Mel-Iza0's picture
Training in progress, epoch 1, checkpoint
e4ea7b5
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 1.1006397468528581,
"eval_steps": 500,
"global_step": 8000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 1.6666666666666667e-05,
"loss": 0.6949,
"step": 10
},
{
"epoch": 0.0,
"learning_rate": 3.3333333333333335e-05,
"loss": 0.7068,
"step": 20
},
{
"epoch": 0.0,
"learning_rate": 5e-05,
"loss": 0.6935,
"step": 30
},
{
"epoch": 0.01,
"learning_rate": 6.666666666666667e-05,
"loss": 0.6943,
"step": 40
},
{
"epoch": 0.01,
"learning_rate": 8.333333333333334e-05,
"loss": 0.6806,
"step": 50
},
{
"epoch": 0.01,
"learning_rate": 0.0001,
"loss": 0.6849,
"step": 60
},
{
"epoch": 0.01,
"learning_rate": 0.00011666666666666668,
"loss": 0.6761,
"step": 70
},
{
"epoch": 0.01,
"learning_rate": 0.00013333333333333334,
"loss": 0.6735,
"step": 80
},
{
"epoch": 0.01,
"learning_rate": 0.00015000000000000001,
"loss": 0.6582,
"step": 90
},
{
"epoch": 0.01,
"learning_rate": 0.0001666666666666667,
"loss": 0.6532,
"step": 100
},
{
"epoch": 0.02,
"learning_rate": 0.00018333333333333334,
"loss": 0.6705,
"step": 110
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 0.6631,
"step": 120
},
{
"epoch": 0.02,
"learning_rate": 0.00021666666666666666,
"loss": 0.6611,
"step": 130
},
{
"epoch": 0.02,
"learning_rate": 0.00023333333333333336,
"loss": 0.6378,
"step": 140
},
{
"epoch": 0.02,
"learning_rate": 0.00025,
"loss": 0.6173,
"step": 150
},
{
"epoch": 0.02,
"learning_rate": 0.0002666666666666667,
"loss": 0.6421,
"step": 160
},
{
"epoch": 0.02,
"learning_rate": 0.00028333333333333335,
"loss": 0.633,
"step": 170
},
{
"epoch": 0.02,
"learning_rate": 0.00030000000000000003,
"loss": 0.6303,
"step": 180
},
{
"epoch": 0.03,
"learning_rate": 0.00031666666666666665,
"loss": 0.6238,
"step": 190
},
{
"epoch": 0.03,
"learning_rate": 0.0003333333333333334,
"loss": 0.6372,
"step": 200
},
{
"epoch": 0.03,
"learning_rate": 0.00035,
"loss": 0.6316,
"step": 210
},
{
"epoch": 0.03,
"learning_rate": 0.00036666666666666667,
"loss": 0.6591,
"step": 220
},
{
"epoch": 0.03,
"learning_rate": 0.00038333333333333334,
"loss": 0.6863,
"step": 230
},
{
"epoch": 0.03,
"learning_rate": 0.0004,
"loss": 0.6636,
"step": 240
},
{
"epoch": 0.03,
"learning_rate": 0.0004,
"loss": 0.6793,
"step": 250
},
{
"epoch": 0.04,
"learning_rate": 0.0004,
"loss": 0.6807,
"step": 260
},
{
"epoch": 0.04,
"learning_rate": 0.0004,
"loss": 0.6521,
"step": 270
},
{
"epoch": 0.04,
"learning_rate": 0.0004,
"loss": 0.6693,
"step": 280
},
{
"epoch": 0.04,
"learning_rate": 0.0004,
"loss": 0.6517,
"step": 290
},
{
"epoch": 0.04,
"learning_rate": 0.0004,
"loss": 0.6745,
"step": 300
},
{
"epoch": 0.04,
"learning_rate": 0.0004,
"loss": 0.6568,
"step": 310
},
{
"epoch": 0.04,
"learning_rate": 0.0004,
"loss": 0.6558,
"step": 320
},
{
"epoch": 0.05,
"learning_rate": 0.0004,
"loss": 0.6553,
"step": 330
},
{
"epoch": 0.05,
"learning_rate": 0.0004,
"loss": 0.6598,
"step": 340
},
{
"epoch": 0.05,
"learning_rate": 0.0004,
"loss": 0.6383,
"step": 350
},
{
"epoch": 0.05,
"learning_rate": 0.0004,
"loss": 0.6406,
"step": 360
},
{
"epoch": 0.05,
"learning_rate": 0.0004,
"loss": 0.6541,
"step": 370
},
{
"epoch": 0.05,
"learning_rate": 0.0004,
"loss": 0.6582,
"step": 380
},
{
"epoch": 0.05,
"learning_rate": 0.0004,
"loss": 0.6467,
"step": 390
},
{
"epoch": 0.06,
"learning_rate": 0.0004,
"loss": 0.6232,
"step": 400
},
{
"epoch": 0.06,
"learning_rate": 0.0004,
"loss": 0.661,
"step": 410
},
{
"epoch": 0.06,
"learning_rate": 0.0004,
"loss": 0.6333,
"step": 420
},
{
"epoch": 0.06,
"learning_rate": 0.0004,
"loss": 0.6432,
"step": 430
},
{
"epoch": 0.06,
"learning_rate": 0.0004,
"loss": 0.6472,
"step": 440
},
{
"epoch": 0.06,
"learning_rate": 0.0004,
"loss": 0.6631,
"step": 450
},
{
"epoch": 0.06,
"learning_rate": 0.0004,
"loss": 0.6306,
"step": 460
},
{
"epoch": 0.06,
"learning_rate": 0.0004,
"loss": 0.6541,
"step": 470
},
{
"epoch": 0.07,
"learning_rate": 0.0004,
"loss": 0.6579,
"step": 480
},
{
"epoch": 0.07,
"learning_rate": 0.0004,
"loss": 0.6301,
"step": 490
},
{
"epoch": 0.07,
"learning_rate": 0.0004,
"loss": 0.651,
"step": 500
},
{
"epoch": 0.07,
"learning_rate": 0.0004,
"loss": 0.6388,
"step": 510
},
{
"epoch": 0.07,
"learning_rate": 0.0004,
"loss": 0.6221,
"step": 520
},
{
"epoch": 0.07,
"learning_rate": 0.0004,
"loss": 0.6293,
"step": 530
},
{
"epoch": 0.07,
"learning_rate": 0.0004,
"loss": 0.6506,
"step": 540
},
{
"epoch": 0.08,
"learning_rate": 0.0004,
"loss": 0.6262,
"step": 550
},
{
"epoch": 0.08,
"learning_rate": 0.0004,
"loss": 0.644,
"step": 560
},
{
"epoch": 0.08,
"learning_rate": 0.0004,
"loss": 0.6097,
"step": 570
},
{
"epoch": 0.08,
"learning_rate": 0.0004,
"loss": 0.6234,
"step": 580
},
{
"epoch": 0.08,
"learning_rate": 0.0004,
"loss": 0.6245,
"step": 590
},
{
"epoch": 0.08,
"learning_rate": 0.0004,
"loss": 0.6496,
"step": 600
},
{
"epoch": 0.08,
"learning_rate": 0.0004,
"loss": 0.6251,
"step": 610
},
{
"epoch": 0.09,
"learning_rate": 0.0004,
"loss": 0.6481,
"step": 620
},
{
"epoch": 0.09,
"learning_rate": 0.0004,
"loss": 0.6158,
"step": 630
},
{
"epoch": 0.09,
"learning_rate": 0.0004,
"loss": 0.6326,
"step": 640
},
{
"epoch": 0.09,
"learning_rate": 0.0004,
"loss": 0.6384,
"step": 650
},
{
"epoch": 0.09,
"learning_rate": 0.0004,
"loss": 0.6417,
"step": 660
},
{
"epoch": 0.09,
"learning_rate": 0.0004,
"loss": 0.624,
"step": 670
},
{
"epoch": 0.09,
"learning_rate": 0.0004,
"loss": 0.641,
"step": 680
},
{
"epoch": 0.09,
"learning_rate": 0.0004,
"loss": 0.6142,
"step": 690
},
{
"epoch": 0.1,
"learning_rate": 0.0004,
"loss": 0.6326,
"step": 700
},
{
"epoch": 0.1,
"learning_rate": 0.0004,
"loss": 0.6176,
"step": 710
},
{
"epoch": 0.1,
"learning_rate": 0.0004,
"loss": 0.629,
"step": 720
},
{
"epoch": 0.1,
"learning_rate": 0.0004,
"loss": 0.6405,
"step": 730
},
{
"epoch": 0.1,
"learning_rate": 0.0004,
"loss": 0.6249,
"step": 740
},
{
"epoch": 0.1,
"learning_rate": 0.0004,
"loss": 0.6179,
"step": 750
},
{
"epoch": 0.1,
"learning_rate": 0.0004,
"loss": 0.6284,
"step": 760
},
{
"epoch": 0.11,
"learning_rate": 0.0004,
"loss": 0.6167,
"step": 770
},
{
"epoch": 0.11,
"learning_rate": 0.0004,
"loss": 0.6131,
"step": 780
},
{
"epoch": 0.11,
"learning_rate": 0.0004,
"loss": 0.6247,
"step": 790
},
{
"epoch": 0.11,
"learning_rate": 0.0004,
"loss": 0.6121,
"step": 800
},
{
"epoch": 0.11,
"learning_rate": 0.0004,
"loss": 0.6237,
"step": 810
},
{
"epoch": 0.11,
"learning_rate": 0.0004,
"loss": 0.613,
"step": 820
},
{
"epoch": 0.11,
"learning_rate": 0.0004,
"loss": 0.6326,
"step": 830
},
{
"epoch": 0.12,
"learning_rate": 0.0004,
"loss": 0.5879,
"step": 840
},
{
"epoch": 0.12,
"learning_rate": 0.0004,
"loss": 0.6274,
"step": 850
},
{
"epoch": 0.12,
"learning_rate": 0.0004,
"loss": 0.6235,
"step": 860
},
{
"epoch": 0.12,
"learning_rate": 0.0004,
"loss": 0.6163,
"step": 870
},
{
"epoch": 0.12,
"learning_rate": 0.0004,
"loss": 0.6305,
"step": 880
},
{
"epoch": 0.12,
"learning_rate": 0.0004,
"loss": 0.632,
"step": 890
},
{
"epoch": 0.12,
"learning_rate": 0.0004,
"loss": 0.6257,
"step": 900
},
{
"epoch": 0.13,
"learning_rate": 0.0004,
"loss": 0.6202,
"step": 910
},
{
"epoch": 0.13,
"learning_rate": 0.0004,
"loss": 0.6007,
"step": 920
},
{
"epoch": 0.13,
"learning_rate": 0.0004,
"loss": 0.6051,
"step": 930
},
{
"epoch": 0.13,
"learning_rate": 0.0004,
"loss": 0.6065,
"step": 940
},
{
"epoch": 0.13,
"learning_rate": 0.0004,
"loss": 0.5804,
"step": 950
},
{
"epoch": 0.13,
"learning_rate": 0.0004,
"loss": 0.5895,
