Hugging Face's logo

notebotIE
/
whisper-large-v2-swiss-german

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
Swiss German
whisper
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
Leo1212 commited on
Commit
c0093dc
•
1 Parent(s): ac0c4ef

End of training

Browse files
Files changed (3) hide show
  1. README.md +12 -11
  2. model-00001-of-00002.safetensors +1 -1
  3. model-00002-of-00002.safetensors +1 -1
README.md CHANGED
@@ -22,7 +22,7 @@ model-index:
22
  metrics:
23
  - name: Wer
24
  type: wer
25
- value: 0.1923865967631639
26
  ---
27
 
28
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -32,10 +32,10 @@ should probably proofread and complete it, then remove this comment. -->
32
 
33
  This model is a fine-tuned version of [openai/whisper-large-v2](https://huggingface.co/openai/whisper-large-v2) on the SwissDialDataset_ETH dataset.
34
  It achieves the following results on the evaluation set:
35
- - Loss: 0.2725
36
- - Wer Ortho: 0.2707
37
- - Wer: 0.1924
38
- - Cer: 0.0743
39
 
40
  ## Model description
41
 
@@ -62,18 +62,19 @@ The following hyperparameters were used during training:
62
  - total_train_batch_size: 16
63
  - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
64
  - lr_scheduler_type: constant_with_warmup
65
- - lr_scheduler_warmup_steps: 50
66
- - training_steps: 1000
67
  - mixed_precision_training: Native AMP
68
 
69
  ### Training results
70
 
71
  | Training Loss | Epoch | Step | Validation Loss | Wer Ortho | Wer | Cer |
72
  |:-------------:|:------:|:----:|:---------------:|:---------:|:------:|:------:|
73
- | 0.1277 | 1.2300 | 250 | 0.2468 | 0.2631 | 0.1855 | 0.0556 |
74
- | 0.0726 | 2.4600 | 500 | 0.2414 | 0.9085 | 0.6793 | 0.5239 |
75
- | 0.0405 | 3.6900 | 750 | 0.2519 | 0.5194 | 0.6558 | 0.4984 |
76
- | 0.0296 | 4.9200 | 1000 | 0.2725 | 0.2707 | 0.1924 | 0.0743 |
 
77
 
78
 
79
  ### Framework versions
 
22
  metrics:
23
  - name: Wer
24
  type: wer
25
+ value: 0.15773877364941874
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 [openai/whisper-large-v2](https://huggingface.co/openai/whisper-large-v2) on the SwissDialDataset_ETH dataset.
34
  It achieves the following results on the evaluation set:
35
+ - Loss: 0.2462
36
+ - Wer Ortho: 0.2459
37
+ - Wer: 0.1577
38
+ - Cer: 0.0373
39
 
40
  ## Model description
41
 
 
62
  - total_train_batch_size: 16
63
  - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
64
  - lr_scheduler_type: constant_with_warmup
65
+ - lr_scheduler_warmup_steps: 5
66
+ - training_steps: 250
67
  - mixed_precision_training: Native AMP
68
 
69
  ### Training results
70
 
71
  | Training Loss | Epoch | Step | Validation Loss | Wer Ortho | Wer | Cer |
72
  |:-------------:|:------:|:----:|:---------------:|:---------:|:------:|:------:|
73
+ | 0.4177 | 0.2460 | 50 | 0.3617 | 0.3915 | 0.3244 | 0.1232 |
74
+ | 0.285 | 0.4920 | 100 | 0.3100 | 0.2905 | 0.2013 | 0.0409 |
75
+ | 0.2659 | 0.7380 | 150 | 0.2632 | 0.3753 | 0.2909 | 0.4770 |
76
+ | 0.2401 | 0.9840 | 200 | 0.2372 | 0.2541 | 0.1568 | 0.0321 |
77
+ | 0.1192 | 1.2300 | 250 | 0.2462 | 0.2459 | 0.1577 | 0.0373 |
78
 
79
 
80
  ### Framework versions
model-00001-of-00002.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:4727554458297192a38a7c989c2b3a97788525669a911618df12248fafa7f29c
3
  size 4992706480
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6289cbc566e98df382e7f9cb668e83c4593ea515b8b92f450bc507d3e7632109
3
  size 4992706480
model-00002-of-00002.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d60dbd8c917156efb9c69f6b035011b19e3309b21749b75d1ad9c34a358d63b1
3
  size 1180663192
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:69b0598214cf6c54127cf981a4cc8ed947c396ffbc267accc5bd8828568d1eaf
3
  size 1180663192