Hugging Face's logo

tartuNLP
/
xls-r-300m-et-children

Automatic Speech Recognition
Transformers
PyTorch
Estonian
wav2vec2
Generated from Trainer
Inference Endpoints
Model card Files Community
agnesluhtaru commited on
Commit
16a9ddb
·
1 Parent(s): a6df402

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +41 -41
README.md CHANGED
@@ -16,8 +16,8 @@ should probably proofread and complete it, then remove this comment. -->
16
 
17
  This model is a fine-tuned version of [TalTechNLP/xls-r-300m-et](https://huggingface.co/TalTechNLP/xls-r-300m-et) on an unknown dataset.
18
  It achieves the following results on the evaluation set:
19
- - Loss: 0.2157
20
- - Wer: 0.1631
21
 
22
  ## Model description
23
 
@@ -36,7 +36,7 @@ More information needed
36
  ### Training hyperparameters
37
 
38
  The following hyperparameters were used during training:
39
- - learning_rate: 0.0003
40
  - train_batch_size: 32
41
  - eval_batch_size: 8
42
  - seed: 42
@@ -44,50 +44,50 @@ The following hyperparameters were used during training:
44
  - total_train_batch_size: 64
45
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
46
  - lr_scheduler_type: linear
47
- - lr_scheduler_warmup_steps: 400
48
  - num_epochs: 60
49
 
50
  ### Training results
51
 
52
  | Training Loss | Epoch | Step | Validation Loss | Wer |
53
  |:-------------:|:-----:|:----:|:---------------:|:------:|
54
- | 0.886 | 1.61 | 100 | 0.1871 | 0.1647 |
55
- | 0.7967 | 3.22 | 200 | 0.1824 | 0.1629 |
56
- | 0.7786 | 4.83 | 300 | 0.1794 | 0.1658 |
57
- | 0.7732 | 6.45 | 400 | 0.1889 | 0.1732 |
58
- | 0.7747 | 8.06 | 500 | 0.1921 | 0.1779 |
59
- | 0.7446 | 9.67 | 600 | 0.1919 | 0.1741 |
60
- | 0.7297 | 11.29 | 700 | 0.2022 | 0.1754 |
61
- | 0.717 | 12.9 | 800 | 0.2039 | 0.1755 |
62
- | 0.7052 | 14.51 | 900 | 0.2070 | 0.1743 |
63
- | 0.6891 | 16.13 | 1000 | 0.1984 | 0.1720 |
64
- | 0.673 | 17.74 | 1100 | 0.2060 | 0.1775 |
65
- | 0.6699 | 19.35 | 1200 | 0.2032 | 0.1726 |
66
- | 0.6504 | 20.96 | 1300 | 0.2047 | 0.1773 |
67
- | 0.6323 | 22.58 | 1400 | 0.2080 | 0.1760 |
68
- | 0.6268 | 24.19 | 1500 | 0.2120 | 0.1744 |
69
- | 0.6056 | 25.8 | 1600 | 0.2057 | 0.1729 |
70
- | 0.5929 | 27.42 | 1700 | 0.2058 | 0.1707 |
71
- | 0.596 | 29.03 | 1800 | 0.2062 | 0.1732 |
72
- | 0.5798 | 30.64 | 1900 | 0.2088 | 0.1716 |
73
- | 0.5688 | 32.26 | 2000 | 0.2145 | 0.1695 |
74
- | 0.5594 | 33.86 | 2100 | 0.2162 | 0.1676 |
75
- | 0.5597 | 35.48 | 2200 | 0.2134 | 0.1713 |
76
- | 0.5481 | 37.1 | 2300 | 0.2156 | 0.1686 |
77
- | 0.5387 | 38.7 | 2400 | 0.2171 | 0.1704 |
78
- | 0.5349 | 40.32 | 2500 | 0.2163 | 0.1687 |
79
- | 0.5229 | 41.93 | 2600 | 0.2176 | 0.1680 |
80
- | 0.5186 | 43.54 | 2700 | 0.2205 | 0.1711 |
81
- | 0.5174 | 45.16 | 2800 | 0.2136 | 0.1611 |
82
- | 0.5098 | 46.77 | 2900 | 0.2169 | 0.1630 |
83
- | 0.5016 | 48.38 | 3000 | 0.2168 | 0.1652 |
84
- | 0.4982 | 49.99 | 3100 | 0.2167 | 0.1648 |
85
- | 0.4952 | 51.61 | 3200 | 0.2187 | 0.1640 |
86
- | 0.4866 | 53.22 | 3300 | 0.2154 | 0.1633 |
87
- | 0.481 | 54.83 | 3400 | 0.2167 | 0.1630 |
88
- | 0.4792 | 56.45 | 3500 | 0.2164 | 0.1624 |
89
- | 0.4781 | 58.06 | 3600 | 0.2164 | 0.1639 |
90
- | 0.4778 | 59.67 | 3700 | 0.2157 | 0.1631 |
91
 
