samantharhay
commited on
Commit
•
6a7698b
1
Parent(s):
5945257
update model card README.md
Browse files
README.md
ADDED
@@ -0,0 +1,87 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
tags:
|
4 |
+
- generated_from_trainer
|
5 |
+
model-index:
|
6 |
+
name: wav2vec2-base-libir-zenodo
|
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 |
+
# wav2vec2-base-libir-zenodo
|
13 |
+
|
14 |
+
This model is a fine-tuned version of [facebook/wav2vec2-base-960h](https://huggingface.co/facebook/wav2vec2-base-960h) on an unknown dataset.
|
15 |
+
It achieves the following results on the evaluation set:
|
16 |
+
- Loss: 1.4238
|
17 |
+
- Wer: 0.4336
|
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: 1e-05
|
37 |
+
- train_batch_size: 2
|
38 |
+
- eval_batch_size: 8
|
39 |
+
- seed: 42
|
40 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
41 |
+
- lr_scheduler_type: linear
|
42 |
+
- lr_scheduler_warmup_steps: 1000
|
43 |
+
- num_epochs: 30
|
44 |
+
- mixed_precision_training: Native AMP
|
45 |
+
|
46 |
+
### Training results
|
47 |
+
|
48 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
49 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
50 |
+
| 3.053 | 1.0 | 31 | 3.1494 | 0.7345 |
|
51 |
+
| 2.9742 | 2.0 | 62 | 3.0527 | 0.7257 |
|
52 |
+
| 2.9139 | 3.0 | 93 | 2.8808 | 0.7257 |
|
53 |
+
| 2.6586 | 4.0 | 124 | 2.6648 | 0.6726 |
|
54 |
+
| 2.7117 | 5.0 | 155 | 2.4695 | 0.6372 |
|
55 |
+
| 2.5173 | 6.0 | 186 | 2.3087 | 0.6195 |
|
56 |
+
| 2.3665 | 7.0 | 217 | 2.2745 | 0.6018 |
|
57 |
+
| 2.1276 | 8.0 | 248 | 2.2180 | 0.5752 |
|
58 |
+
| 2.1624 | 9.0 | 279 | 2.1311 | 0.5752 |
|
59 |
+
| 2.0312 | 10.0 | 310 | 2.0358 | 0.5575 |
|
60 |
+
| 2.0652 | 11.0 | 341 | 1.9146 | 0.5310 |
|
61 |
+
| 1.7963 | 12.0 | 372 | 1.8346 | 0.5221 |
|
62 |
+
| 1.6811 | 13.0 | 403 | 1.8351 | 0.5398 |
|
63 |
+
| 1.5929 | 14.0 | 434 | 1.8256 | 0.4779 |
|
64 |
+
| 1.6644 | 15.0 | 465 | 1.7572 | 0.4779 |
|
65 |
+
| 1.5411 | 16.0 | 496 | 1.8740 | 0.4779 |
|
66 |
+
| 1.4027 | 17.0 | 527 | 1.5143 | 0.4779 |
|
67 |
+
| 1.2634 | 18.0 | 558 | 1.3864 | 0.4867 |
|
68 |
+
| 1.1053 | 19.0 | 589 | 1.3192 | 0.4425 |
|
69 |
+
| 1.0517 | 20.0 | 620 | 1.4705 | 0.4602 |
|
70 |
+
| 1.1033 | 21.0 | 651 | 1.6006 | 0.4956 |
|
71 |
+
| 0.9992 | 22.0 | 682 | 1.4748 | 0.5044 |
|
72 |
+
| 0.8987 | 23.0 | 713 | 1.3544 | 0.4867 |
|
73 |
+
| 0.9656 | 24.0 | 744 | 1.2673 | 0.4336 |
|
74 |
+
| 0.952 | 25.0 | 775 | 1.3955 | 0.4071 |
|
75 |
+
| 0.8507 | 26.0 | 806 | 1.3520 | 0.4425 |
|
76 |
+
| 0.8269 | 27.0 | 837 | 1.8992 | 0.4336 |
|
77 |
+
| 0.7255 | 28.0 | 868 | 1.9850 | 0.4425 |
|
78 |
+
| 0.8269 | 29.0 | 899 | 3.0089 | 0.4425 |
|
79 |
+
| 0.6178 | 30.0 | 930 | 1.4238 | 0.4336 |
|
80 |
+
|
81 |
+
|
82 |
+
### Framework versions
|
83 |
+
|
84 |
+
- Transformers 4.11.3
|
85 |
+
- Pytorch 1.10.0+cu111
|
86 |
+
- Datasets 1.13.3
|
87 |
+
- Tokenizers 0.10.3
|