phi-2-10000 / trainer_state.json
cekal's picture
Upload 14 files
716ec66
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 8.479966080135679,
"eval_steps": 500,
"global_step": 10000,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 1.6593,
"step": 10
},
{
"epoch": 0.02,
"learning_rate": 0.0002,
"loss": 1.7251,
"step": 20
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 1.6415,
"step": 30
},
{
"epoch": 0.03,
"learning_rate": 0.0002,
"loss": 1.7984,
"step": 40
},
{
"epoch": 0.04,
"learning_rate": 0.0002,
"loss": 1.6267,
"step": 50
},
{
"epoch": 0.05,
"learning_rate": 0.0002,
"loss": 1.6327,
"step": 60
},
{
"epoch": 0.06,
"learning_rate": 0.0002,
"loss": 1.4942,
"step": 70
},
{
"epoch": 0.07,
"learning_rate": 0.0002,
"loss": 1.7348,
"step": 80
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 1.6918,
"step": 90
},
{
"epoch": 0.08,
"learning_rate": 0.0002,
"loss": 1.6704,
"step": 100
},
{
"epoch": 0.09,
"learning_rate": 0.0002,
"loss": 1.6685,
"step": 110
},
{
"epoch": 0.1,
"learning_rate": 0.0002,
"loss": 1.7159,
"step": 120
},
{
"epoch": 0.11,
"learning_rate": 0.0002,
"loss": 1.7284,
"step": 130
},
{
"epoch": 0.12,
"learning_rate": 0.0002,
"loss": 1.8045,
"step": 140
},
{
"epoch": 0.13,
"learning_rate": 0.0002,
"loss": 1.6731,
"step": 150
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 1.6264,
"step": 160
},
{
"epoch": 0.14,
"learning_rate": 0.0002,
"loss": 1.673,
"step": 170
},
{
"epoch": 0.15,
"learning_rate": 0.0002,
"loss": 1.6026,
"step": 180
},
{
"epoch": 0.16,
"learning_rate": 0.0002,
"loss": 1.7228,
"step": 190
},
{
"epoch": 0.17,
"learning_rate": 0.0002,
"loss": 1.6999,
"step": 200
},
{
"epoch": 0.18,
"learning_rate": 0.0002,
"loss": 1.6279,
"step": 210
},
{
"epoch": 0.19,
"learning_rate": 0.0002,
"loss": 1.6168,
"step": 220
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 1.7292,
"step": 230
},
{
"epoch": 0.2,
"learning_rate": 0.0002,
"loss": 1.7676,
"step": 240
},
{
"epoch": 0.21,
"learning_rate": 0.0002,
"loss": 1.5916,
"step": 250
},
{
"epoch": 0.22,
"learning_rate": 0.0002,
"loss": 1.6078,
"step": 260
},
{
"epoch": 0.23,
"learning_rate": 0.0002,
"loss": 1.5806,
"step": 270
},
{
"epoch": 0.24,
"learning_rate": 0.0002,
"loss": 1.6083,
"step": 280
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 1.7445,
"step": 290
},
{
"epoch": 0.25,
"learning_rate": 0.0002,
"loss": 1.6042,
"step": 300
},
{
"epoch": 0.26,
"learning_rate": 0.0002,
"loss": 1.5328,
"step": 310
},
{
"epoch": 0.27,
"learning_rate": 0.0002,
"loss": 1.5979,
"step": 320
},
{
"epoch": 0.28,
"learning_rate": 0.0002,
"loss": 1.5787,
"step": 330
},
{
"epoch": 0.29,
"learning_rate": 0.0002,
"loss": 1.7654,
"step": 340
},
{
"epoch": 0.3,
"learning_rate": 0.0002,
"loss": 1.6702,
"step": 350
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 1.6143,
"step": 360
},
{
"epoch": 0.31,
"learning_rate": 0.0002,
"loss": 1.608,
"step": 370
},
{
"epoch": 0.32,
"learning_rate": 0.0002,
"loss": 1.7656,
"step": 380
},
{
"epoch": 0.33,
"learning_rate": 0.0002,
"loss": 1.7336,
"step": 390
},
{
"epoch": 0.34,
"learning_rate": 0.0002,
"loss": 1.6818,
"step": 400
},
{
"epoch": 0.35,
"learning_rate": 0.0002,
"loss": 1.5711,
"step": 410
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 1.4467,
"step": 420
},
{
"epoch": 0.36,
"learning_rate": 0.0002,
"loss": 1.5485,
"step": 430
},
{
"epoch": 0.37,
"learning_rate": 0.0002,
"loss": 1.603,
"step": 440
},
{
"epoch": 0.38,
"learning_rate": 0.0002,
"loss": 1.6327,
"step": 450
},
{
"epoch": 0.39,
"learning_rate": 0.0002,
"loss": 1.6488,
"step": 460
},
{
"epoch": 0.4,
"learning_rate": 0.0002,
"loss": 1.6549,
"step": 470
},
{
"epoch": 0.41,
"learning_rate": 0.0002,
"loss": 1.4652,
"step": 480
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 1.6845,
"step": 490
},
{
"epoch": 0.42,
"learning_rate": 0.0002,
"loss": 1.6544,
"step": 500
},
{
"epoch": 0.43,
"learning_rate": 0.0002,
"loss": 1.6361,
"step": 510
},
{
"epoch": 0.44,
"learning_rate": 0.0002,
"loss": 1.6729,
"step": 520
},
{
"epoch": 0.45,
"learning_rate": 0.0002,
"loss": 1.6131,
"step": 530
},
{
"epoch": 0.46,
"learning_rate": 0.0002,
"loss": 1.6919,
"step": 540
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 1.5626,
"step": 550
},
{
"epoch": 0.47,
"learning_rate": 0.0002,
"loss": 1.6535,
"step": 560
},
{
"epoch": 0.48,
"learning_rate": 0.0002,
"loss": 1.5712,
"step": 570
},
{
"epoch": 0.49,
"learning_rate": 0.0002,
"loss": 1.5684,
"step": 580
},
{
"epoch": 0.5,
"learning_rate": 0.0002,
"loss": 1.685,
"step": 590
},
{
"epoch": 0.51,
"learning_rate": 0.0002,
"loss": 1.6356,
"step": 600
},
{
"epoch": 0.52,
"learning_rate": 0.0002,
"loss": 1.6437,
"step": 610
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 1.5102,
"step": 620
},
{
"epoch": 0.53,
"learning_rate": 0.0002,
"loss": 1.6775,
"step": 630
},
{
"epoch": 0.54,
"learning_rate": 0.0002,
"loss": 1.5823,
"step": 640
},
{
"epoch": 0.55,
"learning_rate": 0.0002,
"loss": 1.6516,
"step": 650
},
{
"epoch": 0.56,
"learning_rate": 0.0002,
"loss": 1.5845,
"step": 660
},
{
"epoch": 0.57,
"learning_rate": 0.0002,
"loss": 1.6506,
"step": 670
},
{
"epoch": 0.58,
"learning_rate": 0.0002,
"loss": 1.6007,
"step": 680
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 1.6858,
"step": 690
},
{
"epoch": 0.59,
"learning_rate": 0.0002,
"loss": 1.5993,
"step": 700
},
{
"epoch": 0.6,
"learning_rate": 0.0002,
"loss": 1.638,
"step": 710
},
{
"epoch": 0.61,
"learning_rate": 0.0002,
"loss": 1.6185,
"step": 720
},
{
"epoch": 0.62,
"learning_rate": 0.0002,
"loss": 1.4716,
"step": 730
},
{
"epoch": 0.63,
"learning_rate": 0.0002,
"loss": 1.5783,
"step": 740
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 1.6938,
"step": 750
},
{
"epoch": 0.64,
"learning_rate": 0.0002,
"loss": 1.5168,
"step": 760
},
{
"epoch": 0.65,
"learning_rate": 0.0002,
"loss": 1.4567,
"step": 770
},
{
"epoch": 0.66,
"learning_rate": 0.0002,
"loss": 1.6481,
"step": 780
},
{
"epoch": 0.67,
"learning_rate": 0.0002,
"loss": 1.7604,
"step": 790
},
{
"epoch": 0.68,
"learning_rate": 0.0002,
"loss": 1.5539,
"step": 800
},
{
"epoch": 0.69,
"learning_rate": 0.0002,
"loss": 1.5588,
"step": 810
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 1.6224,
"step": 820
},
{
"epoch": 0.7,
"learning_rate": 0.0002,
"loss": 1.5466,
"step": 830
},
{
"epoch": 0.71,
"learning_rate": 0.0002,
"loss": 1.6262,
"step": 840
},
{
"epoch": 0.72,
"learning_rate": 0.0002,
"loss": 1.7497,
"step": 850
},
{
"epoch": 0.73,
"learning_rate": 0.0002,
"loss": 1.5336,
"step": 860
},
{
"epoch": 0.74,
"learning_rate": 0.0002,
"loss": 1.5972,
"step": 870
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 1.585,
"step": 880
},
{
"epoch": 0.75,
"learning_rate": 0.0002,
"loss": 1.7212,
"step": 890
},
{
"epoch": 0.76,
"learning_rate": 0.0002,
"loss": 1.5687,
"step": 900
},
{
"epoch": 0.77,
"learning_rate": 0.0002,
"loss": 1.5752,
"step": 910
},
{
"epoch": 0.78,
"learning_rate": 0.0002,
"loss": 1.5069,
"step": 920
},
{
"epoch": 0.79,
"learning_rate": 0.0002,
"loss": 1.4922,
"step": 930
},
{
"epoch": 0.8,
"learning_rate": 0.0002,
"loss": 1.7895,
"step": 940
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 1.6067,
"step": 950
},
{
"epoch": 0.81,
"learning_rate": 0.0002,
"loss": 1.6512,
"step": 960
},
{
"epoch": 0.82,
"learning_rate": 0.0002,
"loss": 1.5318,
"step": 970
},
{
"epoch": 0.83,
"learning_rate": 0.0002,
"loss": 1.5957,
"step": 980
},
{
"epoch": 0.84,
"learning_rate": 0.0002,
"loss": 1.5924,
"step": 990
},
{
"epoch": 0.85,
"learning_rate": 0.0002,
"loss": 1.6206,
"step": 1000
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 1.5302,
"step": 1010
},
{
"epoch": 0.86,
"learning_rate": 0.0002,
"loss": 1.6814,
"step": 1020
},
{
"epoch": 0.87,
"learning_rate": 0.0002,
"loss": 1.5442,
"step": 1030
},
{
"epoch": 0.88,
"learning_rate": 0.0002,
"loss": 1.6821,
"step": 1040
},
{
"epoch": 0.89,
"learning_rate": 0.0002,
"loss": 1.6457,
"step": 1050
},
{
"epoch": 0.9,
"learning_rate": 0.0002,
"loss": 1.6522,
"step": 1060
},
{
"epoch": 0.91,
"learning_rate": 0.0002,
"loss": 1.6522,
"step": 1070
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 1.7023,
"step": 1080
},
{
"epoch": 0.92,
"learning_rate": 0.0002,
"loss": 1.6462,
"step": 1090
},
{
"epoch": 0.93,
"learning_rate": 0.0002,
"loss": 1.5393,
"step": 1100
},
{
"epoch": 0.94,
"learning_rate": 0.0002,
"loss": 1.7156,
"step": 1110
},
{
"epoch": 0.95,
"learning_rate": 0.0002,
"loss": 1.5649,
"step": 1120
},
{
"epoch": 0.96,
"learning_rate": 0.0002,
"loss": 1.6564,
"step": 1130
},
{
"epoch": 0.97,
"learning_rate": 0.0002,
"loss": 1.6574,
"step": 1140
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 1.5514,
"step": 1150
},
{
"epoch": 0.98,
"learning_rate": 0.0002,
"loss": 1.599,
"step": 1160
},
{
"epoch": 0.99,
"learning_rate": 0.0002,
"loss": 1.5292,
"step": 1170
},
{
"epoch": 1.0,
"learning_rate": 0.0002,
"loss": 1.6364,
"step": 1180
},
{
"epoch": 1.01,
"learning_rate": 0.0002,
"loss": 1.6888,
"step": 1190
},
{
"epoch": 1.02,
"learning_rate": 0.0002,
"loss": 1.5532,
"step": 1200
},
{
"epoch": 1.03,
"learning_rate": 0.0002,
"loss": 1.6599,
"step": 1210
},
{
"epoch": 1.03,
"learning_rate": 0.0002,
"loss": 1.7307,
"step": 1220
},
{
"epoch": 1.04,
"learning_rate": 0.0002,
"loss": 1.4995,
"step": 1230
},
{
"epoch": 1.05,
"learning_rate": 0.0002,
"loss": 1.4839,
"step": 1240
},
{
