Commit
•
5c8ca39
1
Parent(s):
815e09c
Model save
Browse files
README.md
ADDED
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
tags:
|
3 |
+
- generated_from_trainer
|
4 |
+
datasets:
|
5 |
+
- timit_asr
|
6 |
+
metrics:
|
7 |
+
- wer
|
8 |
+
model-index:
|
9 |
+
- name: wav2vec2-base-gumbelVQ-timit-fine-tuned
|
10 |
+
results:
|
11 |
+
- task:
|
12 |
+
name: Automatic Speech Recognition
|
13 |
+
type: automatic-speech-recognition
|
14 |
+
dataset:
|
15 |
+
name: timit_asr
|
16 |
+
type: timit_asr
|
17 |
+
config: clean
|
18 |
+
split: test
|
19 |
+
args: clean
|
20 |
+
metrics:
|
21 |
+
- name: Wer
|
22 |
+
type: wer
|
23 |
+
value: 0.4901798635517883
|
24 |
+
---
|
25 |
+
|
26 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
27 |
+
should probably proofread and complete it, then remove this comment. -->
|
28 |
+
|
29 |
+
# wav2vec2-base-gumbelVQ-timit-fine-tuned
|
30 |
+
|
31 |
+
This model was trained from scratch on the timit_asr dataset.
|
32 |
+
It achieves the following results on the evaluation set:
|
33 |
+
- Loss: 0.7549
|
34 |
+
- Wer: 0.4902
|
35 |
+
|
36 |
+
## Model description
|
37 |
+
|
38 |
+
More information needed
|
39 |
+
|
40 |
+
## Intended uses & limitations
|
41 |
+
|
42 |
+
More information needed
|
43 |
+
|
44 |
+
## Training and evaluation data
|
45 |
+
|
46 |
+
More information needed
|
47 |
+
|
48 |
+
## Training procedure
|
49 |
+
|
50 |
+
### Training hyperparameters
|
51 |
+
|
52 |
+
The following hyperparameters were used during training:
|
53 |
+
- learning_rate: 0.0001
|
54 |
+
- train_batch_size: 32
|
55 |
+
- eval_batch_size: 1
|
56 |
+
- seed: 42
|
57 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
58 |
+
- lr_scheduler_type: linear
|
59 |
+
- lr_scheduler_warmup_steps: 1000
|
60 |
+
- num_epochs: 20.0
|
61 |
+
|
62 |
+
### Training results
|
63 |
+
|
64 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
65 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
66 |
+
| 0.4628 | 10.0 | 1450 | 0.6779 | 0.5171 |
|
67 |
+
| 0.3036 | 20.0 | 2900 | 0.7549 | 0.4902 |
|
68 |
+
|
69 |
+
|
70 |
+
### Framework versions
|
71 |
+
|
72 |
+
- Transformers 4.36.2
|
73 |
+
- Pytorch 2.3.0.dev20231229+cu118
|
74 |
+
- Datasets 2.16.0
|
75 |
+
- Tokenizers 0.15.0
|