Hugging Face's logo

techiaith
/
wav2vec2-xlsr-53-ft-cy-en-withlm

Automatic Speech Recognition
Transformers
Safetensors
Welsh
English
wav2vec2
Inference Endpoints
Model card Files Community
DewiBrynJones commited on
Commit
3a473cf
•
1 Parent(s): 7e934a7

Model save

Browse files
Files changed (1) hide show
  1. README.md +27 -45
README.md CHANGED
@@ -1,32 +1,13 @@
1
  ---
2
- language:
3
- - cy
4
  license: apache-2.0
5
  base_model: facebook/wav2vec2-large-xlsr-53
6
  tags:
7
- - automatic-speech-recognition
8
- - python/custom_common_voice.py
9
  - generated_from_trainer
10
- datasets:
11
- - custom_common_voice
12
  metrics:
13
  - wer
14
  model-index:
15
  - name: wav2vec2-xlsr-53-ft-ccv-en-cy
16
- results:
17
- - task:
18
- name: Automatic Speech Recognition
19
- type: automatic-speech-recognition
20
- dataset:
21
- name: PYTHON/CUSTOM_COMMON_VOICE.PY - CY
22
- type: custom_common_voice
23
- config: cy
24
- split: validation
25
- args: 'Config: cy, Training split: train, Eval split: validation'
26
- metrics:
27
- - name: Wer
28
- type: wer
29
- value: 0.21777283505046477
30
  ---
31
 
32
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -34,10 +15,10 @@ should probably proofread and complete it, then remove this comment. -->
34
 
35
  # wav2vec2-xlsr-53-ft-ccv-en-cy
36
 
37
- This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on the PYTHON/CUSTOM_COMMON_VOICE.PY - CY dataset.
38
  It achieves the following results on the evaluation set:
39
- - Loss: 0.2909
40
- - Wer: 0.2178
41
 
42
  ## Model description
43
 
@@ -66,34 +47,35 @@ The following hyperparameters were used during training:
66
  - lr_scheduler_type: linear
67
  - lr_scheduler_warmup_steps: 800
68
  - training_steps: 9000
 
69
 
70
  ### Training results
71
 
72
  | Training Loss | Epoch | Step | Validation Loss | Wer |
73
  |:-------------:|:-----:|:----:|:---------------:|:------:|
74
- | 5.8377 | 0.25 | 500 | 1.2190 | 0.8569 |
75
- | 0.9829 | 0.51 | 1000 | 0.5585 | 0.4701 |
76
- | 0.45 | 0.76 | 1500 | 0.4735 | 0.3901 |
77
- | 0.3151 | 1.01 | 2000 | 0.4125 | 0.3418 |
78
- | 0.2524 | 1.26 | 2500 | 0.3831 | 0.3117 |
79
- | 0.243 | 1.52 | 3000 | 0.3661 | 0.3078 |
80
- | 0.2341 | 1.77 | 3500 | 0.3489 | 0.2883 |
81
- | 0.211 | 2.02 | 4000 | 0.3500 | 0.2738 |
82
- | 0.1702 | 2.27 | 4500 | 0.3459 | 0.2704 |
83
- | 0.1634 | 2.53 | 5000 | 0.3305 | 0.2583 |
84
- | 0.1608 | 2.78 | 5500 | 0.3137 | 0.2479 |
85
- | 0.1481 | 3.03 | 6000 | 0.3288 | 0.2562 |
86
- | 0.1216 | 3.28 | 6500 | 0.3174 | 0.2446 |
87
- | 0.1181 | 3.54 | 7000 | 0.3000 | 0.2325 |
88
- | 0.1143 | 3.79 | 7500 | 0.2929 | 0.2326 |
89
- | 0.1049 | 4.04 | 8000 | 0.2921 | 0.2218 |
90
- | 0.0913 | 4.29 | 8500 | 0.2968 | 0.2208 |
91
- | 0.0883 | 4.55 | 9000 | 0.2909 | 0.2178 |
92
 
93
 
94
  ### Framework versions
95
 
96
- - Transformers 4.33.3
97
- - Pytorch 2.0.1+cu117
98
- - Datasets 2.14.5
99
- - Tokenizers 0.13.3
 
1
  ---
 
 
2
  license: apache-2.0
3
  base_model: facebook/wav2vec2-large-xlsr-53
4
  tags:
 
 
5
  - generated_from_trainer
 
 
6
  metrics:
7
  - wer
8
  model-index:
9
  - name: wav2vec2-xlsr-53-ft-ccv-en-cy
10
+ results: []
 
 
 
 
 
 
 
 
 
 
 
 
 
11
  ---
12
 
13
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
15
 
16
  # wav2vec2-xlsr-53-ft-ccv-en-cy
17
 
18
+ This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on an unknown dataset.
19
  It achieves the following results on the evaluation set:
20
+ - Loss: 0.2765
21
+ - Wer: 0.2115
22
 
23
  ## Model description
24
 
 
47
  - lr_scheduler_type: linear
48
  - lr_scheduler_warmup_steps: 800
49
  - training_steps: 9000
50
+ - mixed_precision_training: Native AMP
51
 
52
  ### Training results
53
 
54
  | Training Loss | Epoch | Step | Validation Loss | Wer |
55
  |:-------------:|:-----:|:----:|:---------------:|:------:|
56
+ | 5.9898 | 0.25 | 500 | 1.3093 | 0.7971 |
57
+ | 1.0749 | 0.5 | 1000 | 0.5816 | 0.4617 |
58
+ | 0.4332 | 0.75 | 1500 | 0.4834 | 0.4091 |
59
+ | 0.3303 | 1.01 | 2000 | 0.4203 | 0.3419 |
60
+ | 0.276 | 1.26 | 2500 | 0.3910 | 0.3186 |
61
+ | 0.2591 | 1.51 | 3000 | 0.3901 | 0.3067 |
62
+ | 0.2501 | 1.76 | 3500 | 0.3646 | 0.2895 |
63
+ | 0.224 | 2.01 | 4000 | 0.3517 | 0.2806 |
64
+ | 0.182 | 2.26 | 4500 | 0.3348 | 0.2656 |
65
+ | 0.1777 | 2.51 | 5000 | 0.3277 | 0.2612 |
66
+ | 0.1734 | 2.77 | 5500 | 0.3323 | 0.2643 |
67
+ | 0.1629 | 3.02 | 6000 | 0.3171 | 0.2485 |
68
+ | 0.1338 | 3.27 | 6500 | 0.3103 | 0.2398 |
69
+ | 0.1292 | 3.52 | 7000 | 0.2934 | 0.2268 |
70
+ | 0.1264 | 3.77 | 7500 | 0.2923 | 0.2248 |
71
+ | 0.118 | 4.02 | 8000 | 0.2880 | 0.2193 |
72
+ | 0.0996 | 4.27 | 8500 | 0.2793 | 0.2124 |
73
+ | 0.0969 | 4.52 | 9000 | 0.2765 | 0.2115 |
74
 
75
 
76
  ### Framework versions
77
 
78
+ - Transformers 4.38.2
79
+ - Pytorch 2.2.1+cu121
80
+ - Datasets 2.18.0
81
+ - Tokenizers 0.15.2