infinitejoy
commited on
Commit
•
89a6c23
1
Parent(s):
4adcb46
update model card README.md
Browse files
README.md
ADDED
@@ -0,0 +1,71 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- hy-AM
|
4 |
+
license: apache-2.0
|
5 |
+
tags:
|
6 |
+
- automatic-speech-recognition
|
7 |
+
- mozilla-foundation/common_voice_7_0
|
8 |
+
- generated_from_trainer
|
9 |
+
datasets:
|
10 |
+
- common_voice
|
11 |
+
model-index:
|
12 |
+
- name: wav2vec2-large-xls-r-300m-armenian
|
13 |
+
results: []
|
14 |
+
---
|
15 |
+
|
16 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
17 |
+
should probably proofread and complete it, then remove this comment. -->
|
18 |
+
|
19 |
+
# wav2vec2-large-xls-r-300m-armenian
|
20 |
+
|
21 |
+
This model is a fine-tuned version of [facebook/wav2vec2-xls-r-300m](https://huggingface.co/facebook/wav2vec2-xls-r-300m) on the MOZILLA-FOUNDATION/COMMON_VOICE_7_0 - HY-AM dataset.
|
22 |
+
It achieves the following results on the evaluation set:
|
23 |
+
- Loss: 0.9669
|
24 |
+
- Wer: 0.6942
|
25 |
+
|
26 |
+
## Model description
|
27 |
+
|
28 |
+
More information needed
|
29 |
+
|
30 |
+
## Intended uses & limitations
|
31 |
+
|
32 |
+
More information needed
|
33 |
+
|
34 |
+
## Training and evaluation data
|
35 |
+
|
36 |
+
More information needed
|
37 |
+
|
38 |
+
## Training procedure
|
39 |
+
|
40 |
+
### Training hyperparameters
|
41 |
+
|
42 |
+
The following hyperparameters were used during training:
|
43 |
+
- learning_rate: 0.0003
|
44 |
+
- train_batch_size: 32
|
45 |
+
- eval_batch_size: 32
|
46 |
+
- seed: 42
|
47 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
48 |
+
- lr_scheduler_type: linear
|
49 |
+
- lr_scheduler_warmup_steps: 500
|
50 |
+
- num_epochs: 200.0
|
51 |
+
- mixed_precision_training: Native AMP
|
52 |
+
|
53 |
+
### Training results
|
54 |
+
|
55 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
56 |
+
|:-------------:|:------:|:----:|:---------------:|:------:|
|
57 |
+
| 1.7294 | 27.78 | 500 | 0.8540 | 0.9944 |
|
58 |
+
| 0.8863 | 55.56 | 1000 | 0.7282 | 0.7312 |
|
59 |
+
| 0.5789 | 83.33 | 1500 | 0.8178 | 0.8102 |
|
60 |
+
| 0.3899 | 111.11 | 2000 | 0.8034 | 0.7701 |
|
61 |
+
| 0.2869 | 138.89 | 2500 | 0.9061 | 0.6999 |
|
62 |
+
| 0.1934 | 166.67 | 3000 | 0.9400 | 0.7105 |
|
63 |
+
| 0.1551 | 194.44 | 3500 | 0.9667 | 0.6955 |
|
64 |
+
|
65 |
+
|
66 |
+
### Framework versions
|
67 |
+
|
68 |
+
- Transformers 4.16.0.dev0
|
69 |
+
- Pytorch 1.10.1+cu102
|
70 |
+
- Datasets 1.17.1.dev0
|
71 |
+
- Tokenizers 0.11.0
|