{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 27.2108843537415,
  "global_step": 3500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.16,
      "learning_rate": 1.9912109375000002e-05,
      "loss": 1.9656,
      "step": 20
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.98193359375e-05,
      "loss": 1.8445,
      "step": 40
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.97216796875e-05,
      "loss": 1.59,
      "step": 60
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.9624023437500002e-05,
      "loss": 1.6275,
      "step": 80
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.9526367187500002e-05,
      "loss": 1.473,
      "step": 100
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.9428710937500003e-05,
      "loss": 1.3701,
      "step": 120
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.93310546875e-05,
      "loss": 1.3224,
      "step": 140
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.92333984375e-05,
      "loss": 1.1423,
      "step": 160
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.91357421875e-05,
      "loss": 1.1652,
      "step": 180
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.9038085937500002e-05,
      "loss": 1.1422,
      "step": 200
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.8940429687500002e-05,
      "loss": 1.002,
      "step": 220
    },
    {
      "epoch": 1.87,
      "learning_rate": 1.8842773437500003e-05,
      "loss": 1.0779,
      "step": 240
    },
    {
      "epoch": 2.02,
      "learning_rate": 1.87451171875e-05,
      "loss": 0.9887,
      "step": 260
    },
    {
      "epoch": 2.18,
      "learning_rate": 1.86474609375e-05,
      "loss": 0.9543,
      "step": 280
    },
    {
      "epoch": 2.33,
      "learning_rate": 1.85498046875e-05,
      "loss": 0.9371,
      "step": 300
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.8452148437500002e-05,
      "loss": 0.8701,
      "step": 320
    },
    {
      "epoch": 2.64,
      "learning_rate": 1.8354492187500003e-05,
      "loss": 0.875,
      "step": 340
    },
    {
      "epoch": 2.8,
      "learning_rate": 1.82568359375e-05,
      "loss": 0.7843,
      "step": 360
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.81591796875e-05,
      "loss": 0.7945,
      "step": 380
    },
    {
      "epoch": 3.11,
      "learning_rate": 1.80615234375e-05,
      "loss": 0.799,
      "step": 400
    },
    {
      "epoch": 3.27,
      "learning_rate": 1.79638671875e-05,
      "loss": 0.7623,
      "step": 420
    },
    {
      "epoch": 3.42,
      "learning_rate": 1.7866210937500002e-05,
      "loss": 0.7263,
      "step": 440
    },
    {
      "epoch": 3.58,
      "learning_rate": 1.7768554687500003e-05,
      "loss": 0.7779,
      "step": 460
    },
    {
      "epoch": 3.73,
      "learning_rate": 1.76708984375e-05,
      "loss": 0.695,
      "step": 480
    },
    {
      "epoch": 3.89,
      "learning_rate": 1.75732421875e-05,
      "loss": 0.7344,
      "step": 500
    },
    {
      "epoch": 4.04,
      "learning_rate": 1.74755859375e-05,
      "loss": 0.695,
      "step": 520
    },
    {
      "epoch": 4.2,
      "learning_rate": 1.7377929687500002e-05,
      "loss": 0.6504,
      "step": 540
    },
    {
      "epoch": 4.35,
      "learning_rate": 1.7280273437500002e-05,
      "loss": 0.6447,
      "step": 560
    },
    {
      "epoch": 4.51,
      "learning_rate": 1.7182617187500003e-05,
      "loss": 0.6931,
      "step": 580
    },
    {
      "epoch": 4.66,
