File size: 5,279 Bytes
ce6f745
9f194b7
89cb67a
9f194b7
 
 
 
 
 
 
 
 
 
09fb966
89cb67a
9f194b7
 
 
 
 
 
 
89cb67a
 
 
ce6f745
 
9f194b7
 
 
89cb67a
9f194b7
 
 
 
89cb67a
 
 
9f194b7
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
89cb67a
9f194b7
 
 
 
 
 
89cb67a
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
9f194b7
 
 
 
 
 
 
 
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
---
license: apache-2.0
base_model: facebook/wav2vec2-xls-r-300m
tags:
- generated_from_trainer
datasets:
- common_voice_17_0
metrics:
- wer
model-index:
- name: xls-r-300-cv17-bulgarian-adap-ru
  results:
  - task:
      name: Automatic Speech Recognition
      type: automatic-speech-recognition
    dataset:
      name: common_voice_17_0
      type: common_voice_17_0
      config: bg
      split: validation
      args: bg
    metrics:
    - name: Wer
      type: wer
      value: 0.3023246994576965
---

<!-- 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. -->

[<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/badr-nlp/xlsr-continual-finetuning-polish/runs/hevbjmzy)
# xls-r-300-cv17-bulgarian-adap-ru

This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the common_voice_17_0 dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3977
- Wer: 0.3023
- Cer: 0.0722

## 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: 0.0003
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 30
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch   | Step | Validation Loss | Wer    | Cer    |
|:-------------:|:-------:|:----:|:---------------:|:------:|:------:|
| 3.1617        | 0.6579  | 100  | 3.1554          | 1.0    | 1.0    |
| 1.0032        | 1.3158  | 200  | 1.0726          | 0.8684 | 0.2419 |
| 0.5552        | 1.9737  | 300  | 0.4924          | 0.5297 | 0.1303 |
| 0.2763        | 2.6316  | 400  | 0.3795          | 0.4442 | 0.1043 |
| 0.2273        | 3.2895  | 500  | 0.3769          | 0.4222 | 0.1014 |
| 0.3216        | 3.9474  | 600  | 0.3611          | 0.3993 | 0.0971 |
| 0.1553        | 4.6053  | 700  | 0.3566          | 0.3927 | 0.0936 |
| 0.1414        | 5.2632  | 800  | 0.3676          | 0.3869 | 0.0923 |
| 0.1774        | 5.9211  | 900  | 0.3680          | 0.3758 | 0.0901 |
| 0.1256        | 6.5789  | 1000 | 0.3637          | 0.3775 | 0.0916 |
| 0.2416        | 7.2368  | 1100 | 0.3893          | 0.3963 | 0.0951 |
| 0.1213        | 7.8947  | 1200 | 0.3677          | 0.3596 | 0.0864 |
| 0.0911        | 8.5526  | 1300 | 0.3850          | 0.3739 | 0.0891 |
| 0.0859        | 9.2105  | 1400 | 0.3962          | 0.3658 | 0.0883 |
| 0.0998        | 9.8684  | 1500 | 0.3608          | 0.3530 | 0.0846 |
| 0.108         | 10.5263 | 1600 | 0.3932          | 0.3908 | 0.0920 |
| 0.0824        | 11.1842 | 1700 | 0.4147          | 0.3591 | 0.0870 |
| 0.0888        | 11.8421 | 1800 | 0.4040          | 0.3660 | 0.0878 |
| 0.0609        | 12.5    | 1900 | 0.4097          | 0.3542 | 0.0857 |
| 0.0692        | 13.1579 | 2000 | 0.4127          | 0.3639 | 0.0874 |
| 0.0513        | 13.8158 | 2100 | 0.4118          | 0.3560 | 0.0870 |
| 0.0752        | 14.4737 | 2200 | 0.4044          | 0.3591 | 0.0888 |
| 0.0833        | 15.1316 | 2300 | 0.3956          | 0.3374 | 0.0812 |
| 0.0826        | 15.7895 | 2400 | 0.3953          | 0.3356 | 0.0811 |
| 0.0934        | 16.4474 | 2500 | 0.4053          | 0.3394 | 0.0819 |
| 0.0562        | 17.1053 | 2600 | 0.4243          | 0.3534 | 0.0843 |
| 0.0661        | 17.7632 | 2700 | 0.4021          | 0.3340 | 0.0791 |
| 0.0496        | 18.4211 | 2800 | 0.4052          | 0.3387 | 0.0818 |
| 0.0599        | 19.0789 | 2900 | 0.4101          | 0.3385 | 0.0806 |
| 0.0446        | 19.7368 | 3000 | 0.3990          | 0.3362 | 0.0810 |
| 0.0482        | 20.3947 | 3100 | 0.4077          | 0.3274 | 0.0781 |
| 0.0309        | 21.0526 | 3200 | 0.4343          | 0.3397 | 0.0817 |
| 0.0757        | 21.7105 | 3300 | 0.4154          | 0.3252 | 0.0781 |
| 0.0377        | 22.3684 | 3400 | 0.4273          | 0.3206 | 0.0770 |
| 0.0282        | 23.0263 | 3500 | 0.3998          | 0.3159 | 0.0751 |
| 0.0676        | 23.6842 | 3600 | 0.3960          | 0.3111 | 0.0745 |
| 0.0673        | 24.3421 | 3700 | 0.3997          | 0.3100 | 0.0741 |
| 0.1793        | 25.0    | 3800 | 0.4065          | 0.3106 | 0.0738 |
| 0.0572        | 25.6579 | 3900 | 0.3951          | 0.3098 | 0.0739 |
| 0.0208        | 26.3158 | 4000 | 0.4097          | 0.3106 | 0.0740 |
| 0.0562        | 26.9737 | 4100 | 0.4016          | 0.3081 | 0.0734 |
| 0.0314        | 27.6316 | 4200 | 0.3939          | 0.3008 | 0.0715 |
| 0.0235        | 28.2895 | 4300 | 0.4008          | 0.3023 | 0.0720 |
| 0.0443        | 28.9474 | 4400 | 0.3963          | 0.3033 | 0.0724 |
| 0.027         | 29.6053 | 4500 | 0.3977          | 0.3023 | 0.0722 |


### Framework versions

- Transformers 4.42.0.dev0
- Pytorch 2.3.1+cu121
- Datasets 2.19.2
- Tokenizers 0.19.1