"epoch": 1.06,
"learning_rate": 0.0002,
"loss": 1.5689,
"step": 1250
},
{
"epoch": 1.07,
"learning_rate": 0.0002,
"loss": 1.5268,
"step": 1260
},
{
"epoch": 1.08,
"learning_rate": 0.0002,
"loss": 1.7108,
"step": 1270
},
{
"epoch": 1.09,
"learning_rate": 0.0002,
"loss": 1.5284,
"step": 1280
},
{
"epoch": 1.09,
"learning_rate": 0.0002,
"loss": 1.6095,
"step": 1290
},
{
"epoch": 1.1,
"learning_rate": 0.0002,
"loss": 1.6331,
"step": 1300
},
{
"epoch": 1.11,
"learning_rate": 0.0002,
"loss": 1.5894,
"step": 1310
},
{
"epoch": 1.12,
"learning_rate": 0.0002,
"loss": 1.6417,
"step": 1320
},
{
"epoch": 1.13,
"learning_rate": 0.0002,
"loss": 1.5079,
"step": 1330
},
{
"epoch": 1.14,
"learning_rate": 0.0002,
"loss": 1.6064,
"step": 1340
},
{
"epoch": 1.14,
"learning_rate": 0.0002,
"loss": 1.5308,
"step": 1350
},
{
"epoch": 1.15,
"learning_rate": 0.0002,
"loss": 1.5682,
"step": 1360
},
{
"epoch": 1.16,
"learning_rate": 0.0002,
"loss": 1.5779,
"step": 1370
},
{
"epoch": 1.17,
"learning_rate": 0.0002,
"loss": 1.5629,
"step": 1380
},
{
"epoch": 1.18,
"learning_rate": 0.0002,
"loss": 1.5669,
"step": 1390
},
{
"epoch": 1.19,
"learning_rate": 0.0002,
"loss": 1.6337,
"step": 1400
},
{
"epoch": 1.2,
"learning_rate": 0.0002,
"loss": 1.5723,
"step": 1410
},
{
"epoch": 1.2,
"learning_rate": 0.0002,
"loss": 1.6695,
"step": 1420
},
{
"epoch": 1.21,
"learning_rate": 0.0002,
"loss": 1.626,
"step": 1430
},
{
"epoch": 1.22,
"learning_rate": 0.0002,
"loss": 1.5941,
"step": 1440
},
{
"epoch": 1.23,
"learning_rate": 0.0002,
"loss": 1.4819,
"step": 1450
},
{
"epoch": 1.24,
"learning_rate": 0.0002,
"loss": 1.6359,
"step": 1460
},
{
"epoch": 1.25,
"learning_rate": 0.0002,
"loss": 1.7286,
"step": 1470
},
{
"epoch": 1.26,
"learning_rate": 0.0002,
"loss": 1.4705,
"step": 1480
},
{
"epoch": 1.26,
"learning_rate": 0.0002,
"loss": 1.5025,
"step": 1490
},
{
"epoch": 1.27,
"learning_rate": 0.0002,
"loss": 1.5677,
"step": 1500
},
{
"epoch": 1.28,
"learning_rate": 0.0002,
"loss": 1.5815,
"step": 1510
},
{
"epoch": 1.29,
"learning_rate": 0.0002,
"loss": 1.6278,
"step": 1520
},
{
"epoch": 1.3,
"learning_rate": 0.0002,
"loss": 1.5388,
"step": 1530
},
{
"epoch": 1.31,
"learning_rate": 0.0002,
"loss": 1.6468,
"step": 1540
},
{
"epoch": 1.31,
"learning_rate": 0.0002,
"loss": 1.664,
"step": 1550
},
{
"epoch": 1.32,
"learning_rate": 0.0002,
"loss": 1.6549,
"step": 1560
},
{
"epoch": 1.33,
"learning_rate": 0.0002,
"loss": 1.6705,
"step": 1570
},
{
"epoch": 1.34,
"learning_rate": 0.0002,
"loss": 1.517,
"step": 1580
},
{
"epoch": 1.35,
"learning_rate": 0.0002,
"loss": 1.6119,
"step": 1590
},
{
"epoch": 1.36,
"learning_rate": 0.0002,
"loss": 1.5735,
"step": 1600
},
{
"epoch": 1.37,
"learning_rate": 0.0002,
"loss": 1.6245,
"step": 1610
},
{
"epoch": 1.37,
"learning_rate": 0.0002,
"loss": 1.6098,
"step": 1620
},
{
"epoch": 1.38,
"learning_rate": 0.0002,
"loss": 1.5175,
"step": 1630
},
{
"epoch": 1.39,
"learning_rate": 0.0002,
"loss": 1.6361,
"step": 1640
},
{
"epoch": 1.4,
"learning_rate": 0.0002,
"loss": 1.4669,
"step": 1650
},
{
"epoch": 1.41,
"learning_rate": 0.0002,
"loss": 1.5289,
"step": 1660
},
{
"epoch": 1.42,
"learning_rate": 0.0002,
"loss": 1.5474,
"step": 1670
},
{
"epoch": 1.42,
"learning_rate": 0.0002,
"loss": 1.4894,
"step": 1680
},
{
"epoch": 1.43,
"learning_rate": 0.0002,
"loss": 1.5975,
"step": 1690
},
{
"epoch": 1.44,
"learning_rate": 0.0002,
"loss": 1.5301,
"step": 1700
},
{
"epoch": 1.45,
"learning_rate": 0.0002,
"loss": 1.6809,
"step": 1710
},
{
"epoch": 1.46,
"learning_rate": 0.0002,
"loss": 1.6261,
"step": 1720
},
{
"epoch": 1.47,
"learning_rate": 0.0002,
"loss": 1.4557,
"step": 1730
},
{
"epoch": 1.48,
"learning_rate": 0.0002,
"loss": 1.568,
"step": 1740
},
{
"epoch": 1.48,
"learning_rate": 0.0002,
"loss": 1.5338,
"step": 1750
},
{
"epoch": 1.49,
"learning_rate": 0.0002,
"loss": 1.5925,
"step": 1760
},
{
"epoch": 1.5,
"learning_rate": 0.0002,
"loss": 1.6949,
"step": 1770
},
{
"epoch": 1.51,
"learning_rate": 0.0002,
"loss": 1.517,
"step": 1780
},
{
"epoch": 1.52,
"learning_rate": 0.0002,
"loss": 1.6104,
"step": 1790
},
{
"epoch": 1.53,
"learning_rate": 0.0002,
"loss": 1.6787,
"step": 1800
},
{
"epoch": 1.53,
"learning_rate": 0.0002,
"loss": 1.5108,
"step": 1810
},
{
"epoch": 1.54,
"learning_rate": 0.0002,
"loss": 1.6884,
"step": 1820
},
{
"epoch": 1.55,
"learning_rate": 0.0002,
"loss": 1.5286,
"step": 1830
},
{
"epoch": 1.56,
"learning_rate": 0.0002,
"loss": 1.557,
"step": 1840
},
{
"epoch": 1.57,
"learning_rate": 0.0002,
"loss": 1.6252,
"step": 1850
},
{
"epoch": 1.58,
"learning_rate": 0.0002,
"loss": 1.6131,
"step": 1860
},
{
"epoch": 1.59,
"learning_rate": 0.0002,
"loss": 1.5481,
"step": 1870
},
{
"epoch": 1.59,
"learning_rate": 0.0002,
"loss": 1.571,
"step": 1880
},
{
"epoch": 1.6,
"learning_rate": 0.0002,
"loss": 1.5579,
"step": 1890
},
{
"epoch": 1.61,
"learning_rate": 0.0002,
"loss": 1.5728,
"step": 1900
},
{
"epoch": 1.62,
"learning_rate": 0.0002,
"loss": 1.5974,
"step": 1910
},
{
"epoch": 1.63,
"learning_rate": 0.0002,
"loss": 1.7084,
"step": 1920
},
{
"epoch": 1.64,
"learning_rate": 0.0002,
"loss": 1.5796,
"step": 1930
},
{
"epoch": 1.65,
"learning_rate": 0.0002,
"loss": 1.5364,
"step": 1940
},
{
"epoch": 1.65,
"learning_rate": 0.0002,
"loss": 1.5484,
"step": 1950
},
{
"epoch": 1.66,
"learning_rate": 0.0002,
"loss": 1.7021,
"step": 1960
},
{
"epoch": 1.67,
"learning_rate": 0.0002,
"loss": 1.6417,
"step": 1970
},
{
"epoch": 1.68,
"learning_rate": 0.0002,
"loss": 1.5349,
"step": 1980
},
{
"epoch": 1.69,
"learning_rate": 0.0002,
"loss": 1.6066,
"step": 1990
},
{
"epoch": 1.7,
"learning_rate": 0.0002,
"loss": 1.6009,
"step": 2000
},
{
"epoch": 1.7,
"learning_rate": 0.0002,
"loss": 1.5731,
"step": 2010
},
{
"epoch": 1.71,
"learning_rate": 0.0002,
"loss": 1.6593,
"step": 2020
},
{
"epoch": 1.72,
"learning_rate": 0.0002,
"loss": 1.5177,
"step": 2030
},
{
"epoch": 1.73,
"learning_rate": 0.0002,
"loss": 1.5316,
"step": 2040
},
{
"epoch": 1.74,
"learning_rate": 0.0002,
"loss": 1.4415,
"step": 2050
},
{
"epoch": 1.75,
"learning_rate": 0.0002,
"loss": 1.5551,
"step": 2060
},
{
"epoch": 1.76,
"learning_rate": 0.0002,
"loss": 1.6137,
"step": 2070
},
{
"epoch": 1.76,
"learning_rate": 0.0002,
"loss": 1.4704,
"step": 2080
},
{
"epoch": 1.77,
"learning_rate": 0.0002,
"loss": 1.5983,
"step": 2090
},
{
"epoch": 1.78,
"learning_rate": 0.0002,
"loss": 1.7158,
"step": 2100
},
{
"epoch": 1.79,
"learning_rate": 0.0002,
"loss": 1.5872,
"step": 2110
},
{
"epoch": 1.8,
"learning_rate": 0.0002,
"loss": 1.6466,
"step": 2120
},
{
"epoch": 1.81,
"learning_rate": 0.0002,
"loss": 1.6144,
"step": 2130
},
{
"epoch": 1.81,
"learning_rate": 0.0002,
"loss": 1.5811,
"step": 2140
},
{
"epoch": 1.82,
"learning_rate": 0.0002,
"loss": 1.577,
"step": 2150
},
{
"epoch": 1.83,
"learning_rate": 0.0002,
"loss": 1.5325,
"step": 2160
},
{
"epoch": 1.84,
"learning_rate": 0.0002,
"loss": 1.6143,
"step": 2170
},
{
"epoch": 1.85,
"learning_rate": 0.0002,
"loss": 1.6564,
"step": 2180
},
{
"epoch": 1.86,
"learning_rate": 0.0002,
"loss": 1.6807,
"step": 2190
},
{
"epoch": 1.87,
"learning_rate": 0.0002,
"loss": 1.53,
"step": 2200
},
{
"epoch": 1.87,
"learning_rate": 0.0002,
"loss": 1.5522,
"step": 2210
},
{
"epoch": 1.88,
"learning_rate": 0.0002,
"loss": 1.7534,
"step": 2220
},
{
"epoch": 1.89,
"learning_rate": 0.0002,
"loss": 1.493,
"step": 2230
},
{
"epoch": 1.9,
"learning_rate": 0.0002,
"loss": 1.5664,
"step": 2240
},
{
"epoch": 1.91,
"learning_rate": 0.0002,
"loss": 1.4852,
"step": 2250
},
{
"epoch": 1.92,
"learning_rate": 0.0002,
"loss": 1.594,
"step": 2260
},
{
"epoch": 1.92,
"learning_rate": 0.0002,
"loss": 1.6135,
"step": 2270
},
{
"epoch": 1.93,
"learning_rate": 0.0002,
"loss": 1.5673,
"step": 2280
},
{
"epoch": 1.94,
"learning_rate": 0.0002,
"loss": 1.5132,
"step": 2290
},
{
"epoch": 1.95,
"learning_rate": 0.0002,
"loss": 1.5225,
"step": 2300
},
{
"epoch": 1.96,
"learning_rate": 0.0002,
"loss": 1.5003,
"step": 2310
},
{
"epoch": 1.97,
"learning_rate": 0.0002,
"loss": 1.6272,
"step": 2320
},
{
"epoch": 1.98,
"learning_rate": 0.0002,
"loss": 1.4995,
"step": 2330
},
{
"epoch": 1.98,
"learning_rate": 0.0002,
"loss": 1.5104,
"step": 2340
},
{
"epoch": 1.99,
"learning_rate": 0.0002,
"loss": 1.5459,
"step": 2350
},
{
"epoch": 2.0,
"learning_rate": 0.0002,
"loss": 1.7678,
"step": 2360
},
{
"epoch": 2.01,
"learning_rate": 0.0002,
"loss": 1.4259,
"step": 2370
},
{
"epoch": 2.02,
"learning_rate": 0.0002,
"loss": 1.5745,
"step": 2380
},
{
"epoch": 2.03,
"learning_rate": 0.0002,
"loss": 1.5728,
"step": 2390
},
{
"epoch": 2.04,
"learning_rate": 0.0002,
"loss": 1.5811,
"step": 2400
},
{
"epoch": 2.04,
"learning_rate": 0.0002,
"loss": 1.3364,
"step": 2410
},
{
"epoch": 2.05,
"learning_rate": 0.0002,
"loss": 1.465,
"step": 2420
},
{
"epoch": 2.06,
"learning_rate": 0.0002,
"loss": 1.4702,
"step": 2430
},
{
"epoch": 2.07,
"learning_rate": 0.0002,
"loss": 1.5371,
"step": 2440
},
{
"epoch": 2.08,
"learning_rate": 0.0002,
"loss": 1.6216,
"step": 2450
},
{
"epoch": 2.09,
"learning_rate": 0.0002,
"loss": 1.4969,
"step": 2460
},
{
"epoch": 2.09,
"learning_rate": 0.0002,
"loss": 1.5256,
"step": 2470
},
{
"epoch": 2.1,
"learning_rate": 0.0002,
"loss": 1.5764,
"step": 2480
},
{
"epoch": 2.11,
"learning_rate": 0.0002,
"loss": 1.5568,
"step": 2490
},
{
"epoch": 2.12,
"learning_rate": 0.0002,
