File size: 3,350 Bytes
c125909
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: apache-2.0
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: wav2vec2-custom-colab
  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. -->

# wav2vec2-custom-colab

This model is a fine-tuned version of [facebook/wav2vec2-base](https://huggingface.co/facebook/wav2vec2-base) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.7785
- Wer: 0.3534

## 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: 0.0001
- train_batch_size: 1
- eval_batch_size: 1
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 10
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step  | Validation Loss | Wer    |
|:-------------:|:-----:|:-----:|:---------------:|:------:|
| 0.4783        | 0.3   | 500   | 0.7199          | 0.5564 |
| 0.4833        | 0.61  | 1000  | 0.8089          | 0.6181 |
| 0.5733        | 0.91  | 1500  | 0.7617          | 0.5530 |
| 0.4641        | 1.21  | 2000  | 0.7937          | 0.5731 |
| 0.4167        | 1.52  | 2500  | 0.7993          | 0.5102 |
| 0.3713        | 1.82  | 3000  | 0.7541          | 0.5437 |
| 0.3395        | 2.12  | 3500  | 0.7658          | 0.5148 |
| 0.2814        | 2.42  | 4000  | 0.7569          | 0.4783 |
| 0.2698        | 2.73  | 4500  | 0.8126          | 0.5174 |
| 0.2767        | 3.03  | 5000  | 0.7838          | 0.4676 |
| 0.2249        | 3.33  | 5500  | 0.8769          | 0.4743 |
| 0.2452        | 3.64  | 6000  | 0.8586          | 0.4778 |
| 0.1828        | 3.94  | 6500  | 0.7695          | 0.4528 |
| 0.1901        | 4.24  | 7000  | 0.7800          | 0.5021 |
| 0.2062        | 4.55  | 7500  | 0.8107          | 0.4567 |
| 0.1614        | 4.85  | 8000  | 0.7941          | 0.4094 |
| 0.1327        | 5.15  | 8500  | 0.7900          | 0.4241 |
| 0.1405        | 5.45  | 9000  | 0.8017          | 0.3992 |
| 0.1219        | 5.76  | 9500  | 0.8099          | 0.4043 |
| 0.1406        | 6.06  | 10000 | 0.8731          | 0.3913 |
| 0.0806        | 6.36  | 10500 | 0.8387          | 0.3868 |
| 0.1039        | 6.67  | 11000 | 0.8105          | 0.3905 |
| 0.0967        | 6.97  | 11500 | 0.7291          | 0.3728 |
| 0.0846        | 7.27  | 12000 | 0.8128          | 0.4201 |
| 0.0722        | 7.58  | 12500 | 0.8204          | 0.3751 |
| 0.0785        | 7.88  | 13000 | 0.7692          | 0.3760 |
| 0.0647        | 8.18  | 13500 | 0.8294          | 0.3752 |
| 0.0523        | 8.48  | 14000 | 0.7646          | 0.3763 |
| 0.0623        | 8.79  | 14500 | 0.7773          | 0.3572 |
| 0.0477        | 9.09  | 15000 | 0.7379          | 0.3635 |
| 0.064         | 9.39  | 15500 | 0.7544          | 0.3538 |
| 0.0321        | 9.7   | 16000 | 0.8118          | 0.3557 |
| 0.0541        | 10.0  | 16500 | 0.7785          | 0.3534 |


### Framework versions

- Transformers 4.26.0
- Pytorch 1.10.0
- Datasets 2.9.0
- Tokenizers 0.13.2