An implementation of the GraphRAG pipeline (based on the 2024 paper "From Local to Global" by Edge et al.) for query-focused summarization of large text corpora.

Raku package used to efficiently schedule and combine multiple LLM generation steps.