ihanif's picture
add checkpoint at 3500
d4e007a
raw
history blame
43.8 kB
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 44.30379746835443,
"global_step": 3500,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.13,
"learning_rate": 3.7499999999999996e-07,
"loss": 23.4903,
"step": 10
},
{
"epoch": 0.25,
"learning_rate": 7.499999999999999e-07,
"loss": 21.1248,
"step": 20
},
{
"epoch": 0.38,
"learning_rate": 1.1249999999999998e-06,
"loss": 22.9317,
"step": 30
},
{
"epoch": 0.51,
"learning_rate": 1.4625e-06,
"loss": 20.6205,
"step": 40
},
{
"epoch": 0.63,
"learning_rate": 1.8375e-06,
"loss": 22.2659,
"step": 50
},
{
"epoch": 0.76,
"learning_rate": 2.2124999999999996e-06,
"loss": 21.4276,
"step": 60
},
{
"epoch": 0.89,
"learning_rate": 2.5875e-06,
"loss": 21.8665,
"step": 70
},
{
"epoch": 1.01,
"learning_rate": 2.9624999999999996e-06,
"loss": 20.8487,
"step": 80
},
{
"epoch": 1.14,
"learning_rate": 3.3374999999999994e-06,
"loss": 21.838,
"step": 90
},
{
"epoch": 1.27,
"learning_rate": 3.7125e-06,
"loss": 19.0875,
"step": 100
},
{
"epoch": 1.39,
"learning_rate": 4.087499999999999e-06,
"loss": 19.3293,
"step": 110
},
{
"epoch": 1.52,
"learning_rate": 4.462499999999999e-06,
"loss": 16.2192,
"step": 120
},
{
"epoch": 1.65,
"learning_rate": 4.8375e-06,
"loss": 15.0126,
"step": 130
},
{
"epoch": 1.77,
"learning_rate": 5.2125e-06,
"loss": 13.5756,
"step": 140
},
{
"epoch": 1.9,
"learning_rate": 5.5874999999999994e-06,
"loss": 12.5467,
"step": 150
},
{
"epoch": 2.03,
"learning_rate": 5.962499999999999e-06,
"loss": 11.8743,
"step": 160
},
{
"epoch": 2.15,
"learning_rate": 6.3375e-06,
"loss": 11.4262,
"step": 170
},
{
"epoch": 2.28,
"learning_rate": 6.712499999999999e-06,
"loss": 10.3319,
"step": 180
},
{
"epoch": 2.41,
"learning_rate": 7.0874999999999995e-06,
"loss": 10.0354,
"step": 190
},
{
"epoch": 2.53,
"learning_rate": 7.4625e-06,
"loss": 9.8568,
"step": 200
},
{
"epoch": 2.66,
"learning_rate": 7.837499999999999e-06,
"loss": 9.4804,
"step": 210
},
{
"epoch": 2.78,
"learning_rate": 8.2125e-06,
"loss": 9.0891,
"step": 220
},
{
"epoch": 2.91,
"learning_rate": 8.5875e-06,
"loss": 8.6768,
"step": 230
},
{
"epoch": 3.04,
"learning_rate": 8.9625e-06,
"loss": 8.5948,
"step": 240
},
{
"epoch": 3.16,
"learning_rate": 9.3375e-06,
"loss": 8.4701,
"step": 250
},
{
"epoch": 3.29,
"learning_rate": 9.712499999999999e-06,
"loss": 8.2693,
"step": 260
},
{
"epoch": 3.42,
"learning_rate": 1.00875e-05,
"loss": 8.0611,
"step": 270
},
{
"epoch": 3.54,
"learning_rate": 1.04625e-05,
"loss": 8.0222,
"step": 280
},
{
"epoch": 3.67,
"learning_rate": 1.0837499999999997e-05,
"loss": 7.5698,
"step": 290
},
{
"epoch": 3.8,
"learning_rate": 1.1212499999999998e-05,
"loss": 7.5901,
"step": 300
},
{
"epoch": 3.92,
"learning_rate": 1.1587499999999999e-05,
"loss": 7.2712,
"step": 310
},
{
"epoch": 4.05,
"learning_rate": 1.19625e-05,
"loss": 7.4485,
"step": 320
},
{
"epoch": 4.18,
"learning_rate": 1.23375e-05,
"loss": 7.1484,
"step": 330
},
{
"epoch": 4.3,
"learning_rate": 1.2712499999999999e-05,
"loss": 7.1105,
"step": 340
},
{
"epoch": 4.43,
"learning_rate": 1.3087499999999998e-05,
"loss": 6.8399,
"step": 350
},
{
"epoch": 4.56,
"learning_rate": 1.3462499999999999e-05,
"loss": 6.8718,
"step": 360
},
{
"epoch": 4.68,
"learning_rate": 1.38375e-05,
"loss": 6.5045,
"step": 370
},
{
"epoch": 4.81,
"learning_rate": 1.4212499999999998e-05,
"loss": 6.574,
"step": 380
},
{
"epoch": 4.94,
"learning_rate": 1.4587499999999999e-05,
"loss": 6.1716,
"step": 390
},
{
"epoch": 5.06,
"learning_rate": 1.49625e-05,
"loss": 6.3618,
"step": 400
},
{
"epoch": 5.19,
"learning_rate": 1.5337499999999997e-05,
"loss": 6.012,
"step": 410
},
{
"epoch": 5.32,
"learning_rate": 1.57125e-05,
"loss": 6.0979,
"step": 420
},
{
"epoch": 5.44,
"learning_rate": 1.6087499999999998e-05,
"loss": 5.7887,
"step": 430
},
{
"epoch": 5.57,
"learning_rate": 1.6462499999999997e-05,
"loss": 5.892,
"step": 440
},
{
"epoch": 5.7,
"learning_rate": 1.68375e-05,
"loss": 5.4672,
"step": 450
},
{
"epoch": 5.82,
"learning_rate": 1.7212499999999998e-05,
