Commit
•
486fa5c
1
Parent(s):
25f096a
update model card README.md
Browse files
README.md
ADDED
@@ -0,0 +1,90 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
tags:
|
3 |
+
- automatic-speech-recognition
|
4 |
+
- timit_asr
|
5 |
+
- generated_from_trainer
|
6 |
+
datasets:
|
7 |
+
- timit_asr
|
8 |
+
model-index:
|
9 |
+
- name: distilhubert-timit
|
10 |
+
results: []
|
11 |
+
---
|
12 |
+
|
13 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
14 |
+
should probably proofread and complete it, then remove this comment. -->
|
15 |
+
|
16 |
+
# distilhubert-timit
|
17 |
+
|
18 |
+
This model is a fine-tuned version of [anton-l/distilhubert](https://huggingface.co/anton-l/distilhubert) on the TIMIT_ASR - NA dataset.
|
19 |
+
It achieves the following results on the evaluation set:
|
20 |
+
- Loss: 1.3688
|
21 |
+
- Wer: 0.6818
|
22 |
+
|
23 |
+
## Model description
|
24 |
+
|
25 |
+
More information needed
|
26 |
+
|
27 |
+
## Intended uses & limitations
|
28 |
+
|
29 |
+
More information needed
|
30 |
+
|
31 |
+
## Training and evaluation data
|
32 |
+
|
33 |
+
More information needed
|
34 |
+
|
35 |
+
## Training procedure
|
36 |
+
|
37 |
+
### Training hyperparameters
|
38 |
+
|
39 |
+
The following hyperparameters were used during training:
|
40 |
+
- learning_rate: 0.0001
|
41 |
+
- train_batch_size: 32
|
42 |
+
- eval_batch_size: 1
|
43 |
+
- seed: 42
|
44 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
45 |
+
- lr_scheduler_type: linear
|
46 |
+
- lr_scheduler_warmup_steps: 1000
|
47 |
+
- num_epochs: 20.0
|
48 |
+
- mixed_precision_training: Native AMP
|
49 |
+
|
50 |
+
### Training results
|
51 |
+
|
52 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
53 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
54 |
+
| 4.2247 | 0.69 | 100 | 3.8607 | 1.0 |
|
55 |
+
| 2.9444 | 1.38 | 200 | 2.9509 | 1.0 |
|
56 |
+
| 2.8858 | 2.07 | 300 | 2.8446 | 1.0 |
|
57 |
+
| 2.2804 | 2.76 | 400 | 2.1985 | 1.0014 |
|
58 |
+
| 1.505 | 3.45 | 500 | 1.4972 | 0.9609 |
|
59 |
+
| 1.06 | 4.14 | 600 | 1.2014 | 0.8058 |
|
60 |
+
| 1.0166 | 4.83 | 700 | 1.0605 | 0.7536 |
|
61 |
+
| 0.966 | 5.52 | 800 | 0.9963 | 0.7101 |
|
62 |
+
| 0.6857 | 6.21 | 900 | 0.9443 | 0.6898 |
|
63 |
+
| 0.5859 | 6.9 | 1000 | 0.9043 | 0.6796 |
|
64 |
+
| 0.6812 | 7.59 | 1100 | 0.9095 | 0.6716 |
|
65 |
+
| 0.6088 | 8.28 | 1200 | 0.9422 | 0.6677 |
|
66 |
+
| 0.4162 | 8.97 | 1300 | 0.9548 | 0.6657 |
|
67 |
+
| 0.3411 | 9.66 | 1400 | 0.9901 | 0.6689 |
|
68 |
+
| 0.3323 | 10.34 | 1500 | 0.9996 | 0.6638 |
|
69 |
+
| 0.431 | 11.03 | 1600 | 1.0521 | 0.6708 |
|
70 |
+
| 0.2029 | 11.72 | 1700 | 1.0946 | 0.6793 |
|
71 |
+
| 0.1424 | 12.41 | 1800 | 1.1288 | 0.6712 |
|
72 |
+
| 0.1922 | 13.1 | 1900 | 1.1456 | 0.6740 |
|
73 |
+
| 0.326 | 13.79 | 2000 | 1.2077 | 0.6915 |
|
74 |
+
| 0.0892 | 14.48 | 2100 | 1.2525 | 0.6796 |
|
75 |
+
| 0.0769 | 15.17 | 2200 | 1.2313 | 0.6736 |
|
76 |
+
| 0.0927 | 15.86 | 2300 | 1.3001 | 0.6864 |
|
77 |
+
| 0.232 | 16.55 | 2400 | 1.3490 | 0.6963 |
|
78 |
+
| 0.0485 | 17.24 | 2500 | 1.3268 | 0.6763 |
|
79 |
+
| 0.0487 | 17.93 | 2600 | 1.3376 | 0.6780 |
|
80 |
+
| 0.0607 | 18.62 | 2700 | 1.3701 | 0.6895 |
|
81 |
+
| 0.1618 | 19.31 | 2800 | 1.3657 | 0.6796 |
|
82 |
+
| 0.0415 | 20.0 | 2900 | 1.3688 | 0.6818 |
|
83 |
+
|
84 |
+
|
85 |
+
### Framework versions
|
86 |
+
|
87 |
+
- Transformers 4.12.0.dev0
|
88 |
+
- Pytorch 1.8.1
|
89 |
+
- Datasets 1.14.1.dev0
|
90 |
+
- Tokenizers 0.10.3
|