"step": 960
},
{
"epoch": 0.13,
"learning_rate": 0.0004,
"loss": 0.6052,
"step": 970
},
{
"epoch": 0.13,
"learning_rate": 0.0004,
"loss": 0.5828,
"step": 980
},
{
"epoch": 0.14,
"learning_rate": 0.0004,
"loss": 0.5873,
"step": 990
},
{
"epoch": 0.14,
"learning_rate": 0.0004,
"loss": 0.608,
"step": 1000
},
{
"epoch": 0.14,
"learning_rate": 0.0004,
"loss": 0.5947,
"step": 1010
},
{
"epoch": 0.14,
"learning_rate": 0.0004,
"loss": 0.5734,
"step": 1020
},
{
"epoch": 0.14,
"learning_rate": 0.0004,
"loss": 0.6063,
"step": 1030
},
{
"epoch": 0.14,
"learning_rate": 0.0004,
"loss": 0.5774,
"step": 1040
},
{
"epoch": 0.14,
"learning_rate": 0.0004,
"loss": 0.5984,
"step": 1050
},
{
"epoch": 0.15,
"learning_rate": 0.0004,
"loss": 0.6127,
"step": 1060
},
{
"epoch": 0.15,
"learning_rate": 0.0004,
"loss": 0.5773,
"step": 1070
},
{
"epoch": 0.15,
"learning_rate": 0.0004,
"loss": 0.5854,
"step": 1080
},
{
"epoch": 0.15,
"learning_rate": 0.0004,
"loss": 0.5822,
"step": 1090
},
{
"epoch": 0.15,
"learning_rate": 0.0004,
"loss": 0.5745,
"step": 1100
},
{
"epoch": 0.15,
"learning_rate": 0.0004,
"loss": 0.5832,
"step": 1110
},
{
"epoch": 0.15,
"learning_rate": 0.0004,
"loss": 0.5926,
"step": 1120
},
{
"epoch": 0.16,
"learning_rate": 0.0004,
"loss": 0.5886,
"step": 1130
},
{
"epoch": 0.16,
"learning_rate": 0.0004,
"loss": 0.5935,
"step": 1140
},
{
"epoch": 0.16,
"learning_rate": 0.0004,
"loss": 0.5899,
"step": 1150
},
{
"epoch": 0.16,
"learning_rate": 0.0004,
"loss": 0.5688,
"step": 1160
},
{
"epoch": 0.16,
"learning_rate": 0.0004,
"loss": 0.5893,
"step": 1170
},
{
"epoch": 0.16,
"learning_rate": 0.0004,
"loss": 0.5961,
"step": 1180
},
{
"epoch": 0.16,
"learning_rate": 0.0004,
"loss": 0.5872,
"step": 1190
},
{
"epoch": 0.17,
"learning_rate": 0.0004,
"loss": 0.5985,
"step": 1200
},
{
"epoch": 0.17,
"learning_rate": 0.0004,
"loss": 0.59,
"step": 1210
},
{
"epoch": 0.17,
"learning_rate": 0.0004,
"loss": 0.5726,
"step": 1220
},
{
"epoch": 0.17,
"learning_rate": 0.0004,
"loss": 0.5858,
"step": 1230
},
{
"epoch": 0.17,
"learning_rate": 0.0004,
"loss": 0.5674,
"step": 1240
},
{
"epoch": 0.17,
"learning_rate": 0.0004,
"loss": 0.5841,
"step": 1250
},
{
"epoch": 0.17,
"learning_rate": 0.0004,
"loss": 0.561,
"step": 1260
},
{
"epoch": 0.17,
"learning_rate": 0.0004,
"loss": 0.5483,
"step": 1270
},
{
"epoch": 0.18,
"learning_rate": 0.0004,
"loss": 0.5779,
"step": 1280
},
{
"epoch": 0.18,
"learning_rate": 0.0004,
"loss": 0.5906,
"step": 1290
},
{
"epoch": 0.18,
"learning_rate": 0.0004,
"loss": 0.5464,
"step": 1300
},
{
"epoch": 0.18,
"learning_rate": 0.0004,
"loss": 0.5706,
"step": 1310
},
{
"epoch": 0.18,
"learning_rate": 0.0004,
"loss": 0.5643,
"step": 1320
},
{
"epoch": 0.18,
"learning_rate": 0.0004,
"loss": 0.5782,
"step": 1330
},
{
"epoch": 0.18,
"learning_rate": 0.0004,
"loss": 0.5911,
"step": 1340
},
{
"epoch": 0.19,
"learning_rate": 0.0004,
"loss": 0.5592,
"step": 1350
},
{
"epoch": 0.19,
"learning_rate": 0.0004,
"loss": 0.5893,
"step": 1360
},
{
"epoch": 0.19,
"learning_rate": 0.0004,
"loss": 0.5783,
"step": 1370
},
{
"epoch": 0.19,
"learning_rate": 0.0004,
"loss": 0.5805,
"step": 1380
},
{
"epoch": 0.19,
"learning_rate": 0.0004,
"loss": 0.5809,
"step": 1390
},
{
"epoch": 0.19,
"learning_rate": 0.0004,
"loss": 0.5757,
"step": 1400
},
{
"epoch": 0.19,
"learning_rate": 0.0004,
"loss": 0.5779,
"step": 1410
},
{
"epoch": 0.2,
"learning_rate": 0.0004,
"loss": 0.5979,
"step": 1420
},
{
"epoch": 0.2,
"learning_rate": 0.0004,
"loss": 0.5741,
"step": 1430
},
{
"epoch": 0.2,
"learning_rate": 0.0004,
"loss": 0.5753,
"step": 1440
},
{
"epoch": 0.2,
"learning_rate": 0.0004,
"loss": 0.5688,
"step": 1450
},
{
"epoch": 0.2,
"learning_rate": 0.0004,
"loss": 0.5575,
"step": 1460
},
{
"epoch": 0.2,
"learning_rate": 0.0004,
"loss": 0.5755,
"step": 1470
},
{
"epoch": 0.2,
"learning_rate": 0.0004,
"loss": 0.5547,
"step": 1480
},
{
"epoch": 0.2,
"learning_rate": 0.0004,
"loss": 0.5733,
"step": 1490
},
{
"epoch": 0.21,
"learning_rate": 0.0004,
"loss": 0.557,
"step": 1500
},
{
"epoch": 0.21,
"learning_rate": 0.0004,
"loss": 0.5737,
"step": 1510
},
{
"epoch": 0.21,
"learning_rate": 0.0004,
"loss": 0.5669,
"step": 1520
},
{
"epoch": 0.21,
"learning_rate": 0.0004,
"loss": 0.5655,
"step": 1530
},
{
"epoch": 0.21,
"learning_rate": 0.0004,
"loss": 0.5851,
"step": 1540
},
{
"epoch": 0.21,
"learning_rate": 0.0004,
"loss": 0.5691,
"step": 1550
},
{
"epoch": 0.21,
"learning_rate": 0.0004,
"loss": 0.5703,
"step": 1560
},
{
"epoch": 0.22,
"learning_rate": 0.0004,
"loss": 0.5765,
"step": 1570
},
{
"epoch": 0.22,
"learning_rate": 0.0004,
"loss": 0.5728,
"step": 1580
},
{
"epoch": 0.22,
"learning_rate": 0.0004,
"loss": 0.5653,
"step": 1590
},
{
"epoch": 0.22,
"learning_rate": 0.0004,
"loss": 0.5577,
"step": 1600
},
{
"epoch": 0.22,
"learning_rate": 0.0004,
"loss": 0.5789,
"step": 1610
},
{
"epoch": 0.22,
"learning_rate": 0.0004,
"loss": 0.5684,
"step": 1620
},
{
"epoch": 0.22,
"learning_rate": 0.0004,
"loss": 0.5595,
"step": 1630
},
{
"epoch": 0.23,
"learning_rate": 0.0004,
"loss": 0.5889,
"step": 1640
},
{
"epoch": 0.23,
"learning_rate": 0.0004,
"loss": 0.5661,
"step": 1650
},
{
"epoch": 0.23,
"learning_rate": 0.0004,
"loss": 0.5541,
"step": 1660
},
{
"epoch": 0.23,
"learning_rate": 0.0004,
"loss": 0.5616,
"step": 1670
},
{
"epoch": 0.23,
"learning_rate": 0.0004,
"loss": 0.565,
"step": 1680
},
{
"epoch": 0.23,
"learning_rate": 0.0004,
"loss": 0.5636,
"step": 1690
},
{
"epoch": 0.23,
"learning_rate": 0.0004,
"loss": 0.5709,
"step": 1700
},
{
"epoch": 0.24,
"learning_rate": 0.0004,
"loss": 0.5428,
"step": 1710
},
{
"epoch": 0.24,
"learning_rate": 0.0004,
"loss": 0.5959,
"step": 1720
},
{
"epoch": 0.24,
"learning_rate": 0.0004,
"loss": 0.5651,
"step": 1730
},
{
"epoch": 0.24,
"learning_rate": 0.0004,
"loss": 0.5485,
"step": 1740
},
{
"epoch": 0.24,
"learning_rate": 0.0004,
"loss": 0.5607,
"step": 1750
},
{
"epoch": 0.24,
"learning_rate": 0.0004,
"loss": 0.5711,
"step": 1760
},
{
"epoch": 0.24,
"learning_rate": 0.0004,
"loss": 0.5536,
"step": 1770
},
{
"epoch": 0.24,
"learning_rate": 0.0004,
"loss": 0.5589,
"step": 1780
},
{
"epoch": 0.25,
"learning_rate": 0.0004,
"loss": 0.5711,
"step": 1790
},
{
"epoch": 0.25,
"learning_rate": 0.0004,
"loss": 0.5448,
"step": 1800
},
{
"epoch": 0.25,
"learning_rate": 0.0004,
"loss": 0.55,
"step": 1810
},
{
"epoch": 0.25,
"learning_rate": 0.0004,
"loss": 0.5519,
"step": 1820
},
{
"epoch": 0.25,
"learning_rate": 0.0004,
"loss": 0.5657,
"step": 1830
},
{
"epoch": 0.25,
"learning_rate": 0.0004,
"loss": 0.558,
"step": 1840
},
{
"epoch": 0.25,
"learning_rate": 0.0004,
"loss": 0.5406,
"step": 1850
},
{
"epoch": 0.26,
"learning_rate": 0.0004,
"loss": 0.5572,
"step": 1860
},
{
"epoch": 0.26,
"learning_rate": 0.0004,
"loss": 0.5391,
"step": 1870
},
{
"epoch": 0.26,
"learning_rate": 0.0004,
"loss": 0.5391,
"step": 1880
},
{
"epoch": 0.26,
"learning_rate": 0.0004,
"loss": 0.5554,
"step": 1890
},
{
"epoch": 0.26,
"learning_rate": 0.0004,
"loss": 0.5667,
"step": 1900
},
{
"epoch": 0.26,
"learning_rate": 0.0004,
"loss": 0.5541,
"step": 1910
},
{
"epoch": 0.26,
"learning_rate": 0.0004,
"loss": 0.5651,
"step": 1920
},
{
"epoch": 0.27,
"learning_rate": 0.0004,
"loss": 0.531,
"step": 1930
},
{
"epoch": 0.27,
"learning_rate": 0.0004,
"loss": 0.5375,
"step": 1940
},
{
"epoch": 0.27,
"learning_rate": 0.0004,
"loss": 0.5323,
"step": 1950
},
{
"epoch": 0.27,
"learning_rate": 0.0004,
"loss": 0.5394,
"step": 1960
},
{
"epoch": 0.27,
"learning_rate": 0.0004,
