Reproducibility Report of "Meta learning with differentiable closed-form solvers"

[reproducibility-org]

Original article: : Meta-learning with differentiable closed-form solvers (https://openreview.net/pdf?id=HyxnZh0ct7)

PDF URL: https://github.com/reproducibility-challenge/iclr_2019/blob/master/papers/LCAX-HyxnZh0ct7/LCAX_camera_ready.pdf Metadata URL: https://github.com/reproducibility-challenge/iclr_2019/blob/master/papers/LCAX-HyxnZh0ct7/ReScience_src/metadata-draft.yaml Code URL: https://zenodo.org/record/2662494#.XMzE-aYpBSM

Scientific domain: Machine Learning Programming language: Python Suggested editor: ICLR Reproducibility Challenge Editors

Corresponding Author: @ArnoutDevos Challenge PR: https://github.com/reproducibility-challenge/iclr_2019/pull/150

ICLR Reproducibility Challenge, 2019

[reproducibility-org]

Reviewer comments:

1. https://github.com/reproducibility-challenge/iclr_2019/pull/150#issuecomment-464272161
2. https://github.com/reproducibility-challenge/iclr_2019/pull/150#issuecomment-464889846
3. https://github.com/reproducibility-challenge/iclr_2019/pull/150#issuecomment-475799455

[rougier]

@ArnoutDevos Can you use the latest rescience template to make sure to use the proper font. Also, you'll need to edit the metadata, reviewer is "anonymous", volume is 5, issue is 2.

[rougier]

Also, the title of the replication must be "[Re] original title" (this will helps people searching for the original article to find yours as well)

[koustuvsinha]

@rougier updated the draft with the recommended changes.

[rougier]

Thanks. Page 8 is mostly empty, do you know why (some extra \clearpage ?)

[koustuvsinha]

updated, also sent you the raw latex source

[rougier]

I'm now editing the paper. I would need a DOI for the code, would it be possible for the authors to deposit their code on Zenodo and give me the corresponding DOI?

[ArnoutDevos]

Available: 10.5281/zenodo.2662494 https://zenodo.org/record/2662494#.XORbLdMzZ24

[rougier]

By the way, you confirm the two last authors have no ORCID?

[ArnoutDevos]

Sylvain Chatel does: 0000-0002-1275-8367

[rougier]

Can you chech everything is ok: https://www.dropbox.com/s/ohi1n1qsyovixsg/article-1.pdf?dl=0

[ArnoutDevos]

Thank you for editing our work.

1. if possible, put small spacing between paragraphs to improve readability (unless there's a strict policy on this).
2. Sylvain Chatel's ORCID is not yet next to his name
3. If possible, shift content end of page 1 to 2, end of 4 to 5
4. page 6 looks very empty

That's all I guess.

Thanks, Arnout

On Wed, May 22, 2019 at 8:36 AM Nicolas P. Rougier notifications@github.com wrote:

Can you chech everything is ok: https://www.dropbox.com/s/ohi1n1qsyovixsg/article-1.pdf?dl=0

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/ReScience/submissions/issues/2?email_source=notifications&email_token=AA6LSMDX5OUWH3WGU3REZCTPWTSYBA5CNFSM4HKXOKZ2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODV6BRQY#issuecomment-494672067, or mute the thread https://github.com/notifications/unsubscribe-auth/AA6LSMGLCSPVYGPFJISKDK3PWTSYBANCNFSM4HKXOKZQ .

[rougier]

Thanks for your comments. At this point, maybe the easiest for me would be to upload the edited version such that you can tweak it.

[rougier]

You can access it at: https://github.com/rougier/ICLR-2019/tree/master/article-1

[koustuvsinha]

Looks good to me!

[rougier]

@ArnoutDevos can you take care of your suggested changes?

[ArnoutDevos]

@rougier will do by next week. Have a deadline this week.

[rougier]

I'll do it then (I want to have them published today).

[ArnoutDevos]

Ok, thanks :)!

[rougier]

@ArnoutDevos I've updated the PDF. Can you have a look?

[ArnoutDevos]

Looks good to me!