patrickvonplaten commited on
Commit
4ae68de
1 Parent(s): 7132dfb

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
+ license: apache-2.0
3
+ tags:
4
+ - automatic-speech-recognition
5
+ - librispeech_asr
6
+ - generated_from_trainer
7
+ model-index:
8
+ - name: sew-d-mid-400k-librispeech-clean-100h-ft
9
+ results: []
10
+ ---
11
+
12
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
13
+ should probably proofread and complete it, then remove this comment. -->
14
+
15
+ # sew-d-mid-400k-librispeech-clean-100h-ft
16
+
17
+ This model is a fine-tuned version of [asapp/sew-d-mid-400k](https://huggingface.co/asapp/sew-d-mid-400k) on the LIBRISPEECH_ASR - CLEAN dataset.
18
+ It achieves the following results on the evaluation set:
19
+ - Loss: 2.3540
20
+ - Wer: 1.0536
21
+
22
+ ## Model description
23
+
24
+ More information needed
25
+
26
+ ## Intended uses & limitations
27
+
28
+ More information needed
29
+
30
+ ## Training and evaluation data
31
+
32
+ More information needed
33
+
34
+ ## Training procedure
35
+
36
+ ### Training hyperparameters
37
+
38
+ The following hyperparameters were used during training:
39
+ - learning_rate: 3e-05
40
+ - train_batch_size: 4
41
+ - eval_batch_size: 8
42
+ - seed: 42
43
+ - distributed_type: multi-GPU
44
+ - num_devices: 8
45
+ - total_train_batch_size: 32
46
+ - total_eval_batch_size: 64
47
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
48
+ - lr_scheduler_type: linear
49
+ - lr_scheduler_warmup_steps: 500
50
+ - num_epochs: 3.0
51
+ - mixed_precision_training: Native AMP
52
+
53
+ ### Training results
54
+
55
+ | Training Loss | Epoch | Step | Validation Loss | Wer |
56
+ |:-------------:|:-----:|:----:|:---------------:|:------:|
57
+ | 7.319 | 0.11 | 100 | 11.0572 | 1.0 |
58
+ | 3.6726 | 0.22 | 200 | 4.2003 | 1.0 |
59
+ | 2.981 | 0.34 | 300 | 3.5742 | 0.9919 |
60
+ | 2.9411 | 0.45 | 400 | 3.2599 | 1.0 |
61
+ | 2.903 | 0.56 | 500 | 2.9350 | 1.0 |
62
+ | 2.8597 | 0.67 | 600 | 2.9514 | 1.0 |
63
+ | 2.7771 | 0.78 | 700 | 2.8521 | 1.0 |
64
+ | 2.7926 | 0.9 | 800 | 2.7821 | 1.0120 |
65
+ | 2.6623 | 1.01 | 900 | 2.7027 | 0.9924 |
66
+ | 2.5893 | 1.12 | 1000 | 2.6667 | 1.0240 |
67
+ | 2.5733 | 1.23 | 1100 | 2.6341 | 1.0368 |
68
+ | 2.5455 | 1.35 | 1200 | 2.5928 | 1.0411 |
69
+ | 2.4919 | 1.46 | 1300 | 2.5695 | 1.0817 |
70
+ | 2.5182 | 1.57 | 1400 | 2.5559 | 1.1072 |
71
+ | 2.4766 | 1.68 | 1500 | 2.5229 | 1.1257 |
72
+ | 2.4267 | 1.79 | 1600 | 2.4991 | 1.1151 |
73
+ | 2.3919 | 1.91 | 1700 | 2.4768 | 1.1139 |
74
+ | 2.3883 | 2.02 | 1800 | 2.4452 | 1.0636 |
75
+ | 2.3737 | 2.13 | 1900 | 2.4304 | 1.0594 |
76
+ | 2.3569 | 2.24 | 2000 | 2.4095 | 1.0539 |
77
+ | 2.3641 | 2.35 | 2100 | 2.3997 | 1.0511 |
78
+ | 2.3281 | 2.47 | 2200 | 2.3856 | 1.0414 |
79
+ | 2.2912 | 2.58 | 2300 | 2.3750 | 1.0696 |
80
+ | 2.3028 | 2.69 | 2400 | 2.3684 | 1.0436 |
81
+ | 2.2906 | 2.8 | 2500 | 2.3613 | 1.0538 |
82
+ | 2.2822 | 2.91 | 2600 | 2.3558 | 1.0506 |
83
+
84
+
85
+ ### Framework versions
86
+
87
+ - Transformers 4.12.0.dev0
88
+ - Pytorch 1.9.0+cu111
89
+ - Datasets 1.13.4.dev0
90
+ - Tokenizers 0.10.3