"loss": 0.5445,
"step": 1970
},
{
"epoch": 0.27,
"learning_rate": 0.0004,
"loss": 0.5217,
"step": 1980
},
{
"epoch": 0.27,
"learning_rate": 0.0004,
"loss": 0.5552,
"step": 1990
},
{
"epoch": 0.28,
"learning_rate": 0.0004,
"loss": 0.5283,
"step": 2000
},
{
"epoch": 0.28,
"learning_rate": 0.0004,
"loss": 0.5494,
"step": 2010
},
{
"epoch": 0.28,
"learning_rate": 0.0004,
"loss": 0.54,
"step": 2020
},
{
"epoch": 0.28,
"learning_rate": 0.0004,
"loss": 0.5343,
"step": 2030
},
{
"epoch": 0.28,
"learning_rate": 0.0004,
"loss": 0.5438,
"step": 2040
},
{
"epoch": 0.28,
"learning_rate": 0.0004,
"loss": 0.5312,
"step": 2050
},
{
"epoch": 0.28,
"learning_rate": 0.0004,
"loss": 0.5381,
"step": 2060
},
{
"epoch": 0.28,
"learning_rate": 0.0004,
"loss": 0.5487,
"step": 2070
},
{
"epoch": 0.29,
"learning_rate": 0.0004,
"loss": 0.5516,
"step": 2080
},
{
"epoch": 0.29,
"learning_rate": 0.0004,
"loss": 0.5437,
"step": 2090
},
{
"epoch": 0.29,
"learning_rate": 0.0004,
"loss": 0.5346,
"step": 2100
},
{
"epoch": 0.29,
"learning_rate": 0.0004,
"loss": 0.5466,
"step": 2110
},
{
"epoch": 0.29,
"learning_rate": 0.0004,
"loss": 0.5629,
"step": 2120
},
{
"epoch": 0.29,
"learning_rate": 0.0004,
"loss": 0.5421,
"step": 2130
},
{
"epoch": 0.29,
"learning_rate": 0.0004,
"loss": 0.5471,
"step": 2140
},
{
"epoch": 0.3,
"learning_rate": 0.0004,
"loss": 0.5287,
"step": 2150
},
{
"epoch": 0.3,
"learning_rate": 0.0004,
"loss": 0.5278,
"step": 2160
},
{
"epoch": 0.3,
"learning_rate": 0.0004,
"loss": 0.5416,
"step": 2170
},
{
"epoch": 0.3,
"learning_rate": 0.0004,
"loss": 0.5411,
"step": 2180
},
{
"epoch": 0.3,
"learning_rate": 0.0004,
"loss": 0.5386,
"step": 2190
},
{
"epoch": 0.3,
"learning_rate": 0.0004,
"loss": 0.5291,
"step": 2200
},
{
"epoch": 0.3,
"learning_rate": 0.0004,
"loss": 0.5298,
"step": 2210
},
{
"epoch": 0.31,
"learning_rate": 0.0004,
"loss": 0.5265,
"step": 2220
},
{
"epoch": 0.31,
"learning_rate": 0.0004,
"loss": 0.5326,
"step": 2230
},
{
"epoch": 0.31,
"learning_rate": 0.0004,
"loss": 0.5283,
"step": 2240
},
{
"epoch": 0.31,
"learning_rate": 0.0004,
"loss": 0.5257,
"step": 2250
},
{
"epoch": 0.31,
"learning_rate": 0.0004,
"loss": 0.5261,
"step": 2260
},
{
"epoch": 0.31,
"learning_rate": 0.0004,
"loss": 0.5243,
"step": 2270
},
{
"epoch": 0.31,
"learning_rate": 0.0004,
"loss": 0.5223,
"step": 2280
},
{
"epoch": 0.32,
"learning_rate": 0.0004,
"loss": 0.5274,
"step": 2290
},
{
"epoch": 0.32,
"learning_rate": 0.0004,
"loss": 0.5287,
"step": 2300
},
{
"epoch": 0.32,
"learning_rate": 0.0004,
"loss": 0.5364,
"step": 2310
},
{
"epoch": 0.32,
"learning_rate": 0.0004,
"loss": 0.5191,
"step": 2320
},
{
"epoch": 0.32,
"learning_rate": 0.0004,
"loss": 0.5405,
"step": 2330
},
{
"epoch": 0.32,
"learning_rate": 0.0004,
"loss": 0.5141,
"step": 2340
},
{
"epoch": 0.32,
"learning_rate": 0.0004,
"loss": 0.5227,
"step": 2350
},
{
"epoch": 0.32,
"learning_rate": 0.0004,
"loss": 0.5287,
"step": 2360
},
{
"epoch": 0.33,
"learning_rate": 0.0004,
"loss": 0.5223,
"step": 2370
},
{
"epoch": 0.33,
"learning_rate": 0.0004,
"loss": 0.5343,
"step": 2380
},
{
"epoch": 0.33,
"learning_rate": 0.0004,
"loss": 0.5299,
"step": 2390
},
{
"epoch": 0.33,
"learning_rate": 0.0004,
"loss": 0.5308,
"step": 2400
},
{
"epoch": 0.33,
"learning_rate": 0.0004,
"loss": 0.5368,
"step": 2410
},
{
"epoch": 0.33,
"learning_rate": 0.0004,
"loss": 0.513,
"step": 2420
},
{
"epoch": 0.33,
"learning_rate": 0.0004,
"loss": 0.5214,
"step": 2430
},
{
"epoch": 0.34,
"learning_rate": 0.0004,
"loss": 0.5253,
"step": 2440
},
{
"epoch": 0.34,
"learning_rate": 0.0004,
"loss": 0.5267,
"step": 2450
},
{
"epoch": 0.34,
"learning_rate": 0.0004,
"loss": 0.5165,
"step": 2460
},
{
"epoch": 0.34,
"learning_rate": 0.0004,
"loss": 0.5337,
"step": 2470
},
{
"epoch": 0.34,
"learning_rate": 0.0004,
"loss": 0.5063,
"step": 2480
},
{
"epoch": 0.34,
"learning_rate": 0.0004,
"loss": 0.5108,
"step": 2490
},
{
"epoch": 0.34,
"learning_rate": 0.0004,
"loss": 0.5269,
"step": 2500
},
{
"epoch": 0.35,
"learning_rate": 0.0004,
"loss": 0.5083,
"step": 2510
},
{
"epoch": 0.35,
"learning_rate": 0.0004,
"loss": 0.51,
"step": 2520
},
{
"epoch": 0.35,
"learning_rate": 0.0004,
"loss": 0.5131,
"step": 2530
},
{
"epoch": 0.35,
"learning_rate": 0.0004,
"loss": 0.53,
"step": 2540
},
{
"epoch": 0.35,
"learning_rate": 0.0004,
"loss": 0.5039,
"step": 2550
},
{
"epoch": 0.35,
"learning_rate": 0.0004,
"loss": 0.5185,
"step": 2560
},
{
"epoch": 0.35,
"learning_rate": 0.0004,
"loss": 0.5178,
"step": 2570
},
{
"epoch": 0.35,
"learning_rate": 0.0004,
"loss": 0.5196,
"step": 2580
},
{
"epoch": 0.36,
"learning_rate": 0.0004,
"loss": 0.5062,
"step": 2590
},
{
"epoch": 0.36,
"learning_rate": 0.0004,
"loss": 0.5194,
"step": 2600
},
{
"epoch": 0.36,
"learning_rate": 0.0004,
"loss": 0.5168,
"step": 2610
},
{
"epoch": 0.36,
"learning_rate": 0.0004,
"loss": 0.5132,
"step": 2620
},
{
"epoch": 0.36,
"learning_rate": 0.0004,
"loss": 0.5064,
"step": 2630
},
{
"epoch": 0.36,
"learning_rate": 0.0004,
"loss": 0.5112,
"step": 2640
},
{
"epoch": 0.36,
"learning_rate": 0.0004,
"loss": 0.5053,
"step": 2650
},
{
"epoch": 0.37,
"learning_rate": 0.0004,
"loss": 0.5218,
"step": 2660
},
{
"epoch": 0.37,
"learning_rate": 0.0004,
"loss": 0.5109,
"step": 2670
},
{
"epoch": 0.37,
"learning_rate": 0.0004,
"loss": 0.5114,
"step": 2680
},
{
"epoch": 0.37,
"learning_rate": 0.0004,
"loss": 0.5102,
"step": 2690
},
{
"epoch": 0.37,
"learning_rate": 0.0004,
"loss": 0.5182,
"step": 2700
},
{
"epoch": 0.37,
"learning_rate": 0.0004,
"loss": 0.5276,
"step": 2710
},
{
"epoch": 0.37,
"learning_rate": 0.0004,
"loss": 0.5218,
"step": 2720
},
{
"epoch": 0.38,
"learning_rate": 0.0004,
"loss": 0.508,
"step": 2730
},
{
"epoch": 0.38,
"learning_rate": 0.0004,
"loss": 0.5229,
"step": 2740
},
{
"epoch": 0.38,
"learning_rate": 0.0004,
"loss": 0.5068,
"step": 2750
},
{
"epoch": 0.38,
"learning_rate": 0.0004,
"loss": 0.5214,
"step": 2760
},
{
"epoch": 0.38,
"learning_rate": 0.0004,
"loss": 0.508,
"step": 2770
},
{
"epoch": 0.38,
"learning_rate": 0.0004,
"loss": 0.5142,
"step": 2780
},
{
"epoch": 0.38,
"learning_rate": 0.0004,
"loss": 0.5039,
"step": 2790
},
{
"epoch": 0.39,
"learning_rate": 0.0004,
"loss": 0.5134,
"step": 2800
},
{
"epoch": 0.39,
"learning_rate": 0.0004,
"loss": 0.4954,
"step": 2810
},
{
"epoch": 0.39,
"learning_rate": 0.0004,
"loss": 0.491,
"step": 2820
},
{
"epoch": 0.39,
"learning_rate": 0.0004,
"loss": 0.4877,
"step": 2830
},
{
"epoch": 0.39,
"learning_rate": 0.0004,
"loss": 0.4993,
"step": 2840
},
{
"epoch": 0.39,
"learning_rate": 0.0004,
"loss": 0.5119,
"step": 2850
},
{
"epoch": 0.39,
"learning_rate": 0.0004,
"loss": 0.5141,
"step": 2860
},
{
"epoch": 0.39,
"learning_rate": 0.0004,
"loss": 0.5003,
"step": 2870
},
{
"epoch": 0.4,
"learning_rate": 0.0004,
"loss": 0.5036,
"step": 2880
},
{
"epoch": 0.4,
"learning_rate": 0.0004,
"loss": 0.5,
"step": 2890
},
{
"epoch": 0.4,
"learning_rate": 0.0004,
"loss": 0.5036,
"step": 2900
},
{
"epoch": 0.4,
"learning_rate": 0.0004,
"loss": 0.4968,
"step": 2910
},
{
"epoch": 0.4,
"learning_rate": 0.0004,
"loss": 0.4959,
"step": 2920
},
{
"epoch": 0.4,
"learning_rate": 0.0004,
"loss": 0.5062,
"step": 2930
},
{
"epoch": 0.4,
"learning_rate": 0.0004,
"loss": 0.5054,
"step": 2940
},
{
"epoch": 0.41,
"learning_rate": 0.0004,
"loss": 0.4962,
"step": 2950
},
{
"epoch": 0.41,
"learning_rate": 0.0004,
"loss": 0.4932,
"step": 2960
},
{
"epoch": 0.41,
"learning_rate": 0.0004,
"loss": 0.5115,
"step": 2970
},
{
"epoch": 0.41,
"learning_rate": 0.0004,
