Hugging Face's logo

jcrkn
/
wav2vec2-large-xls-r-300m-breton-colab

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
wav2vec2
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
jcrkn commited on
Commit
a8a56aa
•
1 Parent(s): d16d7f9

End of training

Browse files
Files changed (2) hide show
  1. README.md +20 -16
  2. pytorch_model.bin +1 -1
README.md CHANGED
@@ -22,7 +22,7 @@ model-index:
22
  metrics:
23
  - name: Wer
24
  type: wer
25
- value: 0.8644540644540645
26
  ---
27
 
28
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -32,8 +32,8 @@ should probably proofread and complete it, then remove this comment. -->
32
 
33
  This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the common_voice_13_0 dataset.
34
  It achieves the following results on the evaluation set:
35
- - Loss: 1.3593
36
- - Wer: 0.8645
37
 
38
  ## Model description
39
 
@@ -56,28 +56,32 @@ The following hyperparameters were used during training:
56
  - train_batch_size: 8
57
  - eval_batch_size: 8
58
  - seed: 42
59
- - gradient_accumulation_steps: 4
60
- - total_train_batch_size: 32
61
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
62
  - lr_scheduler_type: linear
63
- - num_epochs: 5
 
64
 
65
  ### Training results
66
 
67
- | Training Loss | Epoch | Step | Validation Loss | Wer |
68
- |:-------------:|:-----:|:----:|:---------------:|:------:|
69
- | 2.7948 | 0.67 | 100 | 2.5632 | 0.9946 |
70
- | 2.0904 | 1.34 | 200 | 1.9792 | 0.9900 |
71
- | 1.541 | 2.02 | 300 | 1.7199 | 0.9699 |
72
- | 1.1538 | 2.69 | 400 | 1.4724 | 0.9375 |
73
- | 0.8951 | 3.36 | 500 | 1.3501 | 0.8887 |
74
- | 0.6965 | 4.03 | 600 | 1.3517 | 0.8834 |
75
- | 0.4569 | 4.71 | 700 | 1.3593 | 0.8645 |
 
 
 
76
 
77
 
78
  ### Framework versions
79
 
80
- - Transformers 4.32.0
81
  - Pytorch 2.0.1+cu118
82
  - Datasets 2.14.4
83
  - Tokenizers 0.13.3
 
22
  metrics:
23
  - name: Wer
24
  type: wer
25
+ value: 0.508994708994709
26
  ---
27
 
28
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
32
 
33
  This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the common_voice_13_0 dataset.
34
  It achieves the following results on the evaluation set:
35
+ - Loss: 1.2344
36
+ - Wer: 0.5090
37
 
38
  ## Model description
39
 
 
56
  - train_batch_size: 8
57
  - eval_batch_size: 8
58
  - seed: 42
59
+ - gradient_accumulation_steps: 2
60
+ - total_train_batch_size: 16
61
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
62
  - lr_scheduler_type: linear
63
+ - lr_scheduler_warmup_steps: 500
64
+ - num_epochs: 35
65
 
66
  ### Training results
67
 
68
+ | Training Loss | Epoch | Step | Validation Loss | Wer |
69
+ |:-------------:|:-----:|:-----:|:---------------:|:------:|
70
+ | 2.8178 | 3.36 | 1000 | 1.0244 | 0.7207 |
71
+ | 0.5674 | 6.72 | 2000 | 0.9848 | 0.6341 |
72
+ | 0.33 | 10.08 | 3000 | 1.0254 | 0.6014 |
73
+ | 0.2362 | 13.45 | 4000 | 1.1387 | 0.5848 |
74
+ | 0.1777 | 16.81 | 5000 | 1.2125 | 0.5783 |
75
+ | 0.1429 | 20.17 | 6000 | 1.1952 | 0.5572 |
76
+ | 0.1076 | 23.53 | 7000 | 1.2492 | 0.5628 |
77
+ | 0.0842 | 26.89 | 8000 | 1.2103 | 0.5410 |
78
+ | 0.0666 | 30.25 | 9000 | 1.2032 | 0.5128 |
79
+ | 0.051 | 33.61 | 10000 | 1.2344 | 0.5090 |
80
 
81
 
82
  ### Framework versions
83
 
84
+ - Transformers 4.32.1
85
  - Pytorch 2.0.1+cu118
86
  - Datasets 2.14.4
87
  - Tokenizers 0.13.3
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c19891e1e125324aff37f0437af4ac5fea1481d4bbb01001ff73db83350606a0
3
  size 1262078125
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bd6f3c765434120e49b282b44473f5e9f5f8b6861699815412e77a502b8eed34
3
  size 1262078125