File size: 4,817 Bytes
11a32c4
634bc85
11a32c4
 
634bc85
 
 
 
 
 
 
 
11a32c4
 
634bc85
11a32c4
 
634bc85
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
11a32c4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
---
language: mn
license: apache-2.0
tags:
- whisper-event
- hf-asr-leaderboard
- generated_from_multiple_datasets
datasets:
- mozilla-foundation/common_voice_11_0
- google/fleurs
- bayartsogt/ulaanbal-v0
- bayartsogt/youtube-mongolian-v1
metrics:
- wer
- cer
model-index:
- name: whisper-medium-mn-10
  results:
  - task:
      type: automatic-speech-recognition
      name: Automatic Speech Recognition
    dataset:
      name: Common Voice 11.0
      type: mozilla-foundation/common_voice_11_0
      config: mn
      split: test
    metrics:
    - type: wer
      value: 21.258466244264802
      name: Wer
    - type: cer
      value: 6.875610660018193
      name: Cer
---

<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->

# whisper-medium-mn-10

This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2103
- Wer: 21.2585
- Cer: 6.8756

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- training_steps: 40000
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step  | Cer     | Validation Loss | Wer     |
|:-------------:|:-----:|:-----:|:-------:|:---------------:|:-------:|
| 0.4197        | 0.09  | 1000  | 19.0947 | 0.4462          | 53.9600 |
| 0.3288        | 0.17  | 2000  | 14.8016 | 0.3468          | 44.2102 |
| 0.2737        | 0.26  | 3000  | 12.3471 | 0.3020          | 36.1700 |
| 0.2558        | 0.35  | 4000  | 11.7171 | 0.2824          | 34.1709 |
| 0.2406        | 0.43  | 5000  | 10.3551 | 0.2594          | 31.1230 |
| 0.218         | 0.52  | 6000  | 9.7815  | 0.2452          | 29.6865 |
| 0.2253        | 0.61  | 7000  | 9.6712  | 0.2344          | 29.2932 |
| 0.2071        | 0.69  | 8000  | 9.4261  | 0.2283          | 28.5067 |
| 0.2051        | 0.78  | 9000  | 9.0656  | 0.2224          | 27.4033 |
| 0.2064        | 0.87  | 10000 | 8.7851  | 0.2138          | 26.7206 |
| 0.193         | 0.95  | 11000 | 8.5021  | 0.2089          | 25.5790 |
| 0.1577        | 1.04  | 12000 | 8.2873  | 0.2072          | 25.6118 |
| 0.1397        | 1.13  | 13000 | 8.2368  | 0.2046          | 25.1147 |
| 0.1526        | 1.21  | 14000 | 8.7615  | 0.2065          | 26.4638 |
| 0.1497        | 1.3   | 15000 | 0.2004  | 24.4866         | 7.9588  |
| 0.1569        | 1.39  | 16000 | 0.1990  | 24.2244         | 7.9554  |
| 0.1416        | 1.47  | 17000 | 0.2001  | 24.2298         | 7.8754  |
| 0.1371        | 1.56  | 18000 | 0.1932  | 23.6072         | 7.8072  |
| 0.1379        | 1.65  | 19000 | 0.1916  | 23.1320         | 7.5452  |
| 0.1305        | 1.73  | 20000 | 0.1880  | 23.1101         | 7.4290  |
| 0.1395        | 1.82  | 21000 | 0.1877  | 22.9845         | 7.4635  |
| 0.1418        | 1.91  | 22000 | 0.1862  | 22.9080         | 7.5907  |
| 0.1432        | 1.99  | 23000 | 0.1847  | 22.7114         | 7.4290  |
| 0.0965        | 2.08  | 24000 | 0.1931  | 21.7391         | 7.0399  |
| 0.0723        | 2.17  | 25000 | 0.1961  | 22.3236         | 7.2698  |
| 0.0773        | 2.25  | 26000 | 0.1977  | 22.0505         | 7.0752  |
| 0.0862        | 2.34  | 27000 | 0.1959  | 21.9522         | 7.0820  |
| 0.0739        | 2.43  | 28000 | 0.1982  | 21.7719         | 7.1494  |
| 0.0843        | 2.51  | 29000 | 0.1963  | 21.8921         | 7.1241  |
| 0.0734        | 2.6   | 30000 | 0.1980  | 21.7883         | 7.1317  |
| 0.0785        | 2.69  | 31000 | 0.1955  | 21.8757         | 7.1948  |
| 0.0691        | 2.77  | 32000 | 0.1978  | 21.7446         | 7.0938  |
| 0.0834        | 2.86  | 33000 | 0.1953  | 21.3240         | 7.0121  |
| 0.0675        | 2.95  | 34000 | 0.1958  | 21.7719         | 7.0769  |
| 0.042         | 3.03  | 35000 | 0.2053  | 21.3404         | 6.9624  |
| 0.0474        | 3.12  | 36000 | 0.2097  | 21.5534         | 7.0306  |
| 0.0428        | 3.21  | 37000 | 0.2107  | 21.3185         | 6.9809  |
| 0.0343        | 3.29  | 38000 | 0.2111  | 21.3896         | 6.9514  |
| 0.0378        | 3.38  | 39000 | 0.2103  | 21.2585         | 6.8756  |
| 0.0361        | 3.47  | 40000 | 0.2106  | 21.3677         | 6.9009  |


### Framework versions

- Transformers 4.26.0.dev0
- Pytorch 1.13.0+cu117
- Datasets 2.7.1.dev0
- Tokenizers 0.13.2