AWLY is an attempt to curate and organize the most useful Rust ML resources.

The colors

AWLY organizes crates into categories and shelves, and then grades shelves in aggregate, not the individual crates on it: would you choose Rust for this problem, against another language? A shelf full of excellent, production-ready crates is still yellow if together they cover a sliver of the category. There is no formula behind it: each verdict is a judgment, made against the evidence linked from that shelf.

  • green = comprehensive — covers the category, in production; you would not go wrong choosing Rust
  • yellow = partial — solid pieces to build on, but a sliver of the category; another language covers it better
  • red = too early — experimental or nonexistent; not a default choice yet

How crates are ranked

Entries are sorted into four bands. Featured is a handful per topic with demonstrated large impact — something that visibly moved a lot of people, not a well-marketed release. Those are written by hand, with prose and independent links, and re-verified on an 18-month clock. Listed clears a published bar: old enough, still active, used by people other than its author, and described somewhere. New & experimental is where everything starts. Archive is for the dead (yanked, archived, RustSec unmaintained, or two years with no release and no human commit), kept on the page so you stop looking for them.

Downloads and stars decide nothing here; they are shown as facts. The exact clauses, thresholds and archive reasons live in _scraper/src/tier.rs.

Built with heavy AI assistance

Tracking this ecosystem comprehensively and curating it well is more work than one maintainer has, and the site was effectively unmaintained for several years as a result. LLMs change that in both directions: they make the effort possible to scale, and they make it easier to publish something confidently wrong or poorly written. Significant portions of this content were generated and reviewed by agents, then reviewed and edited by humans. All feedback is welcome — curation, content, emphasis, even maintenance strategy — especially from people willing to help.

AI-assisted PRs are welcome, but don't expect maintainers to do significant reading or research to review your PR. Provide proof and provenance for content claims. Convince us you reviewed and editted the content for things LLMs struggle with.

Arguing with it

Everything on this site is a judgment someone can be argued out of, and the fastest way to change one is evidence: a production user, a benchmark, a talk, a maintainer saying something on the record. This site is maintained by Anthony Nowell and whoever shows up, in the open, at github.com/anowell/are-we-learning-yet. Report an issue or missing content — a missing crate, a wrong color, or poorly written content.

Inspired by Are We Web Yet?, which asked the same question about a different problem and answered it honestly.