File size: 3,889 Bytes
3791502
64f4d2d
59ffc43
3791502
 
64f4d2d
 
3791502
c5979dd
 
 
59ffc43
3791502
0f265ca
3791502
c5979dd
 
59ffc43
 
c5979dd
 
 
 
 
 
59ffc43
 
 
 
 
 
 
e49da77
 
 
 
 
 
 
59ffc43
 
 
 
 
 
3791502
 
 
 
 
 
 
64f4d2d
3791502
64f4d2d
 
3791502
e49da77
3791502
9142074
3791502
e49da77
3791502
9142074
3791502
e49da77
3791502
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fe113a9
 
adb4785
fe113a9
 
 
 
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
---
language:
- myv
license: apache-2.0
tags:
- automatic-speech-recognition
- mozilla-foundation/common_voice_8_0
- generated_from_trainer
- myv
- robust-speech-event
- model_for_talk
- hf-asr-leaderboard
datasets:
- mozilla-foundation/common_voice_8_0
model-index:
- name: wav2vec2-xls-r-myv-a1
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: Common Voice 8
      type: mozilla-foundation/common_voice_8_0
      args: myv
    metrics:
    - name: Test WER
      type: wer
      value: 0.6514672686230248
    - name: Test CER
      type: cer
      value: 0.17226131905088124
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: Robust Speech Event - Dev Data
      type: speech-recognition-community-v2/dev_data
      args: vot
    metrics:
    - name: Test WER
      type: wer
      value: NA
    - name: Test CER
      type: cer
      value: NA
---

<!-- 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. -->

# 

This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the MOZILLA-FOUNDATION/COMMON_VOICE_8_0 - MYV dataset.
It achieves the following results on the evaluation set:
- Loss: 1.0356
- Wer: 0.6524

### Evaluation Commands

**1. To evaluate on mozilla-foundation/common_voice_8_0 with test split**

python eval.py  --model_id DrishtiSharma/wav2vec2-xls-r-myv-a1 --dataset mozilla-foundation/common_voice_8_0 --config myv --split test --log_outputs

**2. To evaluate on speech-recognition-community-v2/dev_data**

Erzya language not found in speech-recognition-community-v2/dev_data

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 0.0004
- train_batch_size: 16
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 800
- num_epochs: 200.0
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch  | Step  | Validation Loss | Wer    |
|:-------------:|:------:|:-----:|:---------------:|:------:|
| 5.649         | 9.62   | 500   | 3.0038          | 1.0    |
| 1.6272        | 19.23  | 1000  | 0.7362          | 0.7819 |
| 1.1354        | 28.85  | 1500  | 0.6410          | 0.7111 |
| 1.0424        | 38.46  | 2000  | 0.6907          | 0.7431 |
| 0.9293        | 48.08  | 2500  | 0.7249          | 0.7102 |
| 0.8246        | 57.69  | 3000  | 0.7422          | 0.6966 |
| 0.7837        | 67.31  | 3500  | 0.7413          | 0.6813 |
| 0.7147        | 76.92  | 4000  | 0.7873          | 0.6930 |
| 0.6276        | 86.54  | 4500  | 0.8038          | 0.6677 |
| 0.6041        | 96.15  | 5000  | 0.8240          | 0.6831 |
| 0.5336        | 105.77 | 5500  | 0.8748          | 0.6749 |
| 0.4705        | 115.38 | 6000  | 0.9006          | 0.6497 |
| 0.43          | 125.0  | 6500  | 0.8954          | 0.6551 |
| 0.3859        | 134.62 | 7000  | 0.9074          | 0.6614 |
| 0.3342        | 144.23 | 7500  | 0.9693          | 0.6560 |
| 0.3155        | 153.85 | 8000  | 1.0073          | 0.6691 |
| 0.2673        | 163.46 | 8500  | 1.0170          | 0.6632 |
| 0.2409        | 173.08 | 9000  | 1.0304          | 0.6709 |
| 0.2189        | 182.69 | 9500  | 0.9965          | 0.6546 |
| 0.1973        | 192.31 | 10000 | 1.0360          | 0.6551 |


### Framework versions

- Transformers 4.17.0.dev0
- Pytorch 1.10.2+cu102
- Datasets 1.18.2.dev0
- Tokenizers 0.11.0


### Evaluation Command

!python eval.py \
    --model_id DrishtiSharma/wav2vec2-large-xls-r-300m-myv-v1 \
    --dataset mozilla-foundation/common_voice_8_0 --config myv --split test --log_outputs