Hugging Face's logo

techiaith
/
wav2vec-xlsr-ft-cy-verbatim

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
Welsh
wav2vec2
techiaith/banc-trawsgrifiadau-bangor
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
DewiBrynJones commited on
Commit
a700171
•
1 Parent(s): 3482f63

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +101 -0
README.md ADDED
@@ -0,0 +1,101 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ tags:
4
+ - generated_from_trainer
5
+ datasets:
6
+ - banc-trawsgrifiadau-bangor
7
+ metrics:
8
+ - wer
9
+ model-index:
10
+ - name: wav2vec2-xlsr-ft-btb
11
+ results:
12
+ - task:
13
+ name: Automatic Speech Recognition
14
+ type: automatic-speech-recognition
15
+ dataset:
16
+ name: banc-trawsgrifiadau-bangor
17
+ type: banc-trawsgrifiadau-bangor
18
+ config: default
19
+ split: test
20
+ args: default
21
+ metrics:
22
+ - name: Wer
23
+ type: wer
24
+ value: 0.3264155718657249
25
+ ---
26
+
27
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
28
+ should probably proofread and complete it, then remove this comment. -->
29
+
30
+ # wav2vec2-xlsr-ft-btb
31
+
32
+ This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on the banc-trawsgrifiadau-bangor dataset.
33
+ It achieves the following results on the evaluation set:
34
+ - Loss: 0.4358
35
+ - Wer: 0.3264
36
+
37
+ ## Model description
38
+
39
+ More information needed
40
+
41
+ ## Intended uses & limitations
42
+
43
+ More information needed
44
+
45
+ ## Training and evaluation data
46
+
47
+ More information needed
48
+
49
+ ## Training procedure
50
+
51
+ ### Training hyperparameters
52
+
53
+ The following hyperparameters were used during training:
54
+ - learning_rate: 0.0003
55
+ - train_batch_size: 16
56
+ - eval_batch_size: 8
57
+ - seed: 42
58
+ - gradient_accumulation_steps: 2
59
+ - total_train_batch_size: 32
60
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
61
+ - lr_scheduler_type: linear
62
+ - lr_scheduler_warmup_steps: 500
63
+ - num_epochs: 5.0
64
+ - mixed_precision_training: Native AMP
65
+
66
+ ### Training results
67
+
68
+ | Training Loss | Epoch | Step | Validation Loss | Wer |
69
+ |:-------------:|:-----:|:----:|:---------------:|:------:|
70
+ | No log | 0.21 | 100 | 3.4135 | 1.0 |
71
+ | No log | 0.41 | 200 | 2.9521 | 1.0 |
72
+ | No log | 0.62 | 300 | 2.3339 | 0.9365 |
73
+ | No log | 0.83 | 400 | 1.2433 | 0.8259 |
74
+ | 3.1912 | 1.03 | 500 | 0.8614 | 0.6385 |
75
+ | 3.1912 | 1.24 | 600 | 0.7557 | 0.5612 |
76
+ | 3.1912 | 1.44 | 700 | 0.6781 | 0.5195 |
77
+ | 3.1912 | 1.65 | 800 | 0.6363 | 0.4879 |
78
+ | 3.1912 | 1.86 | 900 | 0.5959 | 0.4559 |
79
+ | 0.8237 | 2.06 | 1000 | 0.5430 | 0.4260 |
80
+ | 0.8237 | 2.27 | 1100 | 0.5293 | 0.4098 |
81
+ | 0.8237 | 2.48 | 1200 | 0.5141 | 0.4056 |
82
+ | 0.8237 | 2.68 | 1300 | 0.4879 | 0.3947 |
83
+ | 0.8237 | 2.89 | 1400 | 0.4697 | 0.3788 |
84
+ | 0.5625 | 3.1 | 1500 | 0.4748 | 0.3780 |
85
+ | 0.5625 | 3.3 | 1600 | 0.4836 | 0.3684 |
86
+ | 0.5625 | 3.51 | 1700 | 0.4796 | 0.3625 |
87
+ | 0.5625 | 3.72 | 1800 | 0.4582 | 0.3515 |
88
+ | 0.5625 | 3.92 | 1900 | 0.4395 | 0.3437 |
89
+ | 0.4267 | 4.13 | 2000 | 0.4410 | 0.3420 |
90
+ | 0.4267 | 4.33 | 2100 | 0.4467 | 0.3382 |
91
+ | 0.4267 | 4.54 | 2200 | 0.4398 | 0.3329 |
92
+ | 0.4267 | 4.75 | 2300 | 0.4383 | 0.3287 |
93
+ | 0.4267 | 4.95 | 2400 | 0.4358 | 0.3264 |
94
+
95
+
96
+ ### Framework versions
97
+
98
+ - Transformers 4.28.1
99
+ - Pytorch 2.0.0+cu117
100
+ - Datasets 2.11.0
101
+ - Tokenizers 0.13.3