"loss": 0.5008,
"step": 2980
},
{
"epoch": 0.41,
"learning_rate": 0.0004,
"loss": 0.5058,
"step": 2990
},
{
"epoch": 0.41,
"learning_rate": 0.0004,
"loss": 0.4981,
"step": 3000
},
{
"epoch": 0.41,
"learning_rate": 0.0004,
"loss": 0.5045,
"step": 3010
},
{
"epoch": 0.42,
"learning_rate": 0.0004,
"loss": 0.5002,
"step": 3020
},
{
"epoch": 0.42,
"learning_rate": 0.0004,
"loss": 0.5061,
"step": 3030
},
{
"epoch": 0.42,
"learning_rate": 0.0004,
"loss": 0.5009,
"step": 3040
},
{
"epoch": 0.42,
"learning_rate": 0.0004,
"loss": 0.5156,
"step": 3050
},
{
"epoch": 0.42,
"learning_rate": 0.0004,
"loss": 0.4986,
"step": 3060
},
{
"epoch": 0.42,
"learning_rate": 0.0004,
"loss": 0.5034,
"step": 3070
},
{
"epoch": 0.42,
"learning_rate": 0.0004,
"loss": 0.5005,
"step": 3080
},
{
"epoch": 0.43,
"learning_rate": 0.0004,
"loss": 0.4729,
"step": 3090
},
{
"epoch": 0.43,
"learning_rate": 0.0004,
"loss": 0.4896,
"step": 3100
},
{
"epoch": 0.43,
"learning_rate": 0.0004,
"loss": 0.4693,
"step": 3110
},
{
"epoch": 0.43,
"learning_rate": 0.0004,
"loss": 0.4824,
"step": 3120
},
{
"epoch": 0.43,
"learning_rate": 0.0004,
"loss": 0.4802,
"step": 3130
},
{
"epoch": 0.43,
"learning_rate": 0.0004,
"loss": 0.4791,
"step": 3140
},
{
"epoch": 0.43,
"learning_rate": 0.0004,
"loss": 0.5013,
"step": 3150
},
{
"epoch": 0.43,
"learning_rate": 0.0004,
"loss": 0.4808,
"step": 3160
},
{
"epoch": 0.44,
"learning_rate": 0.0004,
"loss": 0.4947,
"step": 3170
},
{
"epoch": 0.44,
"learning_rate": 0.0004,
"loss": 0.489,
"step": 3180
},
{
"epoch": 0.44,
"learning_rate": 0.0004,
"loss": 0.4946,
"step": 3190
},
{
"epoch": 0.44,
"learning_rate": 0.0004,
"loss": 0.4866,
"step": 3200
},
{
"epoch": 0.44,
"learning_rate": 0.0004,
"loss": 0.49,
"step": 3210
},
{
"epoch": 0.44,
"learning_rate": 0.0004,
"loss": 0.4861,
"step": 3220
},
{
"epoch": 0.44,
"learning_rate": 0.0004,
"loss": 0.4778,
"step": 3230
},
{
"epoch": 0.45,
"learning_rate": 0.0004,
"loss": 0.5071,
"step": 3240
},
{
"epoch": 0.45,
"learning_rate": 0.0004,
"loss": 0.4665,
"step": 3250
},
{
"epoch": 0.45,
"learning_rate": 0.0004,
"loss": 0.4867,
"step": 3260
},
{
"epoch": 0.45,
"learning_rate": 0.0004,
"loss": 0.488,
"step": 3270
},
{
"epoch": 0.45,
"learning_rate": 0.0004,
"loss": 0.4816,
"step": 3280
},
{
"epoch": 0.45,
"learning_rate": 0.0004,
"loss": 0.4802,
"step": 3290
},
{
"epoch": 0.45,
"learning_rate": 0.0004,
"loss": 0.5,
"step": 3300
},
{
"epoch": 0.46,
"learning_rate": 0.0004,
"loss": 0.4854,
"step": 3310
},
{
"epoch": 0.46,
"learning_rate": 0.0004,
"loss": 0.4913,
"step": 3320
},
{
"epoch": 0.46,
"learning_rate": 0.0004,
"loss": 0.489,
"step": 3330
},
{
"epoch": 0.46,
"learning_rate": 0.0004,
"loss": 0.4834,
"step": 3340
},
{
"epoch": 0.46,
"learning_rate": 0.0004,
"loss": 0.4828,
"step": 3350
},
{
"epoch": 0.46,
"learning_rate": 0.0004,
"loss": 0.4902,
"step": 3360
},
{
"epoch": 0.46,
"learning_rate": 0.0004,
"loss": 0.4873,
"step": 3370
},
{
"epoch": 0.47,
"learning_rate": 0.0004,
"loss": 0.4932,
"step": 3380
},
{
"epoch": 0.47,
"learning_rate": 0.0004,
"loss": 0.463,
"step": 3390
},
{
"epoch": 0.47,
"learning_rate": 0.0004,
"loss": 0.4778,
"step": 3400
},
{
"epoch": 0.47,
"learning_rate": 0.0004,
"loss": 0.4772,
"step": 3410
},
{
"epoch": 0.47,
"learning_rate": 0.0004,
"loss": 0.4492,
"step": 3420
},
{
"epoch": 0.47,
"learning_rate": 0.0004,
"loss": 0.4661,
"step": 3430
},
{
"epoch": 0.47,
"learning_rate": 0.0004,
"loss": 0.4449,
"step": 3440
},
{
"epoch": 0.47,
"learning_rate": 0.0004,
"loss": 0.4912,
"step": 3450
},
{
"epoch": 0.48,
"learning_rate": 0.0004,
"loss": 0.4706,
"step": 3460
},
{
"epoch": 0.48,
"learning_rate": 0.0004,
"loss": 0.4661,
"step": 3470
},
{
"epoch": 0.48,
"learning_rate": 0.0004,
"loss": 0.464,
"step": 3480
},
{
"epoch": 0.48,
"learning_rate": 0.0004,
"loss": 0.48,
"step": 3490
},
{
"epoch": 0.48,
"learning_rate": 0.0004,
"loss": 0.4788,
"step": 3500
},
{
"epoch": 0.48,
"learning_rate": 0.0004,
"loss": 0.4799,
"step": 3510
},
{
"epoch": 0.48,
"learning_rate": 0.0004,
"loss": 0.4778,
"step": 3520
},
{
"epoch": 0.49,
"learning_rate": 0.0004,
"loss": 0.459,
"step": 3530
},
{
"epoch": 0.49,
"learning_rate": 0.0004,
"loss": 0.4604,
"step": 3540
},
{
"epoch": 0.49,
"learning_rate": 0.0004,
"loss": 0.4863,
"step": 3550
},
{
"epoch": 0.49,
"learning_rate": 0.0004,
"loss": 0.486,
"step": 3560
},
{
"epoch": 0.49,
"learning_rate": 0.0004,
"loss": 0.4713,
"step": 3570
},
{
"epoch": 0.49,
"learning_rate": 0.0004,
"loss": 0.4648,
"step": 3580
},
{
"epoch": 0.49,
"learning_rate": 0.0004,
"loss": 0.4807,
"step": 3590
},
{
"epoch": 0.5,
"learning_rate": 0.0004,
"loss": 0.479,
"step": 3600
},
{
"epoch": 0.5,
"learning_rate": 0.0004,
"loss": 0.4594,
"step": 3610
},
{
"epoch": 0.5,
"learning_rate": 0.0004,
"loss": 0.4709,
"step": 3620
},
{
"epoch": 0.5,
"learning_rate": 0.0004,
"loss": 0.4802,
"step": 3630
},
{
"epoch": 0.5,
"learning_rate": 0.0004,
"loss": 0.4538,
"step": 3640
},
{
"epoch": 0.5,
"learning_rate": 0.0004,
"loss": 0.4933,
"step": 3650
},
{
"epoch": 0.5,
"learning_rate": 0.0004,
"loss": 0.473,
"step": 3660
},
{
"epoch": 0.5,
"learning_rate": 0.0004,
"loss": 0.469,
"step": 3670
},
{
"epoch": 0.51,
"learning_rate": 0.0004,
"loss": 0.4627,
"step": 3680
},
{
"epoch": 0.51,
"learning_rate": 0.0004,
"loss": 0.4569,
"step": 3690
},
{
"epoch": 0.51,
"learning_rate": 0.0004,
"loss": 0.4532,
"step": 3700
},
{
"epoch": 0.51,
"learning_rate": 0.0004,
"loss": 0.4878,
"step": 3710
},
{
"epoch": 0.51,
"learning_rate": 0.0004,
"loss": 0.4952,
"step": 3720
},
{
"epoch": 0.51,
"learning_rate": 0.0004,
"loss": 0.4732,
"step": 3730
},
{
"epoch": 0.51,
"learning_rate": 0.0004,
"loss": 0.4654,
"step": 3740
},
{
"epoch": 0.52,
"learning_rate": 0.0004,
"loss": 0.4675,
"step": 3750
},
{
"epoch": 0.52,
"learning_rate": 0.0004,
"loss": 0.4659,
"step": 3760
},
{
"epoch": 0.52,
"learning_rate": 0.0004,
"loss": 0.4491,
"step": 3770
},
{
"epoch": 0.52,
"learning_rate": 0.0004,
"loss": 0.4652,
"step": 3780
},
{
"epoch": 0.52,
"learning_rate": 0.0004,
"loss": 0.4742,
"step": 3790
},
{
"epoch": 0.52,
"learning_rate": 0.0004,
"loss": 0.4626,
"step": 3800
},
{
"epoch": 0.52,
"learning_rate": 0.0004,
"loss": 0.4744,
"step": 3810
},
{
"epoch": 0.53,
"learning_rate": 0.0004,
"loss": 0.4704,
"step": 3820
},
{
"epoch": 0.53,
"learning_rate": 0.0004,
"loss": 0.4792,
"step": 3830
},
{
"epoch": 0.53,
"learning_rate": 0.0004,
"loss": 0.469,
"step": 3840
},
{
"epoch": 0.53,
"learning_rate": 0.0004,
"loss": 0.4749,
"step": 3850
},
{
"epoch": 0.53,
"learning_rate": 0.0004,
"loss": 0.4645,
"step": 3860
},
{
"epoch": 0.53,
"learning_rate": 0.0004,
"loss": 0.4868,
"step": 3870
},
{
"epoch": 0.53,
"learning_rate": 0.0004,
"loss": 0.4618,
"step": 3880
},
{
"epoch": 0.54,
"learning_rate": 0.0004,
"loss": 0.4818,
"step": 3890
},
{
"epoch": 0.54,
"learning_rate": 0.0004,
"loss": 0.4687,
"step": 3900
},
{
"epoch": 0.54,
"learning_rate": 0.0004,
"loss": 0.4569,
"step": 3910
},
{
"epoch": 0.54,
"learning_rate": 0.0004,
"loss": 0.449,
"step": 3920
},
{
"epoch": 0.54,
"learning_rate": 0.0004,
"loss": 0.4807,
"step": 3930
},
{
"epoch": 0.54,
"learning_rate": 0.0004,
"loss": 0.4402,
"step": 3940
},
{
"epoch": 0.54,
"learning_rate": 0.0004,
"loss": 0.4594,
"step": 3950
},
{
"epoch": 0.54,
"learning_rate": 0.0004,
"loss": 0.4603,
"step": 3960
},
{
"epoch": 0.55,
"learning_rate": 0.0004,
"loss": 0.4565,
"step": 3970
},
{
"epoch": 0.55,
"learning_rate": 0.0004,
"loss": 0.4623,
"step": 3980
},
{
"epoch": 0.55,
"learning_rate": 0.0004,