"loss": 5.6353,
"step": 460
},
{
"epoch": 5.95,
"learning_rate": 1.7587499999999997e-05,
"loss": 5.2754,
"step": 470
},
{
"epoch": 6.08,
"learning_rate": 1.7962499999999996e-05,
"loss": 5.3343,
"step": 480
},
{
"epoch": 6.2,
"learning_rate": 1.83375e-05,
"loss": 5.0586,
"step": 490
},
{
"epoch": 6.33,
"learning_rate": 1.8712499999999997e-05,
"loss": 5.0767,
"step": 500
},
{
"epoch": 6.33,
"eval_cer": 1.0,
"eval_loss": 4.878269672393799,
"eval_runtime": 44.5452,
"eval_samples_per_second": 10.798,
"eval_steps_per_second": 1.369,
"eval_wer": 1.0,
"step": 500
},
{
"epoch": 6.46,
"learning_rate": 1.90875e-05,
"loss": 4.7637,
"step": 510
},
{
"epoch": 6.58,
"learning_rate": 1.94625e-05,
"loss": 4.8495,
"step": 520
},
{
"epoch": 6.71,
"learning_rate": 1.9837499999999998e-05,
"loss": 4.5651,
"step": 530
},
{
"epoch": 6.84,
"learning_rate": 2.02125e-05,
"loss": 4.7084,
"step": 540
},
{
"epoch": 6.96,
"learning_rate": 2.05875e-05,
"loss": 4.3947,
"step": 550
},
{
"epoch": 7.09,
"learning_rate": 2.09625e-05,
"loss": 4.4641,
"step": 560
},
{
"epoch": 7.22,
"learning_rate": 2.1337499999999997e-05,
"loss": 4.2175,
"step": 570
},
{
"epoch": 7.34,
"learning_rate": 2.1712499999999996e-05,
"loss": 4.2767,
"step": 580
},
{
"epoch": 7.47,
"learning_rate": 2.2087499999999998e-05,
"loss": 4.0216,
"step": 590
},
{
"epoch": 7.59,
"learning_rate": 2.2462499999999997e-05,
"loss": 4.0459,
"step": 600
},
{
"epoch": 7.72,
"learning_rate": 2.2837499999999996e-05,
"loss": 3.8726,
"step": 610
},
{
"epoch": 7.85,
"learning_rate": 2.32125e-05,
"loss": 3.8982,
"step": 620
},
{
"epoch": 7.97,
"learning_rate": 2.3587499999999997e-05,
"loss": 3.7419,
"step": 630
},
{
"epoch": 8.1,
"learning_rate": 2.39625e-05,
"loss": 3.7623,
"step": 640
},
{
"epoch": 8.23,
"learning_rate": 2.43375e-05,
"loss": 3.6173,
"step": 650
},
{
"epoch": 8.35,
"learning_rate": 2.4712499999999998e-05,
"loss": 3.625,
"step": 660
},
{
"epoch": 8.48,
"learning_rate": 2.50875e-05,
"loss": 3.4981,
"step": 670
},
{
"epoch": 8.61,
"learning_rate": 2.54625e-05,
"loss": 3.5114,
"step": 680
},
{
"epoch": 8.73,
"learning_rate": 2.5837499999999994e-05,
"loss": 3.42,
"step": 690
},
{
"epoch": 8.86,
"learning_rate": 2.6212499999999997e-05,
"loss": 3.4173,
"step": 700
},
{
"epoch": 8.99,
"learning_rate": 2.6587499999999996e-05,
"loss": 3.3501,
"step": 710
},
{
"epoch": 9.11,
"learning_rate": 2.6962499999999998e-05,
"loss": 3.3289,
"step": 720
},
{
"epoch": 9.24,
"learning_rate": 2.7337499999999997e-05,
"loss": 3.2786,
"step": 730
},
{
"epoch": 9.37,
"learning_rate": 2.7712499999999996e-05,
"loss": 3.2836,
"step": 740
},
{
"epoch": 9.49,
"learning_rate": 2.80875e-05,
"loss": 3.2514,
"step": 750
},
{
"epoch": 9.62,
"learning_rate": 2.8462499999999997e-05,
"loss": 3.2778,
"step": 760
},
{
"epoch": 9.75,
"learning_rate": 2.88375e-05,
"loss": 3.2042,
"step": 770
},
{
"epoch": 9.87,
"learning_rate": 2.92125e-05,
"loss": 3.1964,
"step": 780
},
{
"epoch": 10.0,
"learning_rate": 2.9587499999999998e-05,
"loss": 3.2052,
"step": 790
},
{
"epoch": 10.13,
"learning_rate": 2.99625e-05,
"loss": 3.1989,
"step": 800
},
{
"epoch": 10.25,
"learning_rate": 3.03375e-05,
"loss": 3.1823,
"step": 810
},
{
"epoch": 10.38,
"learning_rate": 3.0712499999999994e-05,
"loss": 3.1613,
"step": 820
},
{
"epoch": 10.51,
"learning_rate": 3.10875e-05,
"loss": 3.1659,
"step": 830
},
{
"epoch": 10.63,
"learning_rate": 3.14625e-05,
"loss": 3.1798,
"step": 840
},
{
"epoch": 10.76,
"learning_rate": 3.1837499999999995e-05,
"loss": 3.1711,
"step": 850
},
{
"epoch": 10.89,
"learning_rate": 3.22125e-05,
"loss": 3.1659,
"step": 860
},
{
"epoch": 11.01,
"learning_rate": 3.25875e-05,
"loss": 3.1309,
"step": 870
},
{
"epoch": 11.14,
"learning_rate": 3.2962499999999995e-05,
"loss": 3.1706,
"step": 880
},
{
"epoch": 11.27,
"learning_rate": 3.33375e-05,
"loss": 3.1441,
"step": 890
},
{
"epoch": 11.39,
"learning_rate": 3.37125e-05,
"loss": 3.1341,
"step": 900
},
{
"epoch": 11.52,
"learning_rate": 3.4087499999999995e-05,
"loss": 3.1594,
"step": 910
},
{
"epoch": 11.65,