      "learning_rate": 1.70849609375e-05,
      "loss": 0.6256,
      "step": 600
    },
    {
      "epoch": 4.82,
      "learning_rate": 1.69873046875e-05,
      "loss": 0.6132,
      "step": 620
    },
    {
      "epoch": 4.98,
      "learning_rate": 1.68896484375e-05,
      "loss": 0.6001,
      "step": 640
    },
    {
      "epoch": 5.13,
      "learning_rate": 1.6791992187500002e-05,
      "loss": 0.6176,
      "step": 660
    },
    {
      "epoch": 5.29,
      "learning_rate": 1.6694335937500002e-05,
      "loss": 0.5709,
      "step": 680
    },
    {
      "epoch": 5.44,
      "learning_rate": 1.65966796875e-05,
      "loss": 0.564,
      "step": 700
    },
    {
      "epoch": 5.6,
      "learning_rate": 1.64990234375e-05,
      "loss": 0.5969,
      "step": 720
    },
    {
      "epoch": 5.75,
      "learning_rate": 1.64013671875e-05,
      "loss": 0.5484,
      "step": 740
    },
    {
      "epoch": 5.91,
      "learning_rate": 1.63037109375e-05,
      "loss": 0.5667,
      "step": 760
    },
    {
      "epoch": 6.06,
      "learning_rate": 1.6206054687500002e-05,
      "loss": 0.5442,
      "step": 780
    },
    {
      "epoch": 6.22,
      "learning_rate": 1.6108398437500003e-05,
      "loss": 0.4857,
      "step": 800
    },
    {
      "epoch": 6.38,
      "learning_rate": 1.60107421875e-05,
      "loss": 0.5225,
      "step": 820
    },
    {
      "epoch": 6.53,
      "learning_rate": 1.59130859375e-05,
      "loss": 0.5457,
      "step": 840
    },
    {
      "epoch": 6.69,
      "learning_rate": 1.58154296875e-05,
      "loss": 0.5315,
      "step": 860
    },
    {
      "epoch": 6.84,
      "learning_rate": 1.57177734375e-05,
      "loss": 0.5345,
      "step": 880
    },
    {
      "epoch": 7.0,
      "learning_rate": 1.5620117187500002e-05,
      "loss": 0.5169,
      "step": 900
    },
    {
      "epoch": 7.15,
      "learning_rate": 1.5522460937500003e-05,
      "loss": 0.5115,
      "step": 920
    },
    {
      "epoch": 7.31,
      "learning_rate": 1.54248046875e-05,
      "loss": 0.4913,
      "step": 940
    },
    {
      "epoch": 7.46,
      "learning_rate": 1.53271484375e-05,
      "loss": 0.4868,
      "step": 960
    },
    {
      "epoch": 7.62,
      "learning_rate": 1.5229492187500001e-05,
      "loss": 0.5226,
      "step": 980
    },
    {
      "epoch": 7.77,
      "learning_rate": 1.5131835937500002e-05,
      "loss": 0.4061,
      "step": 1000
    },
    {
      "epoch": 7.93,
      "learning_rate": 1.50341796875e-05,
      "loss": 0.4488,
      "step": 1020
    },
    {
      "epoch": 8.09,
      "learning_rate": 1.4936523437500001e-05,
      "loss": 0.4443,
      "step": 1040
    },
    {
      "epoch": 8.24,
      "learning_rate": 1.4838867187500002e-05,
      "loss": 0.4646,
      "step": 1060
    },
    {
      "epoch": 8.4,
      "learning_rate": 1.47412109375e-05,
      "loss": 0.4513,
      "step": 1080
    },
    {
      "epoch": 8.55,
      "learning_rate": 1.4643554687500001e-05,
      "loss": 0.4394,
      "step": 1100
    },
    {
      "epoch": 8.71,
      "learning_rate": 1.4545898437500002e-05,
      "loss": 0.4337,
      "step": 1120
    },
    {
      "epoch": 8.86,
      "learning_rate": 1.44482421875e-05,
      "loss": 0.5037,
      "step": 1140
    },
    {
      "epoch": 9.02,
      "learning_rate": 1.4350585937500001e-05,
      "loss": 0.4209,
      "step": 1160
    },
    {
      "epoch": 9.17,
      "learning_rate": 1.42529296875e-05,
      "loss": 0.4073,
      "step": 1180
