Skip to content
Discussion options

You must be logged in to vote

We do this quite frequently - most commonly it is because we have chunked a document but still want the document title, etc. to be on each chunk so that any LLM summarizations have that context. Our approach is to pre-chunk the content with a script before running GraphRAG and write each chunk out to its own text file. If you do this, just ensure that your final chunk document tokens remain below the GRAPH_CHUNK_SIZE setting (default 1200). Then when you run GraphRAG it will find that all documents are already chunked, so your chunks will remain intact. Also note that when GraphRAG does chunking it includes a token overlap to ensure good coverage, so you might want to take that into accou…

Replies: 1 comment

Comment options

You must be logged in to vote
0 replies
Answer selected by hyiip
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Category
Q&A
Labels
None yet
2 participants