"loss": 0.4569,
"step": 3990
},
{
"epoch": 0.55,
"learning_rate": 0.0004,
"loss": 0.4534,
"step": 4000
},
{
"epoch": 0.55,
"learning_rate": 0.0004,
"loss": 0.452,
"step": 4010
},
{
"epoch": 0.55,
"learning_rate": 0.0004,
"loss": 0.4521,
"step": 4020
},
{
"epoch": 0.55,
"learning_rate": 0.0004,
"loss": 0.4402,
"step": 4030
},
{
"epoch": 0.56,
"learning_rate": 0.0004,
"loss": 0.4299,
"step": 4040
},
{
"epoch": 0.56,
"learning_rate": 0.0004,
"loss": 0.459,
"step": 4050
},
{
"epoch": 0.56,
"learning_rate": 0.0004,
"loss": 0.4318,
"step": 4060
},
{
"epoch": 0.56,
"learning_rate": 0.0004,
"loss": 0.4625,
"step": 4070
},
{
"epoch": 0.56,
"learning_rate": 0.0004,
"loss": 0.4628,
"step": 4080
},
{
"epoch": 0.56,
"learning_rate": 0.0004,
"loss": 0.4486,
"step": 4090
},
{
"epoch": 0.56,
"learning_rate": 0.0004,
"loss": 0.4581,
"step": 4100
},
{
"epoch": 0.57,
"learning_rate": 0.0004,
"loss": 0.4445,
"step": 4110
},
{
"epoch": 0.57,
"learning_rate": 0.0004,
"loss": 0.453,
"step": 4120
},
{
"epoch": 0.57,
"learning_rate": 0.0004,
"loss": 0.4401,
"step": 4130
},
{
"epoch": 0.57,
"learning_rate": 0.0004,
"loss": 0.4447,
"step": 4140
},
{
"epoch": 0.57,
"learning_rate": 0.0004,
"loss": 0.4488,
"step": 4150
},
{
"epoch": 0.57,
"learning_rate": 0.0004,
"loss": 0.4409,
"step": 4160
},
{
"epoch": 0.57,
"learning_rate": 0.0004,
"loss": 0.4471,
"step": 4170
},
{
"epoch": 0.58,
"learning_rate": 0.0004,
"loss": 0.4484,
"step": 4180
},
{
"epoch": 0.58,
"learning_rate": 0.0004,
"loss": 0.4525,
"step": 4190
},
{
"epoch": 0.58,
"learning_rate": 0.0004,
"loss": 0.4549,
"step": 4200
},
{
"epoch": 0.58,
"learning_rate": 0.0004,
"loss": 0.4514,
"step": 4210
},
{
"epoch": 0.58,
"learning_rate": 0.0004,
"loss": 0.4276,
"step": 4220
},
{
"epoch": 0.58,
"learning_rate": 0.0004,
"loss": 0.4476,
"step": 4230
},
{
"epoch": 0.58,
"learning_rate": 0.0004,
"loss": 0.4536,
"step": 4240
},
{
"epoch": 0.58,
"learning_rate": 0.0004,
"loss": 0.4376,
"step": 4250
},
{
"epoch": 0.59,
"learning_rate": 0.0004,
"loss": 0.4452,
"step": 4260
},
{
"epoch": 0.59,
"learning_rate": 0.0004,
"loss": 0.4404,
"step": 4270
},
{
"epoch": 0.59,
"learning_rate": 0.0004,
"loss": 0.4496,
"step": 4280
},
{
"epoch": 0.59,
"learning_rate": 0.0004,
"loss": 0.4488,
"step": 4290
},
{
"epoch": 0.59,
"learning_rate": 0.0004,
"loss": 0.4623,
"step": 4300
},
{
"epoch": 0.59,
"learning_rate": 0.0004,
"loss": 0.4517,
"step": 4310
},
{
"epoch": 0.59,
"learning_rate": 0.0004,
"loss": 0.4385,
"step": 4320
},
{
"epoch": 0.6,
"learning_rate": 0.0004,
"loss": 0.4544,
"step": 4330
},
{
"epoch": 0.6,
"learning_rate": 0.0004,
"loss": 0.441,
"step": 4340
},
{
"epoch": 0.6,
"learning_rate": 0.0004,
"loss": 0.4279,
"step": 4350
},
{
"epoch": 0.6,
"learning_rate": 0.0004,
"loss": 0.4278,
"step": 4360
},
{
"epoch": 0.6,
"learning_rate": 0.0004,
"loss": 0.4585,
"step": 4370
},
{
"epoch": 0.6,
"learning_rate": 0.0004,
"loss": 0.4484,
"step": 4380
},
{
"epoch": 0.6,
"learning_rate": 0.0004,
"loss": 0.4371,
"step": 4390
},
{
"epoch": 0.61,
"learning_rate": 0.0004,
"loss": 0.4397,
"step": 4400
},
{
"epoch": 0.61,
"learning_rate": 0.0004,
"loss": 0.44,
"step": 4410
},
{
"epoch": 0.61,
"learning_rate": 0.0004,
"loss": 0.4372,
"step": 4420
},
{
"epoch": 0.61,
"learning_rate": 0.0004,
"loss": 0.4462,
"step": 4430
},
{
"epoch": 0.61,
"learning_rate": 0.0004,
"loss": 0.4464,
"step": 4440
},
{
"epoch": 0.61,
"learning_rate": 0.0004,
"loss": 0.4514,
"step": 4450
},
{
"epoch": 0.61,
"learning_rate": 0.0004,
"loss": 0.4286,
"step": 4460
},
{
"epoch": 0.61,
"learning_rate": 0.0004,
"loss": 0.4445,
"step": 4470
},
{
"epoch": 0.62,
"learning_rate": 0.0004,
"loss": 0.4471,
"step": 4480
},
{
"epoch": 0.62,
"learning_rate": 0.0004,
"loss": 0.4378,
"step": 4490
},
{
"epoch": 0.62,
"learning_rate": 0.0004,
"loss": 0.448,
"step": 4500
},
{
"epoch": 0.62,
"learning_rate": 0.0004,
"loss": 0.446,
"step": 4510
},
{
"epoch": 0.62,
"learning_rate": 0.0004,
"loss": 0.4214,
"step": 4520
},
{
"epoch": 0.62,
"learning_rate": 0.0004,
"loss": 0.4332,
"step": 4530
},
{
"epoch": 0.62,
"learning_rate": 0.0004,
"loss": 0.4458,
"step": 4540
},
{
"epoch": 0.63,
"learning_rate": 0.0004,
"loss": 0.4522,
"step": 4550
},
{
"epoch": 0.63,
"learning_rate": 0.0004,
"loss": 0.4438,
"step": 4560
},
{
"epoch": 0.63,
"learning_rate": 0.0004,
"loss": 0.4361,
"step": 4570
},
{
"epoch": 0.63,
"learning_rate": 0.0004,
"loss": 0.4295,
"step": 4580
},
{
"epoch": 0.63,
"learning_rate": 0.0004,
"loss": 0.4161,
"step": 4590
},
{
"epoch": 0.63,
"learning_rate": 0.0004,
"loss": 0.4382,
"step": 4600
},
{
"epoch": 0.63,
"learning_rate": 0.0004,
"loss": 0.4356,
"step": 4610
},
{
"epoch": 0.64,
"learning_rate": 0.0004,
"loss": 0.4395,
"step": 4620
},
{
"epoch": 0.64,
"learning_rate": 0.0004,
"loss": 0.4427,
"step": 4630
},
{
"epoch": 0.64,
"learning_rate": 0.0004,
"loss": 0.4399,
"step": 4640
},
{
"epoch": 0.64,
"learning_rate": 0.0004,
"loss": 0.4292,
"step": 4650
},
{
"epoch": 0.64,
"learning_rate": 0.0004,
"loss": 0.4216,
"step": 4660
},
{
"epoch": 0.64,
"learning_rate": 0.0004,
"loss": 0.4364,
"step": 4670
},
{
"epoch": 0.64,
"learning_rate": 0.0004,
"loss": 0.4364,
"step": 4680
},
{
"epoch": 0.65,
"learning_rate": 0.0004,
"loss": 0.4291,
"step": 4690
},
{
"epoch": 0.65,
"learning_rate": 0.0004,
"loss": 0.4112,
"step": 4700
},
{
"epoch": 0.65,
"learning_rate": 0.0004,
"loss": 0.4398,
"step": 4710
},
{
"epoch": 0.65,
"learning_rate": 0.0004,
"loss": 0.4542,
"step": 4720
},
{
"epoch": 0.65,
"learning_rate": 0.0004,
"loss": 0.4315,
"step": 4730
},
{
"epoch": 0.65,
"learning_rate": 0.0004,
"loss": 0.4526,
"step": 4740
},
{
"epoch": 0.65,
"learning_rate": 0.0004,
"loss": 0.4333,
"step": 4750
},
{
"epoch": 0.65,
"learning_rate": 0.0004,
"loss": 0.4301,
"step": 4760
},
{
"epoch": 0.66,
"learning_rate": 0.0004,
"loss": 0.4304,
"step": 4770
},
{
"epoch": 0.66,
"learning_rate": 0.0004,
"loss": 0.429,
"step": 4780
},
{
"epoch": 0.66,
"learning_rate": 0.0004,
"loss": 0.4361,
"step": 4790
},
{
"epoch": 0.66,
"learning_rate": 0.0004,
"loss": 0.4303,
"step": 4800
},
{
"epoch": 0.66,
"learning_rate": 0.0004,
"loss": 0.4377,
"step": 4810
},
{
"epoch": 0.66,
"learning_rate": 0.0004,
"loss": 0.4356,
"step": 4820
},
{
"epoch": 0.66,
"learning_rate": 0.0004,
"loss": 0.4177,
"step": 4830
},
{
"epoch": 0.67,
"learning_rate": 0.0004,
"loss": 0.4202,
"step": 4840
},
{
"epoch": 0.67,
"learning_rate": 0.0004,
"loss": 0.4351,
"step": 4850
},
{
"epoch": 0.67,
"learning_rate": 0.0004,
"loss": 0.4375,
"step": 4860
},
{
"epoch": 0.67,
"learning_rate": 0.0004,
"loss": 0.4329,
"step": 4870
},
{
"epoch": 0.67,
"learning_rate": 0.0004,
"loss": 0.4275,
"step": 4880
},
{
"epoch": 0.67,
"learning_rate": 0.0004,
"loss": 0.4257,
"step": 4890
},
{
"epoch": 0.67,
"learning_rate": 0.0004,
"loss": 0.4377,
"step": 4900
},
{
"epoch": 0.68,
"learning_rate": 0.0004,
"loss": 0.4393,
"step": 4910
},
{
"epoch": 0.68,
"learning_rate": 0.0004,
"loss": 0.4121,
"step": 4920
},
{
"epoch": 0.68,
"learning_rate": 0.0004,
"loss": 0.4248,
"step": 4930
},
{
"epoch": 0.68,
"learning_rate": 0.0004,
"loss": 0.4161,
"step": 4940
},
{
"epoch": 0.68,
"learning_rate": 0.0004,
"loss": 0.4109,
"step": 4950
},
{
"epoch": 0.68,
"learning_rate": 0.0004,
"loss": 0.4148,
"step": 4960
},
{
"epoch": 0.68,
"learning_rate": 0.0004,
"loss": 0.429,
"step": 4970
},
{
"epoch": 0.69,
"learning_rate": 0.0004,
"loss": 0.4151,
"step": 4980
},
{
"epoch": 0.69,
"learning_rate": 0.0004,
"loss": 0.4164,
"step": 4990
},
{
"epoch": 0.69,
"learning_rate": 0.0004,