    },
    {
      "epoch": 9.33,
      "learning_rate": 1.4155273437500001e-05,
      "loss": 0.3879,
      "step": 1200
    },
    {
      "epoch": 9.48,
      "learning_rate": 1.4057617187500002e-05,
      "loss": 0.4434,
      "step": 1220
    },
    {
      "epoch": 9.64,
      "learning_rate": 1.39599609375e-05,
      "loss": 0.4062,
      "step": 1240
    },
    {
      "epoch": 9.8,
      "learning_rate": 1.3862304687500001e-05,
      "loss": 0.39,
      "step": 1260
    },
    {
      "epoch": 9.95,
      "learning_rate": 1.3764648437500002e-05,
      "loss": 0.4466,
      "step": 1280
    },
    {
      "epoch": 10.11,
      "learning_rate": 1.36669921875e-05,
      "loss": 0.4208,
      "step": 1300
    },
    {
      "epoch": 10.26,
      "learning_rate": 1.3569335937500001e-05,
      "loss": 0.3662,
      "step": 1320
    },
    {
      "epoch": 10.42,
      "learning_rate": 1.3471679687500002e-05,
      "loss": 0.4049,
      "step": 1340
    },
    {
      "epoch": 10.57,
      "learning_rate": 1.33740234375e-05,
      "loss": 0.3989,
      "step": 1360
    },
    {
      "epoch": 10.73,
      "learning_rate": 1.3276367187500001e-05,
      "loss": 0.3839,
      "step": 1380
    },
    {
      "epoch": 10.88,
      "learning_rate": 1.3178710937500002e-05,
      "loss": 0.4065,
      "step": 1400
    },
    {
      "epoch": 11.04,
      "learning_rate": 1.30810546875e-05,
      "loss": 0.3877,
      "step": 1420
    },
    {
      "epoch": 11.2,
      "learning_rate": 1.2983398437500001e-05,
      "loss": 0.4048,
      "step": 1440
    },
    {
      "epoch": 11.35,
      "learning_rate": 1.2885742187500002e-05,
      "loss": 0.3715,
      "step": 1460
    },
    {
      "epoch": 11.51,
      "learning_rate": 1.27880859375e-05,
      "loss": 0.3752,
      "step": 1480
    },
    {
      "epoch": 11.66,
      "learning_rate": 1.2690429687500001e-05,
      "loss": 0.3401,
      "step": 1500
    },
    {
      "epoch": 11.82,
      "learning_rate": 1.25927734375e-05,
      "loss": 0.3545,
      "step": 1520
    },
    {
      "epoch": 11.97,
      "learning_rate": 1.24951171875e-05,
      "loss": 0.3718,
      "step": 1540
    },
    {
      "epoch": 12.13,
      "learning_rate": 1.2397460937500001e-05,
      "loss": 0.3755,
      "step": 1560
    },
    {
      "epoch": 12.28,
      "learning_rate": 1.22998046875e-05,
      "loss": 0.3865,
      "step": 1580
    },
    {
      "epoch": 12.44,
      "learning_rate": 1.2202148437500001e-05,
      "loss": 0.3237,
      "step": 1600
    },
    {
      "epoch": 12.59,
      "learning_rate": 1.2104492187500001e-05,
      "loss": 0.3702,
      "step": 1620
    },
    {
      "epoch": 12.75,
      "learning_rate": 1.20068359375e-05,
      "loss": 0.3238,
      "step": 1640
    },
    {
      "epoch": 12.91,
      "learning_rate": 1.1909179687500001e-05,
      "loss": 0.3373,
      "step": 1660
    },
    {
      "epoch": 13.06,
      "learning_rate": 1.1811523437500002e-05,
      "loss": 0.3486,
      "step": 1680
    },
    {
      "epoch": 13.22,
      "learning_rate": 1.17138671875e-05,
      "loss": 0.362,
      "step": 1700
    },
    {
      "epoch": 13.37,
      "learning_rate": 1.1616210937500001e-05,
      "loss": 0.3257,
      "step": 1720
    },
    {
      "epoch": 13.53,
      "learning_rate": 1.1518554687500002e-05,
      "loss": 0.3414,
      "step": 1740
    },
    {
      "epoch": 13.68,
      "learning_rate": 1.14208984375e-05,
      "loss": 0.3121,
      "step": 1760
    },
    {
