Skip to content

0.1.0

Latest
Compare
Choose a tag to compare
@jackgerrits jackgerrits released this 11 Jul 03:22
· 36 commits to main since this release
4343de4

This release includes the first version of the Python bindings for the project. Check out a quick getting started here.

Notable new features:

Breaking changes

  • Parallelization has been reworked so the train args --read-batch-size and --thread-pool-size were removed and replaced with --num-parse-threads
  • The hash function was changed from MurMur32 to XXHash, so feature/namespace hashes will change
  • Features are now mutable in the reduction stack

All changes

Bug Fixes

  • Update name in gen completions (#35)
  • Remove predict_then_learn impl for coin to match required meaning (#49)
  • Fix python stub file and add CI to verify (#48)

Features

  • Special case thread counts, move example number to metric (#9)
  • Implement sensitivity for coin (#11)
  • Change hash function to xxhash (#12)
  • Implement squarecb (#30)
  • Json parser (#29)
  • Python package (#31)
  • Improve parser ergonomics, accept dict for json (#50)

Performance

  • Use borrowing json parser (#15)
  • Improve interaction performance (#22)
  • Specify thin lto in workspace cargo.toml (#27)

Documentation

  • Update installation instructions on readme
  • Add reductionml homepage (#24)
  • Add initial docs book (#5)
  • Add input docs, add more info for reductions (#36)
  • Add dev dependency info (#38)
  • Fix documentation generation for py package (#53)
  • Fix python docs (#54)
  • Add getting started for python (#51)

Miscellaneous Tasks

  • Fix lints (#10)
  • Update lock file (#16)
  • Add books emoji to readme (#37)
  • Use nextest if available for justfile (#40)
  • Add py install to justfile (#46)
  • Move workflow to correct folder (#47)
  • Prepare for 0.1.0 release (#45)

Refactor

  • Features mutability in reduction signature and AsInner (#13)
  • Dont predict if prediction not used (#17)
  • Use a two queue system for train parallelization (#19)
  • Cleanup (#21)
  • Manage config schemas with checked in copies (#26)
  • Rename schema dir (#28)
  • Integration tests (#33)
  • Split out config creation from workspace itself (#39)

Testing

  • Add simple cb explore test (#32)

Ci

  • Add lock for binary, split build and test (#14)
  • Rename job to fix cache (#20)
  • Caching in site build (#25)
  • Add semantic commit requirement CI (#42)
  • Only trigger python release on v* tags (#43)
  • Fix book generation command (#44)

Full Changelog: v0.0.1...v0.1.0