FAQ - Frequently Asked Questions¶
- How can I test a NeuroLibre submission ?
- Can I submit a non-github repository ?
- What are the hardware limits on NeuroLibre ?
- I want to contribute, how can I do that ?
- Which languages does NeuroLibre support ?
- What type of review are you doing ?
- How do I manage datasets with NeuroLibre ?
- Which version of
repo2datashould I use ?
- How can I cache my experiments ?
- Can I use Dockerfile for my submission ?
We don’t accept non-github submission. Still if you need the interactive binder, you can use any of those providers.
Running time should take less than 8 hours to execute uncached (that includes all notebooks and book build) with or 2 CPU@3GHz. You need use less than ~7.5GB of RAM, take less than 10GB runtime storage and no more than 5GB of data.
It would be a pleasure to include external people on our project, please reach out to us via our mattermost brainhack forum or #TODO:EMAIL! There are fundamentally two ways to contribute to NeuroLibre: as a reviewer or as a developper.
The reviewer team is in charge of checking if submissions execute ok on our servers, there are also exchanging with the author to help them improve it. Developer team works on the binderhub administration, backend workflows and frontend (including github integrations, and JOSS website template).
All languages supported by the jupyter ecosystem, check the following list.
We are a preprint service, and so we stand for minimalistic reviews. This includes basic formatting, and checking if code executes.
We always recommend latest versions for better compatibility.
jupyter-book you are free to use any version you like, since we rely only on the compiled artifcats.
repo2data, we highly advice to match latest version because this is what is used on the backend.
If you have some cache data, you can use
repo2data to make the data available on NeuroLibre.
Then follow the information about jupter-book caching here.