"loss": 0.3987,
"step": 5000
},
{
"epoch": 0.69,
"learning_rate": 0.0004,
"loss": 0.4115,
"step": 5010
},
{
"epoch": 0.69,
"learning_rate": 0.0004,
"loss": 0.4095,
"step": 5020
},
{
"epoch": 0.69,
"learning_rate": 0.0004,
"loss": 0.413,
"step": 5030
},
{
"epoch": 0.69,
"learning_rate": 0.0004,
"loss": 0.418,
"step": 5040
},
{
"epoch": 0.69,
"learning_rate": 0.0004,
"loss": 0.4175,
"step": 5050
},
{
"epoch": 0.7,
"learning_rate": 0.0004,
"loss": 0.413,
"step": 5060
},
{
"epoch": 0.7,
"learning_rate": 0.0004,
"loss": 0.4136,
"step": 5070
},
{
"epoch": 0.7,
"learning_rate": 0.0004,
"loss": 0.4158,
"step": 5080
},
{
"epoch": 0.7,
"learning_rate": 0.0004,
"loss": 0.4082,
"step": 5090
},
{
"epoch": 0.7,
"learning_rate": 0.0004,
"loss": 0.4173,
"step": 5100
},
{
"epoch": 0.7,
"learning_rate": 0.0004,
"loss": 0.4239,
"step": 5110
},
{
"epoch": 0.7,
"learning_rate": 0.0004,
"loss": 0.4106,
"step": 5120
},
{
"epoch": 0.71,
"learning_rate": 0.0004,
"loss": 0.3956,
"step": 5130
},
{
"epoch": 0.71,
"learning_rate": 0.0004,
"loss": 0.4143,
"step": 5140
},
{
"epoch": 0.71,
"learning_rate": 0.0004,
"loss": 0.3961,
"step": 5150
},
{
"epoch": 0.71,
"learning_rate": 0.0004,
"loss": 0.41,
"step": 5160
},
{
"epoch": 0.71,
"learning_rate": 0.0004,
"loss": 0.4221,
"step": 5170
},
{
"epoch": 0.71,
"learning_rate": 0.0004,
"loss": 0.4184,
"step": 5180
},
{
"epoch": 0.71,
"learning_rate": 0.0004,
"loss": 0.4102,
"step": 5190
},
{
"epoch": 0.72,
"learning_rate": 0.0004,
"loss": 0.4359,
"step": 5200
},
{
"epoch": 0.72,
"learning_rate": 0.0004,
"loss": 0.4189,
"step": 5210
},
{
"epoch": 0.72,
"learning_rate": 0.0004,
"loss": 0.4039,
"step": 5220
},
{
"epoch": 0.72,
"learning_rate": 0.0004,
"loss": 0.3979,
"step": 5230
},
{
"epoch": 0.72,
"learning_rate": 0.0004,
"loss": 0.4086,
"step": 5240
},
{
"epoch": 0.72,
"learning_rate": 0.0004,
"loss": 0.3958,
"step": 5250
},
{
"epoch": 0.72,
"learning_rate": 0.0004,
"loss": 0.4079,
"step": 5260
},
{
"epoch": 0.73,
"learning_rate": 0.0004,
"loss": 0.3966,
"step": 5270
},
{
"epoch": 0.73,
"learning_rate": 0.0004,
"loss": 0.4035,
"step": 5280
},
{
"epoch": 0.73,
"learning_rate": 0.0004,
"loss": 0.4007,
"step": 5290
},
{
"epoch": 0.73,
"learning_rate": 0.0004,
"loss": 0.3951,
"step": 5300
},
{
"epoch": 0.73,
"learning_rate": 0.0004,
"loss": 0.42,
"step": 5310
},
{
"epoch": 0.73,
"learning_rate": 0.0004,
"loss": 0.3973,
"step": 5320
},
{
"epoch": 0.73,
"learning_rate": 0.0004,
"loss": 0.4081,
"step": 5330
},
{
"epoch": 0.73,
"learning_rate": 0.0004,
"loss": 0.4195,
"step": 5340
},
{
"epoch": 0.74,
"learning_rate": 0.0004,
"loss": 0.4029,
"step": 5350
},
{
"epoch": 0.74,
"learning_rate": 0.0004,
"loss": 0.4142,
"step": 5360
},
{
"epoch": 0.74,
"learning_rate": 0.0004,
"loss": 0.4101,
"step": 5370
},
{
"epoch": 0.74,
"learning_rate": 0.0004,
"loss": 0.4088,
"step": 5380
},
{
"epoch": 0.74,
"learning_rate": 0.0004,
"loss": 0.406,
"step": 5390
},
{
"epoch": 0.74,
"learning_rate": 0.0004,
"loss": 0.4087,
"step": 5400
},
{
"epoch": 0.74,
"learning_rate": 0.0004,
"loss": 0.4079,
"step": 5410
},
{
"epoch": 0.75,
"learning_rate": 0.0004,
"loss": 0.416,
"step": 5420
},
{
"epoch": 0.75,
"learning_rate": 0.0004,
"loss": 0.3947,
"step": 5430
},
{
"epoch": 0.75,
"learning_rate": 0.0004,
"loss": 0.4028,
"step": 5440
},
{
"epoch": 0.75,
"learning_rate": 0.0004,
"loss": 0.4037,
"step": 5450
},
{
"epoch": 0.75,
"learning_rate": 0.0004,
"loss": 0.391,
"step": 5460
},
{
"epoch": 0.75,
"learning_rate": 0.0004,
"loss": 0.3859,
"step": 5470
},
{
"epoch": 0.75,
"learning_rate": 0.0004,
"loss": 0.4112,
"step": 5480
},
{
"epoch": 0.76,
"learning_rate": 0.0004,
"loss": 0.4048,
"step": 5490
},
{
"epoch": 0.76,
"learning_rate": 0.0004,
"loss": 0.4303,
"step": 5500
},
{
"epoch": 0.76,
"learning_rate": 0.0004,
"loss": 0.3899,
"step": 5510
},
{
"epoch": 0.76,
"learning_rate": 0.0004,
"loss": 0.4212,
"step": 5520
},
{
"epoch": 0.76,
"learning_rate": 0.0004,
"loss": 0.399,
"step": 5530
},
{
"epoch": 0.76,
"learning_rate": 0.0004,
"loss": 0.3984,
"step": 5540
},
{
"epoch": 0.76,
"learning_rate": 0.0004,
"loss": 0.3946,
"step": 5550
},
{
"epoch": 0.76,
"learning_rate": 0.0004,
"loss": 0.4077,
"step": 5560
},
{
"epoch": 0.77,
"learning_rate": 0.0004,
"loss": 0.4097,
"step": 5570
},
{
"epoch": 0.77,
"learning_rate": 0.0004,
"loss": 0.3956,
"step": 5580
},
{
"epoch": 0.77,
"learning_rate": 0.0004,
"loss": 0.3967,
"step": 5590
},
{
"epoch": 0.77,
"learning_rate": 0.0004,
"loss": 0.4002,
"step": 5600
},
{
"epoch": 0.77,
"learning_rate": 0.0004,
"loss": 0.4071,
"step": 5610
},
{
"epoch": 0.77,
"learning_rate": 0.0004,
"loss": 0.405,
"step": 5620
},
{
"epoch": 0.77,
"learning_rate": 0.0004,
"loss": 0.3984,
"step": 5630
},
{
"epoch": 0.78,
"learning_rate": 0.0004,
"loss": 0.3926,
"step": 5640
},
{
"epoch": 0.78,
"learning_rate": 0.0004,
"loss": 0.3949,
"step": 5650
},
{
"epoch": 0.78,
"learning_rate": 0.0004,
"loss": 0.39,
"step": 5660
},
{
"epoch": 0.78,
"learning_rate": 0.0004,
"loss": 0.4172,
"step": 5670
},
{
"epoch": 0.78,
"learning_rate": 0.0004,
"loss": 0.4028,
"step": 5680
},
{
"epoch": 0.78,
"learning_rate": 0.0004,
"loss": 0.4171,
"step": 5690
},
{
"epoch": 0.78,
"learning_rate": 0.0004,
"loss": 0.409,
"step": 5700
},
{
"epoch": 0.79,
"learning_rate": 0.0004,
"loss": 0.4144,
"step": 5710
},
{
"epoch": 0.79,
"learning_rate": 0.0004,
"loss": 0.4068,
"step": 5720
},
{
"epoch": 0.79,
"learning_rate": 0.0004,
"loss": 0.3971,
"step": 5730
},
{
"epoch": 0.79,
"learning_rate": 0.0004,
"loss": 0.3931,
"step": 5740
},
{
"epoch": 0.79,
"learning_rate": 0.0004,
"loss": 0.3818,
"step": 5750
},
{
"epoch": 0.79,
"learning_rate": 0.0004,
"loss": 0.4239,
"step": 5760
},
{
"epoch": 0.79,
"learning_rate": 0.0004,
"loss": 0.401,
"step": 5770
},
{
"epoch": 0.8,
"learning_rate": 0.0004,
"loss": 0.4068,
"step": 5780
},
{
"epoch": 0.8,
"learning_rate": 0.0004,
"loss": 0.4111,
"step": 5790
},
{
"epoch": 0.8,
"learning_rate": 0.0004,
"loss": 0.3967,
"step": 5800
},
{
"epoch": 0.8,
"learning_rate": 0.0004,
"loss": 0.4062,
"step": 5810
},
{
"epoch": 0.8,
"learning_rate": 0.0004,
"loss": 0.3905,
"step": 5820
},
{
"epoch": 0.8,
"learning_rate": 0.0004,
"loss": 0.3924,
"step": 5830
},
{
"epoch": 0.8,
"learning_rate": 0.0004,
"loss": 0.4007,
"step": 5840
},
{
"epoch": 0.8,
"learning_rate": 0.0004,
"loss": 0.3961,
"step": 5850
},
{
"epoch": 0.81,
"learning_rate": 0.0004,
"loss": 0.4056,
"step": 5860
},
{
"epoch": 0.81,
"learning_rate": 0.0004,
"loss": 0.3984,
"step": 5870
},
{
"epoch": 0.81,
"learning_rate": 0.0004,
"loss": 0.3807,
"step": 5880
},
{
"epoch": 0.81,
"learning_rate": 0.0004,
"loss": 0.3842,
"step": 5890
},
{
"epoch": 0.81,
"learning_rate": 0.0004,
"loss": 0.3837,
"step": 5900
},
{
"epoch": 0.81,
"learning_rate": 0.0004,
"loss": 0.3759,
"step": 5910
},
{
"epoch": 0.81,
"learning_rate": 0.0004,
"loss": 0.3961,
"step": 5920
},
{
"epoch": 0.82,
"learning_rate": 0.0004,
"loss": 0.3948,
"step": 5930
},
{
"epoch": 0.82,
"learning_rate": 0.0004,
"loss": 0.369,
"step": 5940
},
{
"epoch": 0.82,
"learning_rate": 0.0004,
"loss": 0.3859,
"step": 5950
},
{
"epoch": 0.82,
"learning_rate": 0.0004,
"loss": 0.4035,
"step": 5960
},
{
"epoch": 0.82,
"learning_rate": 0.0004,
"loss": 0.3937,
"step": 5970
},
{
"epoch": 0.82,
"learning_rate": 0.0004,
"loss": 0.3964,
"step": 5980
},
{
"epoch": 0.82,
"learning_rate": 0.0004,
"loss": 0.3868,
"step": 5990
},
{
"epoch": 0.83,
"learning_rate": 0.0004,
"loss": 0.3884,
"step": 6000
},
{
"epoch": 0.83,
"learning_rate": 0.0004,