      "epoch": 13.84,
      "learning_rate": 1.1323242187500001e-05,
      "loss": 0.3598,
      "step": 1780
    },
    {
      "epoch": 13.99,
      "learning_rate": 1.1225585937500002e-05,
      "loss": 0.2906,
      "step": 1800
    },
    {
      "epoch": 14.15,
      "learning_rate": 1.11279296875e-05,
      "loss": 0.3356,
      "step": 1820
    },
    {
      "epoch": 14.31,
      "learning_rate": 1.1030273437500001e-05,
      "loss": 0.2991,
      "step": 1840
    },
    {
      "epoch": 14.46,
      "learning_rate": 1.09326171875e-05,
      "loss": 0.2987,
      "step": 1860
    },
    {
      "epoch": 14.62,
      "learning_rate": 1.08349609375e-05,
      "loss": 0.3249,
      "step": 1880
    },
    {
      "epoch": 14.77,
      "learning_rate": 1.0737304687500001e-05,
      "loss": 0.3425,
      "step": 1900
    },
    {
      "epoch": 14.93,
      "learning_rate": 1.06396484375e-05,
      "loss": 0.3398,
      "step": 1920
    },
    {
      "epoch": 15.08,
      "learning_rate": 1.05419921875e-05,
      "loss": 0.3018,
      "step": 1940
    },
    {
      "epoch": 15.24,
      "learning_rate": 1.0444335937500001e-05,
      "loss": 0.2475,
      "step": 1960
    },
    {
      "epoch": 15.39,
      "learning_rate": 1.03466796875e-05,
      "loss": 0.3507,
      "step": 1980
    },
    {
      "epoch": 15.55,
      "learning_rate": 1.02490234375e-05,
      "loss": 0.3084,
      "step": 2000
    },
    {
      "epoch": 15.7,
      "learning_rate": 1.0151367187500001e-05,
      "loss": 0.3212,
      "step": 2020
    },
    {
      "epoch": 15.86,
      "learning_rate": 1.00537109375e-05,
      "loss": 0.2831,
      "step": 2040
    },
    {
      "epoch": 16.02,
      "learning_rate": 9.956054687500001e-06,
      "loss": 0.3072,
      "step": 2060
    },
    {
      "epoch": 16.17,
      "learning_rate": 9.858398437500002e-06,
      "loss": 0.3293,
      "step": 2080
    },
    {
      "epoch": 16.33,
      "learning_rate": 9.7607421875e-06,
      "loss": 0.2738,
      "step": 2100
    },
    {
      "epoch": 16.48,
      "learning_rate": 9.663085937500001e-06,
      "loss": 0.3245,
      "step": 2120
    },
    {
      "epoch": 16.64,
      "learning_rate": 9.565429687500002e-06,
      "loss": 0.2846,
      "step": 2140
    },
    {
      "epoch": 16.79,
      "learning_rate": 9.4677734375e-06,
      "loss": 0.2906,
      "step": 2160
    },
    {
      "epoch": 16.95,
      "learning_rate": 9.370117187500001e-06,
      "loss": 0.263,
      "step": 2180
    },
    {
      "epoch": 17.1,
      "learning_rate": 9.2724609375e-06,
      "loss": 0.286,
      "step": 2200
    },
    {
      "epoch": 17.26,
      "learning_rate": 9.1748046875e-06,
      "loss": 0.3152,
      "step": 2220
    },
    {
      "epoch": 17.41,
      "learning_rate": 9.077148437500001e-06,
      "loss": 0.2811,
      "step": 2240
    },
    {
      "epoch": 17.57,
      "learning_rate": 8.9794921875e-06,
      "loss": 0.257,
      "step": 2260
    },
    {
      "epoch": 17.73,
      "learning_rate": 8.8818359375e-06,
      "loss": 0.267,
      "step": 2280
    },
    {
      "epoch": 17.88,
      "learning_rate": 8.784179687500001e-06,
      "loss": 0.3056,
      "step": 2300
    },
    {
      "epoch": 18.04,
      "learning_rate": 8.6865234375e-06,
      "loss": 0.2522,
      "step": 2320
    },
    {
      "epoch": 18.19,
      "learning_rate": 8.5888671875e-06,
      "loss": 0.2806,
      "step": 2340
    },
    {
      "epoch": 18.35,
      "learning_rate": 8.491210937500001e-06,
