How to host your deep learning model trained in flux

Offtopic
1

How to host your saved julia flux model on web? Like Pytorch and Flask are used

1 Like
2 
function predict(req)
    io = IOBuffer(HTTP.payload(req))
    x = JSON3.read(io)
    y = model(x)
    HTTP.Response(200, JSON3.write(y))
end

router = HTTP.Router()
HTTP.@register(router, "POST", "predict", predict)
HTTP.serve(router, "0.0.0.0", port)
1 Like
3

There is also Dash.jl

1 Like
4

I find a lot of features missing in Dash. I would rather use Genie.jl

1 Like
5

But your example was just HTTP right, not genie?

6

This makes me wonder. Are there any ‘example’ pages out there which are based on Genie.jl?

7

Yes I’m working on Genie example/tutorial. This was a quick snippet meanwhile

1 Like
8

I am working on an example with Stipple FastAI.jl(model) and Genie. You might wanna check. However it’s still work in progress

What is Stipple? It’s a all-in-one Streamlit and Plotly solution for Julia. You can make Single page apps and multi-page apps in Julia itself which is then converted in vuejs code …it’s built on top of Genie

Why FastAI? I was already familiar with FastAI python(but not very familiar with Flux.jl) …also SOTA results

https://github.com/AbhimanyuAryan/stipple-fastai