QMD - Quick Markdown Search

xal 6 hours ago | parent | next [-]

hi orange page, author here. QMD is an implementation of the best practices that I picked up in meetings with teams that work in search and retrieval. I tried to make it not overkill and keep things local. Currently working on finetuning better models for query extension and reranking (finetune branch). Feel free to jump in help. Self written and AI PRs welcome.

▲

mpakes 5 hours ago | parent | next [-]

Cranking out open source tools one day, endurance racing in IMSA LMP2 the next — I love it. Condolences about the DNF, but truly an impressive run nonetheless (especially the first-lap recovery). Kudos to you and your team.

	▲	xal 3 hours ago \| parent [-]
		Heh, literally was training qwen3 finetunes during test sessions. Yea cruel sport, but its the low lows allow the high heights.

▲

saltysalt 5 hours ago | parent | prev [-]

Looks neat! Does it require much resources to run locally?

▲

augusteo 3 hours ago | parent | prev | next [-]

Nice work. The hybrid BM25 + vector + LLM reranking approach is exactly what I've been wanting for my notes.

I use Obsidian with hundreds of markdown files. The built-in search is fine for exact matches but misses conceptual connections. Does this handle large vaults well? I'd love to point it at my notes plus meeting transcripts.

▲

ashepp 3 hours ago | parent | prev [-]

Would this be a good candidate for an Obsidian plugin or are there already similar alternatives?

	▲	linkage 15 minutes ago \| parent [-]
		Omnisearch is the closest equivalent: https://github.com/scambier/obsidian-omnisearch