92
 
93
  ### Framework versions
 
16
 
17
  This model is a fine-tuned version of [TalTechNLP/xls-r-300m-et](https://huggingface.co/TalTechNLP/xls-r-300m-et) on an unknown dataset.
18
  It achieves the following results on the evaluation set:
19
+ - Loss: 0.1926
20
+ - Wer: 0.1430
21
 
22
  ## Model description
23
 
 
36
  ### Training hyperparameters
37
 
38
  The following hyperparameters were used during training:
39
+ - learning_rate: 2e-05
40
  - train_batch_size: 32
41
  - eval_batch_size: 8
42
  - seed: 42
 
44
  - total_train_batch_size: 64
45
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
46
  - lr_scheduler_type: linear
47
+ - lr_scheduler_warmup_steps: 200
48
  - num_epochs: 60
49
 
50
  ### Training results
51
 
52
  | Training Loss | Epoch | Step | Validation Loss | Wer |
53
  |:-------------:|:-----:|:----:|:---------------:|:------:|
54
+ | 0.3485 | 1.61 | 100 | 0.2034 | 0.1782 |
55
+ | 0.1914 | 3.22 | 200 | 0.1818 | 0.1606 |
56
+ | 0.1641 | 4.83 | 300 | 0.1770 | 0.1572 |
57
+ | 0.1479 | 6.45 | 400 | 0.1745 | 0.1546 |
58
+ | 0.1374 | 8.06 | 500 | 0.1751 | 0.1538 |
59
+ | 0.1244 | 9.67 | 600 | 0.1734 | 0.1518 |
60
+ | 0.1211 | 11.29 | 700 | 0.1753 | 0.1508 |
61
+ | 0.1096 | 12.9 | 800 | 0.1758 | 0.1483 |
62
+ | 0.1059 | 14.51 | 900 | 0.1771 | 0.1469 |
63
+ | 0.0991 | 16.13 | 1000 | 0.1776 | 0.1469 |
64
+ | 0.0965 | 17.74 | 1100 | 0.1759 | 0.1469 |
65
+ | 0.0944 | 19.35 | 1200 | 0.1784 | 0.1459 |
66
+ | 0.0902 | 20.96 | 1300 | 0.1799 | 0.1469 |
67
+ | 0.0867 | 22.58 | 1400 | 0.1814 | 0.1440 |
68
+ | 0.0817 | 24.19 | 1500 | 0.1828 | 0.1438 |
69
+ | 0.0802 | 25.8 | 1600 | 0.1845 | 0.1438 |
70
+ | 0.0762 | 27.42 | 1700 | 0.1843 | 0.1431 |
71
+ | 0.0774 | 29.03 | 1800 | 0.1839 | 0.1432 |
72
+ | 0.0741 | 30.64 | 1900 | 0.1843 | 0.1442 |
73
+ | 0.0718 | 32.26 | 2000 | 0.1846 | 0.1429 |
74
+ | 0.07 | 33.86 | 2100 | 0.1852 | 0.1429 |
75
+ | 0.0692 | 35.48 | 2200 | 0.1872 | 0.1435 |
76
+ | 0.0671 | 37.1 | 2300 | 0.1874 | 0.1433 |
77
+ | 0.0667 | 38.7 | 2400 | 0.1887 | 0.1435 |
78
+ | 0.066 | 40.32 | 2500 | 0.1880 | 0.1422 |
79
+ | 0.0628 | 41.93 | 2600 | 0.1897 | 0.1426 |
80
+ | 0.0643 | 43.54 | 2700 | 0.1910 | 0.1428 |
81
+ | 0.0643 | 45.16 | 2800 | 0.1900 | 0.1431 |
82
+ | 0.0638 | 46.77 | 2900 | 0.1900 | 0.1427 |
83
+ | 0.0601 | 48.38 | 3000 | 0.1911 | 0.1431 |
84
+ | 0.0593 | 49.99 | 3100 | 0.1914 | 0.1432 |
85
+ | 0.0606 | 51.61 | 3200 | 0.1912 | 0.1433 |
86
+ | 0.0609 | 53.22 | 3300 | 0.1912 | 0.1431 |
87
+ | 0.0587 | 54.83 | 3400 | 0.1921 | 0.1429 |
88
+ | 0.0567 | 56.45 | 3500 | 0.1924 | 0.1430 |
89
+ | 0.0611 | 58.06 | 3600 | 0.1927 | 0.1431 |
90
+ | 0.0581 | 59.67 | 3700 | 0.1926 | 0.1430 |
91
 
92
 
93
  ### Framework versions