 12830
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.9,
      "learning_rate": 0.0004108982412060301,
      "loss": 0.3534,
      "mlm_loss": 0.3534,
      "step": 12840
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.91,
      "learning_rate": 0.00041058417085427135,
      "loss": 0.36,
      "mlm_loss": 0.36,
      "step": 12850
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.92,
      "learning_rate": 0.0004102701005025126,
      "loss": 0.3579,
      "mlm_loss": 0.3579,
      "step": 12860
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.93,
      "learning_rate": 0.0004099560301507538,
      "loss": 0.358,
      "mlm_loss": 0.358,
      "step": 12870
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.93,
      "learning_rate": 0.000409641959798995,
      "loss": 0.3568,
      "mlm_loss": 0.3568,
      "step": 12880
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.94,
      "learning_rate": 0.0004093278894472362,
      "loss": 0.3596,
      "mlm_loss": 0.3596,
      "step": 12890
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.95,
      "learning_rate": 0.00040901381909547743,
      "loss": 0.356,
      "mlm_loss": 0.356,
      "step": 12900
    },
    {
      "epoch": 9.95,
      "eval_ep_loss": -2.3914308547973633,
      "eval_loss": 0.33489990234375,
      "eval_mlm_loss": 0.33489990234375,
      "eval_runtime": 60.1749,
      "eval_samples_per_second": 1161.099,
      "eval_steps_per_second": 0.582,
      "step": 12900
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.96,
      "learning_rate": 0.0004086997487437186,
      "loss": 0.3613,
      "mlm_loss": 0.3613,
      "step": 12910
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.97,
      "learning_rate": 0.0004083856783919598,
      "loss": 0.3561,
      "mlm_loss": 0.3561,
      "step": 12920
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.97,
      "learning_rate": 0.000408071608040201,
      "loss": 0.354,
      "mlm_loss": 0.354,
      "step": 12930
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.98,
      "learning_rate": 0.0004077575376884422,
      "loss": 0.36,
      "mlm_loss": 0.36,
      "step": 12940
    },
    {
      "ep_loss": 0.0,
      "epoch": 9.99,
      "learning_rate": 0.0004074434673366834,
      "loss": 0.3583,
      "mlm_loss": 0.3583,
      "step": 12950
    },
    {
      "ep_loss": 0.0,
      "epoch": 10.0,
      "learning_rate": 0.00040712939698492464,
      "loss": 0.3608,
      "mlm_loss": 0.3608,
      "step": 12960
    },
    {
      "ep_loss": 0.0,
      "epoch": 10.0,
      "learning_rate": 0.0004068153266331658,
      "loss": 0.3596,
      "mlm_loss": 0.3596,
      "step": 12970
    },
    {
      "ep_loss": 0.0,
      "epoch": 10.01,
      "learning_rate": 0.00040650125628140704,
      "loss": 0.3559,
      "mlm_loss": 0.3559,
      "step": 12980
    },
    {
      "ep_loss": 0.0,
      "epoch": 10.02,
      "learning_rate": 0.00040618718592964827,
      "loss": 0.3582,
      "mlm_loss": 0.3582,
      "step": 12990
    },
    {
      "ep_loss": 0.0,
      "epoch": 10.03,
      "learning_rate": 0.0004058731155778895,
      "loss": 0.3516,
      "mlm_loss": 0.3516,
      "step": 13000
    },
    {
      "epoch": 10.03,
      "eval_ep_loss": -2.4109866619110107,
      "eval_loss": 0.3338833749294281,
      "eval_mlm_loss": 0.3338833749294281,
      "eval_runtime": 59.7838,
      "eval_samples_per_second": 1168.695,
      "eval_steps_per_second": 0.585,
      "step": 13000
    }
  ],
  "max_steps": 25920,
  "num_train_epochs": 20,
  "total_flos": 3.5279344821600256e+18,
  "trial_name": null,
  "trial_params": null
}