"loss": 1.6166,
"step": 2500
},
{
"epoch": 2.13,
"learning_rate": 0.0002,
"loss": 1.3921,
"step": 2510
},
{
"epoch": 2.14,
"learning_rate": 0.0002,
"loss": 1.6939,
"step": 2520
},
{
"epoch": 2.15,
"learning_rate": 0.0002,
"loss": 1.5767,
"step": 2530
},
{
"epoch": 2.15,
"learning_rate": 0.0002,
"loss": 1.5017,
"step": 2540
},
{
"epoch": 2.16,
"learning_rate": 0.0002,
"loss": 1.5693,
"step": 2550
},
{
"epoch": 2.17,
"learning_rate": 0.0002,
"loss": 1.5234,
"step": 2560
},
{
"epoch": 2.18,
"learning_rate": 0.0002,
"loss": 1.5158,
"step": 2570
},
{
"epoch": 2.19,
"learning_rate": 0.0002,
"loss": 1.4764,
"step": 2580
},
{
"epoch": 2.2,
"learning_rate": 0.0002,
"loss": 1.4859,
"step": 2590
},
{
"epoch": 2.2,
"learning_rate": 0.0002,
"loss": 1.7474,
"step": 2600
},
{
"epoch": 2.21,
"learning_rate": 0.0002,
"loss": 1.5771,
"step": 2610
},
{
"epoch": 2.22,
"learning_rate": 0.0002,
"loss": 1.5878,
"step": 2620
},
{
"epoch": 2.23,
"learning_rate": 0.0002,
"loss": 1.5759,
"step": 2630
},
{
"epoch": 2.24,
"learning_rate": 0.0002,
"loss": 1.6315,
"step": 2640
},
{
"epoch": 2.25,
"learning_rate": 0.0002,
"loss": 1.6279,
"step": 2650
},
{
"epoch": 2.26,
"learning_rate": 0.0002,
"loss": 1.5291,
"step": 2660
},
{
"epoch": 2.26,
"learning_rate": 0.0002,
"loss": 1.6262,
"step": 2670
},
{
"epoch": 2.27,
"learning_rate": 0.0002,
"loss": 1.4154,
"step": 2680
},
{
"epoch": 2.28,
"learning_rate": 0.0002,
"loss": 1.6217,
"step": 2690
},
{
"epoch": 2.29,
"learning_rate": 0.0002,
"loss": 1.6745,
"step": 2700
},
{
"epoch": 2.3,
"learning_rate": 0.0002,
"loss": 1.5109,
"step": 2710
},
{
"epoch": 2.31,
"learning_rate": 0.0002,
"loss": 1.4785,
"step": 2720
},
{
"epoch": 2.32,
"learning_rate": 0.0002,
"loss": 1.5506,
"step": 2730
},
{
"epoch": 2.32,
"learning_rate": 0.0002,
"loss": 1.5805,
"step": 2740
},
{
"epoch": 2.33,
"learning_rate": 0.0002,
"loss": 1.6135,
"step": 2750
},
{
"epoch": 2.34,
"learning_rate": 0.0002,
"loss": 1.462,
"step": 2760
},
{
"epoch": 2.35,
"learning_rate": 0.0002,
"loss": 1.5708,
"step": 2770
},
{
"epoch": 2.36,
"learning_rate": 0.0002,
"loss": 1.447,
"step": 2780
},
{
"epoch": 2.37,
"learning_rate": 0.0002,
"loss": 1.5131,
"step": 2790
},
{
"epoch": 2.37,
"learning_rate": 0.0002,
"loss": 1.5665,
"step": 2800
},
{
"epoch": 2.38,
"learning_rate": 0.0002,
"loss": 1.4354,
"step": 2810
},
{
"epoch": 2.39,
"learning_rate": 0.0002,
"loss": 1.5063,
"step": 2820
},
{
"epoch": 2.4,
"learning_rate": 0.0002,
"loss": 1.69,
"step": 2830
},
{
"epoch": 2.41,
"learning_rate": 0.0002,
"loss": 1.5463,
"step": 2840
},
{
"epoch": 2.42,
"learning_rate": 0.0002,
"loss": 1.6239,
"step": 2850
},
{
"epoch": 2.43,
"learning_rate": 0.0002,
"loss": 1.4631,
"step": 2860
},
{
"epoch": 2.43,
"learning_rate": 0.0002,
"loss": 1.5087,
"step": 2870
},
{
"epoch": 2.44,
"learning_rate": 0.0002,
"loss": 1.6203,
"step": 2880
},
{
"epoch": 2.45,
"learning_rate": 0.0002,
"loss": 1.5759,
"step": 2890
},
{
"epoch": 2.46,
"learning_rate": 0.0002,
"loss": 1.7263,
"step": 2900
},
{
"epoch": 2.47,
"learning_rate": 0.0002,
"loss": 1.4421,
"step": 2910
},
{
"epoch": 2.48,
"learning_rate": 0.0002,
"loss": 1.6171,
"step": 2920
},
{
"epoch": 2.48,
"learning_rate": 0.0002,
"loss": 1.607,
"step": 2930
},
{
"epoch": 2.49,
"learning_rate": 0.0002,
"loss": 1.5498,
"step": 2940
},
{
"epoch": 2.5,
"learning_rate": 0.0002,
"loss": 1.6086,
"step": 2950
},
{
"epoch": 2.51,
"learning_rate": 0.0002,
"loss": 1.4261,
"step": 2960
},
{
"epoch": 2.52,
"learning_rate": 0.0002,
"loss": 1.6494,
"step": 2970
},
{
"epoch": 2.53,
"learning_rate": 0.0002,
"loss": 1.62,
"step": 2980
},
{
"epoch": 2.54,
"learning_rate": 0.0002,
"loss": 1.5495,
"step": 2990
},
{
"epoch": 2.54,
"learning_rate": 0.0002,
"loss": 1.5679,
"step": 3000
},
{
"epoch": 2.55,
"learning_rate": 0.0002,
"loss": 1.4961,
"step": 3010
},
{
"epoch": 2.56,
"learning_rate": 0.0002,
"loss": 1.7111,
"step": 3020
},
{
"epoch": 2.57,
"learning_rate": 0.0002,
"loss": 1.6501,
"step": 3030
},
{
"epoch": 2.58,
"learning_rate": 0.0002,
"loss": 1.5388,
"step": 3040
},
{
"epoch": 2.59,
"learning_rate": 0.0002,
"loss": 1.5594,
"step": 3050
},
{
"epoch": 2.59,
"learning_rate": 0.0002,
"loss": 1.5439,
"step": 3060
},
{
"epoch": 2.6,
"learning_rate": 0.0002,
"loss": 1.5835,
"step": 3070
},
{
"epoch": 2.61,
"learning_rate": 0.0002,
"loss": 1.5374,
"step": 3080
},
{
"epoch": 2.62,
"learning_rate": 0.0002,
"loss": 1.6671,
"step": 3090
},
{
"epoch": 2.63,
"learning_rate": 0.0002,
"loss": 1.5936,
"step": 3100
},
{
"epoch": 2.64,
"learning_rate": 0.0002,
"loss": 1.4547,
"step": 3110
},
{
"epoch": 2.65,
"learning_rate": 0.0002,
"loss": 1.5545,
"step": 3120
},
{
"epoch": 2.65,
"learning_rate": 0.0002,
"loss": 1.5129,
"step": 3130
},
{
"epoch": 2.66,
"learning_rate": 0.0002,
"loss": 1.5686,
"step": 3140
},
{
"epoch": 2.67,
"learning_rate": 0.0002,
"loss": 1.612,
"step": 3150
},
{
"epoch": 2.68,
"learning_rate": 0.0002,
"loss": 1.3861,
"step": 3160
},
{
"epoch": 2.69,
"learning_rate": 0.0002,
"loss": 1.5814,
"step": 3170
},
{
"epoch": 2.7,
"learning_rate": 0.0002,
"loss": 1.6375,
"step": 3180
},
{
"epoch": 2.71,
"learning_rate": 0.0002,
"loss": 1.3922,
"step": 3190
},
{
"epoch": 2.71,
"learning_rate": 0.0002,
"loss": 1.6834,
"step": 3200
},
{
"epoch": 2.72,
"learning_rate": 0.0002,
"loss": 1.5478,
"step": 3210
},
{
"epoch": 2.73,
"learning_rate": 0.0002,
"loss": 1.5009,
"step": 3220
},
{
"epoch": 2.74,
"learning_rate": 0.0002,
"loss": 1.5824,
"step": 3230
},
{
"epoch": 2.75,
"learning_rate": 0.0002,
"loss": 1.546,
"step": 3240
},
{
"epoch": 2.76,
"learning_rate": 0.0002,
"loss": 1.6216,
"step": 3250
},
{
"epoch": 2.76,
"learning_rate": 0.0002,
"loss": 1.4876,
"step": 3260
},
{
"epoch": 2.77,
"learning_rate": 0.0002,
"loss": 1.5886,
"step": 3270
},
{
"epoch": 2.78,
"learning_rate": 0.0002,
"loss": 1.5217,
"step": 3280
},
{
"epoch": 2.79,
"learning_rate": 0.0002,
"loss": 1.5341,
"step": 3290
},
{
"epoch": 2.8,
"learning_rate": 0.0002,
"loss": 1.6274,
"step": 3300
},
{
"epoch": 2.81,
"learning_rate": 0.0002,
"loss": 1.4551,
"step": 3310
},
{
"epoch": 2.82,
"learning_rate": 0.0002,
"loss": 1.5824,
"step": 3320
},
{
"epoch": 2.82,
"learning_rate": 0.0002,
"loss": 1.3791,
"step": 3330
},
{
"epoch": 2.83,
"learning_rate": 0.0002,
"loss": 1.5234,
"step": 3340
},
{
"epoch": 2.84,
"learning_rate": 0.0002,
"loss": 1.6358,
"step": 3350
},
{
"epoch": 2.85,
"learning_rate": 0.0002,
"loss": 1.404,
"step": 3360
},
{
"epoch": 2.86,
"learning_rate": 0.0002,
"loss": 1.5611,
"step": 3370
},
{
"epoch": 2.87,
"learning_rate": 0.0002,
"loss": 1.6879,
"step": 3380
},
{
"epoch": 2.87,
"learning_rate": 0.0002,
"loss": 1.5839,
"step": 3390
},
{
"epoch": 2.88,
"learning_rate": 0.0002,
"loss": 1.52,
"step": 3400
},
{
"epoch": 2.89,
"learning_rate": 0.0002,
"loss": 1.4931,
"step": 3410
},
{
"epoch": 2.9,
"learning_rate": 0.0002,
"loss": 1.5241,
"step": 3420
},
{
"epoch": 2.91,
"learning_rate": 0.0002,
"loss": 1.541,
"step": 3430
},
{
"epoch": 2.92,
"learning_rate": 0.0002,
"loss": 1.5124,
"step": 3440
},
{
"epoch": 2.93,
"learning_rate": 0.0002,
"loss": 1.6682,
"step": 3450
},
{
"epoch": 2.93,
"learning_rate": 0.0002,
"loss": 1.4485,
"step": 3460
},
{
"epoch": 2.94,
"learning_rate": 0.0002,
"loss": 1.529,
"step": 3470
},
{
"epoch": 2.95,
"learning_rate": 0.0002,
"loss": 1.5178,
"step": 3480
},
{
"epoch": 2.96,
"learning_rate": 0.0002,
"loss": 1.6435,
"step": 3490
},
{
"epoch": 2.97,
"learning_rate": 0.0002,
"loss": 1.5454,
"step": 3500
},
{
"epoch": 2.98,
"learning_rate": 0.0002,
"loss": 1.4709,
"step": 3510
},
{
"epoch": 2.98,
"learning_rate": 0.0002,
"loss": 1.5763,
"step": 3520
},
{
"epoch": 2.99,
"learning_rate": 0.0002,
"loss": 1.5715,
"step": 3530
},
{
"epoch": 3.0,
"learning_rate": 0.0002,
"loss": 1.5391,
"step": 3540
},
{
"epoch": 3.01,
"learning_rate": 0.0002,
"loss": 1.5817,
"step": 3550
},
{
"epoch": 3.02,
"learning_rate": 0.0002,
"loss": 1.528,
"step": 3560
},
{
"epoch": 3.03,
"learning_rate": 0.0002,
"loss": 1.5876,
"step": 3570
},
{
"epoch": 3.04,
"learning_rate": 0.0002,
"loss": 1.5378,
"step": 3580
},
{
"epoch": 3.04,
"learning_rate": 0.0002,
"loss": 1.4347,
"step": 3590
},
{
"epoch": 3.05,
"learning_rate": 0.0002,
"loss": 1.4383,
"step": 3600
},
{
"epoch": 3.06,
"learning_rate": 0.0002,
"loss": 1.5023,
"step": 3610
},
{
"epoch": 3.07,
"learning_rate": 0.0002,
"loss": 1.5704,
"step": 3620
},
{
"epoch": 3.08,
"learning_rate": 0.0002,
"loss": 1.5496,
"step": 3630
},
{
"epoch": 3.09,
"learning_rate": 0.0002,
"loss": 1.4265,
"step": 3640
},
{
"epoch": 3.1,
"learning_rate": 0.0002,
"loss": 1.505,
"step": 3650
},
{
"epoch": 3.1,
"learning_rate": 0.0002,
"loss": 1.5635,
"step": 3660
},
{
"epoch": 3.11,
"learning_rate": 0.0002,
"loss": 1.521,
"step": 3670
},
{
"epoch": 3.12,
"learning_rate": 0.0002,
"loss": 1.4984,
"step": 3680
},
{
"epoch": 3.13,
"learning_rate": 0.0002,
"loss": 1.4338,
"step": 3690
},
{
"epoch": 3.14,
"learning_rate": 0.0002,
"loss": 1.6499,
"step": 3700
},
{
"epoch": 3.15,
"learning_rate": 0.0002,
"loss": 1.5462,
"step": 3710
},
{
"epoch": 3.15,
"learning_rate": 0.0002,
"loss": 1.5312,
"step": 3720
},
{
"epoch": 3.16,
"learning_rate": 0.0002,
"loss": 1.5251,
"step": 3730
},
{
"epoch": 3.17,
"learning_rate": 0.0002,
"loss": 1.281,
"step": 3740
},
{
"epoch": 3.18,
"learning_rate": 0.0002,
"loss": 1.588,
"step": 3750
},
{
"epoch": 3.19,