"learning_rate": 3.44625e-05,
"loss": 3.1262,
"step": 920
},
{
"epoch": 11.77,
"learning_rate": 3.48375e-05,
"loss": 3.1541,
"step": 930
},
{
"epoch": 11.9,
"learning_rate": 3.5212499999999995e-05,
"loss": 3.1393,
"step": 940
},
{
"epoch": 12.03,
"learning_rate": 3.55875e-05,
"loss": 3.1398,
"step": 950
},
{
"epoch": 12.15,
"learning_rate": 3.596249999999999e-05,
"loss": 3.1551,
"step": 960
},
{
"epoch": 12.28,
"learning_rate": 3.6337499999999996e-05,
"loss": 3.1391,
"step": 970
},
{
"epoch": 12.41,
"learning_rate": 3.67125e-05,
"loss": 3.1236,
"step": 980
},
{
"epoch": 12.53,
"learning_rate": 3.7087499999999993e-05,
"loss": 3.1327,
"step": 990
},
{
"epoch": 12.66,
"learning_rate": 3.7462499999999996e-05,
"loss": 3.1156,
"step": 1000
},
{
"epoch": 12.66,
"eval_cer": 1.0,
"eval_loss": 3.0990264415740967,
"eval_runtime": 44.9243,
"eval_samples_per_second": 10.707,
"eval_steps_per_second": 1.358,
"eval_wer": 1.0,
"step": 1000
},
{
"epoch": 12.78,
"learning_rate": 3.783749999999999e-05,
"loss": 3.1365,
"step": 1010
},
{
"epoch": 12.91,
"learning_rate": 3.8212499999999994e-05,
"loss": 3.1123,
"step": 1020
},
{
"epoch": 13.04,
"learning_rate": 3.8587499999999996e-05,
"loss": 3.132,
"step": 1030
},
{
"epoch": 13.16,
"learning_rate": 3.896249999999999e-05,
"loss": 3.1218,
"step": 1040
},
{
"epoch": 13.29,
"learning_rate": 3.9337499999999994e-05,
"loss": 3.1266,
"step": 1050
},
{
"epoch": 13.42,
"learning_rate": 3.9712499999999996e-05,
"loss": 3.1247,
"step": 1060
},
{
"epoch": 13.54,
"learning_rate": 4.008749999999999e-05,
"loss": 3.1203,
"step": 1070
},
{
"epoch": 13.67,
"learning_rate": 4.0462499999999994e-05,
"loss": 3.1071,
"step": 1080
},
{
"epoch": 13.8,
"learning_rate": 4.0837499999999997e-05,
"loss": 3.1095,
"step": 1090
},
{
"epoch": 13.92,
"learning_rate": 4.12125e-05,
"loss": 3.0724,
"step": 1100
},
{
"epoch": 14.05,
"learning_rate": 4.1587499999999994e-05,
"loss": 3.1,
"step": 1110
},
{
"epoch": 14.18,
"learning_rate": 4.19625e-05,
"loss": 3.0862,
"step": 1120
},
{
"epoch": 14.3,
"learning_rate": 4.23375e-05,
"loss": 3.1141,
"step": 1130
},
{
"epoch": 14.43,
"learning_rate": 4.2712499999999995e-05,
"loss": 3.0847,
"step": 1140
},
{
"epoch": 14.56,
"learning_rate": 4.30875e-05,
"loss": 3.0845,
"step": 1150
},
{
"epoch": 14.68,
"learning_rate": 4.34625e-05,
"loss": 3.0537,
"step": 1160
},
{
"epoch": 14.81,
"learning_rate": 4.3837499999999995e-05,
"loss": 3.0811,
"step": 1170
},
{
"epoch": 14.94,
"learning_rate": 4.42125e-05,
"loss": 3.031,
"step": 1180
},
{
"epoch": 15.06,
"learning_rate": 4.45875e-05,
"loss": 3.0431,
"step": 1190
},
{
"epoch": 15.19,
"learning_rate": 4.4962499999999995e-05,
"loss": 2.9891,
"step": 1200
},
{
"epoch": 15.32,
"learning_rate": 4.53375e-05,
"loss": 2.9511,
"step": 1210
},
{
"epoch": 15.44,
"learning_rate": 4.57125e-05,
"loss": 2.8874,
"step": 1220
},
{
"epoch": 15.57,
"learning_rate": 4.60875e-05,
"loss": 2.8216,
"step": 1230
},
{
"epoch": 15.7,
"learning_rate": 4.64625e-05,
"loss": 2.7211,
"step": 1240
},
{
"epoch": 15.82,
"learning_rate": 4.68375e-05,
"loss": 2.6755,
"step": 1250
},
{
"epoch": 15.95,
"learning_rate": 4.721249999999999e-05,
"loss": 2.5301,
"step": 1260
},
{
"epoch": 16.08,
"learning_rate": 4.758749999999999e-05,
"loss": 2.4484,
"step": 1270
},
{
"epoch": 16.2,
"learning_rate": 4.7962499999999994e-05,
"loss": 2.2522,
"step": 1280
},
{
"epoch": 16.33,
"learning_rate": 4.8337499999999996e-05,
"loss": 2.1895,
"step": 1290
},
{
"epoch": 16.46,
"learning_rate": 4.871249999999999e-05,
"loss": 2.0274,
"step": 1300
},
{
"epoch": 16.58,
"learning_rate": 4.9087499999999994e-05,
"loss": 1.9528,
"step": 1310
},
{
"epoch": 16.71,
"learning_rate": 4.9462499999999996e-05,
"loss": 1.8849,
"step": 1320
},
{
"epoch": 16.84,
"learning_rate": 4.983749999999999e-05,
"loss": 1.8432,
"step": 1330
},
{
"epoch": 16.96,
"learning_rate": 5.0212499999999994e-05,
"loss": 1.7181,
"step": 1340
},
{
"epoch": 17.09,
"learning_rate": 5.0587499999999996e-05,
"loss": 1.7099,
"step": 1350
},
{
"epoch": 17.22,
"learning_rate": 5.096249999999999e-05,
