File size: 3,625 Bytes
180d66b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
---
license: apache-2.0
base_model: facebook/wav2vec2-xls-r-300m
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: wav2vec2_XLSR_darija_maroc
  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_XLSR_darija_maroc

This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2860
- Wer: 0.3290

## 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

### Training results

| Training Loss | Epoch | Step  | Validation Loss | Wer    |
|:-------------:|:-----:|:-----:|:---------------:|:------:|
| 4.9354        | 0.83  | 400   | 2.0492          | 1.0371 |
| 0.8236        | 1.66  | 800   | 0.4434          | 0.5832 |
| 0.4821        | 2.49  | 1200  | 0.3597          | 0.5114 |
| 0.3823        | 3.32  | 1600  | 0.3265          | 0.4758 |
| 0.3231        | 4.15  | 2000  | 0.3149          | 0.4526 |
| 0.2854        | 4.97  | 2400  | 0.2797          | 0.4237 |
| 0.2529        | 5.8   | 2800  | 0.3027          | 0.4415 |
| 0.2493        | 6.63  | 3200  | 0.2926          | 0.4264 |
| 0.2138        | 7.46  | 3600  | 0.2857          | 0.4169 |
| 0.2067        | 8.29  | 4000  | 0.2743          | 0.4099 |
| 0.1898        | 9.12  | 4400  | 0.2798          | 0.3993 |
| 0.1755        | 9.95  | 4800  | 0.2800          | 0.3913 |
| 0.1603        | 10.78 | 5200  | 0.2709          | 0.3860 |
| 0.1608        | 11.61 | 5600  | 0.2716          | 0.3872 |
| 0.1462        | 12.44 | 6000  | 0.2697          | 0.3825 |
| 0.137         | 13.26 | 6400  | 0.2855          | 0.3819 |
| 0.1326        | 14.09 | 6800  | 0.2860          | 0.3733 |
| 0.123         | 14.92 | 7200  | 0.2677          | 0.3813 |
| 0.1168        | 15.75 | 7600  | 0.2780          | 0.3740 |
| 0.1113        | 16.58 | 8000  | 0.2926          | 0.3719 |
| 0.1057        | 17.41 | 8400  | 0.2927          | 0.3704 |
| 0.0996        | 18.24 | 8800  | 0.2825          | 0.3602 |
| 0.0967        | 19.07 | 9200  | 0.2983          | 0.3641 |
| 0.0925        | 19.9  | 9600  | 0.2843          | 0.3576 |
| 0.0894        | 20.73 | 10000 | 0.2726          | 0.3668 |
| 0.0836        | 21.55 | 10400 | 0.2829          | 0.3560 |
| 0.0789        | 22.38 | 10800 | 0.2806          | 0.3508 |
| 0.0778        | 23.21 | 11200 | 0.2849          | 0.3540 |
| 0.0742        | 24.04 | 11600 | 0.2770          | 0.3436 |
| 0.0679        | 24.87 | 12000 | 0.2850          | 0.3425 |
| 0.063         | 25.7  | 12400 | 0.2846          | 0.3366 |
| 0.0593        | 26.53 | 12800 | 0.2811          | 0.3351 |
| 0.0586        | 27.36 | 13200 | 0.2863          | 0.3322 |
| 0.0555        | 28.19 | 13600 | 0.2819          | 0.3311 |
| 0.053         | 29.02 | 14000 | 0.2874          | 0.3301 |
| 0.0498        | 29.84 | 14400 | 0.2860          | 0.3290 |


### Framework versions

- Transformers 4.34.1
- Pytorch 2.1.0+cu118
- Datasets 2.14.6
- Tokenizers 0.14.1