Update README.md
Browse files
README.md
CHANGED
@@ -11,10 +11,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
11 |
|
12 |
# checkpoints2
|
13 |
|
14 |
-
This model
|
15 |
-
It achieves the following results on the evaluation set:
|
16 |
-
- Loss: 0.1593
|
17 |
-
- Wer: 0.0579
|
18 |
|
19 |
## Model description
|
20 |
|
@@ -45,61 +42,6 @@ The following hyperparameters were used during training:
|
|
45 |
- num_epochs: 50
|
46 |
- mixed_precision_training: Native AMP
|
47 |
|
48 |
-
### Training results
|
49 |
-
|
50 |
-
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
51 |
-
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
52 |
-
| 0.2075 | 1.0 | 124 | 0.1170 | 0.0525 |
|
53 |
-
| 0.0978 | 2.0 | 248 | 0.0875 | 0.0507 |
|
54 |
-
| 0.046 | 3.0 | 372 | 0.1017 | 0.0534 |
|
55 |
-
| 0.031 | 4.0 | 496 | 0.1167 | 0.0543 |
|
56 |
-
| 0.0353 | 5.0 | 620 | 0.1194 | 0.0534 |
|
57 |
-
| 0.0345 | 6.0 | 744 | 0.1217 | 0.0597 |
|
58 |
-
| 0.0248 | 7.0 | 868 | 0.1366 | 0.0633 |
|
59 |
-
| 0.02 | 8.0 | 992 | 0.1193 | 0.0597 |
|
60 |
-
| 0.0339 | 9.0 | 1116 | 0.1341 | 0.0624 |
|
61 |
-
| 0.0172 | 10.0 | 1240 | 0.1200 | 0.0570 |
|
62 |
-
| 0.022 | 11.0 | 1364 | 0.1323 | 0.0688 |
|
63 |
-
| 0.0221 | 12.0 | 1488 | 0.1219 | 0.0543 |
|
64 |
-
| 0.0168 | 13.0 | 1612 | 0.1567 | 0.0769 |
|
65 |
-
| 0.0145 | 14.0 | 1736 | 0.1337 | 0.0606 |
|
66 |
-
| 0.0114 | 15.0 | 1860 | 0.1299 | 0.0688 |
|
67 |
-
| 0.0209 | 16.0 | 1984 | 0.1266 | 0.0661 |
|
68 |
-
| 0.0183 | 17.0 | 2108 | 0.1412 | 0.0697 |
|
69 |
-
| 0.0135 | 18.0 | 2232 | 0.1394 | 0.0670 |
|
70 |
-
| 0.0101 | 19.0 | 2356 | 0.1429 | 0.0724 |
|
71 |
-
| 0.0125 | 20.0 | 2480 | 0.1296 | 0.0733 |
|
72 |
-
| 0.0102 | 21.0 | 2604 | 0.1233 | 0.0697 |
|
73 |
-
| 0.006 | 22.0 | 2728 | 0.1368 | 0.0769 |
|
74 |
-
| 0.0058 | 23.0 | 2852 | 0.1225 | 0.0624 |
|
75 |
-
| 0.0046 | 24.0 | 2976 | 0.1220 | 0.0606 |
|
76 |
-
| 0.0081 | 25.0 | 3100 | 0.1425 | 0.0751 |
|
77 |
-
| 0.0088 | 26.0 | 3224 | 0.1311 | 0.0670 |
|
78 |
-
| 0.0055 | 27.0 | 3348 | 0.1286 | 0.0597 |
|
79 |
-
| 0.0052 | 28.0 | 3472 | 0.1228 | 0.0561 |
|
80 |
-
| 0.0026 | 29.0 | 3596 | 0.1437 | 0.0606 |
|
81 |
-
| 0.0052 | 30.0 | 3720 | 0.1430 | 0.0679 |
|
82 |
-
| 0.0096 | 31.0 | 3844 | 0.1458 | 0.0733 |
|
83 |
-
| 0.0051 | 32.0 | 3968 | 0.1393 | 0.0606 |
|
84 |
-
| 0.0031 | 33.0 | 4092 | 0.1441 | 0.0633 |
|
85 |
-
| 0.0042 | 34.0 | 4216 | 0.1539 | 0.0679 |
|
86 |
-
| 0.0032 | 35.0 | 4340 | 0.1496 | 0.0661 |
|
87 |
-
| 0.0035 | 36.0 | 4464 | 0.1486 | 0.0615 |
|
88 |
-
| 0.0017 | 37.0 | 4588 | 0.1489 | 0.0643 |
|
89 |
-
| 0.0021 | 38.0 | 4712 | 0.1583 | 0.0615 |
|
90 |
-
| 0.0031 | 39.0 | 4836 | 0.1544 | 0.0597 |
|
91 |
-
| 0.0021 | 40.0 | 4960 | 0.1638 | 0.0633 |
|
92 |
-
| 0.0017 | 41.0 | 5084 | 0.1630 | 0.0633 |
|
93 |
-
| 0.0013 | 42.0 | 5208 | 0.1618 | 0.0561 |
|
94 |
-
| 0.0015 | 43.0 | 5332 | 0.1656 | 0.0597 |
|
95 |
-
| 0.0015 | 44.0 | 5456 | 0.1662 | 0.0588 |
|
96 |
-
| 0.0018 | 45.0 | 5580 | 0.1594 | 0.0597 |
|
97 |
-
| 0.0007 | 46.0 | 5704 | 0.1603 | 0.0597 |
|
98 |
-
| 0.0005 | 47.0 | 5828 | 0.1614 | 0.0615 |
|
99 |
-
| 0.0011 | 48.0 | 5952 | 0.1599 | 0.0579 |
|
100 |
-
| 0.0007 | 49.0 | 6076 | 0.1601 | 0.0579 |
|
101 |
-
| 0.0009 | 50.0 | 6200 | 0.1593 | 0.0579 |
|
102 |
-
|
103 |
|
104 |
### Framework versions
|
105 |
|
|
|
11 |
|
12 |
# checkpoints2
|
13 |
|
14 |
+
This model is a fine-tuned version of ttop324/wav2vec2-live-japanese on the None dataset.
|
|
|
|
|
|
|
15 |
|
16 |
## Model description
|
17 |
|
|
|
42 |
- num_epochs: 50
|
43 |
- mixed_precision_training: Native AMP
|
44 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
45 |
|
46 |
### Framework versions
|
47 |
|