"loss": 1.6276,
"step": 1360
},
{
"epoch": 17.34,
"learning_rate": 5.1337499999999994e-05,
"loss": 1.6456,
"step": 1370
},
{
"epoch": 17.47,
"learning_rate": 5.1712499999999997e-05,
"loss": 1.5613,
"step": 1380
},
{
"epoch": 17.59,
"learning_rate": 5.20875e-05,
"loss": 1.5553,
"step": 1390
},
{
"epoch": 17.72,
"learning_rate": 5.2462499999999994e-05,
"loss": 1.4874,
"step": 1400
},
{
"epoch": 17.85,
"learning_rate": 5.28375e-05,
"loss": 1.5079,
"step": 1410
},
{
"epoch": 17.97,
"learning_rate": 5.32125e-05,
"loss": 1.4935,
"step": 1420
},
{
"epoch": 18.1,
"learning_rate": 5.3587499999999995e-05,
"loss": 1.4843,
"step": 1430
},
{
"epoch": 18.23,
"learning_rate": 5.39625e-05,
"loss": 1.416,
"step": 1440
},
{
"epoch": 18.35,
"learning_rate": 5.43375e-05,
"loss": 1.3815,
"step": 1450
},
{
"epoch": 18.48,
"learning_rate": 5.4712499999999995e-05,
"loss": 1.4005,
"step": 1460
},
{
"epoch": 18.61,
"learning_rate": 5.50875e-05,
"loss": 1.3824,
"step": 1470
},
{
"epoch": 18.73,
"learning_rate": 5.54625e-05,
"loss": 1.3038,
"step": 1480
},
{
"epoch": 18.86,
"learning_rate": 5.58375e-05,
"loss": 1.4168,
"step": 1490
},
{
"epoch": 18.99,
"learning_rate": 5.62125e-05,
"loss": 1.3506,
"step": 1500
},
{
"epoch": 18.99,
"eval_cer": 0.28892684849736266,
"eval_loss": 1.1056294441223145,
"eval_runtime": 44.5929,
"eval_samples_per_second": 10.786,
"eval_steps_per_second": 1.368,
"eval_wer": 0.7031036834924966,
"step": 1500
},
{
"epoch": 19.11,
"learning_rate": 5.658749999999999e-05,
"loss": 1.3407,
"step": 1510
},
{
"epoch": 19.24,
"learning_rate": 5.696249999999999e-05,
"loss": 1.2849,
"step": 1520
},
{
"epoch": 19.37,
"learning_rate": 5.733749999999999e-05,
"loss": 1.3141,
"step": 1530
},
{
"epoch": 19.49,
"learning_rate": 5.771249999999999e-05,
"loss": 1.2858,
"step": 1540
},
{
"epoch": 19.62,
"learning_rate": 5.8087499999999996e-05,
"loss": 1.2842,
"step": 1550
},
{
"epoch": 19.75,
"learning_rate": 5.846249999999999e-05,
"loss": 1.2537,
"step": 1560
},
{
"epoch": 19.87,
"learning_rate": 5.8837499999999994e-05,
"loss": 1.304,
"step": 1570
},
{
"epoch": 20.0,
"learning_rate": 5.9212499999999996e-05,
"loss": 1.243,
"step": 1580
},
{
"epoch": 20.13,
"learning_rate": 5.958749999999999e-05,
"loss": 1.2636,
"step": 1590
},
{
"epoch": 20.25,
"learning_rate": 5.9962499999999994e-05,
"loss": 1.2239,
"step": 1600
},
{
"epoch": 20.38,
"learning_rate": 6.0337499999999996e-05,
"loss": 1.267,
"step": 1610
},
{
"epoch": 20.51,
"learning_rate": 6.071249999999999e-05,
"loss": 1.216,
"step": 1620
},
{
"epoch": 20.63,
"learning_rate": 6.10875e-05,
"loss": 1.253,
"step": 1630
},
{
"epoch": 20.76,
"learning_rate": 6.14625e-05,
"loss": 1.2137,
"step": 1640
},
{
"epoch": 20.89,
"learning_rate": 6.183749999999999e-05,
"loss": 1.1845,
"step": 1650
},
{
"epoch": 21.01,
"learning_rate": 6.22125e-05,
"loss": 1.1501,
"step": 1660
},
{
"epoch": 21.14,
"learning_rate": 6.25875e-05,
"loss": 1.1904,
"step": 1670
},
{
"epoch": 21.27,
"learning_rate": 6.296249999999999e-05,
"loss": 1.1456,
"step": 1680
},
{
"epoch": 21.39,
"learning_rate": 6.33375e-05,
"loss": 1.1798,
"step": 1690
},
{
"epoch": 21.52,
"learning_rate": 6.37125e-05,
"loss": 1.1122,
"step": 1700
},
{
"epoch": 21.65,
"learning_rate": 6.408749999999999e-05,
"loss": 1.1512,
"step": 1710
},
{
"epoch": 21.77,
"learning_rate": 6.44625e-05,
"loss": 1.1413,
"step": 1720
},
{
"epoch": 21.9,
"learning_rate": 6.48375e-05,
"loss": 1.1652,
"step": 1730
},
{
"epoch": 22.03,
"learning_rate": 6.521249999999999e-05,
"loss": 1.177,
"step": 1740
},
{
"epoch": 22.15,
"learning_rate": 6.55875e-05,
"loss": 1.1326,
"step": 1750
},
{
"epoch": 22.28,
"learning_rate": 6.596249999999998e-05,
"loss": 1.1082,
"step": 1760
},
{
"epoch": 22.41,
"learning_rate": 6.633749999999999e-05,
"loss": 1.1321,
"step": 1770
},
{
"epoch": 22.53,
"learning_rate": 6.671249999999999e-05,
"loss": 1.0721,
"step": 1780
},
{
"epoch": 22.66,
"learning_rate": 6.70875e-05,
"loss": 1.1199,
"step": 1790
},
{
"epoch": 22.78,
"learning_rate": 6.746249999999999e-05,
"loss": 1.1436,
