update model card README.md
Browse files
README.md
ADDED
@@ -0,0 +1,69 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
tags:
|
3 |
+
- generated_from_trainer
|
4 |
+
model-index:
|
5 |
+
- name: ai-light-dance_singing_ft_wav2vec2-large-xlsr-53-5gram-v4
|
6 |
+
results: []
|
7 |
+
---
|
8 |
+
|
9 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
10 |
+
should probably proofread and complete it, then remove this comment. -->
|
11 |
+
|
12 |
+
# ai-light-dance_singing_ft_wav2vec2-large-xlsr-53-5gram-v4
|
13 |
+
|
14 |
+
This model is a fine-tuned version of [gary109/ai-light-dance_singing_ft_wav2vec2-large-xlsr-53-5gram-v2](https://huggingface.co/gary109/ai-light-dance_singing_ft_wav2vec2-large-xlsr-53-5gram-v2) on the None dataset.
|
15 |
+
It achieves the following results on the evaluation set:
|
16 |
+
- Loss: 0.4328
|
17 |
+
- Wer: 0.1575
|
18 |
+
|
19 |
+
## Model description
|
20 |
+
|
21 |
+
More information needed
|
22 |
+
|
23 |
+
## Intended uses & limitations
|
24 |
+
|
25 |
+
More information needed
|
26 |
+
|
27 |
+
## Training and evaluation data
|
28 |
+
|
29 |
+
More information needed
|
30 |
+
|
31 |
+
## Training procedure
|
32 |
+
|
33 |
+
### Training hyperparameters
|
34 |
+
|
35 |
+
The following hyperparameters were used during training:
|
36 |
+
- learning_rate: 4e-06
|
37 |
+
- train_batch_size: 8
|
38 |
+
- eval_batch_size: 8
|
39 |
+
- seed: 42
|
40 |
+
- gradient_accumulation_steps: 16
|
41 |
+
- total_train_batch_size: 128
|
42 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
43 |
+
- lr_scheduler_type: linear
|
44 |
+
- lr_scheduler_warmup_steps: 500
|
45 |
+
- num_epochs: 10.0
|
46 |
+
- mixed_precision_training: Native AMP
|
47 |
+
|
48 |
+
### Training results
|
49 |
+
|
50 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
51 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
52 |
+
| 0.1335 | 1.0 | 138 | 0.4256 | 0.1605 |
|
53 |
+
| 0.1288 | 2.0 | 276 | 0.4234 | 0.1602 |
|
54 |
+
| 0.1278 | 3.0 | 414 | 0.4243 | 0.1597 |
|
55 |
+
| 0.1345 | 4.0 | 552 | 0.4231 | 0.1597 |
|
56 |
+
| 0.1344 | 5.0 | 690 | 0.4246 | 0.1597 |
|
57 |
+
| 0.1237 | 6.0 | 828 | 0.4279 | 0.1595 |
|
58 |
+
| 0.1109 | 7.0 | 966 | 0.4354 | 0.1573 |
|
59 |
+
| 0.1247 | 8.0 | 1104 | 0.4318 | 0.1570 |
|
60 |
+
| 0.1372 | 9.0 | 1242 | 0.4341 | 0.1573 |
|
61 |
+
| 0.1256 | 10.0 | 1380 | 0.4328 | 0.1575 |
|
62 |
+
|
63 |
+
|
64 |
+
### Framework versions
|
65 |
+
|
66 |
+
- Transformers 4.21.0.dev0
|
67 |
+
- Pytorch 1.9.1+cu102
|
68 |
+
- Datasets 2.3.3.dev0
|
69 |
+
- Tokenizers 0.12.1
|