TastyPancakes/Flux.jl
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Remove unnecessary file

Browse Source
This commit is contained in:
Avik Pal 2018-07-13 14:27:04 +05:30
parent 2664a16556
commit f57db22abe
1 changed files with 0 additions and 21 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

21
docs/src/models/convolution.md
View File

@ -1,21 +0,0 @@
# Additional Convolution Layers
## Depthwise Convolutions
Using Depthwise Convolutions is pretty straightforword and much similar
to the usage of normal Convolutions. So simply we can swap in a
Depthwise Convolution in place of a Convolution.
Lets say we have to define a simple convolution layer like
```julia
m = Conv((3, 3), 3=>64, pad = (1, 1))
```
The alternative to this using a Depthwise Convolution would be
```julia
m = Chain(DepthwiseConv((3, 3), 3=>2, pad = (1, 1)),
Conv((1, 1), 6=>64))
```
Incase the second argument to `DepthwiseConv` is an `Integer` instead of a
`Pair` the channel multiplier is taken to be 1.