V0.0.1 packaging - Githubissues

bazingagin / npc_gzip

Code for Paper: “Low-Resource” Text Classification: A Parameter-Free Classification Method with Compressors

MIT License

1.77k stars 156 forks source link

V0.0.1 packaging #30

Closed zbloss closed 1 year ago

zbloss commented 1 year ago

Resolves Issue #24

Adds:

Compressor classes - gzip, bz2, & lzma.
Distance class to compute distance metrics.
Several Exceptions.
KnnCompressor class to do predictioning.
Github Actions pipelines to:
- Run unit tests.
- Deploy package to pypi.
Unit tests

Changes:

Moves the original codebase to original_codebase

zbloss commented 1 year ago

Alright we're extremely close, if not ready to go. @bazingagin I fixed the bug causing the low accuracy by adding a new sample_data method. The accuracy on the AGNews example increased drastically.

Also I updated the load_filipino function

EliahKagan commented 1 year ago

Expanding the CI test matrix to include other operating systems could be deferred until some point after this PR is merged. Is it okay if I mark all my comments about that as resolved?

zbloss commented 1 year ago

Expanding the CI test matrix to include other operating systems could be deferred until some point after this PR is merged. Is it okay if I mark all my comments about that as resolved?

Yep go ahead, we'll tackle that in the future

EliahKagan commented 1 year ago

To speed things up, I'll do another pass through my old unresolved comments, marking things I feel I don't need to ask you about as resolved, and opening new comments for anything that remains.