Hugging Face's logo

hamidov02
/
wav2vec2-large-xls-hun-53h-colab

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
wav2vec2
generated_from_trainer
Inference Endpoints
Model card Files Metrics Community
hamidov02 commited on
Commit
66afc1b
•
1 Parent(s): 9c14545

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +96 -0
README.md ADDED
@@ -0,0 +1,96 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ tags:
4
+ - generated_from_trainer
5
+ datasets:
6
+ - common_voice
7
+ model-index:
8
+ - name: wav2vec2-large-xls-hun-53h-colab
9
+ results: []
10
+ ---
11
+
12
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
13
+ should probably proofread and complete it, then remove this comment. -->
14
+
15
+ # wav2vec2-large-xls-hun-53h-colab
16
+
17
+ This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on the common_voice dataset.
18
+ It achieves the following results on the evaluation set:
19
+ - Loss: 0.6027
20
+ - Wer: 0.4618
21
+
22
+ ## Model description
23
+
24
+ More information needed
25
+
26
+ ## Intended uses & limitations
27
+
28
+ More information needed
29
+
30
+ ## Training and evaluation data
31
+
32
+ More information needed
33
+
34
+ ## Training procedure
35
+
36
+ ### Training hyperparameters
37
+
38
+ The following hyperparameters were used during training:
39
+ - learning_rate: 0.0003
40
+ - train_batch_size: 16
41
+ - eval_batch_size: 8
42
+ - seed: 42
43
+ - gradient_accumulation_steps: 2
44
+ - total_train_batch_size: 32
45
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
46
+ - lr_scheduler_type: linear
47
+ - lr_scheduler_warmup_steps: 500
48
+ - num_epochs: 23
49
+ - mixed_precision_training: Native AMP
50
+
51
+ ### Training results
52
+
53
+ | Training Loss | Epoch | Step | Validation Loss | Wer |
54
+ |:-------------:|:-----:|:----:|:---------------:|:------:|
55
+ | 13.4225 | 0.67 | 100 | 3.7750 | 1.0 |
56
+ | 3.4121 | 1.34 | 200 | 3.3166 | 1.0 |
57
+ | 3.2263 | 2.01 | 300 | 3.1403 | 1.0 |
58
+ | 3.0038 | 2.68 | 400 | 2.2474 | 0.9990 |
59
+ | 1.2243 | 3.35 | 500 | 0.8174 | 0.7666 |
60
+ | 0.6368 | 4.03 | 600 | 0.6306 | 0.6633 |
61
+ | 0.4426 | 4.7 | 700 | 0.6151 | 0.6648 |
62
+ | 0.3821 | 5.37 | 800 | 0.5765 | 0.6138 |
63
+ | 0.3337 | 6.04 | 900 | 0.5522 | 0.5785 |
64
+ | 0.2832 | 6.71 | 1000 | 0.5822 | 0.5691 |
65
+ | 0.2485 | 7.38 | 1100 | 0.5626 | 0.5449 |
66
+ | 0.2335 | 8.05 | 1200 | 0.5866 | 0.5662 |
67
+ | 0.2031 | 8.72 | 1300 | 0.5574 | 0.5420 |
68
+ | 0.1925 | 9.39 | 1400 | 0.5572 | 0.5297 |
69
+ | 0.1793 | 10.07 | 1500 | 0.5878 | 0.5185 |
70
+ | 0.1652 | 10.74 | 1600 | 0.6173 | 0.5243 |
71
+ | 0.1663 | 11.41 | 1700 | 0.5807 | 0.5133 |
72
+ | 0.1544 | 12.08 | 1800 | 0.5979 | 0.5154 |
73
+ | 0.148 | 12.75 | 1900 | 0.5545 | 0.4986 |
74
+ | 0.138 | 13.42 | 2000 | 0.5798 | 0.4947 |
75
+ | 0.1353 | 14.09 | 2100 | 0.5670 | 0.5028 |
76
+ | 0.1283 | 14.76 | 2200 | 0.5862 | 0.4957 |
77
+ | 0.1271 | 15.43 | 2300 | 0.6009 | 0.4961 |
78
+ | 0.1108 | 16.11 | 2400 | 0.5873 | 0.4975 |
79
+ | 0.1182 | 16.78 | 2500 | 0.6013 | 0.4893 |
80
+ | 0.103 | 17.45 | 2600 | 0.6165 | 0.4898 |
81
+ | 0.1084 | 18.12 | 2700 | 0.6186 | 0.4838 |
82
+ | 0.1014 | 18.79 | 2800 | 0.6122 | 0.4767 |
83
+ | 0.1009 | 19.46 | 2900 | 0.5981 | 0.4793 |
84
+ | 0.1004 | 20.13 | 3000 | 0.6034 | 0.4770 |
85
+ | 0.0922 | 20.8 | 3100 | 0.6127 | 0.4663 |
86
+ | 0.09 | 21.47 | 3200 | 0.5967 | 0.4672 |
87
+ | 0.0893 | 22.15 | 3300 | 0.6051 | 0.4611 |
88
+ | 0.0817 | 22.82 | 3400 | 0.6027 | 0.4618 |
89
+
90
+
91
+ ### Framework versions
92
+
93
+ - Transformers 4.11.3
94
+ - Pytorch 1.10.0+cu113
95
+ - Datasets 1.18.3
96
+ - Tokenizers 0.10.3