      "loss": 0.266,
      "step": 2360
    },
    {
      "epoch": 18.5,
      "learning_rate": 8.3935546875e-06,
      "loss": 0.288,
      "step": 2380
    },
    {
      "epoch": 18.66,
      "learning_rate": 8.2958984375e-06,
      "loss": 0.2712,
      "step": 2400
    },
    {
      "epoch": 18.81,
      "learning_rate": 8.198242187500001e-06,
      "loss": 0.3099,
      "step": 2420
    },
    {
      "epoch": 18.97,
      "learning_rate": 8.1005859375e-06,
      "loss": 0.2205,
      "step": 2440
    },
    {
      "epoch": 19.13,
      "learning_rate": 8.002929687500001e-06,
      "loss": 0.253,
      "step": 2460
    },
    {
      "epoch": 19.28,
      "learning_rate": 7.905273437500001e-06,
      "loss": 0.2885,
      "step": 2480
    },
    {
      "epoch": 19.44,
      "learning_rate": 7.8076171875e-06,
      "loss": 0.2326,
      "step": 2500
    },
    {
      "epoch": 19.59,
      "learning_rate": 7.709960937500001e-06,
      "loss": 0.255,
      "step": 2520
    },
    {
      "epoch": 19.75,
      "learning_rate": 7.612304687500001e-06,
      "loss": 0.2698,
      "step": 2540
    },
    {
      "epoch": 19.9,
      "learning_rate": 7.5146484375000004e-06,
      "loss": 0.2532,
      "step": 2560
    },
    {
      "epoch": 20.06,
      "learning_rate": 7.4169921875e-06,
      "loss": 0.2562,
      "step": 2580
    },
    {
      "epoch": 20.21,
      "learning_rate": 7.319335937500001e-06,
      "loss": 0.2717,
      "step": 2600
    },
    {
      "epoch": 20.37,
      "learning_rate": 7.2216796875000005e-06,
      "loss": 0.2652,
      "step": 2620
    },
    {
      "epoch": 20.52,
      "learning_rate": 7.1240234375e-06,
      "loss": 0.2654,
      "step": 2640
    },
    {
      "epoch": 20.68,
      "learning_rate": 7.026367187500001e-06,
      "loss": 0.2766,
      "step": 2660
    },
    {
      "epoch": 20.84,
      "learning_rate": 6.928710937500001e-06,
      "loss": 0.2246,
      "step": 2680
    },
    {
      "epoch": 20.99,
      "learning_rate": 6.8310546875e-06,
      "loss": 0.2546,
      "step": 2700
    },
    {
      "epoch": 21.15,
      "learning_rate": 6.733398437500001e-06,
      "loss": 0.2708,
      "step": 2720
    },
    {
      "epoch": 21.3,
      "learning_rate": 6.635742187500001e-06,
      "loss": 0.2385,
      "step": 2740
    },
    {
      "epoch": 21.46,
      "learning_rate": 6.5380859375e-06,
      "loss": 0.2271,
      "step": 2760
    },
    {
      "epoch": 21.61,
      "learning_rate": 6.4404296875e-06,
      "loss": 0.2874,
      "step": 2780
    },
    {
      "epoch": 21.77,
      "learning_rate": 6.342773437500001e-06,
      "loss": 0.2447,
      "step": 2800
    },
    {
      "epoch": 21.92,
      "learning_rate": 6.2451171875000005e-06,
      "loss": 0.2309,
      "step": 2820
    },
    {
      "epoch": 22.08,
      "learning_rate": 6.1474609375e-06,
      "loss": 0.2587,
      "step": 2840
    },
    {
      "epoch": 22.24,
      "learning_rate": 6.049804687500001e-06,
      "loss": 0.2303,
      "step": 2860
    },
    {
      "epoch": 22.39,
      "learning_rate": 5.952148437500001e-06,
      "loss": 0.2376,
      "step": 2880
    },
    {
      "epoch": 22.55,
      "learning_rate": 5.8544921875e-06,
      "loss": 0.2696,
      "step": 2900
    },
    {
      "epoch": 22.7,
      "learning_rate": 5.7568359375e-06,
      "loss": 0.2459,
      "step": 2920
    },
    {
      "epoch": 22.86,
      "learning_rate": 5.659179687500001e-06,
      "loss": 0.2032,
