Hugging Face's logo

patrickvonplaten
/
unispeech-sat-base-plus-timit-ft

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
unispeech-sat
timit_asr
generated_from_trainer
Inference Endpoints
Model card Files Metrics Community
1
patrickvonplaten commited on
Commit
67b7c36
•
1 Parent(s): 5cb092a

update model card README.md

Browse files
Files changed (1) hide show
  1. README.md +90 -0
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: unispeech-sat-base-plus-timit-ft
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
+ # unispeech-sat-base-plus-timit-ft
17
+
18
+ This model is a fine-tuned version of [microsoft/unispeech-sat-base-plus](https://huggingface.co/microsoft/unispeech-sat-base-plus) on the TIMIT_ASR - NA dataset.
19
+ It achieves the following results on the evaluation set:
20
+ - Loss: 0.6549
21
+ - Wer: 0.4051
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
+ | 3.3838 | 0.69 | 100 | 3.2528 | 1.0 |
55
+ | 2.9608 | 1.38 | 200 | 2.9682 | 1.0 |
56
+ | 2.9574 | 2.07 | 300 | 2.9346 | 1.0 |
57
+ | 2.8555 | 2.76 | 400 | 2.7612 | 1.0 |
58
+ | 1.7418 | 3.45 | 500 | 1.5732 | 0.9857 |
59
+ | 0.9606 | 4.14 | 600 | 1.0014 | 0.7052 |
60
+ | 0.8334 | 4.83 | 700 | 0.7691 | 0.6161 |
61
+ | 0.852 | 5.52 | 800 | 0.7169 | 0.5997 |
62
+ | 0.5707 | 6.21 | 900 | 0.6821 | 0.5527 |
63
+ | 0.4235 | 6.9 | 1000 | 0.6078 | 0.5140 |
64
+ | 0.4357 | 7.59 | 1100 | 0.5927 | 0.4982 |
65
+ | 0.5004 | 8.28 | 1200 | 0.5814 | 0.4826 |
66
+ | 0.3757 | 8.97 | 1300 | 0.5951 | 0.4643 |
67
+ | 0.2579 | 9.66 | 1400 | 0.5990 | 0.4581 |
68
+ | 0.2087 | 10.34 | 1500 | 0.5864 | 0.4488 |
69
+ | 0.3155 | 11.03 | 1600 | 0.5836 | 0.4464 |
70
+ | 0.2701 | 11.72 | 1700 | 0.6045 | 0.4348 |
71
+ | 0.172 | 12.41 | 1800 | 0.6494 | 0.4344 |
72
+ | 0.1529 | 13.1 | 1900 | 0.5915 | 0.4241 |
73
+ | 0.2411 | 13.79 | 2000 | 0.6156 | 0.4246 |
74
+ | 0.2348 | 14.48 | 2100 | 0.6363 | 0.4206 |
75
+ | 0.1429 | 15.17 | 2200 | 0.6394 | 0.4161 |
76
+ | 0.1151 | 15.86 | 2300 | 0.6186 | 0.4167 |
77
+ | 0.1723 | 16.55 | 2400 | 0.6498 | 0.4124 |
78
+ | 0.1997 | 17.24 | 2500 | 0.6541 | 0.4076 |
79
+ | 0.1297 | 17.93 | 2600 | 0.6546 | 0.4117 |
80
+ | 0.101 | 18.62 | 2700 | 0.6471 | 0.4075 |
81
+ | 0.1272 | 19.31 | 2800 | 0.6586 | 0.4065 |
82
+ | 0.1901 | 20.0 | 2900 | 0.6549 | 0.4051 |
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