Skip to content

v0.5.1

Choose a tag to compare

View all tags
@Tejas-TA Tejas-TA released this 21 Jul 22:04
· 5 commits to main since this release
v0.5.1
729bcd8

Fixed

  1. ModelTool.to_langchain() now wires ainvoke() in as the LangChain StructuredTool's coroutine, so async LangChain agents actually run the model prediction asynchronously instead of silently falling back to sync execution.

Added

  1. Warning when confidence_threshold is set on a classifier whose underlying model has no predict_proba method (mirrors the existing warning for regressors) — previously this failed silently with the threshold never taking effect.
  2. Test coverage for the LangChain exporter (tests/test_exporters_langchain.py) — schema shape, sync/async invocation, and missing-dependency error path. Previously untested.
  3. Test coverage for both confidence_threshold construction-time warning paths.
Assets 2
Loading