Collected sources and patterns will appear here. Add from search, explore, or the patterns library.
LLMResponse, SourceText -> VerifiedEntitySpans
Resolve hallucinated or slightly altered LLM-extracted entity spans by locating their closest matching substring boundaries in the original source text.
Problem it solves
Generative LLMs often output clean entity names that do not exactly match the original sentence casing or spelling.
Consumes
Emits
The real projects this mechanism was found in. Attribution is the point — this is how the best teams actually do it.