"learning_rate": 0.0002,
"loss": 1.4185,
"step": 3760
},
{
"epoch": 3.2,
"learning_rate": 0.0002,
"loss": 1.4537,
"step": 3770
},
{
"epoch": 3.21,
"learning_rate": 0.0002,
"loss": 1.6018,
"step": 3780
},
{
"epoch": 3.21,
"learning_rate": 0.0002,
"loss": 1.4299,
"step": 3790
},
{
"epoch": 3.22,
"learning_rate": 0.0002,
"loss": 1.5186,
"step": 3800
},
{
"epoch": 3.23,
"learning_rate": 0.0002,
"loss": 1.5442,
"step": 3810
},
{
"epoch": 3.24,
"learning_rate": 0.0002,
"loss": 1.5184,
"step": 3820
},
{
"epoch": 3.25,
"learning_rate": 0.0002,
"loss": 1.513,
"step": 3830
},
{
"epoch": 3.26,
"learning_rate": 0.0002,
"loss": 1.5123,
"step": 3840
},
{
"epoch": 3.26,
"learning_rate": 0.0002,
"loss": 1.5207,
"step": 3850
},
{
"epoch": 3.27,
"learning_rate": 0.0002,
"loss": 1.5201,
"step": 3860
},
{
"epoch": 3.28,
"learning_rate": 0.0002,
"loss": 1.577,
"step": 3870
},
{
"epoch": 3.29,
"learning_rate": 0.0002,
"loss": 1.5026,
"step": 3880
},
{
"epoch": 3.3,
"learning_rate": 0.0002,
"loss": 1.3096,
"step": 3890
},
{
"epoch": 3.31,
"learning_rate": 0.0002,
"loss": 1.5535,
"step": 3900
},
{
"epoch": 3.32,
"learning_rate": 0.0002,
"loss": 1.597,
"step": 3910
},
{
"epoch": 3.32,
"learning_rate": 0.0002,
"loss": 1.5472,
"step": 3920
},
{
"epoch": 3.33,
"learning_rate": 0.0002,
"loss": 1.5457,
"step": 3930
},
{
"epoch": 3.34,
"learning_rate": 0.0002,
"loss": 1.4,
"step": 3940
},
{
"epoch": 3.35,
"learning_rate": 0.0002,
"loss": 1.5405,
"step": 3950
},
{
"epoch": 3.36,
"learning_rate": 0.0002,
"loss": 1.6487,
"step": 3960
},
{
"epoch": 3.37,
"learning_rate": 0.0002,
"loss": 1.6052,
"step": 3970
},
{
"epoch": 3.38,
"learning_rate": 0.0002,
"loss": 1.5562,
"step": 3980
},
{
"epoch": 3.38,
"learning_rate": 0.0002,
"loss": 1.3533,
"step": 3990
},
{
"epoch": 3.39,
"learning_rate": 0.0002,
"loss": 1.5525,
"step": 4000
},
{
"epoch": 3.4,
"learning_rate": 0.0002,
"loss": 1.6005,
"step": 4010
},
{
"epoch": 3.41,
"learning_rate": 0.0002,
"loss": 1.5411,
"step": 4020
},
{
"epoch": 3.42,
"learning_rate": 0.0002,
"loss": 1.6184,
"step": 4030
},
{
"epoch": 3.43,
"learning_rate": 0.0002,
"loss": 1.4801,
"step": 4040
},
{
"epoch": 3.43,
"learning_rate": 0.0002,
"loss": 1.6219,
"step": 4050
},
{
"epoch": 3.44,
"learning_rate": 0.0002,
"loss": 1.5613,
"step": 4060
},
{
"epoch": 3.45,
"learning_rate": 0.0002,
"loss": 1.5599,
"step": 4070
},
{
"epoch": 3.46,
"learning_rate": 0.0002,
"loss": 1.5572,
"step": 4080
},
{
"epoch": 3.47,
"learning_rate": 0.0002,
"loss": 1.3984,
"step": 4090
},
{
"epoch": 3.48,
"learning_rate": 0.0002,
"loss": 1.6092,
"step": 4100
},
{
"epoch": 3.49,
"learning_rate": 0.0002,
"loss": 1.6004,
"step": 4110
},
{
"epoch": 3.49,
"learning_rate": 0.0002,
"loss": 1.5043,
"step": 4120
},
{
"epoch": 3.5,
"learning_rate": 0.0002,
"loss": 1.5486,
"step": 4130
},
{
"epoch": 3.51,
"learning_rate": 0.0002,
"loss": 1.3519,
"step": 4140
},
{
"epoch": 3.52,
"learning_rate": 0.0002,
"loss": 1.5584,
"step": 4150
},
{
"epoch": 3.53,
"learning_rate": 0.0002,
"loss": 1.5317,
"step": 4160
},
{
"epoch": 3.54,
"learning_rate": 0.0002,
"loss": 1.5425,
"step": 4170
},
{
"epoch": 3.54,
"learning_rate": 0.0002,
"loss": 1.5687,
"step": 4180
},
{
"epoch": 3.55,
"learning_rate": 0.0002,
"loss": 1.3501,
"step": 4190
},
{
"epoch": 3.56,
"learning_rate": 0.0002,
"loss": 1.4635,
"step": 4200
},
{
"epoch": 3.57,
"learning_rate": 0.0002,
"loss": 1.5026,
"step": 4210
},
{
"epoch": 3.58,
"learning_rate": 0.0002,
"loss": 1.464,
"step": 4220
},
{
"epoch": 3.59,
"learning_rate": 0.0002,
"loss": 1.5452,
"step": 4230
},
{
"epoch": 3.6,
"learning_rate": 0.0002,
"loss": 1.3647,
"step": 4240
},
{
"epoch": 3.6,
"learning_rate": 0.0002,
"loss": 1.5876,
"step": 4250
},
{
"epoch": 3.61,
"learning_rate": 0.0002,
"loss": 1.4965,
"step": 4260
},
{
"epoch": 3.62,
"learning_rate": 0.0002,
"loss": 1.54,
"step": 4270
},
{
"epoch": 3.63,
"learning_rate": 0.0002,
"loss": 1.552,
"step": 4280
},
{
"epoch": 3.64,
"learning_rate": 0.0002,
"loss": 1.3977,
"step": 4290
},
{
"epoch": 3.65,
"learning_rate": 0.0002,
"loss": 1.5418,
"step": 4300
},
{
"epoch": 3.65,
"learning_rate": 0.0002,
"loss": 1.6027,
"step": 4310
},
{
"epoch": 3.66,
"learning_rate": 0.0002,
"loss": 1.5297,
"step": 4320
},
{
"epoch": 3.67,
"learning_rate": 0.0002,
"loss": 1.6881,
"step": 4330
},
{
"epoch": 3.68,
"learning_rate": 0.0002,
"loss": 1.4749,
"step": 4340
},
{
"epoch": 3.69,
"learning_rate": 0.0002,
"loss": 1.6126,
"step": 4350
},
{
"epoch": 3.7,
"learning_rate": 0.0002,
"loss": 1.5104,
"step": 4360
},
{
"epoch": 3.71,
"learning_rate": 0.0002,
"loss": 1.6391,
"step": 4370
},
{
"epoch": 3.71,
"learning_rate": 0.0002,
"loss": 1.6787,
"step": 4380
},
{
"epoch": 3.72,
"learning_rate": 0.0002,
"loss": 1.4109,
"step": 4390
},
{
"epoch": 3.73,
"learning_rate": 0.0002,
"loss": 1.4054,
"step": 4400
},
{
"epoch": 3.74,
"learning_rate": 0.0002,
"loss": 1.563,
"step": 4410
},
{
"epoch": 3.75,
"learning_rate": 0.0002,
"loss": 1.5424,
"step": 4420
},
{
"epoch": 3.76,
"learning_rate": 0.0002,
"loss": 1.5182,
"step": 4430
},
{
"epoch": 3.77,
"learning_rate": 0.0002,
"loss": 1.3578,
"step": 4440
},
{
"epoch": 3.77,
"learning_rate": 0.0002,
"loss": 1.5236,
"step": 4450
},
{
"epoch": 3.78,
"learning_rate": 0.0002,
"loss": 1.5169,
"step": 4460
},
{
"epoch": 3.79,
"learning_rate": 0.0002,
"loss": 1.5977,
"step": 4470
},
{
"epoch": 3.8,
"learning_rate": 0.0002,
"loss": 1.6173,
"step": 4480
},
{
"epoch": 3.81,
"learning_rate": 0.0002,
"loss": 1.4385,
"step": 4490
},
{
"epoch": 3.82,
"learning_rate": 0.0002,
"loss": 1.5392,
"step": 4500
},
{
"epoch": 3.82,
"learning_rate": 0.0002,
"loss": 1.4421,
"step": 4510
},
{
"epoch": 3.83,
"learning_rate": 0.0002,
"loss": 1.5976,
"step": 4520
},
{
"epoch": 3.84,
"learning_rate": 0.0002,
"loss": 1.5677,
"step": 4530
},
{
"epoch": 3.85,
"learning_rate": 0.0002,
"loss": 1.3807,
"step": 4540
},
{
"epoch": 3.86,
"learning_rate": 0.0002,
"loss": 1.5043,
"step": 4550
},
{
"epoch": 3.87,
"learning_rate": 0.0002,
"loss": 1.481,
"step": 4560
},
{
"epoch": 3.88,
"learning_rate": 0.0002,
"loss": 1.5422,
"step": 4570
},
{
"epoch": 3.88,
"learning_rate": 0.0002,
"loss": 1.5324,
"step": 4580
},
{
"epoch": 3.89,
"learning_rate": 0.0002,
"loss": 1.4074,
"step": 4590
},
{
"epoch": 3.9,
"learning_rate": 0.0002,
"loss": 1.5912,
"step": 4600
},
{
"epoch": 3.91,
"learning_rate": 0.0002,
"loss": 1.4961,
"step": 4610
},
{
"epoch": 3.92,
"learning_rate": 0.0002,
"loss": 1.5955,
"step": 4620
},
{
"epoch": 3.93,
"learning_rate": 0.0002,
"loss": 1.5646,
"step": 4630
},
{
"epoch": 3.93,
"learning_rate": 0.0002,
"loss": 1.4775,
"step": 4640
},
{
"epoch": 3.94,
"learning_rate": 0.0002,
"loss": 1.5562,
"step": 4650
},
{
"epoch": 3.95,
"learning_rate": 0.0002,
"loss": 1.3141,
"step": 4660
},
{
"epoch": 3.96,
"learning_rate": 0.0002,
"loss": 1.643,
"step": 4670
},
{
"epoch": 3.97,
"learning_rate": 0.0002,
"loss": 1.4919,
"step": 4680
},
{
"epoch": 3.98,
"learning_rate": 0.0002,
"loss": 1.3392,
"step": 4690
},
{
"epoch": 3.99,
"learning_rate": 0.0002,
"loss": 1.5918,
"step": 4700
},
{
"epoch": 3.99,
"learning_rate": 0.0002,
"loss": 1.4699,
"step": 4710
},
{
"epoch": 4.0,
"learning_rate": 0.0002,
"loss": 1.4662,
"step": 4720
},
{
"epoch": 4.01,
"learning_rate": 0.0002,
"loss": 1.489,
"step": 4730
},
{
"epoch": 4.02,
"learning_rate": 0.0002,
"loss": 1.4084,
"step": 4740
},
{
"epoch": 4.03,
"learning_rate": 0.0002,
"loss": 1.5435,
"step": 4750
},
{
"epoch": 4.04,
"learning_rate": 0.0002,
"loss": 1.5578,
"step": 4760
},
{
"epoch": 4.04,
"learning_rate": 0.0002,
"loss": 1.267,
"step": 4770
},
{
"epoch": 4.05,
"learning_rate": 0.0002,
"loss": 1.6063,
"step": 4780
},
{
"epoch": 4.06,
"learning_rate": 0.0002,
"loss": 1.5703,
"step": 4790
},
{
"epoch": 4.07,
"learning_rate": 0.0002,
"loss": 1.5615,
"step": 4800
},
{
"epoch": 4.08,
"learning_rate": 0.0002,
"loss": 1.5027,
"step": 4810
},
{
"epoch": 4.09,
"learning_rate": 0.0002,
"loss": 1.2511,
"step": 4820
},
{
"epoch": 4.1,
"learning_rate": 0.0002,
"loss": 1.4244,
"step": 4830
},
{
"epoch": 4.1,
"learning_rate": 0.0002,
"loss": 1.5168,
"step": 4840
},
{
"epoch": 4.11,
"learning_rate": 0.0002,
"loss": 1.5112,
"step": 4850
},
{
"epoch": 4.12,
"learning_rate": 0.0002,
"loss": 1.522,
"step": 4860
},
{
"epoch": 4.13,
"learning_rate": 0.0002,
"loss": 1.3662,
"step": 4870
},
{
"epoch": 4.14,
"learning_rate": 0.0002,
"loss": 1.5248,
"step": 4880
},
{
"epoch": 4.15,
"learning_rate": 0.0002,
"loss": 1.5016,
"step": 4890
},
{
"epoch": 4.16,
"learning_rate": 0.0002,
"loss": 1.4813,
"step": 4900
},
{
"epoch": 4.16,
"learning_rate": 0.0002,
"loss": 1.4656,
"step": 4910
},
{
"epoch": 4.17,
"learning_rate": 0.0002,
"loss": 1.3701,
"step": 4920
},
{
"epoch": 4.18,
"learning_rate": 0.0002,
"loss": 1.4482,
"step": 4930
},
{
"epoch": 4.19,
"learning_rate": 0.0002,
"loss": 1.4892,
"step": 4940
},
{
"epoch": 4.2,
"learning_rate": 0.0002,
"loss": 1.5361,
"step": 4950
},
{
"epoch": 4.21,
"learning_rate": 0.0002,
"loss": 1.5442,
"step": 4960
},
{
"epoch": 4.21,
"learning_rate": 0.0002,
"loss": 1.3156,
"step": 4970
},
{
"epoch": 4.22,
"learning_rate": 0.0002,
"loss": 1.586,
"step": 4980
},
{
"epoch": 4.23,
"learning_rate": 0.0002,
"loss": 1.4893,
"step": 4990
},
{
"epoch": 4.24,
"learning_rate": 0.0002,
"loss": 1.4546,
"step": 5000
},
{
"epoch": 4.25,
"learning_rate": 0.0002,
"loss": 1.5129,