"step": 1800
},
{
"epoch": 22.91,
"learning_rate": 6.783749999999999e-05,
"loss": 1.0941,
"step": 1810
},
{
"epoch": 23.04,
"learning_rate": 6.82125e-05,
"loss": 1.0631,
"step": 1820
},
{
"epoch": 23.16,
"learning_rate": 6.85875e-05,
"loss": 1.064,
"step": 1830
},
{
"epoch": 23.29,
"learning_rate": 6.896249999999999e-05,
"loss": 1.0739,
"step": 1840
},
{
"epoch": 23.42,
"learning_rate": 6.93375e-05,
"loss": 1.0354,
"step": 1850
},
{
"epoch": 23.54,
"learning_rate": 6.97125e-05,
"loss": 1.0343,
"step": 1860
},
{
"epoch": 23.67,
"learning_rate": 7.008749999999999e-05,
"loss": 1.0724,
"step": 1870
},
{
"epoch": 23.8,
"learning_rate": 7.04625e-05,
"loss": 1.0982,
"step": 1880
},
{
"epoch": 23.92,
"learning_rate": 7.08375e-05,
"loss": 1.065,
"step": 1890
},
{
"epoch": 24.05,
"learning_rate": 7.121249999999999e-05,
"loss": 1.0754,
"step": 1900
},
{
"epoch": 24.18,
"learning_rate": 7.15875e-05,
"loss": 1.0708,
"step": 1910
},
{
"epoch": 24.3,
"learning_rate": 7.19625e-05,
"loss": 1.0165,
"step": 1920
},
{
"epoch": 24.43,
"learning_rate": 7.233749999999999e-05,
"loss": 1.02,
"step": 1930
},
{
"epoch": 24.56,
"learning_rate": 7.27125e-05,
"loss": 1.0985,
"step": 1940
},
{
"epoch": 24.68,
"learning_rate": 7.30875e-05,
"loss": 0.9746,
"step": 1950
},
{
"epoch": 24.81,
"learning_rate": 7.346249999999999e-05,
"loss": 1.0644,
"step": 1960
},
{
"epoch": 24.94,
"learning_rate": 7.38375e-05,
"loss": 1.0104,
"step": 1970
},
{
"epoch": 25.06,
"learning_rate": 7.42125e-05,
"loss": 1.028,
"step": 1980
},
{
"epoch": 25.19,
"learning_rate": 7.45875e-05,
"loss": 1.0107,
"step": 1990
},
{
"epoch": 25.32,
"learning_rate": 7.49625e-05,
"loss": 0.9997,
"step": 2000
},
{
"epoch": 25.32,
"eval_cer": 0.2301165693690988,
"eval_loss": 0.919084906578064,
"eval_runtime": 44.703,
"eval_samples_per_second": 10.76,
"eval_steps_per_second": 1.365,
"eval_wer": 0.5943894952251023,
"step": 2000
},
{
"epoch": 25.44,
"learning_rate": 7.465384615384615e-05,
"loss": 0.9571,
"step": 2010
},
{
"epoch": 25.57,
"learning_rate": 7.426923076923075e-05,
"loss": 0.9801,
"step": 2020
},
{
"epoch": 25.7,
"learning_rate": 7.388461538461538e-05,
"loss": 0.9779,
"step": 2030
},
{
"epoch": 25.82,
"learning_rate": 7.35e-05,
"loss": 1.0168,
"step": 2040
},
{
"epoch": 25.95,
"learning_rate": 7.31153846153846e-05,
"loss": 0.9302,
"step": 2050
},
{
"epoch": 26.08,
"learning_rate": 7.273076923076923e-05,
"loss": 0.989,
"step": 2060
},
{
"epoch": 26.2,
"learning_rate": 7.234615384615385e-05,
"loss": 0.9357,
"step": 2070
},
{
"epoch": 26.33,
"learning_rate": 7.196153846153846e-05,
"loss": 0.9858,
"step": 2080
},
{
"epoch": 26.46,
"learning_rate": 7.157692307692307e-05,
"loss": 0.9813,
"step": 2090
},
{
"epoch": 26.58,
"learning_rate": 7.119230769230769e-05,
"loss": 0.9554,
"step": 2100
},
{
"epoch": 26.71,
"learning_rate": 7.08076923076923e-05,
"loss": 0.8935,
"step": 2110
},
{
"epoch": 26.84,
"learning_rate": 7.042307692307692e-05,
"loss": 0.9955,
"step": 2120
},
{
"epoch": 26.96,
"learning_rate": 7.003846153846154e-05,
"loss": 0.9205,
"step": 2130
},
{
"epoch": 27.09,
"learning_rate": 6.965384615384615e-05,
"loss": 0.9527,
"step": 2140
},
{
"epoch": 27.22,
"learning_rate": 6.926923076923075e-05,
"loss": 0.8899,
"step": 2150
},
{
"epoch": 27.34,
"learning_rate": 6.888461538461538e-05,
"loss": 0.9594,
"step": 2160
},
{
"epoch": 27.47,
"learning_rate": 6.85e-05,
"loss": 0.9061,
"step": 2170
},
{
"epoch": 27.59,
"learning_rate": 6.81153846153846e-05,
"loss": 0.94,
"step": 2180
},
{
"epoch": 27.72,
"learning_rate": 6.773076923076923e-05,
"loss": 0.8611,
"step": 2190
},
{
"epoch": 27.85,
"learning_rate": 6.734615384615385e-05,
"loss": 0.9391,
"step": 2200
},
{
"epoch": 27.97,
"learning_rate": 6.696153846153846e-05,
"loss": 0.8905,
"step": 2210
},
{
"epoch": 28.1,
"learning_rate": 6.657692307692307e-05,
"loss": 0.888,
"step": 2220
},
{
"epoch": 28.23,
"learning_rate": 6.619230769230769e-05,
"loss": 0.8749,
"step": 2230
},
{
"epoch": 28.35,
"learning_rate": 6.580769230769231e-05,
"loss": 0.9331,
"step": 2240
},
{
"epoch": 28.48,
