mlcommons / mlcommons/modelbench

Improve how locales and prompt file name interact with each other

Open
#781 1 comment 0 reactions 1 assignee View on GitHub

@rogthefrog is already working on this.

Since Jan 9, 2025.

Dominant language
Python
Stars
134
Forks
36
Avg merge
1d 11h
Merged PRs (30d)
17

Description

The safe test downloads prompt files from modellab. It supports multiple locales in a given file, but doesn't do well with different files for different locales.

Workarounds exist, but they're hacky.

Options:

  • Improve file naming conventions to support locales. Proposal: if a file includes one locale, include that locale in the filename. English prompt files don't include en_us and that makes the code messy.
  • Ensure that we don't try to make test items in locale X using a file that does not include locale X (e.g. in the smoke test).
  • Handle file not found errors better, especially in unit or smoke tests.
  • If any unit tests rely on downloading an asset from modellab, replace that with a local file.

References:

Workaround for test items

Workaround for filenames with and without locales

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.