"step": 5010
},
{
"epoch": 4.26,
"learning_rate": 0.0002,
"loss": 1.209,
"step": 5020
},
{
"epoch": 4.27,
"learning_rate": 0.0002,
"loss": 1.623,
"step": 5030
},
{
"epoch": 4.27,
"learning_rate": 0.0002,
"loss": 1.6484,
"step": 5040
},
{
"epoch": 4.28,
"learning_rate": 0.0002,
"loss": 1.5791,
"step": 5050
},
{
"epoch": 4.29,
"learning_rate": 0.0002,
"loss": 1.4645,
"step": 5060
},
{
"epoch": 4.3,
"learning_rate": 0.0002,
"loss": 1.2736,
"step": 5070
},
{
"epoch": 4.31,
"learning_rate": 0.0002,
"loss": 1.5257,
"step": 5080
},
{
"epoch": 4.32,
"learning_rate": 0.0002,
"loss": 1.5646,
"step": 5090
},
{
"epoch": 4.32,
"learning_rate": 0.0002,
"loss": 1.5087,
"step": 5100
},
{
"epoch": 4.33,
"learning_rate": 0.0002,
"loss": 1.3825,
"step": 5110
},
{
"epoch": 4.34,
"learning_rate": 0.0002,
"loss": 1.2006,
"step": 5120
},
{
"epoch": 4.35,
"learning_rate": 0.0002,
"loss": 1.5815,
"step": 5130
},
{
"epoch": 4.36,
"learning_rate": 0.0002,
"loss": 1.4795,
"step": 5140
},
{
"epoch": 4.37,
"learning_rate": 0.0002,
"loss": 1.5083,
"step": 5150
},
{
"epoch": 4.38,
"learning_rate": 0.0002,
"loss": 1.5342,
"step": 5160
},
{
"epoch": 4.38,
"learning_rate": 0.0002,
"loss": 1.325,
"step": 5170
},
{
"epoch": 4.39,
"learning_rate": 0.0002,
"loss": 1.5091,
"step": 5180
},
{
"epoch": 4.4,
"learning_rate": 0.0002,
"loss": 1.6294,
"step": 5190
},
{
"epoch": 4.41,
"learning_rate": 0.0002,
"loss": 1.4513,
"step": 5200
},
{
"epoch": 4.42,
"learning_rate": 0.0002,
"loss": 1.5546,
"step": 5210
},
{
"epoch": 4.43,
"learning_rate": 0.0002,
"loss": 1.4262,
"step": 5220
},
{
"epoch": 4.44,
"learning_rate": 0.0002,
"loss": 1.535,
"step": 5230
},
{
"epoch": 4.44,
"learning_rate": 0.0002,
"loss": 1.4684,
"step": 5240
},
{
"epoch": 4.45,
"learning_rate": 0.0002,
"loss": 1.4877,
"step": 5250
},
{
"epoch": 4.46,
"learning_rate": 0.0002,
"loss": 1.4953,
"step": 5260
},
{
"epoch": 4.47,
"learning_rate": 0.0002,
"loss": 1.349,
"step": 5270
},
{
"epoch": 4.48,
"learning_rate": 0.0002,
"loss": 1.4479,
"step": 5280
},
{
"epoch": 4.49,
"learning_rate": 0.0002,
"loss": 1.52,
"step": 5290
},
{
"epoch": 4.49,
"learning_rate": 0.0002,
"loss": 1.5108,
"step": 5300
},
{
"epoch": 4.5,
"learning_rate": 0.0002,
"loss": 1.499,
"step": 5310
},
{
"epoch": 4.51,
"learning_rate": 0.0002,
"loss": 1.3457,
"step": 5320
},
{
"epoch": 4.52,
"learning_rate": 0.0002,
"loss": 1.5064,
"step": 5330
},
{
"epoch": 4.53,
"learning_rate": 0.0002,
"loss": 1.5746,
"step": 5340
},
{
"epoch": 4.54,
"learning_rate": 0.0002,
"loss": 1.5319,
"step": 5350
},
{
"epoch": 4.55,
"learning_rate": 0.0002,
"loss": 1.4537,
"step": 5360
},
{
"epoch": 4.55,
"learning_rate": 0.0002,
"loss": 1.2479,
"step": 5370
},
{
"epoch": 4.56,
"learning_rate": 0.0002,
"loss": 1.587,
"step": 5380
},
{
"epoch": 4.57,
"learning_rate": 0.0002,
"loss": 1.5436,
"step": 5390
},
{
"epoch": 4.58,
"learning_rate": 0.0002,
"loss": 1.4625,
"step": 5400
},
{
"epoch": 4.59,
"learning_rate": 0.0002,
"loss": 1.4104,
"step": 5410
},
{
"epoch": 4.6,
"learning_rate": 0.0002,
"loss": 1.275,
"step": 5420
},
{
"epoch": 4.6,
"learning_rate": 0.0002,
"loss": 1.482,
"step": 5430
},
{
"epoch": 4.61,
"learning_rate": 0.0002,
"loss": 1.4538,
"step": 5440
},
{
"epoch": 4.62,
"learning_rate": 0.0002,
"loss": 1.5787,
"step": 5450
},
{
"epoch": 4.63,
"learning_rate": 0.0002,
"loss": 1.5594,
"step": 5460
},
{
"epoch": 4.64,
"learning_rate": 0.0002,
"loss": 1.1875,
"step": 5470
},
{
"epoch": 4.65,
"learning_rate": 0.0002,
"loss": 1.5711,
"step": 5480
},
{
"epoch": 4.66,
"learning_rate": 0.0002,
"loss": 1.5763,
"step": 5490
},
{
"epoch": 4.66,
"learning_rate": 0.0002,
"loss": 1.5577,
"step": 5500
},
{
"epoch": 4.67,
"learning_rate": 0.0002,
"loss": 1.6211,
"step": 5510
},
{
"epoch": 4.68,
"learning_rate": 0.0002,
"loss": 1.3429,
"step": 5520
},
{
"epoch": 4.69,
"learning_rate": 0.0002,
"loss": 1.469,
"step": 5530
},
{
"epoch": 4.7,
"learning_rate": 0.0002,
"loss": 1.6777,
"step": 5540
},
{
"epoch": 4.71,
"learning_rate": 0.0002,
"loss": 1.5508,
"step": 5550
},
{
"epoch": 4.71,
"learning_rate": 0.0002,
"loss": 1.5462,
"step": 5560
},
{
"epoch": 4.72,
"learning_rate": 0.0002,
"loss": 1.423,
"step": 5570
},
{
"epoch": 4.73,
"learning_rate": 0.0002,
"loss": 1.5179,
"step": 5580
},
{
"epoch": 4.74,
"learning_rate": 0.0002,
"loss": 1.3995,
"step": 5590
},
{
"epoch": 4.75,
"learning_rate": 0.0002,
"loss": 1.511,
"step": 5600
},
{
"epoch": 4.76,
"learning_rate": 0.0002,
"loss": 1.3894,
"step": 5610
},
{
"epoch": 4.77,
"learning_rate": 0.0002,
"loss": 1.4058,
"step": 5620
},
{
"epoch": 4.77,
"learning_rate": 0.0002,
"loss": 1.5193,
"step": 5630
},
{
"epoch": 4.78,
"learning_rate": 0.0002,
"loss": 1.5863,
"step": 5640
},
{
"epoch": 4.79,
"learning_rate": 0.0002,
"loss": 1.6402,
"step": 5650
},
{
"epoch": 4.8,
"learning_rate": 0.0002,
"loss": 1.5331,
"step": 5660
},
{
"epoch": 4.81,
"learning_rate": 0.0002,
"loss": 1.3937,
"step": 5670
},
{
"epoch": 4.82,
"learning_rate": 0.0002,
"loss": 1.5331,
"step": 5680
},
{
"epoch": 4.83,
"learning_rate": 0.0002,
"loss": 1.5579,
"step": 5690
},
{
"epoch": 4.83,
"learning_rate": 0.0002,
"loss": 1.4868,
"step": 5700
},
{
"epoch": 4.84,
"learning_rate": 0.0002,
"loss": 1.5125,
"step": 5710
},
{
"epoch": 4.85,
"learning_rate": 0.0002,
"loss": 1.2138,
"step": 5720
},
{
"epoch": 4.86,
"learning_rate": 0.0002,
"loss": 1.4658,
"step": 5730
},
{
"epoch": 4.87,
"learning_rate": 0.0002,
"loss": 1.535,
"step": 5740
},
{
"epoch": 4.88,
"learning_rate": 0.0002,
"loss": 1.5209,
"step": 5750
},
{
"epoch": 4.88,
"learning_rate": 0.0002,
"loss": 1.5414,
"step": 5760
},
{
"epoch": 4.89,
"learning_rate": 0.0002,
"loss": 1.3484,
"step": 5770
},
{
"epoch": 4.9,
"learning_rate": 0.0002,
"loss": 1.5072,
"step": 5780
},
{
"epoch": 4.91,
"learning_rate": 0.0002,
"loss": 1.533,
"step": 5790
},
{
"epoch": 4.92,
"learning_rate": 0.0002,
"loss": 1.5316,
"step": 5800
},
{
"epoch": 4.93,
"learning_rate": 0.0002,
"loss": 1.4447,
"step": 5810
},
{
"epoch": 4.94,
"learning_rate": 0.0002,
"loss": 1.4276,
"step": 5820
},
{
"epoch": 4.94,
"learning_rate": 0.0002,
"loss": 1.6619,
"step": 5830
},
{
"epoch": 4.95,
"learning_rate": 0.0002,
"loss": 1.4997,
"step": 5840
},
{
"epoch": 4.96,
"learning_rate": 0.0002,
"loss": 1.6259,
"step": 5850
},
{
"epoch": 4.97,
"learning_rate": 0.0002,
"loss": 1.6372,
"step": 5860
},
{
"epoch": 4.98,
"learning_rate": 0.0002,
"loss": 1.3913,
"step": 5870
},
{
"epoch": 4.99,
"learning_rate": 0.0002,
"loss": 1.649,
"step": 5880
},
{
"epoch": 4.99,
"learning_rate": 0.0002,
"loss": 1.4818,
"step": 5890
},
{
"epoch": 5.0,
"learning_rate": 0.0002,
"loss": 1.3887,
"step": 5900
},
{
"epoch": 5.01,
"learning_rate": 0.0002,
"loss": 1.5075,
"step": 5910
},
{
"epoch": 5.02,
"learning_rate": 0.0002,
"loss": 1.4915,
"step": 5920
},
{
"epoch": 5.03,
"learning_rate": 0.0002,
"loss": 1.461,
"step": 5930
},
{
"epoch": 5.04,
"learning_rate": 0.0002,
"loss": 1.4279,
"step": 5940
},
{
"epoch": 5.05,
"learning_rate": 0.0002,
"loss": 1.2986,
"step": 5950
},
{
"epoch": 5.05,
"learning_rate": 0.0002,
"loss": 1.4594,
"step": 5960
},
{
"epoch": 5.06,
"learning_rate": 0.0002,
"loss": 1.5978,
"step": 5970
},
{
"epoch": 5.07,
"learning_rate": 0.0002,
"loss": 1.6462,
"step": 5980
},
{
"epoch": 5.08,
"learning_rate": 0.0002,
"loss": 1.4229,
"step": 5990
},
{
"epoch": 5.09,
"learning_rate": 0.0002,
"loss": 1.1898,
"step": 6000
},
{
"epoch": 5.1,
"learning_rate": 0.0002,
"loss": 1.52,
"step": 6010
},
{
"epoch": 5.1,
"learning_rate": 0.0002,
"loss": 1.6185,
"step": 6020
},
{
"epoch": 5.11,
"learning_rate": 0.0002,
"loss": 1.5638,
"step": 6030
},
{
"epoch": 5.12,
"learning_rate": 0.0002,
"loss": 1.4536,
"step": 6040
},
{
"epoch": 5.13,
"learning_rate": 0.0002,
"loss": 1.2273,
"step": 6050
},
{
"epoch": 5.14,
"learning_rate": 0.0002,
"loss": 1.5153,
"step": 6060
},
{
"epoch": 5.15,
"learning_rate": 0.0002,
"loss": 1.5322,
"step": 6070
},
{
"epoch": 5.16,
"learning_rate": 0.0002,
"loss": 1.5884,
"step": 6080
},
{
"epoch": 5.16,
"learning_rate": 0.0002,
"loss": 1.4536,
"step": 6090
},
{
"epoch": 5.17,
"learning_rate": 0.0002,
"loss": 1.3191,
"step": 6100
},
{
"epoch": 5.18,
"learning_rate": 0.0002,
"loss": 1.5591,
"step": 6110
},
{
"epoch": 5.19,
"learning_rate": 0.0002,
"loss": 1.5249,
"step": 6120
},
{
"epoch": 5.2,
"learning_rate": 0.0002,
"loss": 1.5196,
"step": 6130
},
{
"epoch": 5.21,
"learning_rate": 0.0002,
"loss": 1.3867,
"step": 6140
},
{
"epoch": 5.22,
"learning_rate": 0.0002,
"loss": 1.234,
"step": 6150
},
{
"epoch": 5.22,
"learning_rate": 0.0002,
"loss": 1.5363,
"step": 6160
},
{
"epoch": 5.23,
"learning_rate": 0.0002,
"loss": 1.3658,
"step": 6170
},
{
"epoch": 5.24,
"learning_rate": 0.0002,
"loss": 1.4364,
"step": 6180
},
{
"epoch": 5.25,
"learning_rate": 0.0002,
"loss": 1.3715,
"step": 6190
},
{
"epoch": 5.26,
"learning_rate": 0.0002,
"loss": 1.1881,
"step": 6200
},
{
"epoch": 5.27,
"learning_rate": 0.0002,
"loss": 1.55,
"step": 6210
},
{
"epoch": 5.27,
"learning_rate": 0.0002,
"loss": 1.531,
"step": 6220
},
{
"epoch": 5.28,
"learning_rate": 0.0002,
"loss": 1.449,
"step": 6230
},
{
"epoch": 5.29,
"learning_rate": 0.0002,
"loss": 1.4631,
"step": 6240
},
{
"epoch": 5.3,
"learning_rate": 0.0002,
"loss": 1.2799,
"step": 6250
},
{
"epoch": 5.31,
"learning_rate": 0.0002,
"loss": 1.5619,
"step": 6260
},
{
"epoch": 5.32,
"learning_rate": 0.0002,