"learning_rate": 6.542307692307692e-05,
"loss": 0.8135,
"step": 2250
},
{
"epoch": 28.61,
"learning_rate": 6.503846153846154e-05,
"loss": 0.9121,
"step": 2260
},
{
"epoch": 28.73,
"learning_rate": 6.465384615384615e-05,
"loss": 0.859,
"step": 2270
},
{
"epoch": 28.86,
"learning_rate": 6.426923076923076e-05,
"loss": 0.8726,
"step": 2280
},
{
"epoch": 28.99,
"learning_rate": 6.388461538461538e-05,
"loss": 0.8497,
"step": 2290
},
{
"epoch": 29.11,
"learning_rate": 6.35e-05,
"loss": 0.8673,
"step": 2300
},
{
"epoch": 29.24,
"learning_rate": 6.31153846153846e-05,
"loss": 0.8349,
"step": 2310
},
{
"epoch": 29.37,
"learning_rate": 6.273076923076923e-05,
"loss": 0.8946,
"step": 2320
},
{
"epoch": 29.49,
"learning_rate": 6.234615384615384e-05,
"loss": 0.8805,
"step": 2330
},
{
"epoch": 29.62,
"learning_rate": 6.196153846153846e-05,
"loss": 0.8752,
"step": 2340
},
{
"epoch": 29.75,
"learning_rate": 6.157692307692307e-05,
"loss": 0.8197,
"step": 2350
},
{
"epoch": 29.87,
"learning_rate": 6.119230769230769e-05,
"loss": 0.8332,
"step": 2360
},
{
"epoch": 30.0,
"learning_rate": 6.08076923076923e-05,
"loss": 0.7933,
"step": 2370
},
{
"epoch": 30.13,
"learning_rate": 6.0423076923076924e-05,
"loss": 0.8712,
"step": 2380
},
{
"epoch": 30.25,
"learning_rate": 6.003846153846153e-05,
"loss": 0.824,
"step": 2390
},
{
"epoch": 30.38,
"learning_rate": 5.965384615384615e-05,
"loss": 0.8158,
"step": 2400
},
{
"epoch": 30.51,
"learning_rate": 5.926923076923076e-05,
"loss": 0.8218,
"step": 2410
},
{
"epoch": 30.63,
"learning_rate": 5.888461538461538e-05,
"loss": 0.8403,
"step": 2420
},
{
"epoch": 30.76,
"learning_rate": 5.85e-05,
"loss": 0.7986,
"step": 2430
},
{
"epoch": 30.89,
"learning_rate": 5.8115384615384614e-05,
"loss": 0.8391,
"step": 2440
},
{
"epoch": 31.01,
"learning_rate": 5.773076923076922e-05,
"loss": 0.7736,
"step": 2450
},
{
"epoch": 31.14,
"learning_rate": 5.734615384615384e-05,
"loss": 0.8478,
"step": 2460
},
{
"epoch": 31.27,
"learning_rate": 5.696153846153846e-05,
"loss": 0.7728,
"step": 2470
},
{
"epoch": 31.39,
"learning_rate": 5.6576923076923073e-05,
"loss": 0.8231,
"step": 2480
},
{
"epoch": 31.52,
"learning_rate": 5.619230769230769e-05,
"loss": 0.7602,
"step": 2490
},
{
"epoch": 31.65,
"learning_rate": 5.58076923076923e-05,
"loss": 0.7838,
"step": 2500
},
{
"epoch": 31.65,
"eval_cer": 0.2152122088112177,
"eval_loss": 0.8952043056488037,
"eval_runtime": 45.1763,
"eval_samples_per_second": 10.647,
"eval_steps_per_second": 1.35,
"eval_wer": 0.555593451568895,
"step": 2500
},
{
"epoch": 31.77,
"learning_rate": 5.542307692307691e-05,
"loss": 0.8065,
"step": 2510
},
{
"epoch": 31.9,
"learning_rate": 5.503846153846153e-05,
"loss": 0.773,
"step": 2520
},
{
"epoch": 32.03,
"learning_rate": 5.465384615384615e-05,
"loss": 0.7854,
"step": 2530
},
{
"epoch": 32.15,
"learning_rate": 5.426923076923076e-05,
"loss": 0.7724,
"step": 2540
},
{
"epoch": 32.28,
"learning_rate": 5.3884615384615384e-05,
"loss": 0.7639,
"step": 2550
},
{
"epoch": 32.41,
"learning_rate": 5.35e-05,
"loss": 0.7993,
"step": 2560
},
{
"epoch": 32.53,
"learning_rate": 5.311538461538461e-05,
"loss": 0.7957,
"step": 2570
},
{
"epoch": 32.66,
"learning_rate": 5.273076923076922e-05,
"loss": 0.7686,
"step": 2580
},
{
"epoch": 32.78,
"learning_rate": 5.234615384615384e-05,
"loss": 0.8096,
"step": 2590
},
{
"epoch": 32.91,
"learning_rate": 5.196153846153846e-05,
"loss": 0.7357,
"step": 2600
},
{
"epoch": 33.04,
"learning_rate": 5.1576923076923074e-05,
"loss": 0.7674,
"step": 2610
},
{
"epoch": 33.16,
"learning_rate": 5.119230769230769e-05,
"loss": 0.7989,
"step": 2620
},
{
"epoch": 33.29,
"learning_rate": 5.08076923076923e-05,
"loss": 0.7474,
"step": 2630
},
{
"epoch": 33.42,
"learning_rate": 5.042307692307692e-05,
"loss": 0.7153,
"step": 2640
},
{
"epoch": 33.54,
"learning_rate": 5.0038461538461533e-05,
"loss": 0.7109,
"step": 2650
},
{
"epoch": 33.67,
"learning_rate": 4.965384615384615e-05,
"loss": 0.7841,
"step": 2660
},
{
"epoch": 33.8,
"learning_rate": 4.926923076923076e-05,
"loss": 0.762,
"step": 2670
