HyperGAN
Search…
ez_norm
1
description: 'layer ez_norm for configurable component'
2
---
3
4
# ez_norm layer
5
6
`ez_norm` is a custom normalization technique that uses a conv of the input by a linear projection of a style vector.
7
8
## Optional arguments
9
10
`style` - The name of the style vector to use. Defaults to "w"
11
12
## input size
13
14
Any 4-d tensor
15
16
## output size
17
18
Same as input size
19
20
## syntax
21
22
```json
23
"ez_norm style=[style vector name]"
24
```
25
26
## examples
27
28
```json
29
"latent name=w",
30
...
31
"cat self (ez_norm style=w)"
32
```
Copied!
Last modified 1yr ago
Copy link