Hugging Face's logo

patrickvonplaten
/
unispeech-sat-large-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
db46bdc
·
1 Parent(s): 2578cf7

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-large-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-large-timit-ft
17
+
18
+ This model is a fine-tuned version of [microsoft/unispeech-sat-large](https://huggingface.co/microsoft/unispeech-sat-large) on the TIMIT_ASR - NA dataset.
19
+ It achieves the following results on the evaluation set:
20
+ - Loss: 0.6074
21
+ - Wer: 0.3880
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
+ | 6.2516 | 0.69 | 100 | 5.8638 | 1.0 |
55
+ | 2.9596 | 1.38 | 200 | 2.9550 | 1.0 |
56
+ | 2.8831 | 2.07 | 300 | 2.8547 | 1.0 |
57
+ | 2.3223 | 2.76 | 400 | 2.2044 | 1.0063 |
58
+ | 1.2104 | 3.45 | 500 | 1.0845 | 0.7706 |
59
+ | 0.6779 | 4.14 | 600 | 0.7342 | 0.5663 |
60
+ | 0.6319 | 4.83 | 700 | 0.6054 | 0.4881 |
61
+ | 0.664 | 5.52 | 800 | 0.5808 | 0.4913 |
62
+ | 0.402 | 6.21 | 900 | 0.5647 | 0.4611 |
63
+ | 0.3176 | 6.9 | 1000 | 0.5211 | 0.4440 |
64
+ | 0.3392 | 7.59 | 1100 | 0.5187 | 0.4359 |
65
+ | 0.3888 | 8.28 | 1200 | 0.5501 | 0.4391 |
66
+ | 0.2874 | 8.97 | 1300 | 0.5249 | 0.4148 |
67
+ | 0.208 | 9.66 | 1400 | 0.5407 | 0.4152 |
68
+ | 0.1457 | 10.34 | 1500 | 0.5722 | 0.4155 |
69
+ | 0.2375 | 11.03 | 1600 | 0.5780 | 0.4059 |
70
+ | 0.2111 | 11.72 | 1700 | 0.5823 | 0.4094 |
71
+ | 0.1422 | 12.41 | 1800 | 0.5754 | 0.3977 |
72
+ | 0.125 | 13.1 | 1900 | 0.5784 | 0.4031 |
73
+ | 0.1996 | 13.79 | 2000 | 0.5630 | 0.3956 |
74
+ | 0.1747 | 14.48 | 2100 | 0.5880 | 0.3964 |
75
+ | 0.1263 | 15.17 | 2200 | 0.5987 | 0.3951 |
76
+ | 0.11 | 15.86 | 2300 | 0.5688 | 0.3964 |
77
+ | 0.1411 | 16.55 | 2400 | 0.6223 | 0.3906 |
78
+ | 0.1647 | 17.24 | 2500 | 0.6135 | 0.3960 |
79
+ | 0.1162 | 17.93 | 2600 | 0.6224 | 0.3960 |
80
+ | 0.098 | 18.62 | 2700 | 0.6017 | 0.3907 |
81
+ | 0.1183 | 19.31 | 2800 | 0.6121 | 0.3885 |
82
+ | 0.1717 | 20.0 | 2900 | 0.6074 | 0.3880 |
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