Hugging Face's logo

nrshoudi
/
wav2vec2-base-960h-Arabic

Automatic Speech Recognition
Transformers
PyTorch
wav2vec2
Generated from Trainer
Inference Endpoints
Model card Files Community
1
nrshoudi commited on
Commit
9c9b9d6
•
1 Parent(s): 539b3a7

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +42 -32
README.md CHANGED
@@ -16,7 +16,7 @@ should probably proofread and complete it, then remove this comment. -->
16
 
17
  This model is a fine-tuned version of [facebook/wav2vec2-base-960h](https://huggingface.co/facebook/wav2vec2-base-960h) on the None dataset.
18
  It achieves the following results on the evaluation set:
19
- - Loss: 0.6429
20
  - Wer: 1.0
21
  - Cer: 1.0
22
 
@@ -46,43 +46,53 @@ The following hyperparameters were used during training:
46
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
47
  - lr_scheduler_type: linear
48
  - lr_scheduler_warmup_steps: 250
49
- - num_epochs: 30.0
50
  - mixed_precision_training: Native AMP
51
 
52
  ### Training results
53
 
54
  | Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
55
  |:-------------:|:-----:|:----:|:---------------:|:---:|:---:|
56
- | 8.0949 | 1.0 | 51 | 2.5659 | 1.0 | 1.0 |
57
- | 1.6592 | 2.0 | 102 | 3.0708 | 1.0 | 1.0 |
58
- | 1.498 | 3.0 | 153 | 1.1693 | 1.0 | 1.0 |
59
- | 1.2662 | 4.0 | 204 | 0.6669 | 1.0 | 1.0 |
60
- | 2.0861 | 5.0 | 255 | 2.4768 | 1.0 | 1.0 |
61
- | 2.0145 | 6.0 | 306 | 0.6489 | 1.0 | 1.0 |
62
- | 1.0447 | 7.0 | 357 | 0.6436 | 1.0 | 1.0 |
63
- | 1.04 | 8.0 | 408 | 0.9009 | 1.0 | 1.0 |
64
- | 1.0605 | 9.0 | 459 | 0.9958 | 1.0 | 1.0 |
65
- | 0.927 | 10.0 | 510 | 0.6903 | 1.0 | 1.0 |
66
- | 0.8545 | 11.0 | 561 | 1.1133 | 1.0 | 1.0 |
67
- | 0.8015 | 12.0 | 612 | 0.6406 | 1.0 | 1.0 |
68
- | 0.7248 | 13.0 | 663 | 0.7234 | 1.0 | 1.0 |
69
- | 0.7024 | 14.0 | 714 | 0.6867 | 1.0 | 1.0 |
70
- | 0.785 | 15.0 | 765 | 0.9036 | 1.0 | 1.0 |
71
- | 0.6882 | 16.0 | 816 | 0.8451 | 1.0 | 1.0 |
72
- | 0.6904 | 17.0 | 867 | 0.7125 | 1.0 | 1.0 |
73
- | 0.6959 | 18.0 | 918 | 0.7053 | 1.0 | 1.0 |
74
- | 0.6743 | 19.0 | 969 | 0.6756 | 1.0 | 1.0 |
75
- | 0.8035 | 20.0 | 1020 | 0.7716 | 1.0 | 1.0 |
76
- | 0.809 | 21.0 | 1071 | 0.7694 | 1.0 | 1.0 |
77
- | 0.7035 | 22.0 | 1122 | 0.6683 | 1.0 | 1.0 |
78
- | 0.7812 | 23.0 | 1173 | 0.6530 | 1.0 | 1.0 |
79
- | 0.7244 | 24.0 | 1224 | 0.6563 | 1.0 | 1.0 |
80
- | 0.7568 | 25.0 | 1275 | 0.6359 | 1.0 | 1.0 |
81
- | 0.6387 | 26.0 | 1326 | 0.6359 | 1.0 | 1.0 |
82
- | 0.6316 | 27.0 | 1377 | 0.6882 | 1.0 | 1.0 |
83
- | 0.7283 | 28.0 | 1428 | 0.6538 | 1.0 | 1.0 |
84
- | 0.6035 | 29.0 | 1479 | 0.6484 | 1.0 | 1.0 |
85
- | 0.6 | 30.0 | 1530 | 0.6429 | 1.0 | 1.0 |
 
 
 
 
 
 
 
 
 
 
86
 
87
 
88
  ### Framework versions
 
16
 
17
  This model is a fine-tuned version of [facebook/wav2vec2-base-960h](https://huggingface.co/facebook/wav2vec2-base-960h) on the None dataset.
18
  It achieves the following results on the evaluation set:
19
+ - Loss: 0.6190
20
  - Wer: 1.0
21
  - Cer: 1.0
22
 
