A few days ago, I opened BOOTH for Hacktoberfest contributors.
BOOTH is a lightweight, provider-agnostic Python library that adds a checkpoint before an LLM output reaches your application.
check() / acheck() handle ambiguity and confidence checks, while check_with_evidence() lets you compare an answer against evidence retrieved by your RAG pipeline.
Since Hacktoberfest started, developers have begun contributing integrations across different LLM providers and models.
More contributions and feedback are always welcome. 🙌
🛠️ Open issues: https://github.com/Vedantgitbot/booth/issues
🔗 Repo: https://github.com/Vedantgitbot/booth
Really happy to see BOOTH growing beyond just me.
Top comments (0)