Hugging Face's logo

patrickvonplaten
/
wav2vec2-base-repro-timit

Automatic Speech Recognition
Transformers
PyTorch
TensorBoard
wav2vec2
timit_asr
generated_from_trainer
Inference Endpoints
Model card Files Metrics Community
1
patrickvonplaten commited on
Commit
394cd6b
•
1 Parent(s): 46d0f4f

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: wav2vec2-base-repro-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
+ # wav2vec2-base-repro-timit
17
+
18
+ This model is a fine-tuned version of [patrickvonplaten/wav2vec2-base-repro-960h-libri-85k-steps](https://huggingface.co/patrickvonplaten/wav2vec2-base-repro-960h-libri-85k-steps) on the TIMIT_ASR - NA dataset.
19
+ It achieves the following results on the evaluation set:
20
+ - Loss: 0.8562
21
+ - Wer: 0.5484
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
+ | 5.9793 | 0.69 | 100 | 5.4532 | 1.0 |
55
+ | 2.9066 | 1.38 | 200 | 2.9070 | 1.0 |
56
+ | 2.2562 | 2.07 | 300 | 2.0323 | 1.0 |
57
+ | 1.5273 | 2.76 | 400 | 1.1510 | 0.8001 |
58
+ | 1.1085 | 3.45 | 500 | 0.9521 | 0.7053 |
59
+ | 0.813 | 4.14 | 600 | 0.8617 | 0.6702 |
60
+ | 0.8434 | 4.83 | 700 | 0.8068 | 0.6393 |
61
+ | 0.9631 | 5.52 | 800 | 0.7863 | 0.6248 |
62
+ | 0.707 | 6.21 | 900 | 0.7476 | 0.5973 |
63
+ | 0.5568 | 6.9 | 1000 | 0.7350 | 0.5911 |
64
+ | 0.6171 | 7.59 | 1100 | 0.7171 | 0.5841 |
65
+ | 0.7011 | 8.28 | 1200 | 0.7318 | 0.5798 |
66
+ | 0.5546 | 8.97 | 1300 | 0.7447 | 0.5767 |
67
+ | 0.4278 | 9.66 | 1400 | 0.7481 | 0.5650 |
68
+ | 0.3576 | 10.34 | 1500 | 0.7443 | 0.5713 |
69
+ | 0.5506 | 11.03 | 1600 | 0.7574 | 0.5664 |
70
+ | 0.4127 | 11.72 | 1700 | 0.8043 | 0.5631 |
71
+ | 0.3251 | 12.41 | 1800 | 0.7738 | 0.5550 |
72
+ | 0.3119 | 13.1 | 1900 | 0.7829 | 0.5516 |
73
+ | 0.4371 | 13.79 | 2000 | 0.8025 | 0.5556 |
74
+ | 0.3772 | 14.48 | 2100 | 0.8451 | 0.5559 |
75
+ | 0.2942 | 15.17 | 2200 | 0.8300 | 0.5556 |
76
+ | 0.2503 | 15.86 | 2300 | 0.8417 | 0.5541 |
77
+ | 0.3671 | 16.55 | 2400 | 0.8568 | 0.5528 |
78
+ | 0.3867 | 17.24 | 2500 | 0.8521 | 0.5510 |
79
+ | 0.2614 | 17.93 | 2600 | 0.8479 | 0.5523 |
80
+ | 0.2441 | 18.62 | 2700 | 0.8558 | 0.5494 |
81
+ | 0.3059 | 19.31 | 2800 | 0.8553 | 0.5474 |
82
+ | 0.3734 | 20.0 | 2900 | 0.8562 | 0.5484 |
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