patrickvonplaten
commited on
Commit
•
0437868
1
Parent(s):
c872564
update model card README.md
Browse files
README.md
ADDED
@@ -0,0 +1,91 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
tags:
|
4 |
+
- automatic-speech-recognition
|
5 |
+
- timit_asr
|
6 |
+
- generated_from_trainer
|
7 |
+
datasets:
|
8 |
+
- timit_asr
|
9 |
+
model-index:
|
10 |
+
- name: sew-d-small-100k-ft-timit
|
11 |
+
results: []
|
12 |
+
---
|
13 |
+
|
14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
15 |
+
should probably proofread and complete it, then remove this comment. -->
|
16 |
+
|
17 |
+
# sew-d-small-100k-ft-timit
|
18 |
+
|
19 |
+
This model is a fine-tuned version of [asapp/sew-d-small-100k](https://huggingface.co/asapp/sew-d-small-100k) on the TIMIT_ASR - NA dataset.
|
20 |
+
It achieves the following results on the evaluation set:
|
21 |
+
- Loss: 1.7430
|
22 |
+
- Wer: 0.8090
|
23 |
+
|
24 |
+
## Model description
|
25 |
+
|
26 |
+
More information needed
|
27 |
+
|
28 |
+
## Intended uses & limitations
|
29 |
+
|
30 |
+
More information needed
|
31 |
+
|
32 |
+
## Training and evaluation data
|
33 |
+
|
34 |
+
More information needed
|
35 |
+
|
36 |
+
## Training procedure
|
37 |
+
|
38 |
+
### Training hyperparameters
|
39 |
+
|
40 |
+
The following hyperparameters were used during training:
|
41 |
+
- learning_rate: 0.0001
|
42 |
+
- train_batch_size: 32
|
43 |
+
- eval_batch_size: 1
|
44 |
+
- seed: 42
|
45 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
46 |
+
- lr_scheduler_type: linear
|
47 |
+
- lr_scheduler_warmup_steps: 1000
|
48 |
+
- num_epochs: 20.0
|
49 |
+
- mixed_precision_training: Native AMP
|
50 |
+
|
51 |
+
### Training results
|
52 |
+
|
53 |
+
| Training Loss | Epoch | Step | Validation Loss | Wer |
|
54 |
+
|:-------------:|:-----:|:----:|:---------------:|:------:|
|
55 |
+
| 4.2068 | 0.69 | 100 | 4.0802 | 1.0 |
|
56 |
+
| 2.9806 | 1.38 | 200 | 2.9792 | 1.0 |
|
57 |
+
| 2.9781 | 2.07 | 300 | 2.9408 | 1.0 |
|
58 |
+
| 2.9656 | 2.76 | 400 | 2.9143 | 1.0 |
|
59 |
+
| 2.8953 | 3.45 | 500 | 2.8774 | 1.0 |
|
60 |
+
| 2.7714 | 4.14 | 600 | 2.7774 | 0.9999 |
|
61 |
+
| 2.6883 | 4.83 | 700 | 2.6552 | 0.9854 |
|
62 |
+
| 2.6477 | 5.52 | 800 | 2.5424 | 1.0114 |
|
63 |
+
| 2.3607 | 6.21 | 900 | 2.4291 | 1.1012 |
|
64 |
+
| 2.1048 | 6.9 | 1000 | 2.2526 | 0.9744 |
|
65 |
+
| 2.2448 | 7.59 | 1100 | 2.1697 | 0.9332 |
|
66 |
+
| 2.3525 | 8.28 | 1200 | 2.0429 | 0.8721 |
|
67 |
+
| 2.099 | 8.97 | 1300 | 2.0076 | 0.9183 |
|
68 |
+
| 1.8236 | 9.66 | 1400 | 1.9695 | 0.8263 |
|
69 |
+
| 1.6602 | 10.34 | 1500 | 1.9328 | 0.8537 |
|
70 |
+
| 2.2146 | 11.03 | 1600 | 1.8795 | 0.8443 |
|
71 |
+
| 1.9278 | 11.72 | 1700 | 1.8749 | 0.8075 |
|
72 |
+
| 1.6324 | 12.41 | 1800 | 1.8504 | 0.8091 |
|
73 |
+
| 1.6517 | 13.1 | 1900 | 1.8228 | 0.8233 |
|
74 |
+
| 2.0463 | 13.79 | 2000 | 1.8194 | 0.8564 |
|
75 |
+
| 1.8736 | 14.48 | 2100 | 1.7950 | 0.8225 |
|
76 |
+
| 1.6233 | 15.17 | 2200 | 1.7896 | 0.8429 |
|
77 |
+
| 1.4982 | 15.86 | 2300 | 1.7793 | 0.8441 |
|
78 |
+
| 1.8955 | 16.55 | 2400 | 1.7640 | 0.8058 |
|
79 |
+
| 1.8253 | 17.24 | 2500 | 1.7668 | 0.8134 |
|
80 |
+
| 1.5332 | 17.93 | 2600 | 1.7611 | 0.8015 |
|
81 |
+
| 1.452 | 18.62 | 2700 | 1.7497 | 0.8072 |
|
82 |
+
| 1.7609 | 19.31 | 2800 | 1.7464 | 0.8062 |
|
83 |
+
| 1.7743 | 20.0 | 2900 | 1.7430 | 0.8090 |
|
84 |
+
|
85 |
+
|
86 |
+
### Framework versions
|
87 |
+
|
88 |
+
- Transformers 4.12.0.dev0
|
89 |
+
- Pytorch 1.8.1
|
90 |
+
- Datasets 1.14.1.dev0
|
91 |
+
- Tokenizers 0.10.3
|