"loss": 1.6012,
"step": 6270
},
{
"epoch": 5.33,
"learning_rate": 0.0002,
"loss": 1.5947,
"step": 6280
},
{
"epoch": 5.33,
"learning_rate": 0.0002,
"loss": 1.4251,
"step": 6290
},
{
"epoch": 5.34,
"learning_rate": 0.0002,
"loss": 1.3763,
"step": 6300
},
{
"epoch": 5.35,
"learning_rate": 0.0002,
"loss": 1.4792,
"step": 6310
},
{
"epoch": 5.36,
"learning_rate": 0.0002,
"loss": 1.5086,
"step": 6320
},
{
"epoch": 5.37,
"learning_rate": 0.0002,
"loss": 1.5464,
"step": 6330
},
{
"epoch": 5.38,
"learning_rate": 0.0002,
"loss": 1.4114,
"step": 6340
},
{
"epoch": 5.38,
"learning_rate": 0.0002,
"loss": 1.2575,
"step": 6350
},
{
"epoch": 5.39,
"learning_rate": 0.0002,
"loss": 1.5501,
"step": 6360
},
{
"epoch": 5.4,
"learning_rate": 0.0002,
"loss": 1.4801,
"step": 6370
},
{
"epoch": 5.41,
"learning_rate": 0.0002,
"loss": 1.4182,
"step": 6380
},
{
"epoch": 5.42,
"learning_rate": 0.0002,
"loss": 1.5154,
"step": 6390
},
{
"epoch": 5.43,
"learning_rate": 0.0002,
"loss": 1.2175,
"step": 6400
},
{
"epoch": 5.44,
"learning_rate": 0.0002,
"loss": 1.516,
"step": 6410
},
{
"epoch": 5.44,
"learning_rate": 0.0002,
"loss": 1.3966,
"step": 6420
},
{
"epoch": 5.45,
"learning_rate": 0.0002,
"loss": 1.5535,
"step": 6430
},
{
"epoch": 5.46,
"learning_rate": 0.0002,
"loss": 1.3998,
"step": 6440
},
{
"epoch": 5.47,
"learning_rate": 0.0002,
"loss": 1.2663,
"step": 6450
},
{
"epoch": 5.48,
"learning_rate": 0.0002,
"loss": 1.6977,
"step": 6460
},
{
"epoch": 5.49,
"learning_rate": 0.0002,
"loss": 1.4275,
"step": 6470
},
{
"epoch": 5.5,
"learning_rate": 0.0002,
"loss": 1.5268,
"step": 6480
},
{
"epoch": 5.5,
"learning_rate": 0.0002,
"loss": 1.4411,
"step": 6490
},
{
"epoch": 5.51,
"learning_rate": 0.0002,
"loss": 1.1291,
"step": 6500
},
{
"epoch": 5.52,
"learning_rate": 0.0002,
"loss": 1.5772,
"step": 6510
},
{
"epoch": 5.53,
"learning_rate": 0.0002,
"loss": 1.578,
"step": 6520
},
{
"epoch": 5.54,
"learning_rate": 0.0002,
"loss": 1.5055,
"step": 6530
},
{
"epoch": 5.55,
"learning_rate": 0.0002,
"loss": 1.511,
"step": 6540
},
{
"epoch": 5.55,
"learning_rate": 0.0002,
"loss": 1.3824,
"step": 6550
},
{
"epoch": 5.56,
"learning_rate": 0.0002,
"loss": 1.555,
"step": 6560
},
{
"epoch": 5.57,
"learning_rate": 0.0002,
"loss": 1.4135,
"step": 6570
},
{
"epoch": 5.58,
"learning_rate": 0.0002,
"loss": 1.5468,
"step": 6580
},
{
"epoch": 5.59,
"learning_rate": 0.0002,
"loss": 1.4291,
"step": 6590
},
{
"epoch": 5.6,
"learning_rate": 0.0002,
"loss": 1.3226,
"step": 6600
},
{
"epoch": 5.61,
"learning_rate": 0.0002,
"loss": 1.493,
"step": 6610
},
{
"epoch": 5.61,
"learning_rate": 0.0002,
"loss": 1.63,
"step": 6620
},
{
"epoch": 5.62,
"learning_rate": 0.0002,
"loss": 1.5038,
"step": 6630
},
{
"epoch": 5.63,
"learning_rate": 0.0002,
"loss": 1.3837,
"step": 6640
},
{
"epoch": 5.64,
"learning_rate": 0.0002,
"loss": 1.1463,
"step": 6650
},
{
"epoch": 5.65,
"learning_rate": 0.0002,
"loss": 1.4743,
"step": 6660
},
{
"epoch": 5.66,
"learning_rate": 0.0002,
"loss": 1.447,
"step": 6670
},
{
"epoch": 5.66,
"learning_rate": 0.0002,
"loss": 1.4767,
"step": 6680
},
{
"epoch": 5.67,
"learning_rate": 0.0002,
"loss": 1.3807,
"step": 6690
},
{
"epoch": 5.68,
"learning_rate": 0.0002,
"loss": 1.3412,
"step": 6700
},
{
"epoch": 5.69,
"learning_rate": 0.0002,
"loss": 1.6119,
"step": 6710
},
{
"epoch": 5.7,
"learning_rate": 0.0002,
"loss": 1.4933,
"step": 6720
},
{
"epoch": 5.71,
"learning_rate": 0.0002,
"loss": 1.4865,
"step": 6730
},
{
"epoch": 5.72,
"learning_rate": 0.0002,
"loss": 1.4406,
"step": 6740
},
{
"epoch": 5.72,
"learning_rate": 0.0002,
"loss": 1.2174,
"step": 6750
},
{
"epoch": 5.73,
"learning_rate": 0.0002,
"loss": 1.4711,
"step": 6760
},
{
"epoch": 5.74,
"learning_rate": 0.0002,
"loss": 1.5218,
"step": 6770
},
{
"epoch": 5.75,
"learning_rate": 0.0002,
"loss": 1.4247,
"step": 6780
},
{
"epoch": 5.76,
"learning_rate": 0.0002,
"loss": 1.4435,
"step": 6790
},
{
"epoch": 5.77,
"learning_rate": 0.0002,
"loss": 1.297,
"step": 6800
},
{
"epoch": 5.77,
"learning_rate": 0.0002,
"loss": 1.5848,
"step": 6810
},
{
"epoch": 5.78,
"learning_rate": 0.0002,
"loss": 1.5184,
"step": 6820
},
{
"epoch": 5.79,
"learning_rate": 0.0002,
"loss": 1.5139,
"step": 6830
},
{
"epoch": 5.8,
"learning_rate": 0.0002,
"loss": 1.5817,
"step": 6840
},
{
"epoch": 5.81,
"learning_rate": 0.0002,
"loss": 1.2052,
"step": 6850
},
{
"epoch": 5.82,
"learning_rate": 0.0002,
"loss": 1.4788,
"step": 6860
},
{
"epoch": 5.83,
"learning_rate": 0.0002,
"loss": 1.6348,
"step": 6870
},
{
"epoch": 5.83,
"learning_rate": 0.0002,
"loss": 1.4991,
"step": 6880
},
{
"epoch": 5.84,
"learning_rate": 0.0002,
"loss": 1.4767,
"step": 6890
},
{
"epoch": 5.85,
"learning_rate": 0.0002,
"loss": 1.3738,
"step": 6900
},
{
"epoch": 5.86,
"learning_rate": 0.0002,
"loss": 1.5229,
"step": 6910
},
{
"epoch": 5.87,
"learning_rate": 0.0002,
"loss": 1.4112,
"step": 6920
},
{
"epoch": 5.88,
"learning_rate": 0.0002,
"loss": 1.3928,
"step": 6930
},
{
"epoch": 5.89,
"learning_rate": 0.0002,
"loss": 1.4323,
"step": 6940
},
{
"epoch": 5.89,
"learning_rate": 0.0002,
"loss": 1.2127,
"step": 6950
},
{
"epoch": 5.9,
"learning_rate": 0.0002,
"loss": 1.5399,
"step": 6960
},
{
"epoch": 5.91,
"learning_rate": 0.0002,
"loss": 1.4532,
"step": 6970
},
{
"epoch": 5.92,
"learning_rate": 0.0002,
"loss": 1.5573,
"step": 6980
},
{
"epoch": 5.93,
"learning_rate": 0.0002,
"loss": 1.4063,
"step": 6990
},
{
"epoch": 5.94,
"learning_rate": 0.0002,
"loss": 1.1604,
"step": 7000
},
{
"epoch": 5.94,
"learning_rate": 0.0002,
"loss": 1.5204,
"step": 7010
},
{
"epoch": 5.95,
"learning_rate": 0.0002,
"loss": 1.4892,
"step": 7020
},
{
"epoch": 5.96,
"learning_rate": 0.0002,
"loss": 1.4432,
"step": 7030
},
{
"epoch": 5.97,
"learning_rate": 0.0002,
"loss": 1.3795,
"step": 7040
},
{
"epoch": 5.98,
"learning_rate": 0.0002,
"loss": 1.2603,
"step": 7050
},
{
"epoch": 5.99,
"learning_rate": 0.0002,
"loss": 1.5511,
"step": 7060
},
{
"epoch": 6.0,
"learning_rate": 0.0002,
"loss": 1.4368,
"step": 7070
},
{
"epoch": 6.0,
"learning_rate": 0.0002,
"loss": 1.3983,
"step": 7080
},
{
"epoch": 6.01,
"learning_rate": 0.0002,
"loss": 1.6457,
"step": 7090
},
{
"epoch": 6.02,
"learning_rate": 0.0002,
"loss": 1.4911,
"step": 7100
},
{
"epoch": 6.03,
"learning_rate": 0.0002,
"loss": 1.4557,
"step": 7110
},
{
"epoch": 6.04,
"learning_rate": 0.0002,
"loss": 1.3691,
"step": 7120
},
{
"epoch": 6.05,
"learning_rate": 0.0002,
"loss": 1.258,
"step": 7130
},
{
"epoch": 6.05,
"learning_rate": 0.0002,
"loss": 1.6249,
"step": 7140
},
{
"epoch": 6.06,
"learning_rate": 0.0002,
"loss": 1.4699,
"step": 7150
},
{
"epoch": 6.07,
"learning_rate": 0.0002,
"loss": 1.4954,
"step": 7160
},
{
"epoch": 6.08,
"learning_rate": 0.0002,
"loss": 1.3533,
"step": 7170
},
{
"epoch": 6.09,
"learning_rate": 0.0002,
"loss": 1.2077,
"step": 7180
},
{
"epoch": 6.1,
"learning_rate": 0.0002,
"loss": 1.3844,
"step": 7190
},
{
"epoch": 6.11,
"learning_rate": 0.0002,
"loss": 1.4854,
"step": 7200
},
{
"epoch": 6.11,
"learning_rate": 0.0002,
"loss": 1.3921,
"step": 7210
},
{
"epoch": 6.12,
"learning_rate": 0.0002,
"loss": 1.3416,
"step": 7220
},
{
"epoch": 6.13,
"learning_rate": 0.0002,
"loss": 1.2056,
"step": 7230
},
{
"epoch": 6.14,
"learning_rate": 0.0002,
"loss": 1.58,
"step": 7240
},
{
"epoch": 6.15,
"learning_rate": 0.0002,
"loss": 1.5476,
"step": 7250
},
{
"epoch": 6.16,
"learning_rate": 0.0002,
"loss": 1.5085,
"step": 7260
},
{
"epoch": 6.16,
"learning_rate": 0.0002,
"loss": 1.436,
"step": 7270
},
{
"epoch": 6.17,
"learning_rate": 0.0002,
"loss": 1.1278,
"step": 7280
},
{
"epoch": 6.18,
"learning_rate": 0.0002,
"loss": 1.6121,
"step": 7290
},
{
"epoch": 6.19,
"learning_rate": 0.0002,
"loss": 1.4968,
"step": 7300
},
{
"epoch": 6.2,
"learning_rate": 0.0002,
"loss": 1.5557,
"step": 7310
},
{
"epoch": 6.21,
"learning_rate": 0.0002,
"loss": 1.2907,
"step": 7320
},
{
"epoch": 6.22,
"learning_rate": 0.0002,
"loss": 1.0883,
"step": 7330
},
{
"epoch": 6.22,
"learning_rate": 0.0002,
"loss": 1.4494,
"step": 7340
},
{
"epoch": 6.23,
"learning_rate": 0.0002,
"loss": 1.5522,
"step": 7350
},
{
"epoch": 6.24,
"learning_rate": 0.0002,
"loss": 1.4535,
"step": 7360
},
{
"epoch": 6.25,
"learning_rate": 0.0002,
"loss": 1.3859,
"step": 7370
},
{
"epoch": 6.26,
"learning_rate": 0.0002,
"loss": 1.2116,
"step": 7380
},
{
"epoch": 6.27,
"learning_rate": 0.0002,
"loss": 1.6045,
"step": 7390
},
{
"epoch": 6.28,
"learning_rate": 0.0002,
"loss": 1.5672,
"step": 7400
},
{
"epoch": 6.28,
"learning_rate": 0.0002,
"loss": 1.4037,
"step": 7410
},
{
"epoch": 6.29,
"learning_rate": 0.0002,
"loss": 1.5343,
"step": 7420
},
{
"epoch": 6.3,
"learning_rate": 0.0002,
"loss": 1.2923,
"step": 7430
},
{
"epoch": 6.31,
"learning_rate": 0.0002,
"loss": 1.5149,
"step": 7440
},
{
"epoch": 6.32,
"learning_rate": 0.0002,
"loss": 1.4317,
"step": 7450
},
{
"epoch": 6.33,
"learning_rate": 0.0002,
"loss": 1.5348,
"step": 7460
},
{
"epoch": 6.33,
"learning_rate": 0.0002,
"loss": 1.3813,
"step": 7470
},
{
"epoch": 6.34,
"learning_rate": 0.0002,
"loss": 1.2469,
"step": 7480
},
{
"epoch": 6.35,
"learning_rate": 0.0002,
"loss": 1.584,
"step": 7490
},
{
"epoch": 6.36,
"learning_rate": 0.0002,
"loss": 1.3528,
"step": 7500
},
{
"epoch": 6.37,
"learning_rate": 0.0002,
"loss": 1.5387,
"step": 7510
},
{
"epoch": 6.38,
"learning_rate": 0.0002,
"loss": 1.4305,
"step": 7520