 
46
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
47
  - lr_scheduler_type: linear
48
  - lr_scheduler_warmup_steps: 250
49
+ - num_epochs: 40
50
  - mixed_precision_training: Native AMP
51
 
52
  ### Training results
53
 
54
  | Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
55
  |:-------------:|:-----:|:----:|:---------------:|:---:|:---:|
56
+ | 8.1025 | 1.0 | 51 | 4.9004 | 1.0 | 1.0 |
57
+ | 1.3634 | 2.0 | 102 | 1.2766 | 1.0 | 1.0 |
58
+ | 1.1559 | 3.0 | 153 | 1.1179 | 1.0 | 1.0 |
59
+ | 1.0742 | 4.0 | 204 | 1.4401 | 1.0 | 1.0 |
60
+ | 1.0112 | 5.0 | 255 | 0.7570 | 1.0 | 1.0 |
61
+ | 0.9263 | 6.0 | 306 | 0.6297 | 1.0 | 1.0 |
62
+ | 0.9983 | 7.0 | 357 | 1.7016 | 1.0 | 1.0 |
63
+ | 0.7949 | 8.0 | 408 | 0.6406 | 1.0 | 1.0 |
64
+ | 0.9205 | 9.0 | 459 | 1.0371 | 1.0 | 1.0 |
65
+ | 0.7783 | 10.0 | 510 | 0.8300 | 1.0 | 1.0 |
66
+ | 0.8202 | 11.0 | 561 | 0.7223 | 1.0 | 1.0 |
67
+ | 0.7737 | 12.0 | 612 | 0.7909 | 1.0 | 1.0 |
68
+ | 0.7426 | 13.0 | 663 | 0.7968 | 1.0 | 1.0 |
69
+ | 0.7211 | 14.0 | 714 | 0.7648 | 1.0 | 1.0 |
70
+ | 0.7526 | 15.0 | 765 | 0.6257 | 1.0 | 1.0 |
71
+ | 0.7361 | 16.0 | 816 | 0.6500 | 1.0 | 1.0 |
72
+ | 0.716 | 17.0 | 867 | 0.9519 | 1.0 | 1.0 |
73
+ | 0.7595 | 18.0 | 918 | 0.6324 | 1.0 | 1.0 |
74
+ | 0.8078 | 19.0 | 969 | 0.8474 | 1.0 | 1.0 |
75
+ | 0.7761 | 20.0 | 1020 | 0.6274 | 1.0 | 1.0 |
76
+ | 0.6514 | 21.0 | 1071 | 0.7698 | 1.0 | 1.0 |
77
+ | 0.8607 | 22.0 | 1122 | 0.6179 | 1.0 | 1.0 |
78
+ | 0.7999 | 23.0 | 1173 | 0.6416 | 1.0 | 1.0 |
79
+ | 0.7267 | 24.0 | 1224 | 0.6506 | 1.0 | 1.0 |
80
+ | 0.6705 | 25.0 | 1275 | 0.6232 | 1.0 | 1.0 |
81
+ | 0.6669 | 26.0 | 1326 | 0.6472 | 1.0 | 1.0 |
82
+ | 0.6731 | 27.0 | 1377 | 0.6190 | 1.0 | 1.0 |
83
+ | 0.6532 | 28.0 | 1428 | 0.6197 | 1.0 | 1.0 |
84
+ | 0.6423 | 29.0 | 1479 | 0.6608 | 1.0 | 1.0 |
85
+ | 0.6574 | 30.0 | 1530 | 0.6175 | 1.0 | 1.0 |
86
+ | 0.6586 | 31.0 | 1581 | 0.6320 | 1.0 | 1.0 |
87
+ | 0.6339 | 32.0 | 1632 | 0.6196 | 1.0 | 1.0 |
88
+ | 0.6628 | 33.0 | 1683 | 0.6176 | 1.0 | 1.0 |
89
+ | 0.6222 | 34.0 | 1734 | 0.6434 | 1.0 | 1.0 |
90
+ | 0.6293 | 35.0 | 1785 | 0.6301 | 1.0 | 1.0 |
91
+ | 0.6337 | 36.0 | 1836 | 0.6357 | 1.0 | 1.0 |
92
+ | 0.6168 | 37.0 | 1887 | 0.6179 | 1.0 | 1.0 |
93
+ | 0.6093 | 38.0 | 1938 | 0.6197 | 1.0 | 1.0 |
94
+ | 0.6053 | 39.0 | 1989 | 0.6188 | 1.0 | 1.0 |
95
+ | 0.6014 | 40.0 | 2040 | 0.6190 | 1.0 | 1.0 |
96
 
97
 
98
  ### Framework versions