Hugging Face's logo

jethrowang
/
vanilla-whisper-tiny_tat

TensorBoard
Safetensors
Chinese
whisper
Generated from Trainer
Model card Files Metrics Community
jethrowang commited on
Commit
a20004f
·
verified ·
1 Parent(s): 97ad61d

End of training

Browse files
Files changed (1) hide show
  1. README.md +27 -17
README.md CHANGED
@@ -22,7 +22,7 @@ model-index:
22
  metrics:
23
  - name: Wer
24
  type: wer
25
- value: 50.77879890051919
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 [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the TAT ASR Aligned dataset.
34
  It achieves the following results on the evaluation set:
35
- - Loss: 0.9571
36
- - Wer: 50.7788
37
 
38
  ## Model description
39
 
@@ -58,24 +58,34 @@ The following hyperparameters were used during training:
58
  - seed: 42
59
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
60
  - lr_scheduler_type: linear
61
- - lr_scheduler_warmup_steps: 681
62
- - training_steps: 6810
63
  - mixed_precision_training: Native AMP
64
 
65
  ### Training results
66
 
67
- | Training Loss | Epoch | Step | Validation Loss | Wer |
68
- |:-------------:|:------:|:----:|:---------------:|:-------:|
69
- | 0.5845 | 0.9985 | 681 | 0.7811 | 73.2571 |
70
- | 0.3661 | 1.9971 | 1362 | 0.6189 | 64.0836 |
71
- | 0.246 | 2.9956 | 2043 | 0.5674 | 60.3548 |
72
- | 0.1568 | 3.9941 | 2724 | 0.5329 | 55.8153 |
73
- | 0.1081 | 4.9927 | 3405 | 0.5543 | 55.5654 |
74
- | 0.064 | 5.9912 | 4086 | 0.6285 | 55.5793 |
75
- | 0.0397 | 6.9897 | 4767 | 0.6875 | 53.7385 |
76
- | 0.0204 | 7.9883 | 5448 | 0.7531 | 52.5585 |
77
- | 0.0071 | 8.9868 | 6129 | 0.8651 | 51.1647 |
78
- | 0.0014 | 9.9853 | 6810 | 0.9571 | 50.7788 |
 
 
 
 
 
 
 
 
 
 
79
 
80
 
81
  ### Framework versions
 
22
  metrics:
23
  - name: Wer
24
  type: wer
25
+ value: 50.65108143376739
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-tiny](https://huggingface.co/openai/whisper-tiny) on the TAT ASR Aligned dataset.
34
  It achieves the following results on the evaluation set:
35
+ - Loss: 1.0814
36
+ - Wer: 50.6511
37
 
38
  ## Model description
39
 
 
58
  - seed: 42
59
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
60
  - lr_scheduler_type: linear
61
+ - lr_scheduler_warmup_steps: 1362
62
+ - training_steps: 13620
63
  - mixed_precision_training: Native AMP
64
 
65
  ### Training results
66
 
67
+ | Training Loss | Epoch | Step | Validation Loss | Wer |
68
+ |:-------------:|:-------:|:-----:|:---------------:|:-------:|
69
+ | 0.4229 | 0.9985 | 681 | 0.6344 | 65.8772 |
70
+ | 0.4529 | 1.9971 | 1362 | 0.7311 | 71.3135 |
71
+ | 0.3271 | 2.9956 | 2043 | 0.6324 | 64.7777 |
72
+ | 0.2257 | 3.9941 | 2724 | 0.5510 | 58.4141 |
73
+ | 0.1554 | 4.9927 | 3405 | 0.5350 | 57.6394 |
74
+ | 0.122 | 5.9912 | 4086 | 0.5911 | 56.8148 |
75
+ | 0.0935 | 6.9897 | 4767 | 0.6122 | 56.1179 |
76
+ | 0.0718 | 7.9883 | 5448 | 0.6492 | 54.7158 |
77
+ | 0.0588 | 8.9868 | 6129 | 0.6623 | 55.5599 |
78
+ | 0.0466 | 9.9853 | 6810 | 0.6883 | 56.9481 |
79
+ | 0.0349 | 10.9839 | 7491 | 0.7069 | 54.6770 |
80
+ | 0.0298 | 11.9824 | 8172 | 0.7441 | 54.2272 |
81
+ | 0.0215 | 12.9809 | 8853 | 0.7937 | 54.9491 |
82
+ | 0.0141 | 13.9795 | 9534 | 0.8062 | 52.9278 |
83
+ | 0.01 | 14.9780 | 10215 | 0.8717 | 53.2804 |
84
+ | 0.0067 | 15.9765 | 10896 | 0.9279 | 52.6029 |
85
+ | 0.0031 | 16.9751 | 11577 | 0.9783 | 52.0393 |
86
+ | 0.0012 | 17.9736 | 12258 | 1.0311 | 50.6622 |
87
+ | 0.0004 | 18.9721 | 12939 | 1.0574 | 50.7566 |
88
+ | 0.0001 | 19.9707 | 13620 | 1.0814 | 50.6511 |
89
 
90
 
91
  ### Framework versions