automl/promptolution v2.2.3

Release v2.2.3

on GitHub

Release v2.2.3

What's changed

Bug fixes:

• Fixed list handling in local LLM response parsing — single-prompt batches were not unwrapped correctly
• Fixed chat template handling for local LLMs (#72)
• Fixed ExperimentConfig not being applied before resolving x_column in tasks (#68)
• Fixed keyword handling in RandomSearchSelector.select_exemplars (#70)
• Fixed various typos

Further changes:

• Added CONTRIBUTING.md with workflow, code quality, and testing guidelines
• Added accelerate as a dependency of the [transformers] extra, enabling local inference without manual installation
• Overhauled README: added quickstart code example, PyPI badge, installation guide, scientific publications section, and citation block
• New tests now account for 96% test coverage

Full Changelog: here