Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Update the tutorial of Neural Graph ODE #635

Merged
merged 35 commits into from
Jun 23, 2022
Merged

Update the tutorial of Neural Graph ODE #635

merged 35 commits into from
Jun 23, 2022

Conversation

YichengDWu
Copy link
Contributor

@YichengDWu YichengDWu commented Jun 19, 2022 •
edited
Loading

@YichengDWu
Copy link
Contributor Author

YichengDWu commented Jun 19, 2022 •
edited
Loading

Should update this PR once SciML/DiffEqFlux.jl#727 is fixed. initialparameters(rng::AbstractRNG, node::NeuralODE) and initialstates(rng::AbstractRNG, node::NeuralODE) could be removed.

@ChrisRackauckas
Copy link
Member

For now this will run with CPU on the CI machine, but we will switch the CI to GPU shortly after.

@YichengDWu YichengDWu requested a review from avik-pal June 21, 2022 18:46
YichengDWu and others added 4 commits June 21, 2022 14:12
@YichengDWu @ChrisRackauckas
Update docs/src/neural_ode/neural_gde.md
24cb318 
Co-authored-by: Christopher Rackauckas <accounts@chrisrackauckas.com>
@YichengDWu
Update neural_gde.md
b6acd68 
delete the output
@YichengDWu
Update neural_gde.md
bba9449 
try blocking the output of downloading Cora
@YichengDWu
Update Documentation.yml
7ebc3d8 
add env vari
@YichengDWu
Copy link
Contributor Author

Bummer, we will have to add DATADEPS_ALWAYS_ACCEPT in the environment

@YichengDWu
Copy link
Contributor Author

Let's take this because it's at least an improvement (with proper doctesting), and we can update the optimizer setup later.

Very glad to hear that

@ChrisRackauckas
Copy link
Member

Bummer, we will have to add DATADEPS_ALWAYS_ACCEPT in the environment

That's fine, that's expected for CI.

@YichengDWu
Copy link
Contributor Author

It looks like I omitted the train() wrapper in the Step by Step Explanation section. Since it should be exactly the same as the copy-and-paste code above, do we need to repeat the test?

@ChrisRackauckas
Copy link
Member

Since it should be exactly the same as the copy-and-paste code above, do we need to repeat the test?

I assume it might be nice to really show some of the intermediate print out results. But I was thinking these docs would have an issue being all built, and indeed it looks like they will come out to a few hours in total 😅, so I need to find a way to split the docs build here across multiple CI machines and merge it.

@YichengDWu
Copy link
Contributor Author

YichengDWu commented Jun 22, 2022 •
edited
Loading

Totally agreed. But what if I want to go inside a function and explain things? Is it possible to do so? The scope might be a problem

@YichengDWu
Copy link
Contributor Author

YichengDWu commented Jun 22, 2022 •
edited
Loading

I haven't changed that btw, but I think it is kinda nice to break train() into parts to explain them. I'm thinking maybe we could disable the doctest on the final block since it would just print the same thing as the copy-paste code does, and we will still have all the intermediate results printed out.

@ChrisRackauckas
Copy link
Member

ChrisRackauckas commented Jun 23, 2022 •
edited
Loading

That's a good point. It does have the downside of not testing that part, but it would decrease times a bunch and give a bit more control over what is "displayed". Feel free to go either way.

@YichengDWu
Update neural_gde.md
a6125c0 
remove doctest on the last block
@ChrisRackauckas ChrisRackauckas mentioned this pull request Jun 23, 2022
Closed
@ChrisRackauckas
Copy link
Member

Awesome thanks. We'll circle back around to the other bits but for now, it's working so that's at least a lot better.

@ChrisRackauckas ChrisRackauckas merged commit f06856f into SciML:master Jun 23, 2022
@YichengDWu YichengDWu deleted the graph-node branch June 23, 2022 18:56
@YichengDWu
Copy link
Contributor Author

DifferentialEquations is repeated in project.toml
image

@ChrisRackauckas
Copy link
Member

oh hot damn what a week 😅

@frankschae
Copy link
Member

seen in PR #662 too. I removed it now.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@CarloLucibello CarloLucibello CarloLucibello left review comments

@ChrisRackauckas ChrisRackauckas ChrisRackauckas approved these changes

@avik-pal avik-pal Awaiting requested review from avik-pal

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

Neural Graph Differential Equations example not working
6 participants
@YichengDWu @ChrisRackauckas @ba2tro @frankschae @CarloLucibello @avik-pal