• Added strip_thinking function to better support deepseek-r1
• Added a probability meets criteria evaluation
• Minor bug fixes and improvements
• Updated dependencies

• Updated to not-again-ai v.0.15
• Updated to Poetry v2 and updated pyproject.toml accordingly

• Added MMLU-Pro and IFEval evals
• Many other improvements

Refactors the common abstractions to make it creating evaluations more flexible, while still providing some common logic.

• Added an LLM-based evaluation, meets criteria.
• Other quality of life script improvements

Initial release