Hugging Face's logo

rmacek
/
ORF-small-de

PEFT
TensorBoard
Safetensors
German
Generated from Trainer
Eval Results
Model card Files Metrics Community
rmacek commited on
Commit
24babce
·
verified ·
1 Parent(s): 472a6bd

End of training

Browse files
Files changed (2) hide show
  1. README.md +12 -13
  2. adapter_model.safetensors +1 -1
README.md CHANGED
@@ -22,7 +22,7 @@ model-index:
22
  args: 'config: de, split: test'
23
  metrics:
24
  - type: wer
25
- value: 29.46806728721495
26
  name: Wer
27
  ---
28
 
@@ -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/whisper-small](https://huggingface.co/openai/whisper-small) on the ZIB2 Common Voice dataset.
35
  It achieves the following results on the evaluation set:
36
- - Loss: 0.7149
37
- - Wer: 29.4681
38
 
39
  ## Model description
40
 
@@ -58,8 +58,7 @@ The following hyperparameters were used during training:
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: 8000
64
  - mixed_precision_training: Native AMP
65
 
@@ -67,14 +66,14 @@ The following hyperparameters were used during training:
67
 
68
  | Training Loss | Epoch | Step | Validation Loss | Wer |
69
  |:-------------:|:-------:|:----:|:---------------:|:-------:|
70
- | 0.5717 | 1.7153 | 1000 | 0.6098 | 35.2851 |
71
- | 0.41 | 3.4305 | 2000 | 0.6053 | 32.7056 |
72
- | 0.386 | 5.1458 | 3000 | 0.6148 | 26.9699 |
73
- | 0.3123 | 6.8611 | 4000 | 0.6345 | 26.4842 |
74
- | 0.2183 | 8.5763 | 5000 | 0.6622 | 28.1329 |
75
- | 0.2231 | 10.2916 | 6000 | 0.6901 | 28.5253 |
76
- | 0.2259 | 12.0069 | 7000 | 0.7028 | 27.4413 |
77
- | 0.1708 | 13.7221 | 8000 | 0.7149 | 29.4681 |
78
 
79
 
80
  ### Framework versions
 
22
  args: 'config: de, split: test'
23
  metrics:
24
  - type: wer
25
+ value: 27.268895060503866
26
  name: Wer
27
  ---
28
 
 
33
 
34
  This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the ZIB2 Common Voice dataset.
35
  It achieves the following results on the evaluation set:
36
+ - Loss: 0.7038
37
+ - Wer: 27.2689
38
 
39
  ## Model description
40
 
 
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: cosine
 
62
  - training_steps: 8000
63
  - mixed_precision_training: Native AMP
64
 
 
66
 
67
  | Training Loss | Epoch | Step | Validation Loss | Wer |
68
  |:-------------:|:-------:|:----:|:---------------:|:-------:|
69
+ | 0.4896 | 1.7153 | 1000 | 0.6019 | 26.6961 |
70
+ | 0.3559 | 3.4305 | 2000 | 0.6038 | 26.6192 |
71
+ | 0.259 | 5.1458 | 3000 | 0.6216 | 33.8450 |
72
+ | 0.3272 | 6.8611 | 4000 | 0.6382 | 27.0730 |
73
+ | 0.2413 | 8.5763 | 5000 | 0.6704 | 31.3207 |
74
+ | 0.1691 | 10.2916 | 6000 | 0.6922 | 27.2466 |
75
+ | 0.1702 | 12.0069 | 7000 | 0.7008 | 27.3284 |
76
+ | 0.1726 | 13.7221 | 8000 | 0.7038 | 27.2689 |
77
 
78
 
79
  ### Framework versions
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d65b34ecd718df11e249cb3d52097a133fedefe174f6b06a66a783cd768313de
3
  size 14176064
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cd164ecba99e337365e9c84cfda59a6190f66fa7d30734c28d92e36f58a7b3d4
3
  size 14176064