Halve Your Multi-Vector Index With HierarchicalTokenPooling at 100.6% BEIR
tomaarsen · x · 2026-08-18
Late interaction is the state of the art for visual document retrieval: text queries against page images with charts and tables intact, no OCR — ColPali-family checkpoints run through the exact same two calls, with MaxSim scoring query text tokens against image patches. The post also covers HierarchicalTokenPooling (Clavié, Chaffin & Adams), which clusters each document's token vectors with Ward linkage and keeps 1/poolfactor of them: poolfactor=2 halves the index at 100.6% of unpooled BEIR performance. It can be applied per call, standalone, or baked into the model.
Related event: Sentence Transformers v6.0 Ships Late-Interaction Multi-Vector Models(27 posts)→
More from coding & agent
- Practical Prompt: Search and buy domains using GoDaddy MCP — ifioknkem · 2026-08-18
- Prompt 7: deploy to Netlify, self-test like a visitor, then ship — ifioknkem · 2026-08-18
- Practical Prompt: Have Claude audit and refine your website design — ifioknkem · 2026-08-18
- Practical Prompt: Build a complete responsive website with Claude — ifioknkem · 2026-08-18
- Prompt 2: have Claude write all website copy, section by section — ifioknkem · 2026-08-18
- Prompt 1 of the Claude website series: full site strategy and structure — ifioknkem · 2026-08-18