},
{
"epoch": 6.39,
"learning_rate": 0.0002,
"loss": 1.2885,
"step": 7530
},
{
"epoch": 6.39,
"learning_rate": 0.0002,
"loss": 1.6225,
"step": 7540
},
{
"epoch": 6.4,
"learning_rate": 0.0002,
"loss": 1.4623,
"step": 7550
},
{
"epoch": 6.41,
"learning_rate": 0.0002,
"loss": 1.5429,
"step": 7560
},
{
"epoch": 6.42,
"learning_rate": 0.0002,
"loss": 1.3256,
"step": 7570
},
{
"epoch": 6.43,
"learning_rate": 0.0002,
"loss": 1.2111,
"step": 7580
},
{
"epoch": 6.44,
"learning_rate": 0.0002,
"loss": 1.4675,
"step": 7590
},
{
"epoch": 6.44,
"learning_rate": 0.0002,
"loss": 1.4037,
"step": 7600
},
{
"epoch": 6.45,
"learning_rate": 0.0002,
"loss": 1.4552,
"step": 7610
},
{
"epoch": 6.46,
"learning_rate": 0.0002,
"loss": 1.4485,
"step": 7620
},
{
"epoch": 6.47,
"learning_rate": 0.0002,
"loss": 1.1495,
"step": 7630
},
{
"epoch": 6.48,
"learning_rate": 0.0002,
"loss": 1.4942,
"step": 7640
},
{
"epoch": 6.49,
"learning_rate": 0.0002,
"loss": 1.4281,
"step": 7650
},
{
"epoch": 6.5,
"learning_rate": 0.0002,
"loss": 1.4942,
"step": 7660
},
{
"epoch": 6.5,
"learning_rate": 0.0002,
"loss": 1.2583,
"step": 7670
},
{
"epoch": 6.51,
"learning_rate": 0.0002,
"loss": 1.2829,
"step": 7680
},
{
"epoch": 6.52,
"learning_rate": 0.0002,
"loss": 1.4964,
"step": 7690
},
{
"epoch": 6.53,
"learning_rate": 0.0002,
"loss": 1.6005,
"step": 7700
},
{
"epoch": 6.54,
"learning_rate": 0.0002,
"loss": 1.4124,
"step": 7710
},
{
"epoch": 6.55,
"learning_rate": 0.0002,
"loss": 1.2546,
"step": 7720
},
{
"epoch": 6.56,
"learning_rate": 0.0002,
"loss": 1.1718,
"step": 7730
},
{
"epoch": 6.56,
"learning_rate": 0.0002,
"loss": 1.4129,
"step": 7740
},
{
"epoch": 6.57,
"learning_rate": 0.0002,
"loss": 1.4846,
"step": 7750
},
{
"epoch": 6.58,
"learning_rate": 0.0002,
"loss": 1.5016,
"step": 7760
},
{
"epoch": 6.59,
"learning_rate": 0.0002,
"loss": 1.3891,
"step": 7770
},
{
"epoch": 6.6,
"learning_rate": 0.0002,
"loss": 1.2621,
"step": 7780
},
{
"epoch": 6.61,
"learning_rate": 0.0002,
"loss": 1.4797,
"step": 7790
},
{
"epoch": 6.61,
"learning_rate": 0.0002,
"loss": 1.5409,
"step": 7800
},
{
"epoch": 6.62,
"learning_rate": 0.0002,
"loss": 1.4809,
"step": 7810
},
{
"epoch": 6.63,
"learning_rate": 0.0002,
"loss": 1.3902,
"step": 7820
},
{
"epoch": 6.64,
"learning_rate": 0.0002,
"loss": 1.2547,
"step": 7830
},
{
"epoch": 6.65,
"learning_rate": 0.0002,
"loss": 1.5441,
"step": 7840
},
{
"epoch": 6.66,
"learning_rate": 0.0002,
"loss": 1.5562,
"step": 7850
},
{
"epoch": 6.67,
"learning_rate": 0.0002,
"loss": 1.4046,
"step": 7860
},
{
"epoch": 6.67,
"learning_rate": 0.0002,
"loss": 1.4045,
"step": 7870
},
{
"epoch": 6.68,
"learning_rate": 0.0002,
"loss": 1.2597,
"step": 7880
},
{
"epoch": 6.69,
"learning_rate": 0.0002,
"loss": 1.4591,
"step": 7890
},
{
"epoch": 6.7,
"learning_rate": 0.0002,
"loss": 1.4758,
"step": 7900
},
{
"epoch": 6.71,
"learning_rate": 0.0002,
"loss": 1.478,
"step": 7910
},
{
"epoch": 6.72,
"learning_rate": 0.0002,
"loss": 1.2977,
"step": 7920
},
{
"epoch": 6.72,
"learning_rate": 0.0002,
"loss": 1.1349,
"step": 7930
},
{
"epoch": 6.73,
"learning_rate": 0.0002,
"loss": 1.4475,
"step": 7940
},
{
"epoch": 6.74,
"learning_rate": 0.0002,
"loss": 1.4823,
"step": 7950
},
{
"epoch": 6.75,
"learning_rate": 0.0002,
"loss": 1.4775,
"step": 7960
},
{
"epoch": 6.76,
"learning_rate": 0.0002,
"loss": 1.3841,
"step": 7970
},
{
"epoch": 6.77,
"learning_rate": 0.0002,
"loss": 1.1831,
"step": 7980
},
{
"epoch": 6.78,
"learning_rate": 0.0002,
"loss": 1.6087,
"step": 7990
},
{
"epoch": 6.78,
"learning_rate": 0.0002,
"loss": 1.5955,
"step": 8000
},
{
"epoch": 6.79,
"learning_rate": 0.0002,
"loss": 1.4423,
"step": 8010
},
{
"epoch": 6.8,
"learning_rate": 0.0002,
"loss": 1.3646,
"step": 8020
},
{
"epoch": 6.81,
"learning_rate": 0.0002,
"loss": 1.3033,
"step": 8030
},
{
"epoch": 6.82,
"learning_rate": 0.0002,
"loss": 1.4167,
"step": 8040
},
{
"epoch": 6.83,
"learning_rate": 0.0002,
"loss": 1.484,
"step": 8050
},
{
"epoch": 6.83,
"learning_rate": 0.0002,
"loss": 1.5335,
"step": 8060
},
{
"epoch": 6.84,
"learning_rate": 0.0002,
"loss": 1.3122,
"step": 8070
},
{
"epoch": 6.85,
"learning_rate": 0.0002,
"loss": 1.1209,
"step": 8080
},
{
"epoch": 6.86,
"learning_rate": 0.0002,
"loss": 1.5175,
"step": 8090
},
{
"epoch": 6.87,
"learning_rate": 0.0002,
"loss": 1.5588,
"step": 8100
},
{
"epoch": 6.88,
"learning_rate": 0.0002,
"loss": 1.3836,
"step": 8110
},
{
"epoch": 6.89,
"learning_rate": 0.0002,
"loss": 1.3939,
"step": 8120
},
{
"epoch": 6.89,
"learning_rate": 0.0002,
"loss": 1.2827,
"step": 8130
},
{
"epoch": 6.9,
"learning_rate": 0.0002,
"loss": 1.552,
"step": 8140
},
{
"epoch": 6.91,
"learning_rate": 0.0002,
"loss": 1.4809,
"step": 8150
},
{
"epoch": 6.92,
"learning_rate": 0.0002,
"loss": 1.4922,
"step": 8160
},
{
"epoch": 6.93,
"learning_rate": 0.0002,
"loss": 1.382,
"step": 8170
},
{
"epoch": 6.94,
"learning_rate": 0.0002,
"loss": 1.1526,
"step": 8180
},
{
"epoch": 6.95,
"learning_rate": 0.0002,
"loss": 1.5794,
"step": 8190
},
{
"epoch": 6.95,
"learning_rate": 0.0002,
"loss": 1.4828,
"step": 8200
},
{
"epoch": 6.96,
"learning_rate": 0.0002,
"loss": 1.4657,
"step": 8210
},
{
"epoch": 6.97,
"learning_rate": 0.0002,
"loss": 1.3454,
"step": 8220
},
{
"epoch": 6.98,
"learning_rate": 0.0002,
"loss": 1.1721,
"step": 8230
},
{
"epoch": 6.99,
"learning_rate": 0.0002,
"loss": 1.4019,
"step": 8240
},
{
"epoch": 7.0,
"learning_rate": 0.0002,
"loss": 1.4363,
"step": 8250
},
{
"epoch": 7.0,
"learning_rate": 0.0002,
"loss": 1.203,
"step": 8260
},
{
"epoch": 7.01,
"learning_rate": 0.0002,
"loss": 1.6892,
"step": 8270
},
{
"epoch": 7.02,
"learning_rate": 0.0002,
"loss": 1.4186,
"step": 8280
},
{
"epoch": 7.03,
"learning_rate": 0.0002,
"loss": 1.4372,
"step": 8290
},
{
"epoch": 7.04,
"learning_rate": 0.0002,
"loss": 1.2173,
"step": 8300
},
{
"epoch": 7.05,
"learning_rate": 0.0002,
"loss": 1.2425,
"step": 8310
},
{
"epoch": 7.06,
"learning_rate": 0.0002,
"loss": 1.4162,
"step": 8320
},
{
"epoch": 7.06,
"learning_rate": 0.0002,
"loss": 1.5368,
"step": 8330
},
{
"epoch": 7.07,
"learning_rate": 0.0002,
"loss": 1.4177,
"step": 8340
},
{
"epoch": 7.08,
"learning_rate": 0.0002,
"loss": 1.3548,
"step": 8350
},
{
"epoch": 7.09,
"learning_rate": 0.0002,
"loss": 1.2481,
"step": 8360
},
{
"epoch": 7.1,
"learning_rate": 0.0002,
"loss": 1.4746,
"step": 8370
},
{
"epoch": 7.11,
"learning_rate": 0.0002,
"loss": 1.4281,
"step": 8380
},
{
"epoch": 7.11,
"learning_rate": 0.0002,
"loss": 1.4201,
"step": 8390
},
{
"epoch": 7.12,
"learning_rate": 0.0002,
"loss": 1.2541,
"step": 8400
},
{
"epoch": 7.13,
"learning_rate": 0.0002,
"loss": 1.1694,
"step": 8410
},
{
"epoch": 7.14,
"learning_rate": 0.0002,
"loss": 1.4667,
"step": 8420
},
{
"epoch": 7.15,
"learning_rate": 0.0002,
"loss": 1.3491,
"step": 8430
},
{
"epoch": 7.16,
"learning_rate": 0.0002,
"loss": 1.3915,
"step": 8440
},
{
"epoch": 7.17,
"learning_rate": 0.0002,
"loss": 1.1878,
"step": 8450
},
{
"epoch": 7.17,
"learning_rate": 0.0002,
"loss": 1.0918,
"step": 8460
},
{
"epoch": 7.18,
"learning_rate": 0.0002,
"loss": 1.4782,
"step": 8470
},
{
"epoch": 7.19,
"learning_rate": 0.0002,
"loss": 1.5082,
"step": 8480
},
{
"epoch": 7.2,
"learning_rate": 0.0002,
"loss": 1.4576,
"step": 8490
},
{
"epoch": 7.21,
"learning_rate": 0.0002,
"loss": 1.2424,
"step": 8500
},
{
"epoch": 7.22,
"learning_rate": 0.0002,
"loss": 1.2462,
"step": 8510
},
{
"epoch": 7.22,
"learning_rate": 0.0002,
"loss": 1.512,
"step": 8520
},
{
"epoch": 7.23,
"learning_rate": 0.0002,
"loss": 1.398,
"step": 8530
},
{
"epoch": 7.24,
"learning_rate": 0.0002,
"loss": 1.3696,
"step": 8540
},
{
"epoch": 7.25,
"learning_rate": 0.0002,
"loss": 1.2943,
"step": 8550
},
{
"epoch": 7.26,
"learning_rate": 0.0002,
"loss": 1.2604,
"step": 8560
},
{
"epoch": 7.27,
"learning_rate": 0.0002,
"loss": 1.3822,
"step": 8570
},
{
"epoch": 7.28,
"learning_rate": 0.0002,
"loss": 1.4384,
"step": 8580
},
{
"epoch": 7.28,
"learning_rate": 0.0002,
"loss": 1.3567,
"step": 8590
},
{
"epoch": 7.29,
"learning_rate": 0.0002,
"loss": 1.1633,
"step": 8600
},
{
"epoch": 7.3,
"learning_rate": 0.0002,
"loss": 1.2239,
"step": 8610
},
{
"epoch": 7.31,
"learning_rate": 0.0002,
"loss": 1.286,
"step": 8620
},
{
"epoch": 7.32,
"learning_rate": 0.0002,
"loss": 1.4776,
"step": 8630
},
{
"epoch": 7.33,
"learning_rate": 0.0002,
"loss": 1.4323,
"step": 8640
},
{
"epoch": 7.34,
"learning_rate": 0.0002,
"loss": 1.2915,
"step": 8650
},
{
"epoch": 7.34,
"learning_rate": 0.0002,
"loss": 1.1991,
"step": 8660
},
{
"epoch": 7.35,
"learning_rate": 0.0002,
"loss": 1.4924,
"step": 8670
},
{
"epoch": 7.36,
"learning_rate": 0.0002,
"loss": 1.52,
"step": 8680
},
{
"epoch": 7.37,
"learning_rate": 0.0002,
"loss": 1.3564,
"step": 8690
},
{
"epoch": 7.38,
"learning_rate": 0.0002,
"loss": 1.2837,
"step": 8700
},
{
"epoch": 7.39,
"learning_rate": 0.0002,
"loss": 1.24,
"step": 8710
},
{
"epoch": 7.39,
"learning_rate": 0.0002,
"loss": 1.5955,
"step": 8720
},
{
"epoch": 7.4,
"learning_rate": 0.0002,
"loss": 1.5096,
"step": 8730
},
{
"epoch": 7.41,
"learning_rate": 0.0002,
"loss": 1.4383,
"step": 8740
},
{
"epoch": 7.42,
"learning_rate": 0.0002,
"loss": 1.1425,
"step": 8750
},
{
"epoch": 7.43,
"learning_rate": 0.0002,
"loss": 1.1201,
"step": 8760
},
{
"epoch": 7.44,
"learning_rate": 0.0002,
"loss": 1.4737,
"step": 8770
},
{
"epoch": 7.45,
"learning_rate": 0.0002,
