Repository navigation
Visualizing Docling HybridChunker + contextualize() vs. Naive Splitters for pgvector Hybrid RRF #5336
muhammadtalha072
started this conversation in
Show and tell
Replies: 0 comments
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Hi everyone! We built pgRAG-Inspector (
pgrag-eval-lite), a local-first diagnostic workbench that pairs Docling (DocumentConverter+ structure-aware chunking) with PostgreSQL 17 +pgvector0.8+ Hybrid RRF.What It Solves for Docling Users
BROKEN_MARKDOWN_TABLE),>80%sliding-window token overlap, and mid-sentence cutoffs.Docling HybridChunker + contextualize().pgvectorcosine distance (<=>), Postgrests_rank_cd, and Hybrid Reciprocal Rank Fusion (RRF) at$0.00LLM judge cost.Quickstart via PyPI
Would love any feedback from the Docling community on additional table/layout health heuristics you'd like to see in
pgrag check!All reactions