Comments (3)
Dense(in, out)
is more consistent with the Chain
constructor reading left-to-right (and chaining in general), and is also standard in the rest of the deep learning frameworks I'm familiar with.
from flux.jl.
while I also tend to think in matrix dimensions, I'd claim that Dense(in, out)
is more user friendly
from flux.jl.
Dense(in, out) is more consistent with the Chain constructor reading left-to-right (and chaining in general), and is also standard in the rest of the deep learning frameworks I'm familiar with.
This is a strong enough argument. It's just my brain that keeps visualizing layers as functions (well, actually actually they are) and place the input on the right and the output on the left, as in y=f(x)
. On the same line I prefer to write a chain as
Chain( f,
g,
h)
which resembles the actual computation
x = f(x)
x = g(x)
x = h(x)
At least in julia the composition operator |> has a clear direction x |> f |> g
, it would be nice if we could use it to form a Chain
from flux.jl.
Related Issues (20)
- move the codebase to 4 spaces indentation HOT 6
- Why Flux is Significantly Slower than Pytorch? HOT 2
- Implement data movement across GPU devices. HOT 1
- Flux.jl Documentation (Training API Reference)
- Create how-to guide for custom layers HOT 2
- Huber Loss Fails with Metal GPU
- `train!` using Metal and stateful optimizers fails HOT 5
- Warning: `sort(d::Dict; args...)` is deprecated, use `sort!(OrderedDict(d); args...)` instead. HOT 2
- Does `withgradient` have lower precicision than simply calling the function? HOT 3
- Warning: sort(d::Dict; args...) is deprecated, use sort!(OrderedDict(d); args...) instead. HOT 1
- `show` is confused by shared parameters
- No cudnn implementation of Conv((1,) N=>M) HOT 12
- Flux docs missing withgradient() call for multi-objective loss functions HOT 6
- can't use masks in multi-head-attention layer HOT 6
- Adapt saving & loading example to CuArrays HOT 2
- Segmentation fault when doing a forward pass with a model saved with BSON HOT 2
- Flux new explicit API does not work but old implicit API works for a simple RNN HOT 4
- Intel Arc GPU support. HOT 1
- `using Flux, cuDNN` freezes, but `using Flux, CUDA, cuDNN` works HOT 1
- Convolutional network slower than tensorflow on CPU HOT 13
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from flux.jl.