},
{
"epoch": 33.92,
"learning_rate": 4.8884615384615385e-05,
"loss": 0.7414,
"step": 2680
},
{
"epoch": 34.05,
"learning_rate": 4.849999999999999e-05,
"loss": 0.7544,
"step": 2690
},
{
"epoch": 34.18,
"learning_rate": 4.811538461538461e-05,
"loss": 0.7338,
"step": 2700
},
{
"epoch": 34.3,
"learning_rate": 4.773076923076922e-05,
"loss": 0.7266,
"step": 2710
},
{
"epoch": 34.43,
"learning_rate": 4.7346153846153845e-05,
"loss": 0.7131,
"step": 2720
},
{
"epoch": 34.56,
"learning_rate": 4.696153846153846e-05,
"loss": 0.7291,
"step": 2730
},
{
"epoch": 34.68,
"learning_rate": 4.6576923076923074e-05,
"loss": 0.7051,
"step": 2740
},
{
"epoch": 34.81,
"learning_rate": 4.619230769230769e-05,
"loss": 0.7643,
"step": 2750
},
{
"epoch": 34.94,
"learning_rate": 4.5807692307692304e-05,
"loss": 0.727,
"step": 2760
},
{
"epoch": 35.06,
"learning_rate": 4.542307692307692e-05,
"loss": 0.7142,
"step": 2770
},
{
"epoch": 35.19,
"learning_rate": 4.5038461538461534e-05,
"loss": 0.7055,
"step": 2780
},
{
"epoch": 35.32,
"learning_rate": 4.465384615384615e-05,
"loss": 0.7339,
"step": 2790
},
{
"epoch": 35.44,
"learning_rate": 4.426923076923077e-05,
"loss": 0.6956,
"step": 2800
},
{
"epoch": 35.57,
"learning_rate": 4.3884615384615385e-05,
"loss": 0.7508,
"step": 2810
},
{
"epoch": 35.7,
"learning_rate": 4.3499999999999993e-05,
"loss": 0.7072,
"step": 2820
},
{
"epoch": 35.82,
"learning_rate": 4.311538461538461e-05,
"loss": 0.7103,
"step": 2830
},
{
"epoch": 35.95,
"learning_rate": 4.273076923076923e-05,
"loss": 0.6783,
"step": 2840
},
{
"epoch": 36.08,
"learning_rate": 4.2346153846153845e-05,
"loss": 0.7419,
"step": 2850
},
{
"epoch": 36.2,
"learning_rate": 4.196153846153846e-05,
"loss": 0.7091,
"step": 2860
},
{
"epoch": 36.33,
"learning_rate": 4.1576923076923075e-05,
"loss": 0.7073,
"step": 2870
},
{
"epoch": 36.46,
"learning_rate": 4.119230769230768e-05,
"loss": 0.6937,
"step": 2880
},
{
"epoch": 36.58,
"learning_rate": 4.0807692307692305e-05,
"loss": 0.756,
"step": 2890
},
{
"epoch": 36.71,
"learning_rate": 4.042307692307692e-05,
"loss": 0.6744,
"step": 2900
},
{
"epoch": 36.84,
"learning_rate": 4.0038461538461534e-05,
"loss": 0.7165,
"step": 2910
},
{
"epoch": 36.96,
"learning_rate": 3.9653846153846156e-05,
"loss": 0.6831,
"step": 2920
},
{
"epoch": 37.09,
"learning_rate": 3.926923076923077e-05,
"loss": 0.6894,
"step": 2930
},
{
"epoch": 37.22,
"learning_rate": 3.888461538461538e-05,
"loss": 0.6419,
"step": 2940
},
{
"epoch": 37.34,
"learning_rate": 3.8499999999999994e-05,
"loss": 0.7187,
"step": 2950
},
{
"epoch": 37.47,
"learning_rate": 3.811538461538461e-05,
"loss": 0.677,
"step": 2960
},
{
"epoch": 37.59,
"learning_rate": 3.773076923076923e-05,
"loss": 0.7263,
"step": 2970
},
{
"epoch": 37.72,
"learning_rate": 3.734615384615384e-05,
"loss": 0.6257,
"step": 2980
},
{
"epoch": 37.85,
"learning_rate": 3.696153846153846e-05,
"loss": 0.7051,
"step": 2990
},
{
"epoch": 37.97,
"learning_rate": 3.6576923076923075e-05,
"loss": 0.6665,
"step": 3000
},
{
"epoch": 37.97,
"eval_cer": 0.2016781484053836,
"eval_loss": 0.8907838463783264,
"eval_runtime": 45.6242,
"eval_samples_per_second": 10.543,
"eval_steps_per_second": 1.337,
"eval_wer": 0.5251534788540245,
"step": 3000
},
{
"epoch": 38.1,
"learning_rate": 3.619230769230769e-05,
"loss": 0.7016,
"step": 3010
},
{
"epoch": 38.23,
"learning_rate": 3.5807692307692305e-05,
"loss": 0.6585,
"step": 3020
},
{
"epoch": 38.35,
"learning_rate": 3.542307692307692e-05,
"loss": 0.6673,
"step": 3030
},
{
"epoch": 38.48,
"learning_rate": 3.5038461538461535e-05,
"loss": 0.6411,
"step": 3040
},
{
"epoch": 38.61,
"learning_rate": 3.465384615384615e-05,
"loss": 0.7038,
"step": 3050
},
{
"epoch": 38.73,
"learning_rate": 3.4269230769230765e-05,
"loss": 0.6458,
"step": 3060
},
{
"epoch": 38.86,
"learning_rate": 3.3884615384615386e-05,
"loss": 0.7231,
"step": 3070
},
{
"epoch": 38.99,
"learning_rate": 3.3499999999999994e-05,
"loss": 0.6495,
"step": 3080
},
{
"epoch": 39.11,
"learning_rate": 3.3115384615384616e-05,
"loss": 0.6788,
"step": 3090
},
{
