feat(ingest): describe PDF figures with vision model#55

Open

scaleborg wants to merge 7 commits intomainfrom

feat/pdf-figure-descriptions

Commits on Mar 14, 2026

feat(ingest): describe PDF figures with vision model during OCR extraction
scaleborg
committed
refactor(ingest): improve figure description prompt for better signal extraction
scaleborg
committed
test(ingest): add figure retrieval eval script
scaleborg
committed
test(eval): add figure-dependent cases to golden RAG dataset
scaleborg
committed
docs: document PDF figure description pipeline and eval coverage
scaleborg
committed
docs(architecture): add multimodal PDF ingestion diagram
scaleborg
committed
test(eval): add multimodal chat integration test
scaleborg
committed