File size: 4,030 Bytes
4f1efb4
7bc4bab
 
4f1efb4
 
 
7bc4bab
 
 
4f1efb4
 
 
 
 
 
 
 
 
 
 
 
7bc4bab
4f1efb4
 
 
7bc4bab
4f1efb4
 
 
7bc4bab
4f1efb4
 
 
 
 
 
 
7bc4bab
4f1efb4
7bc4bab
 
4f1efb4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
language:
- hi
license: cc-by-nc-4.0
base_model: facebook/mms-1b-all
tags:
- automatic-speech-recognition
- mozilla-foundation/common_voice_16_0
- mms
- generated_from_trainer
datasets:
- common_voice_16_0
metrics:
- wer
model-index:
- name: wav2vec2-common_voice-hi-mms-demo
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: MOZILLA-FOUNDATION/COMMON_VOICE_16_0 - HI
      type: common_voice_16_0
      config: hi
      split: test
      args: 'Config: hi, Training split: train+validation, Eval split: test'
    metrics:
    - name: Wer
      type: wer
      value: 0.2516432655283731
---

<!-- 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. -->

# wav2vec2-common_voice-hi-mms-demo

This model is a fine-tuned version of [facebook/mms-1b-all](https://huggingface.co/facebook/mms-1b-all) on the MOZILLA-FOUNDATION/COMMON_VOICE_16_0 - HI dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2672
- Wer: 0.2516

## 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: 0.001
- 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: 100
- num_epochs: 4.0
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Wer    |
|:-------------:|:-----:|:----:|:---------------:|:------:|
| No log        | 0.11  | 100  | 0.4487          | 0.3565 |
| No log        | 0.23  | 200  | 0.3544          | 0.3317 |
| No log        | 0.34  | 300  | 0.3693          | 0.3088 |
| No log        | 0.45  | 400  | 0.3404          | 0.3040 |
| 1.5084        | 0.56  | 500  | 0.3346          | 0.2995 |
| 1.5084        | 0.68  | 600  | 0.3411          | 0.2936 |
| 1.5084        | 0.79  | 700  | 0.3175          | 0.2887 |
| 1.5084        | 0.9   | 800  | 0.3159          | 0.2898 |
| 1.5084        | 1.02  | 900  | 0.3139          | 0.3045 |
| 0.3485        | 1.13  | 1000 | 0.3067          | 0.2958 |
| 0.3485        | 1.24  | 1100 | 0.2969          | 0.2767 |
| 0.3485        | 1.35  | 1200 | 0.2916          | 0.2714 |
| 0.3485        | 1.47  | 1300 | 0.2893          | 0.2663 |
| 0.3485        | 1.58  | 1400 | 0.3183          | 0.2985 |
| 0.3152        | 1.69  | 1500 | 0.2961          | 0.2688 |
| 0.3152        | 1.81  | 1600 | 0.2848          | 0.2665 |
| 0.3152        | 1.92  | 1700 | 0.2844          | 0.2656 |
| 0.3152        | 2.03  | 1800 | 0.2855          | 0.2707 |
| 0.3152        | 2.14  | 1900 | 0.2887          | 0.2686 |
| 0.3058        | 2.26  | 2000 | 0.2858          | 0.2657 |
| 0.3058        | 2.37  | 2100 | 0.2814          | 0.2629 |
| 0.3058        | 2.48  | 2200 | 0.2809          | 0.2633 |
| 0.3058        | 2.6   | 2300 | 0.2779          | 0.2613 |
| 0.3058        | 2.71  | 2400 | 0.2745          | 0.2581 |
| 0.2861        | 2.82  | 2500 | 0.2769          | 0.2618 |
| 0.2861        | 2.93  | 2600 | 0.2742          | 0.2576 |
| 0.2861        | 3.05  | 2700 | 0.2730          | 0.2575 |
| 0.2861        | 3.16  | 2800 | 0.2727          | 0.2564 |
| 0.2861        | 3.27  | 2900 | 0.2726          | 0.2563 |
| 0.2839        | 3.39  | 3000 | 0.2713          | 0.2576 |
| 0.2839        | 3.5   | 3100 | 0.2690          | 0.2537 |
| 0.2839        | 3.61  | 3200 | 0.2706          | 0.2540 |
| 0.2839        | 3.72  | 3300 | 0.2687          | 0.2542 |
| 0.2839        | 3.84  | 3400 | 0.2671          | 0.2521 |
| 0.2706        | 3.95  | 3500 | 0.2673          | 0.2522 |


### Framework versions

- Transformers 4.38.0.dev0
- Pytorch 2.1.2+cu121
- Datasets 2.16.1
- Tokenizers 0.15.1