      "step": 2940
    },
    {
      "epoch": 23.01,
      "learning_rate": 5.5615234375e-06,
      "loss": 0.2491,
      "step": 2960
    },
    {
      "epoch": 23.17,
      "learning_rate": 5.4638671875e-06,
      "loss": 0.2395,
      "step": 2980
    },
    {
      "epoch": 23.32,
      "learning_rate": 5.366210937500001e-06,
      "loss": 0.2573,
      "step": 3000
    },
    {
      "epoch": 23.48,
      "learning_rate": 5.2685546875000005e-06,
      "loss": 0.2016,
      "step": 3020
    },
    {
      "epoch": 23.63,
      "learning_rate": 5.1708984375e-06,
      "loss": 0.2486,
      "step": 3040
    },
    {
      "epoch": 23.79,
      "learning_rate": 5.073242187500001e-06,
      "loss": 0.2209,
      "step": 3060
    },
    {
      "epoch": 23.95,
      "learning_rate": 4.9755859375000006e-06,
      "loss": 0.2373,
      "step": 3080
    },
    {
      "epoch": 24.1,
      "learning_rate": 4.8779296875e-06,
      "loss": 0.2433,
      "step": 3100
    },
    {
      "epoch": 24.26,
      "learning_rate": 4.7802734375e-06,
      "loss": 0.1869,
      "step": 3120
    },
    {
      "epoch": 24.41,
      "learning_rate": 4.682617187500001e-06,
      "loss": 0.2299,
      "step": 3140
    },
    {
      "epoch": 24.57,
      "learning_rate": 4.5849609375e-06,
      "loss": 0.2556,
      "step": 3160
    },
    {
      "epoch": 24.72,
      "learning_rate": 4.4873046875e-06,
      "loss": 0.2569,
      "step": 3180
    },
    {
      "epoch": 24.88,
      "learning_rate": 4.389648437500001e-06,
      "loss": 0.2298,
      "step": 3200
    },
    {
      "epoch": 25.03,
      "learning_rate": 4.2919921875000005e-06,
      "loss": 0.2368,
      "step": 3220
    },
    {
      "epoch": 25.19,
      "learning_rate": 4.1943359375e-06,
      "loss": 0.1928,
      "step": 3240
    },
    {
      "epoch": 25.34,
      "learning_rate": 4.0966796875e-06,
      "loss": 0.2253,
      "step": 3260
    },
    {
      "epoch": 25.5,
      "learning_rate": 3.9990234375000005e-06,
      "loss": 0.2369,
      "step": 3280
    },
    {
      "epoch": 25.66,
      "learning_rate": 3.9013671875e-06,
      "loss": 0.2117,
      "step": 3300
    },
    {
      "epoch": 25.81,
      "learning_rate": 3.8037109375000004e-06,
      "loss": 0.2407,
      "step": 3320
    },
    {
      "epoch": 25.97,
      "learning_rate": 3.7060546875e-06,
      "loss": 0.2451,
      "step": 3340
    },
    {
      "epoch": 26.12,
      "learning_rate": 3.6083984375000004e-06,
      "loss": 0.235,
      "step": 3360
    },
    {
      "epoch": 26.28,
      "learning_rate": 3.5107421875e-06,
      "loss": 0.2446,
      "step": 3380
    },
    {
      "epoch": 26.43,
      "learning_rate": 3.4130859375000003e-06,
      "loss": 0.1815,
      "step": 3400
    },
    {
      "epoch": 26.59,
      "learning_rate": 3.3154296875000004e-06,
      "loss": 0.2283,
      "step": 3420
    },
    {
      "epoch": 26.74,
      "learning_rate": 3.2177734375e-06,
      "loss": 0.2136,
      "step": 3440
    },
    {
      "epoch": 26.9,
      "learning_rate": 3.1201171875000003e-06,
      "loss": 0.2438,
      "step": 3460
    },
    {
      "epoch": 27.06,
      "learning_rate": 3.0224609375e-06,
      "loss": 0.2097,
      "step": 3480
    },
    {
      "epoch": 27.21,
      "learning_rate": 2.9248046875000003e-06,
      "loss": 0.2361,
      "step": 3500
    }
  ],
  "max_steps": 4096,
  "num_train_epochs": 32,
  "total_flos": 4.5484736274432e+18,
  "trial_name": null,
  "trial_params": null
}