"epoch": 39.24,
"learning_rate": 3.273076923076923e-05,
"loss": 0.6452,
"step": 3100
},
{
"epoch": 39.37,
"learning_rate": 3.2346153846153846e-05,
"loss": 0.7015,
"step": 3110
},
{
"epoch": 39.49,
"learning_rate": 3.196153846153846e-05,
"loss": 0.6518,
"step": 3120
},
{
"epoch": 39.62,
"learning_rate": 3.1576923076923076e-05,
"loss": 0.6757,
"step": 3130
},
{
"epoch": 39.75,
"learning_rate": 3.119230769230769e-05,
"loss": 0.6495,
"step": 3140
},
{
"epoch": 39.87,
"learning_rate": 3.0807692307692305e-05,
"loss": 0.6434,
"step": 3150
},
{
"epoch": 40.0,
"learning_rate": 3.0423076923076924e-05,
"loss": 0.6132,
"step": 3160
},
{
"epoch": 40.13,
"learning_rate": 3.0038461538461535e-05,
"loss": 0.6959,
"step": 3170
},
{
"epoch": 40.25,
"learning_rate": 2.965384615384615e-05,
"loss": 0.6468,
"step": 3180
},
{
"epoch": 40.38,
"learning_rate": 2.926923076923077e-05,
"loss": 0.6681,
"step": 3190
},
{
"epoch": 40.51,
"learning_rate": 2.888461538461538e-05,
"loss": 0.6446,
"step": 3200
},
{
"epoch": 40.63,
"learning_rate": 2.8499999999999998e-05,
"loss": 0.6554,
"step": 3210
},
{
"epoch": 40.76,
"learning_rate": 2.8115384615384613e-05,
"loss": 0.6204,
"step": 3220
},
{
"epoch": 40.89,
"learning_rate": 2.7730769230769228e-05,
"loss": 0.677,
"step": 3230
},
{
"epoch": 41.01,
"learning_rate": 2.7346153846153843e-05,
"loss": 0.5961,
"step": 3240
},
{
"epoch": 41.14,
"learning_rate": 2.696153846153846e-05,
"loss": 0.665,
"step": 3250
},
{
"epoch": 41.27,
"learning_rate": 2.6576923076923073e-05,
"loss": 0.6753,
"step": 3260
},
{
"epoch": 41.39,
"learning_rate": 2.619230769230769e-05,
"loss": 0.6387,
"step": 3270
},
{
"epoch": 41.52,
"learning_rate": 2.5807692307692306e-05,
"loss": 0.6281,
"step": 3280
},
{
"epoch": 41.65,
"learning_rate": 2.542307692307692e-05,
"loss": 0.6287,
"step": 3290
},
{
"epoch": 41.77,
"learning_rate": 2.5038461538461536e-05,
"loss": 0.6413,
"step": 3300
},
{
"epoch": 41.9,
"learning_rate": 2.4653846153846154e-05,
"loss": 0.6061,
"step": 3310
},
{
"epoch": 42.03,
"learning_rate": 2.4269230769230765e-05,
"loss": 0.648,
"step": 3320
},
{
"epoch": 42.15,
"learning_rate": 2.3884615384615384e-05,
"loss": 0.5926,
"step": 3330
},
{
"epoch": 42.28,
"learning_rate": 2.35e-05,
"loss": 0.6366,
"step": 3340
},
{
"epoch": 42.41,
"learning_rate": 2.3115384615384614e-05,
"loss": 0.6625,
"step": 3350
},
{
"epoch": 42.53,
"learning_rate": 2.273076923076923e-05,
"loss": 0.634,
"step": 3360
},
{
"epoch": 42.66,
"learning_rate": 2.2346153846153847e-05,
"loss": 0.618,
"step": 3370
},
{
"epoch": 42.78,
"learning_rate": 2.1961538461538458e-05,
"loss": 0.5911,
"step": 3380
},
{
"epoch": 42.91,
"learning_rate": 2.1576923076923076e-05,
"loss": 0.5936,
"step": 3390
},
{
"epoch": 43.04,
"learning_rate": 2.119230769230769e-05,
"loss": 0.6267,
"step": 3400
},
{
"epoch": 43.16,
"learning_rate": 2.0807692307692303e-05,
"loss": 0.6123,
"step": 3410
},
{
"epoch": 43.29,
"learning_rate": 2.042307692307692e-05,
"loss": 0.6398,
"step": 3420
},
{
"epoch": 43.42,
"learning_rate": 2.003846153846154e-05,
"loss": 0.606,
"step": 3430
},
{
"epoch": 43.54,
"learning_rate": 1.965384615384615e-05,
"loss": 0.6253,
"step": 3440
},
{
"epoch": 43.67,
"learning_rate": 1.9269230769230766e-05,
"loss": 0.5847,
"step": 3450
},
{
"epoch": 43.8,
"learning_rate": 1.8884615384615384e-05,
"loss": 0.6248,
"step": 3460
},
{
"epoch": 43.92,
"learning_rate": 1.85e-05,
"loss": 0.5884,
"step": 3470
},
{
"epoch": 44.05,
"learning_rate": 1.8115384615384614e-05,
"loss": 0.6038,
"step": 3480
},
{
"epoch": 44.18,
"learning_rate": 1.773076923076923e-05,
"loss": 0.5888,
"step": 3490
},
{
"epoch": 44.3,
"learning_rate": 1.7346153846153844e-05,
"loss": 0.6265,
"step": 3500
},
{
"epoch": 44.3,
"eval_cer": 0.19540855592889456,
"eval_loss": 0.9062958359718323,
"eval_runtime": 44.6904,
"eval_samples_per_second": 10.763,
"eval_steps_per_second": 1.365,
"eval_wer": 0.5133015006821282,
"step": 3500
}
],
"max_steps": 3950,
"num_train_epochs": 50,
"total_flos": 4.029580130869645e+19,
"trial_name": null,
"trial_params": null
}