"loss": 0.3758,
"step": 6010
},
{
"epoch": 0.83,
"learning_rate": 0.0004,
"loss": 0.4014,
"step": 6020
},
{
"epoch": 0.83,
"learning_rate": 0.0004,
"loss": 0.3892,
"step": 6030
},
{
"epoch": 0.83,
"learning_rate": 0.0004,
"loss": 0.3872,
"step": 6040
},
{
"epoch": 0.83,
"learning_rate": 0.0004,
"loss": 0.3921,
"step": 6050
},
{
"epoch": 0.83,
"learning_rate": 0.0004,
"loss": 0.3856,
"step": 6060
},
{
"epoch": 0.84,
"learning_rate": 0.0004,
"loss": 0.3942,
"step": 6070
},
{
"epoch": 0.84,
"learning_rate": 0.0004,
"loss": 0.3841,
"step": 6080
},
{
"epoch": 0.84,
"learning_rate": 0.0004,
"loss": 0.3757,
"step": 6090
},
{
"epoch": 0.84,
"learning_rate": 0.0004,
"loss": 0.3954,
"step": 6100
},
{
"epoch": 0.84,
"learning_rate": 0.0004,
"loss": 0.3788,
"step": 6110
},
{
"epoch": 0.84,
"learning_rate": 0.0004,
"loss": 0.4091,
"step": 6120
},
{
"epoch": 0.84,
"learning_rate": 0.0004,
"loss": 0.4002,
"step": 6130
},
{
"epoch": 0.84,
"learning_rate": 0.0004,
"loss": 0.4041,
"step": 6140
},
{
"epoch": 0.85,
"learning_rate": 0.0004,
"loss": 0.3847,
"step": 6150
},
{
"epoch": 0.85,
"learning_rate": 0.0004,
"loss": 0.3801,
"step": 6160
},
{
"epoch": 0.85,
"learning_rate": 0.0004,
"loss": 0.3922,
"step": 6170
},
{
"epoch": 0.85,
"learning_rate": 0.0004,
"loss": 0.382,
"step": 6180
},
{
"epoch": 0.85,
"learning_rate": 0.0004,
"loss": 0.3793,
"step": 6190
},
{
"epoch": 0.85,
"learning_rate": 0.0004,
"loss": 0.3829,
"step": 6200
},
{
"epoch": 0.85,
"learning_rate": 0.0004,
"loss": 0.3644,
"step": 6210
},
{
"epoch": 0.86,
"learning_rate": 0.0004,
"loss": 0.384,
"step": 6220
},
{
"epoch": 0.86,
"learning_rate": 0.0004,
"loss": 0.3675,
"step": 6230
},
{
"epoch": 0.86,
"learning_rate": 0.0004,
"loss": 0.3854,
"step": 6240
},
{
"epoch": 0.86,
"learning_rate": 0.0004,
"loss": 0.375,
"step": 6250
},
{
"epoch": 0.86,
"learning_rate": 0.0004,
"loss": 0.3809,
"step": 6260
},
{
"epoch": 0.86,
"learning_rate": 0.0004,
"loss": 0.3612,
"step": 6270
},
{
"epoch": 0.86,
"learning_rate": 0.0004,
"loss": 0.3598,
"step": 6280
},
{
"epoch": 0.87,
"learning_rate": 0.0004,
"loss": 0.3761,
"step": 6290
},
{
"epoch": 0.87,
"learning_rate": 0.0004,
"loss": 0.3718,
"step": 6300
},
{
"epoch": 0.87,
"learning_rate": 0.0004,
"loss": 0.3886,
"step": 6310
},
{
"epoch": 0.87,
"learning_rate": 0.0004,
"loss": 0.3765,
"step": 6320
},
{
"epoch": 0.87,
"learning_rate": 0.0004,
"loss": 0.368,
"step": 6330
},
{
"epoch": 0.87,
"learning_rate": 0.0004,
"loss": 0.3775,
"step": 6340
},
{
"epoch": 0.87,
"learning_rate": 0.0004,
"loss": 0.3858,
"step": 6350
},
{
"epoch": 0.88,
"learning_rate": 0.0004,
"loss": 0.385,
"step": 6360
},
{
"epoch": 0.88,
"learning_rate": 0.0004,
"loss": 0.3817,
"step": 6370
},
{
"epoch": 0.88,
"learning_rate": 0.0004,
"loss": 0.3724,
"step": 6380
},
{
"epoch": 0.88,
"learning_rate": 0.0004,
"loss": 0.3778,
"step": 6390
},
{
"epoch": 0.88,
"learning_rate": 0.0004,
"loss": 0.3795,
"step": 6400
},
{
"epoch": 0.88,
"learning_rate": 0.0004,
"loss": 0.3818,
"step": 6410
},
{
"epoch": 0.88,
"learning_rate": 0.0004,
"loss": 0.3825,
"step": 6420
},
{
"epoch": 0.88,
"learning_rate": 0.0004,
"loss": 0.3936,
"step": 6430
},
{
"epoch": 0.89,
"learning_rate": 0.0004,
"loss": 0.3926,
"step": 6440
},
{
"epoch": 0.89,
"learning_rate": 0.0004,
"loss": 0.3776,
"step": 6450
},
{
"epoch": 0.89,
"learning_rate": 0.0004,
"loss": 0.3815,
"step": 6460
},
{
"epoch": 0.89,
"learning_rate": 0.0004,
"loss": 0.3674,
"step": 6470
},
{
"epoch": 0.89,
"learning_rate": 0.0004,
"loss": 0.3758,
"step": 6480
},
{
"epoch": 0.89,
"learning_rate": 0.0004,
"loss": 0.367,
"step": 6490
},
{
"epoch": 0.89,
"learning_rate": 0.0004,
"loss": 0.3715,
"step": 6500
},
{
"epoch": 0.9,
"learning_rate": 0.0004,
"loss": 0.3905,
"step": 6510
},
{
"epoch": 0.9,
"learning_rate": 0.0004,
"loss": 0.3566,
"step": 6520
},
{
"epoch": 0.9,
"learning_rate": 0.0004,
"loss": 0.3742,
"step": 6530
},
{
"epoch": 0.9,
"learning_rate": 0.0004,
"loss": 0.3702,
"step": 6540
},
{
"epoch": 0.9,
"learning_rate": 0.0004,
"loss": 0.3758,
"step": 6550
},
{
"epoch": 0.9,
"learning_rate": 0.0004,
"loss": 0.3677,
"step": 6560
},
{
"epoch": 0.9,
"learning_rate": 0.0004,
"loss": 0.3847,
"step": 6570
},
{
"epoch": 0.91,
"learning_rate": 0.0004,
"loss": 0.3611,
"step": 6580
},
{
"epoch": 0.91,
"learning_rate": 0.0004,
"loss": 0.3809,
"step": 6590
},
{
"epoch": 0.91,
"learning_rate": 0.0004,
"loss": 0.3778,
"step": 6600
},
{
"epoch": 0.91,
"learning_rate": 0.0004,
"loss": 0.3753,
"step": 6610
},
{
"epoch": 0.91,
"learning_rate": 0.0004,
"loss": 0.3845,
"step": 6620
},
{
"epoch": 0.91,
"learning_rate": 0.0004,
"loss": 0.3734,
"step": 6630
},
{
"epoch": 0.91,
"learning_rate": 0.0004,
"loss": 0.3685,
"step": 6640
},
{
"epoch": 0.91,
"learning_rate": 0.0004,
"loss": 0.3758,
"step": 6650
},
{
"epoch": 0.92,
"learning_rate": 0.0004,
"loss": 0.3603,
"step": 6660
},
{
"epoch": 0.92,
"learning_rate": 0.0004,
"loss": 0.3794,
"step": 6670
},
{
"epoch": 0.92,
"learning_rate": 0.0004,
"loss": 0.3561,
"step": 6680
},
{
"epoch": 0.92,
"learning_rate": 0.0004,
"loss": 0.3707,
"step": 6690
},
{
"epoch": 0.92,
"learning_rate": 0.0004,
"loss": 0.3694,
"step": 6700
},
{
"epoch": 0.92,
"learning_rate": 0.0004,
"loss": 0.3694,
"step": 6710
},
{
"epoch": 0.92,
"learning_rate": 0.0004,
"loss": 0.3762,
"step": 6720
},
{
"epoch": 0.93,
"learning_rate": 0.0004,
"loss": 0.3795,
"step": 6730
},
{
"epoch": 0.93,
"learning_rate": 0.0004,
"loss": 0.3562,
"step": 6740
},
{
"epoch": 0.93,
"learning_rate": 0.0004,
"loss": 0.3614,
"step": 6750
},
{
"epoch": 0.93,
"learning_rate": 0.0004,
"loss": 0.3698,
"step": 6760
},
{
"epoch": 0.93,
"learning_rate": 0.0004,
"loss": 0.3564,
"step": 6770
},
{
"epoch": 0.93,
"learning_rate": 0.0004,
"loss": 0.3809,
"step": 6780
},
{
"epoch": 0.93,
"learning_rate": 0.0004,
"loss": 0.3643,
"step": 6790
},
{
"epoch": 0.94,
"learning_rate": 0.0004,
"loss": 0.3515,
"step": 6800
},
{
"epoch": 0.94,
"learning_rate": 0.0004,
"loss": 0.3523,
"step": 6810
},
{
"epoch": 0.94,
"learning_rate": 0.0004,
"loss": 0.3586,
"step": 6820
},
{
"epoch": 0.94,
"learning_rate": 0.0004,
"loss": 0.3637,
"step": 6830
},
{
"epoch": 0.94,
"learning_rate": 0.0004,
"loss": 0.3618,
"step": 6840
},
{
"epoch": 0.94,
"learning_rate": 0.0004,
"loss": 0.369,
"step": 6850
},
{
"epoch": 0.94,
"learning_rate": 0.0004,
"loss": 0.3588,
"step": 6860
},
{
"epoch": 0.95,
"learning_rate": 0.0004,
"loss": 0.3658,
"step": 6870
},
{
"epoch": 0.95,
"learning_rate": 0.0004,
"loss": 0.3638,
"step": 6880
},
{
"epoch": 0.95,
"learning_rate": 0.0004,
"loss": 0.3474,
"step": 6890
},
{
"epoch": 0.95,
"learning_rate": 0.0004,
"loss": 0.3614,
"step": 6900
},
{
"epoch": 0.95,
"learning_rate": 0.0004,
"loss": 0.3731,
"step": 6910
},
{
"epoch": 0.95,
"learning_rate": 0.0004,
"loss": 0.3641,
"step": 6920
},
{
"epoch": 0.95,
"learning_rate": 0.0004,
"loss": 0.3683,
"step": 6930
},
{
"epoch": 0.95,
"learning_rate": 0.0004,
"loss": 0.3634,
"step": 6940
},
{
"epoch": 0.96,
"learning_rate": 0.0004,
"loss": 0.3673,
"step": 6950
},
{
"epoch": 0.96,
"learning_rate": 0.0004,
"loss": 0.3704,
"step": 6960
},
{
"epoch": 0.96,
"learning_rate": 0.0004,
"loss": 0.3731,
"step": 6970
},
{
"epoch": 0.96,
"learning_rate": 0.0004,
"loss": 0.3715,
"step": 6980
},
{
"epoch": 0.96,
"learning_rate": 0.0004,
"loss": 0.3591,
"step": 6990
},
{
"epoch": 0.96,
"learning_rate": 0.0004,
"loss": 0.3671,
"step": 7000
},
{
"epoch": 0.96,
"learning_rate": 0.0004,
"loss": 0.3865,
"step": 7010
},
{
"epoch": 0.97,
