Hugging Face's logo

Padomin
/
t5-base-TEDxJP-0front-1body-3rear

Transformers
PyTorch
TensorBoard
t5
text2text-generation
Generated from Trainer
text-generation-inference
Model card Files Metrics Community
Padomin commited on
Commit
d4991a0
·
1 Parent(s): 7bd7238

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +20 -20
README.md CHANGED
@@ -16,16 +16,16 @@ should probably proofread and complete it, then remove this comment. -->
16
 
17
  This model is a fine-tuned version of [sonoisa/t5-base-japanese](https://huggingface.co/sonoisa/t5-base-japanese) on the te_dx_jp dataset.
18
  It achieves the following results on the evaluation set:
19
- - Loss: 0.4719
20
- - Wer: 0.1748
21
- - Mer: 0.1690
22
- - Wil: 0.2568
23
- - Wip: 0.7432
24
- - Hits: 55504
25
- - Substitutions: 6470
26
- - Deletions: 2613
27
- - Insertions: 2206
28
- - Cer: 0.1395
29
 
30
  ## Model description
31
 
@@ -57,16 +57,16 @@ The following hyperparameters were used during training:
57
 
58
  | Training Loss | Epoch | Step | Validation Loss | Wer | Mer | Wil | Wip | Hits | Substitutions | Deletions | Insertions | Cer |
59
  |:-------------:|:-----:|:-----:|:---------------:|:------:|:------:|:------:|:------:|:-----:|:-------------:|:---------:|:----------:|:------:|
60
- | 0.6161 | 1.0 | 1457 | 0.4938 | 0.2067 | 0.1968 | 0.2860 | 0.7140 | 54512 | 6651 | 3424 | 3278 | 0.1736 |
61
- | 0.5668 | 2.0 | 2914 | 0.4580 | 0.1897 | 0.1818 | 0.2709 | 0.7291 | 55150 | 6620 | 2817 | 2817 | 0.1538 |
62
- | 0.4918 | 3.0 | 4371 | 0.4463 | 0.1791 | 0.1729 | 0.2615 | 0.7385 | 55305 | 6537 | 2745 | 2283 | 0.1428 |
63
- | 0.3842 | 4.0 | 5828 | 0.4526 | 0.1744 | 0.1689 | 0.2562 | 0.7438 | 55449 | 6429 | 2709 | 2127 | 0.1364 |
64
- | 0.4054 | 5.0 | 7285 | 0.4524 | 0.1748 | 0.1691 | 0.2568 | 0.7432 | 55501 | 6467 | 2619 | 2207 | 0.1382 |
65
- | 0.4111 | 6.0 | 8742 | 0.4517 | 0.1758 | 0.1700 | 0.2578 | 0.7422 | 55449 | 6474 | 2664 | 2219 | 0.1402 |
66
- | 0.3628 | 7.0 | 10199 | 0.4563 | 0.1777 | 0.1715 | 0.2595 | 0.7405 | 55453 | 6502 | 2632 | 2342 | 0.1419 |
67
- | 0.3509 | 8.0 | 11656 | 0.4645 | 0.1752 | 0.1694 | 0.2572 | 0.7428 | 55480 | 6472 | 2635 | 2210 | 0.1411 |
68
- | 0.322 | 9.0 | 13113 | 0.4707 | 0.1756 | 0.1697 | 0.2582 | 0.7418 | 55494 | 6537 | 2556 | 2249 | 0.1400 |
69
- | 0.3529 | 10.0 | 14570 | 0.4719 | 0.1748 | 0.1690 | 0.2568 | 0.7432 | 55504 | 6470 | 2613 | 2206 | 0.1395 |
70
 
71
 
72
  ### Framework versions
 
16
 
17
  This model is a fine-tuned version of [sonoisa/t5-base-japanese](https://huggingface.co/sonoisa/t5-base-japanese) on the te_dx_jp dataset.
18
  It achieves the following results on the evaluation set:
19
+ - Loss: 0.4700
20
+ - Wer: 0.1779
21
+ - Mer: 0.1718
22
+ - Wil: 0.2600
23
+ - Wip: 0.7400
24
+ - Hits: 55384
25
+ - Substitutions: 6510
26
+ - Deletions: 2693
27
+ - Insertions: 2287
28
+ - Cer: 0.1398
29
 
30
  ## Model description
31
 
 
57
 
58
  | Training Loss | Epoch | Step | Validation Loss | Wer | Mer | Wil | Wip | Hits | Substitutions | Deletions | Insertions | Cer |
59
  |:-------------:|:-----:|:-----:|:---------------:|:------:|:------:|:------:|:------:|:-----:|:-------------:|:---------:|:----------:|:------:|
60
+ | 0.6519 | 1.0 | 1457 | 0.4991 | 0.2099 | 0.1985 | 0.2891 | 0.7109 | 54721 | 6807 | 3059 | 3689 | 0.1855 |
61
+ | 0.5507 | 2.0 | 2914 | 0.4589 | 0.1827 | 0.1764 | 0.2653 | 0.7347 | 55094 | 6566 | 2927 | 2305 | 0.1504 |
62
+ | 0.5097 | 3.0 | 4371 | 0.4493 | 0.1797 | 0.1734 | 0.2615 | 0.7385 | 55330 | 6503 | 2754 | 2352 | 0.1428 |
63
+ | 0.4457 | 4.0 | 5828 | 0.4458 | 0.1757 | 0.1702 | 0.2581 | 0.7419 | 55319 | 6463 | 2805 | 2078 | 0.1376 |
64
+ | 0.3913 | 5.0 | 7285 | 0.4486 | 0.1774 | 0.1716 | 0.2600 | 0.7400 | 55324 | 6525 | 2738 | 2195 | 0.1414 |
65
+ | 0.3641 | 6.0 | 8742 | 0.4553 | 0.1764 | 0.1706 | 0.2595 | 0.7405 | 55397 | 6566 | 2624 | 2202 | 0.1378 |
66
+ | 0.4101 | 7.0 | 10199 | 0.4596 | 0.1770 | 0.1711 | 0.2596 | 0.7404 | 55360 | 6528 | 2699 | 2202 | 0.1387 |
67
+ | 0.3305 | 8.0 | 11656 | 0.4654 | 0.1783 | 0.1722 | 0.2606 | 0.7394 | 55358 | 6528 | 2701 | 2288 | 0.1393 |
68
+ | 0.317 | 9.0 | 13113 | 0.4671 | 0.1782 | 0.1720 | 0.2604 | 0.7396 | 55386 | 6524 | 2677 | 2307 | 0.1400 |
69
+ | 0.3232 | 10.0 | 14570 | 0.4700 | 0.1779 | 0.1718 | 0.2600 | 0.7400 | 55384 | 6510 | 2693 | 2287 | 0.1398 |
70
 
71
 
72
  ### Framework versions