File size: 3,707 Bytes
36642a6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7ca79df
 
36642a6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7ca79df
36642a6
 
 
 
 
4c7e894
36642a6
 
 
7ca79df
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
36642a6
 
 
 
 
 
 
 
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
---
license: apache-2.0
tags:
- generated_from_trainer
model-index:
- name: kids_phoneme_sm_model
  results: []
---

<!-- 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. -->

# kids_phoneme_sm_model

This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 1.0236
- Cer: 0.4987

## 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: 0.0004
- train_batch_size: 2
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 30

### Training results

| Training Loss | Epoch | Step  | Validation Loss | Cer    |
|:-------------:|:-----:|:-----:|:---------------:|:------:|
| 2.9442        | 0.74  | 500   | 4.9380          | 1.0    |
| 2.8667        | 1.48  | 1000  | 3.3662          | 1.0    |
| 2.8316        | 2.22  | 1500  | 4.4512          | 1.0    |
| 2.8011        | 2.96  | 2000  | 4.1959          | 1.0    |
| 2.7941        | 3.7   | 2500  | 3.2025          | 1.0    |
| 2.5078        | 4.44  | 3000  | 2.1864          | 1.0    |
| 1.8915        | 5.19  | 3500  | 1.6942          | 0.9979 |
| 1.5858        | 5.93  | 4000  | 1.4032          | 0.9707 |
| 1.3097        | 6.67  | 4500  | 1.1950          | 0.9264 |
| 1.134         | 7.41  | 5000  | 1.0407          | 0.8629 |
| 1.0054        | 8.15  | 5500  | 0.9647          | 0.8089 |
| 0.9141        | 8.89  | 6000  | 0.8932          | 0.7713 |
| 0.7902        | 9.63  | 6500  | 0.8355          | 0.7111 |
| 0.7334        | 10.37 | 7000  | 0.8343          | 0.6986 |
| 0.7315        | 11.11 | 7500  | 0.7893          | 0.6806 |
| 0.6443        | 11.85 | 8000  | 0.7572          | 0.6572 |
| 0.5798        | 12.59 | 8500  | 0.7501          | 0.6522 |
| 0.5845        | 13.33 | 9000  | 0.7337          | 0.6166 |
| 0.5366        | 14.07 | 9500  | 0.8090          | 0.6066 |
| 0.5046        | 14.81 | 10000 | 0.7767          | 0.5924 |
| 0.4569        | 15.56 | 10500 | 0.7593          | 0.6074 |
| 0.425         | 16.3  | 11000 | 0.7844          | 0.5832 |
| 0.4421        | 17.04 | 11500 | 0.7757          | 0.5836 |
| 0.3839        | 17.78 | 12000 | 0.8051          | 0.5782 |
| 0.3483        | 18.52 | 12500 | 0.7850          | 0.5715 |
| 0.3499        | 19.26 | 13000 | 0.8381          | 0.5531 |
| 0.3124        | 20.0  | 13500 | 0.7887          | 0.5527 |
| 0.2715        | 20.74 | 14000 | 0.8220          | 0.5581 |
| 0.2823        | 21.48 | 14500 | 0.8489          | 0.5426 |
| 0.257         | 22.22 | 15000 | 0.8818          | 0.5322 |
| 0.2529        | 22.96 | 15500 | 0.9106          | 0.5259 |
| 0.2219        | 23.7  | 16000 | 0.9197          | 0.5184 |
| 0.2003        | 24.44 | 16500 | 0.9177          | 0.5226 |
| 0.202         | 25.19 | 17000 | 0.9586          | 0.5167 |
| 0.1753        | 25.93 | 17500 | 0.9617          | 0.5159 |
| 0.1781        | 26.67 | 18000 | 0.9664          | 0.5063 |
| 0.1619        | 27.41 | 18500 | 1.0026          | 0.5100 |
| 0.16          | 28.15 | 19000 | 1.0088          | 0.4987 |
| 0.1471        | 28.89 | 19500 | 1.0207          | 0.5033 |
| 0.1459        | 29.63 | 20000 | 1.0236          | 0.4987 |


### Framework versions

- Transformers 4.30.1
- Pytorch 2.0.0
- Datasets 2.12.0
- Tokenizers 0.13.3