Hugging Face's logo

michaelszhu
/
wav2vec2-base-finetuned-radio-ASR-2

Automatic Speech Recognition
Transformers
Safetensors
English
wav2vec2
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Community
michaelszhu commited on
Commit
b3d8b23
•
1 Parent(s): 98c1984

End of training

Browse files
Files changed (3) hide show
  1. README.md +20 -14
  2. model.safetensors +1 -1
  3. training_args.bin +1 -1
README.md CHANGED
@@ -23,7 +23,7 @@ model-index:
23
  metrics:
24
  - name: Wer
25
  type: wer
26
- value: 100.0
27
  ---
28
 
29
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -33,8 +33,8 @@ should probably proofread and complete it, then remove this comment. -->
33
 
34
  This model is a fine-tuned version of [openai/wav2vec2](https://huggingface.co/openai/wav2vec2) on the Radio-Modified Common Voice 11.0 dataset.
35
  It achieves the following results on the evaluation set:
36
- - Loss: 382.1391
37
- - Wer: 100.0
38
 
39
  ## Model description
40
 
@@ -53,31 +53,37 @@ More information needed
53
  ### Training hyperparameters
54
 
55
  The following hyperparameters were used during training:
56
- - learning_rate: 0.0001
57
  - train_batch_size: 8
58
  - eval_batch_size: 8
59
  - seed: 42
 
 
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
- - training_steps: 6000
64
  - mixed_precision_training: Native AMP
65
 
66
  ### Training results
67
 
68
- | Training Loss | Epoch | Step | Validation Loss | Wer |
69
- |:-------------:|:------:|:----:|:---------------:|:-----:|
70
- | 506.6064 | 0.1667 | 1000 | 442.2693 | 100.0 |
71
- | 435.214 | 0.3333 | 2000 | 379.3636 | 100.0 |
72
- | 460.0346 | 0.5 | 3000 | 379.4807 | 100.0 |
73
- | 416.7359 | 0.6667 | 4000 | 382.4834 | 100.0 |
74
- | 443.2083 | 0.8333 | 5000 | 380.2564 | 100.0 |
75
- | 407.3617 | 1.0 | 6000 | 382.1391 | 100.0 |
 
 
 
 
76
 
77
 
78
  ### Framework versions
79
 
80
  - Transformers 4.41.2
81
- - Pytorch 2.3.0+cu121
82
  - Datasets 2.19.2
83
  - Tokenizers 0.19.1
 
23
  metrics:
24
  - name: Wer
25
  type: wer
26
+ value: 93.31727244160469
27
  ---
28
 
29
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
33
 
34
  This model is a fine-tuned version of [openai/wav2vec2](https://huggingface.co/openai/wav2vec2) on the Radio-Modified Common Voice 11.0 dataset.
35
  It achieves the following results on the evaluation set:
36
+ - Loss: 274.6956
37
+ - Wer: 93.3173
38
 
39
  ## Model description
40
 
 
53
  ### Training hyperparameters
54
 
55
  The following hyperparameters were used during training:
56
+ - learning_rate: 1e-05
57
  - train_batch_size: 8
58
  - eval_batch_size: 8
59
  - seed: 42
60
+ - gradient_accumulation_steps: 2
61
+ - total_train_batch_size: 16
62
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
63
  - lr_scheduler_type: linear
64
  - lr_scheduler_warmup_steps: 500
65
+ - training_steps: 10000
66
  - mixed_precision_training: Native AMP
67
 
68
  ### Training results
69
 
70
+ | Training Loss | Epoch | Step | Validation Loss | Wer |
71
+ |:-------------:|:-----:|:-----:|:---------------:|:-------:|
72
+ | 454.7647 | 0.1 | 1000 | 432.4697 | 96.7955 |
73
+ | 381.9785 | 0.2 | 2000 | 345.0900 | 97.3774 |
74
+ | 346.4287 | 0.3 | 3000 | 345.3813 | 96.7005 |
75
+ | 313.4983 | 0.4 | 4000 | 334.5611 | 95.2668 |
76
+ | 344.4422 | 0.5 | 5000 | 422.9466 | 96.1369 |
77
+ | 349.3033 | 0.6 | 6000 | 337.1495 | 91.9877 |
78
+ | 337.6954 | 0.7 | 7000 | 299.1713 | 95.9529 |
79
+ | 313.1935 | 0.8 | 8000 | 283.5478 | 95.4207 |
80
+ | 348.8207 | 0.9 | 9000 | 275.7858 | 92.5278 |
81
+ | 325.8637 | 1.0 | 10000 | 274.6956 | 93.3173 |
82
 
83
 
84
  ### Framework versions
85
 
86
  - Transformers 4.41.2
87
+ - Pytorch 2.3.1+cu121
88
  - Datasets 2.19.2
89
  - Tokenizers 0.19.1
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a54e85fd5ed766deadceeecf35c2ba1375d85b121c90f41ab5f527d61c7b7ff3
3
  size 377611120
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8dd7b3b3e95c65bff047a53687889dbe9164db3f93cfd7dad051d200e0a7dc59
3
  size 377611120
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f15056c56eaaa934b15edc47c28523cfec498d7e58ee9909a7f34e4f81af1296
3
  size 5048
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1b477aae325973fc97f326529b6d51f2ef0286007caefdce8d143ba8e36ecd8b
3
  size 5048