"loss": 1.4821,
"step": 8780
},
{
"epoch": 7.45,
"learning_rate": 0.0002,
"loss": 1.3373,
"step": 8790
},
{
"epoch": 7.46,
"learning_rate": 0.0002,
"loss": 1.2833,
"step": 8800
},
{
"epoch": 7.47,
"learning_rate": 0.0002,
"loss": 1.2052,
"step": 8810
},
{
"epoch": 7.48,
"learning_rate": 0.0002,
"loss": 1.4898,
"step": 8820
},
{
"epoch": 7.49,
"learning_rate": 0.0002,
"loss": 1.552,
"step": 8830
},
{
"epoch": 7.5,
"learning_rate": 0.0002,
"loss": 1.4002,
"step": 8840
},
{
"epoch": 7.5,
"learning_rate": 0.0002,
"loss": 1.3512,
"step": 8850
},
{
"epoch": 7.51,
"learning_rate": 0.0002,
"loss": 1.1392,
"step": 8860
},
{
"epoch": 7.52,
"learning_rate": 0.0002,
"loss": 1.6019,
"step": 8870
},
{
"epoch": 7.53,
"learning_rate": 0.0002,
"loss": 1.412,
"step": 8880
},
{
"epoch": 7.54,
"learning_rate": 0.0002,
"loss": 1.4842,
"step": 8890
},
{
"epoch": 7.55,
"learning_rate": 0.0002,
"loss": 1.4081,
"step": 8900
},
{
"epoch": 7.56,
"learning_rate": 0.0002,
"loss": 1.2884,
"step": 8910
},
{
"epoch": 7.56,
"learning_rate": 0.0002,
"loss": 1.5398,
"step": 8920
},
{
"epoch": 7.57,
"learning_rate": 0.0002,
"loss": 1.4211,
"step": 8930
},
{
"epoch": 7.58,
"learning_rate": 0.0002,
"loss": 1.4608,
"step": 8940
},
{
"epoch": 7.59,
"learning_rate": 0.0002,
"loss": 1.3281,
"step": 8950
},
{
"epoch": 7.6,
"learning_rate": 0.0002,
"loss": 1.1872,
"step": 8960
},
{
"epoch": 7.61,
"learning_rate": 0.0002,
"loss": 1.4193,
"step": 8970
},
{
"epoch": 7.62,
"learning_rate": 0.0002,
"loss": 1.5137,
"step": 8980
},
{
"epoch": 7.62,
"learning_rate": 0.0002,
"loss": 1.4427,
"step": 8990
},
{
"epoch": 7.63,
"learning_rate": 0.0002,
"loss": 1.2066,
"step": 9000
},
{
"epoch": 7.64,
"learning_rate": 0.0002,
"loss": 1.1746,
"step": 9010
},
{
"epoch": 7.65,
"learning_rate": 0.0002,
"loss": 1.5112,
"step": 9020
},
{
"epoch": 7.66,
"learning_rate": 0.0002,
"loss": 1.4964,
"step": 9030
},
{
"epoch": 7.67,
"learning_rate": 0.0002,
"loss": 1.4684,
"step": 9040
},
{
"epoch": 7.67,
"learning_rate": 0.0002,
"loss": 1.2576,
"step": 9050
},
{
"epoch": 7.68,
"learning_rate": 0.0002,
"loss": 1.1935,
"step": 9060
},
{
"epoch": 7.69,
"learning_rate": 0.0002,
"loss": 1.4668,
"step": 9070
},
{
"epoch": 7.7,
"learning_rate": 0.0002,
"loss": 1.5314,
"step": 9080
},
{
"epoch": 7.71,
"learning_rate": 0.0002,
"loss": 1.5164,
"step": 9090
},
{
"epoch": 7.72,
"learning_rate": 0.0002,
"loss": 1.3777,
"step": 9100
},
{
"epoch": 7.73,
"learning_rate": 0.0002,
"loss": 1.2683,
"step": 9110
},
{
"epoch": 7.73,
"learning_rate": 0.0002,
"loss": 1.5755,
"step": 9120
},
{
"epoch": 7.74,
"learning_rate": 0.0002,
"loss": 1.4987,
"step": 9130
},
{
"epoch": 7.75,
"learning_rate": 0.0002,
"loss": 1.4834,
"step": 9140
},
{
"epoch": 7.76,
"learning_rate": 0.0002,
"loss": 1.2279,
"step": 9150
},
{
"epoch": 7.77,
"learning_rate": 0.0002,
"loss": 1.246,
"step": 9160
},
{
"epoch": 7.78,
"learning_rate": 0.0002,
"loss": 1.4722,
"step": 9170
},
{
"epoch": 7.78,
"learning_rate": 0.0002,
"loss": 1.4459,
"step": 9180
},
{
"epoch": 7.79,
"learning_rate": 0.0002,
"loss": 1.4356,
"step": 9190
},
{
"epoch": 7.8,
"learning_rate": 0.0002,
"loss": 1.2317,
"step": 9200
},
{
"epoch": 7.81,
"learning_rate": 0.0002,
"loss": 1.2911,
"step": 9210
},
{
"epoch": 7.82,
"learning_rate": 0.0002,
"loss": 1.5216,
"step": 9220
},
{
"epoch": 7.83,
"learning_rate": 0.0002,
"loss": 1.4873,
"step": 9230
},
{
"epoch": 7.84,
"learning_rate": 0.0002,
"loss": 1.5899,
"step": 9240
},
{
"epoch": 7.84,
"learning_rate": 0.0002,
"loss": 1.3099,
"step": 9250
},
{
"epoch": 7.85,
"learning_rate": 0.0002,
"loss": 1.2797,
"step": 9260
},
{
"epoch": 7.86,
"learning_rate": 0.0002,
"loss": 1.5825,
"step": 9270
},
{
"epoch": 7.87,
"learning_rate": 0.0002,
"loss": 1.483,
"step": 9280
},
{
"epoch": 7.88,
"learning_rate": 0.0002,
"loss": 1.5077,
"step": 9290
},
{
"epoch": 7.89,
"learning_rate": 0.0002,
"loss": 1.3687,
"step": 9300
},
{
"epoch": 7.89,
"learning_rate": 0.0002,
"loss": 1.2189,
"step": 9310
},
{
"epoch": 7.9,
"learning_rate": 0.0002,
"loss": 1.5236,
"step": 9320
},
{
"epoch": 7.91,
"learning_rate": 0.0002,
"loss": 1.3547,
"step": 9330
},
{
"epoch": 7.92,
"learning_rate": 0.0002,
"loss": 1.5465,
"step": 9340
},
{
"epoch": 7.93,
"learning_rate": 0.0002,
"loss": 1.2757,
"step": 9350
},
{
"epoch": 7.94,
"learning_rate": 0.0002,
"loss": 1.3276,
"step": 9360
},
{
"epoch": 7.95,
"learning_rate": 0.0002,
"loss": 1.7065,
"step": 9370
},
{
"epoch": 7.95,
"learning_rate": 0.0002,
"loss": 1.5325,
"step": 9380
},
{
"epoch": 7.96,
"learning_rate": 0.0002,
"loss": 1.3239,
"step": 9390
},
{
"epoch": 7.97,
"learning_rate": 0.0002,
"loss": 1.3755,
"step": 9400
},
{
"epoch": 7.98,
"learning_rate": 0.0002,
"loss": 1.2675,
"step": 9410
},
{
"epoch": 7.99,
"learning_rate": 0.0002,
"loss": 1.4128,
"step": 9420
},
{
"epoch": 8.0,
"learning_rate": 0.0002,
"loss": 1.3858,
"step": 9430
},
{
"epoch": 8.01,
"learning_rate": 0.0002,
"loss": 1.2553,
"step": 9440
},
{
"epoch": 8.01,
"learning_rate": 0.0002,
"loss": 1.3909,
"step": 9450
},
{
"epoch": 8.02,
"learning_rate": 0.0002,
"loss": 1.4506,
"step": 9460
},
{
"epoch": 8.03,
"learning_rate": 0.0002,
"loss": 1.404,
"step": 9470
},
{
"epoch": 8.04,
"learning_rate": 0.0002,
"loss": 1.2636,
"step": 9480
},
{
"epoch": 8.05,
"learning_rate": 0.0002,
"loss": 1.1642,
"step": 9490
},
{
"epoch": 8.06,
"learning_rate": 0.0002,
"loss": 1.5091,
"step": 9500
},
{
"epoch": 8.06,
"learning_rate": 0.0002,
"loss": 1.4799,
"step": 9510
},
{
"epoch": 8.07,
"learning_rate": 0.0002,
"loss": 1.4367,
"step": 9520
},
{
"epoch": 8.08,
"learning_rate": 0.0002,
"loss": 1.0981,
"step": 9530
},
{
"epoch": 8.09,
"learning_rate": 0.0002,
"loss": 1.2105,
"step": 9540
},
{
"epoch": 8.1,
"learning_rate": 0.0002,
"loss": 1.5877,
"step": 9550
},
{
"epoch": 8.11,
"learning_rate": 0.0002,
"loss": 1.4701,
"step": 9560
},
{
"epoch": 8.12,
"learning_rate": 0.0002,
"loss": 1.3886,
"step": 9570
},
{
"epoch": 8.12,
"learning_rate": 0.0002,
"loss": 1.2352,
"step": 9580
},
{
"epoch": 8.13,
"learning_rate": 0.0002,
"loss": 1.1406,
"step": 9590
},
{
"epoch": 8.14,
"learning_rate": 0.0002,
"loss": 1.4667,
"step": 9600
},
{
"epoch": 8.15,
"learning_rate": 0.0002,
"loss": 1.4823,
"step": 9610
},
{
"epoch": 8.16,
"learning_rate": 0.0002,
"loss": 1.3255,
"step": 9620
},
{
"epoch": 8.17,
"learning_rate": 0.0002,
"loss": 1.204,
"step": 9630
},
{
"epoch": 8.17,
"learning_rate": 0.0002,
"loss": 1.0877,
"step": 9640
},
{
"epoch": 8.18,
"learning_rate": 0.0002,
"loss": 1.4449,
"step": 9650
},
{
"epoch": 8.19,
"learning_rate": 0.0002,
"loss": 1.532,
"step": 9660
},
{
"epoch": 8.2,
"learning_rate": 0.0002,
"loss": 1.4103,
"step": 9670
},
{
"epoch": 8.21,
"learning_rate": 0.0002,
"loss": 1.2474,
"step": 9680
},
{
"epoch": 8.22,
"learning_rate": 0.0002,
"loss": 1.2159,
"step": 9690
},
{
"epoch": 8.23,
"learning_rate": 0.0002,
"loss": 1.4667,
"step": 9700
},
{
"epoch": 8.23,
"learning_rate": 0.0002,
"loss": 1.5464,
"step": 9710
},
{
"epoch": 8.24,
"learning_rate": 0.0002,
"loss": 1.479,
"step": 9720
},
{
"epoch": 8.25,
"learning_rate": 0.0002,
"loss": 1.1258,
"step": 9730
},
{
"epoch": 8.26,
"learning_rate": 0.0002,
"loss": 1.1221,
"step": 9740
},
{
"epoch": 8.27,
"learning_rate": 0.0002,
"loss": 1.5416,
"step": 9750
},
{
"epoch": 8.28,
"learning_rate": 0.0002,
"loss": 1.4036,
"step": 9760
},
{
"epoch": 8.28,
"learning_rate": 0.0002,
"loss": 1.4446,
"step": 9770
},
{
"epoch": 8.29,
"learning_rate": 0.0002,
"loss": 1.1907,
"step": 9780
},
{
"epoch": 8.3,
"learning_rate": 0.0002,
"loss": 1.217,
"step": 9790
},
{
"epoch": 8.31,
"learning_rate": 0.0002,
"loss": 1.4939,
"step": 9800
},
{
"epoch": 8.32,
"learning_rate": 0.0002,
"loss": 1.282,
"step": 9810
},
{
"epoch": 8.33,
"learning_rate": 0.0002,
"loss": 1.3567,
"step": 9820
},
{
"epoch": 8.34,
"learning_rate": 0.0002,
"loss": 1.0911,
"step": 9830
},
{
"epoch": 8.34,
"learning_rate": 0.0002,
"loss": 1.2465,
"step": 9840
},
{
"epoch": 8.35,
"learning_rate": 0.0002,
"loss": 1.4912,
"step": 9850
},
{
"epoch": 8.36,
"learning_rate": 0.0002,
"loss": 1.4224,
"step": 9860
},
{
"epoch": 8.37,
"learning_rate": 0.0002,
"loss": 1.3764,
"step": 9870
},
{
"epoch": 8.38,
"learning_rate": 0.0002,
"loss": 1.1824,
"step": 9880
},
{
"epoch": 8.39,
"learning_rate": 0.0002,
"loss": 1.2522,
"step": 9890
},
{
"epoch": 8.4,
"learning_rate": 0.0002,
"loss": 1.4143,
"step": 9900
},
{
"epoch": 8.4,
"learning_rate": 0.0002,
"loss": 1.5434,
"step": 9910
},
{
"epoch": 8.41,
"learning_rate": 0.0002,
"loss": 1.4295,
"step": 9920
},
{
"epoch": 8.42,
"learning_rate": 0.0002,
"loss": 1.2302,
"step": 9930
},
{
"epoch": 8.43,
"learning_rate": 0.0002,
"loss": 1.2449,
"step": 9940
},
{
"epoch": 8.44,
"learning_rate": 0.0002,
"loss": 1.5495,
"step": 9950
},
{
"epoch": 8.45,
"learning_rate": 0.0002,
"loss": 1.4749,
"step": 9960
},
{
"epoch": 8.45,
"learning_rate": 0.0002,
"loss": 1.3933,
"step": 9970
},
{
"epoch": 8.46,
"learning_rate": 0.0002,
"loss": 1.3481,
"step": 9980
},
{
"epoch": 8.47,
"learning_rate": 0.0002,
"loss": 1.1948,
"step": 9990
},
{
"epoch": 8.48,
"learning_rate": 0.0002,
"loss": 1.4184,
"step": 10000
}
],
"logging_steps": 10,
"max_steps": 10000,
"num_input_tokens_seen": 0,
"num_train_epochs": 9,
"save_steps": 500,
"total_flos": 8.063861670697574e+17,
"train_batch_size": 1,
"trial_name": null,
"trial_params": null
}