facebookresearch / recipes

Recipes are a standard, well supported set of blueprints for machine learning engineers to rapidly train models using the latest research techniques without significant engineering overhead.Specifically, recipes aims to provide- Consistent access to pre-trained SOTA models ready for production- Reference implementations for SOTA research reproducibility, and infrastructure to guarantee correctness, efficiency, and interoperability.
BSD 3-Clause "New" or "Revised" License
284 stars 35 forks source link

Updated README #8

Closed laurencer closed 2 years ago

laurencer commented 2 years ago

Updated the README to be more user oriented in explaining the benefits of torchrecipes.

facebook-github-bot commented 2 years ago

@kandluis has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

laurencer commented 2 years ago

As an aside for completeness - I think it would be good to add some tutorials around common & expected use-cases on how people should get started.

E.g.

I found the current story a bit complex to be honest to try and figure out the best/easiest/recommended way to approach some of the above tasks. Also might be useful to consider usage patterns (e.g. do I fork and clone this repo; do I modify in place or make a copy of a recipe; do I add/update config or modify code; etc).

facebook-github-bot commented 2 years ago

@laurencer has updated the pull request. You must reimport the pull request before landing.

facebook-github-bot commented 2 years ago

@kandluis has imported this pull request. If you are a Facebook employee, you can view this diff on Phabricator.

kandluis commented 2 years ago

Yep, these are in development as we speak for both vision and text. We're also planning to create some easy-ish templates and try to provide a standardize way to contribute.

The primarily principle here will be to minimize unnecessary overhead for users of recipes.

Also might be useful to consider usage patterns (e.g. do I fork and clone this repo; do I modify in place or make a copy of a recipe; do I add/update config or modify code; etc).

We need to jot this down, but generally, "users" would pip install torchrecipes and use as a library (inheriting, creating their own, plugging into their custom pipelies, etc.). Advanced users would fork and tweak to their needs, with a subset of these becoming contributors back to the main library (especially if they're developing recipes with wide interest/adoption, or new utlities/tooling).

cc @Nayef211 @sallysyw