"learning_rate": 0.0004,
"loss": 0.3836,
"step": 7020
},
{
"epoch": 0.97,
"learning_rate": 0.0004,
"loss": 0.3523,
"step": 7030
},
{
"epoch": 0.97,
"learning_rate": 0.0004,
"loss": 0.373,
"step": 7040
},
{
"epoch": 0.97,
"learning_rate": 0.0004,
"loss": 0.3716,
"step": 7050
},
{
"epoch": 0.97,
"learning_rate": 0.0004,
"loss": 0.3591,
"step": 7060
},
{
"epoch": 0.97,
"learning_rate": 0.0004,
"loss": 0.3596,
"step": 7070
},
{
"epoch": 0.97,
"learning_rate": 0.0004,
"loss": 0.3608,
"step": 7080
},
{
"epoch": 0.98,
"learning_rate": 0.0004,
"loss": 0.343,
"step": 7090
},
{
"epoch": 0.98,
"learning_rate": 0.0004,
"loss": 0.3612,
"step": 7100
},
{
"epoch": 0.98,
"learning_rate": 0.0004,
"loss": 0.3479,
"step": 7110
},
{
"epoch": 0.98,
"learning_rate": 0.0004,
"loss": 0.367,
"step": 7120
},
{
"epoch": 0.98,
"learning_rate": 0.0004,
"loss": 0.3535,
"step": 7130
},
{
"epoch": 0.98,
"learning_rate": 0.0004,
"loss": 0.3434,
"step": 7140
},
{
"epoch": 0.98,
"learning_rate": 0.0004,
"loss": 0.3608,
"step": 7150
},
{
"epoch": 0.99,
"learning_rate": 0.0004,
"loss": 0.3591,
"step": 7160
},
{
"epoch": 0.99,
"learning_rate": 0.0004,
"loss": 0.3376,
"step": 7170
},
{
"epoch": 0.99,
"learning_rate": 0.0004,
"loss": 0.3559,
"step": 7180
},
{
"epoch": 0.99,
"learning_rate": 0.0004,
"loss": 0.3544,
"step": 7190
},
{
"epoch": 0.99,
"learning_rate": 0.0004,
"loss": 0.3688,
"step": 7200
},
{
"epoch": 0.99,
"learning_rate": 0.0004,
"loss": 0.3573,
"step": 7210
},
{
"epoch": 0.99,
"learning_rate": 0.0004,
"loss": 0.3505,
"step": 7220
},
{
"epoch": 0.99,
"learning_rate": 0.0004,
"loss": 0.3662,
"step": 7230
},
{
"epoch": 1.0,
"learning_rate": 0.0004,
"loss": 0.3657,
"step": 7240
},
{
"epoch": 1.0,
"learning_rate": 0.0004,
"loss": 0.3768,
"step": 7250
},
{
"epoch": 1.0,
"learning_rate": 0.0004,
"loss": 0.3532,
"step": 7260
},
{
"epoch": 1.0,
"learning_rate": 0.0004,
"loss": 0.3642,
"step": 7270
},
{
"epoch": 1.0,
"learning_rate": 0.0004,
"loss": 0.3595,
"step": 7280
},
{
"epoch": 1.0,
"learning_rate": 0.0004,
"loss": 0.3642,
"step": 7290
},
{
"epoch": 1.0,
"learning_rate": 0.0004,
"loss": 0.3561,
"step": 7300
},
{
"epoch": 1.01,
"learning_rate": 0.0004,
"loss": 0.3494,
"step": 7310
},
{
"epoch": 1.01,
"learning_rate": 0.0004,
"loss": 0.3601,
"step": 7320
},
{
"epoch": 1.01,
"learning_rate": 0.0004,
"loss": 0.3735,
"step": 7330
},
{
"epoch": 1.01,
"learning_rate": 0.0004,
"loss": 0.3556,
"step": 7340
},
{
"epoch": 1.01,
"learning_rate": 0.0004,
"loss": 0.3539,
"step": 7350
},
{
"epoch": 1.01,
"learning_rate": 0.0004,
"loss": 0.3641,
"step": 7360
},
{
"epoch": 1.01,
"learning_rate": 0.0004,
"loss": 0.3453,
"step": 7370
},
{
"epoch": 1.02,
"learning_rate": 0.0004,
"loss": 0.3333,
"step": 7380
},
{
"epoch": 1.02,
"learning_rate": 0.0004,
"loss": 0.3561,
"step": 7390
},
{
"epoch": 1.02,
"learning_rate": 0.0004,
"loss": 0.3439,
"step": 7400
},
{
"epoch": 1.02,
"learning_rate": 0.0004,
"loss": 0.3508,
"step": 7410
},
{
"epoch": 1.02,
"learning_rate": 0.0004,
"loss": 0.3442,
"step": 7420
},
{
"epoch": 1.02,
"learning_rate": 0.0004,
"loss": 0.3146,
"step": 7430
},
{
"epoch": 1.02,
"learning_rate": 0.0004,
"loss": 0.3417,
"step": 7440
},
{
"epoch": 1.02,
"learning_rate": 0.0004,
"loss": 0.3473,
"step": 7450
},
{
"epoch": 1.03,
"learning_rate": 0.0004,
"loss": 0.3494,
"step": 7460
},
{
"epoch": 1.03,
"learning_rate": 0.0004,
"loss": 0.3633,
"step": 7470
},
{
"epoch": 1.03,
"learning_rate": 0.0004,
"loss": 0.3235,
"step": 7480
},
{
"epoch": 1.03,
"learning_rate": 0.0004,
"loss": 0.3489,
"step": 7490
},
{
"epoch": 1.03,
"learning_rate": 0.0004,
"loss": 0.34,
"step": 7500
},
{
"epoch": 1.03,
"learning_rate": 0.0004,
"loss": 0.348,
"step": 7510
},
{
"epoch": 1.03,
"learning_rate": 0.0004,
"loss": 0.3579,
"step": 7520
},
{
"epoch": 1.04,
"learning_rate": 0.0004,
"loss": 0.3494,
"step": 7530
},
{
"epoch": 1.04,
"learning_rate": 0.0004,
"loss": 0.3275,
"step": 7540
},
{
"epoch": 1.04,
"learning_rate": 0.0004,
"loss": 0.3514,
"step": 7550
},
{
"epoch": 1.04,
"learning_rate": 0.0004,
"loss": 0.3292,
"step": 7560
},
{
"epoch": 1.04,
"learning_rate": 0.0004,
"loss": 0.3435,
"step": 7570
},
{
"epoch": 1.04,
"learning_rate": 0.0004,
"loss": 0.3297,
"step": 7580
},
{
"epoch": 1.04,
"learning_rate": 0.0004,
"loss": 0.3513,
"step": 7590
},
{
"epoch": 1.05,
"learning_rate": 0.0004,
"loss": 0.3542,
"step": 7600
},
{
"epoch": 1.05,
"learning_rate": 0.0004,
"loss": 0.341,
"step": 7610
},
{
"epoch": 1.05,
"learning_rate": 0.0004,
"loss": 0.3381,
"step": 7620
},
{
"epoch": 1.05,
"learning_rate": 0.0004,
"loss": 0.3608,
"step": 7630
},
{
"epoch": 1.05,
"learning_rate": 0.0004,
"loss": 0.3511,
"step": 7640
},
{
"epoch": 1.05,
"learning_rate": 0.0004,
"loss": 0.361,
"step": 7650
},
{
"epoch": 1.05,
"learning_rate": 0.0004,
"loss": 0.3581,
"step": 7660
},
{
"epoch": 1.06,
"learning_rate": 0.0004,
"loss": 0.3535,
"step": 7670
},
{
"epoch": 1.06,
"learning_rate": 0.0004,
"loss": 0.3333,
"step": 7680
},
{
"epoch": 1.06,
"learning_rate": 0.0004,
"loss": 0.3469,
"step": 7690
},
{
"epoch": 1.06,
"learning_rate": 0.0004,
"loss": 0.3446,
"step": 7700
},
{
"epoch": 1.06,
"learning_rate": 0.0004,
"loss": 0.3385,
"step": 7710
},
{
"epoch": 1.06,
"learning_rate": 0.0004,
"loss": 0.3492,
"step": 7720
},
{
"epoch": 1.06,
"learning_rate": 0.0004,
"loss": 0.3351,
"step": 7730
},
{
"epoch": 1.06,
"learning_rate": 0.0004,
"loss": 0.3531,
"step": 7740
},
{
"epoch": 1.07,
"learning_rate": 0.0004,
"loss": 0.3391,
"step": 7750
},
{
"epoch": 1.07,
"learning_rate": 0.0004,
"loss": 0.3366,
"step": 7760
},
{
"epoch": 1.07,
"learning_rate": 0.0004,
"loss": 0.3305,
"step": 7770
},
{
"epoch": 1.07,
"learning_rate": 0.0004,
"loss": 0.3389,
"step": 7780
},
{
"epoch": 1.07,
"learning_rate": 0.0004,
"loss": 0.3457,
"step": 7790
},
{
"epoch": 1.07,
"learning_rate": 0.0004,
"loss": 0.3352,
"step": 7800
},
{
"epoch": 1.07,
"learning_rate": 0.0004,
"loss": 0.3502,
"step": 7810
},
{
"epoch": 1.08,
"learning_rate": 0.0004,
"loss": 0.3435,
"step": 7820
},
{
"epoch": 1.08,
"learning_rate": 0.0004,
"loss": 0.3393,
"step": 7830
},
{
"epoch": 1.08,
"learning_rate": 0.0004,
"loss": 0.3469,
"step": 7840
},
{
"epoch": 1.08,
"learning_rate": 0.0004,
"loss": 0.3491,
"step": 7850
},
{
"epoch": 1.08,
"learning_rate": 0.0004,
"loss": 0.3598,
"step": 7860
},
{
"epoch": 1.08,
"learning_rate": 0.0004,
"loss": 0.3488,
"step": 7870
},
{
"epoch": 1.08,
"learning_rate": 0.0004,
"loss": 0.3479,
"step": 7880
},
{
"epoch": 1.09,
"learning_rate": 0.0004,
"loss": 0.3504,
"step": 7890
},
{
"epoch": 1.09,
"learning_rate": 0.0004,
"loss": 0.3574,
"step": 7900
},
{
"epoch": 1.09,
"learning_rate": 0.0004,
"loss": 0.3533,
"step": 7910
},
{
"epoch": 1.09,
"learning_rate": 0.0004,
"loss": 0.354,
"step": 7920
},
{
"epoch": 1.09,
"learning_rate": 0.0004,
"loss": 0.3341,
"step": 7930
},
{
"epoch": 1.09,
"learning_rate": 0.0004,
"loss": 0.3581,
"step": 7940
},
{
"epoch": 1.09,
"learning_rate": 0.0004,
"loss": 0.3507,
"step": 7950
},
{
"epoch": 1.1,
"learning_rate": 0.0004,
"loss": 0.3649,
"step": 7960
},
{
"epoch": 1.1,
"learning_rate": 0.0004,
"loss": 0.3384,
"step": 7970
},
{
"epoch": 1.1,
"learning_rate": 0.0004,
"loss": 0.3485,
"step": 7980
},
{
"epoch": 1.1,
"learning_rate": 0.0004,
"loss": 0.3465,
"step": 7990
},
{
"epoch": 1.1,
"learning_rate": 0.0004,
"loss": 0.336,
"step": 8000
}
],
"logging_steps": 10,
"max_steps": 8000,
"num_train_epochs": 2,
"save_steps": 500,
"total_flos": 2.806745558679552e+18,
"trial_name": null,
"trial_params": null
}