Hacker News

Turbovec – Google's TurboQuant for vector search in Rust

259 points by fittingopposite 18 hours ago | 32 comments

Eridrus 15 hours ago

FAISS is no longer close to SoTA:

https://ann-benchmarks.com/index.html https://vector-index-bench.github.io/ https://big-ann-benchmarks.com/neurips23.html

nl 12 hours ago

I think their point is the size/performance tradeoff rather than outright performance. The point of TurboQuant is the size savings, while still giving high accuracy.

It's been a while, but I do recall some high-performing vector matching indexes being very large.

ehsanu1 9 hours ago

Surprised that usearch isn't in any of these, it's pretty fast.

ghm2199 17 hours ago

Wow! 4GB for 10 million documents. This means one could build a reverse index much faster than before and devx processes like debugging, performance testing would become much smoother. Can't wait for the sqlite bindings to come out!

ghm2199 17 hours ago

Also the removal latency is on a log scale. Which is quite insane.

nharada 17 hours ago

It would be nice to have the README be a little more human written for a project where you actually want people to adopt it

badatnames 16 hours ago

Anthropic employee. This is what your brain on kool aid looks like

deeviant 16 hours ago

Then again, if the only thing the human doing is bitching about AI use, it's not really that comparatively useful.

righthand 12 hours ago

Sure it is useful, the bitching is canary in the shit software mine. How do you know the software isnt shit if the Readme is shit?

bobmarleybiceps 14 hours ago

people should read turboquant's open review comments: https://openreview.net/forum?id=tO3ASKZlok

esafak 12 hours ago

tl,dr: there is an allegedly better alternative, and it's already implemented everywhere: https://github.com/VectorDB-NTU/RaBitQ-Library#rabitq-in-ind...

sp1982 17 hours ago

If anyone is looking to retrofit to an existing pipeline, I use similar ideas to compress vectors for job search, getting roughly 8x compression with about a 3.5% drop in quality. My experiment: https://corvi.careers/blog/vector-search-embedding-compressi...

lmeyerov 9 hours ago

Interestingly, while we don't fine-tune generative models for Louie.ai, we found fine-tuning embedding models to be a major $ saver. Instead of 1K-2K wide frontier embedding vector lens... Just 64. Huge savings on vector DB $$$.

I'm curious how that works with something like turboquant. Not needed any more, still dominant, better together, ... .

anishvarghese 17 hours ago

This looks perfect for local, privacy first search, but since it's built in Rust, has anyone tried compiling it to WASM to run directly inside a browser extension?

westurner 16 hours ago

oxirs does embeddings and GraphRAG, and full text search with Tantivy; oxirs-vec, oxirs-graphrag

There's an oxirs-wasm with RDF and SPARQL bindings with a query budget. Tantivy-wasm says that the release WASM bundle is 1.5 MB.

cool-japan/oxirs: https://github.com/cool-japan/oxirs

oxirs-wasm: https://crates.io/crates/oxirs-wasm

tantivy-wasm: https://github.com/phiresky/tantivy-wasm

Is there an advantage to adding an MCP local memory interface over agent instructions on how to use a rust CLI?

And then write Markdown documents with Google OKF-like frontmatter YAML metadata for agents that work with tokens not linked data graphs; https://github.com/GoogleCloudPlatform/knowledge-catalog/blo...

coredog64 15 hours ago

Can WASM use AVX512-VNNI?

m00dy 15 minutes ago

nope

LtdJorge 14 hours ago

No, WASM only has 128b SIMD instructions, for now.

cpursley 17 hours ago

Also interested.

mskkm 6 hours ago

There are already several openreview comments alleging academic misconduct around TurboQuant: https://openreview.net/forum?id=tO3ASKZlok

Some write-ups argue that this was deliberate rather than a good-faith mistake: https://dev.to/gaoj0017/turboquant-and-rabitq-what-the-publi...

And now this. Pretty bold AI slop.

cat-whisperer 12 hours ago

What's a good embedding model and search to run locally? something fast and lightweight.

beernet 15 hours ago

Why not just use Qdrant? They've been integrating TurboQuant for months, works well.

kanungle 10 hours ago

Integrated in 5 weeks and just expanded data types for turbo4 in last release. No longer need to store fp32 vectors if you don't need them

OutOfHere 12 hours ago

I am not convinced that Turbovec yields better retrieval than the same amount of bits of a Matryoshka embedding.

burgerboii 17 hours ago

Who is this co-author called t <t@t>?

cute_boi 15 hours ago

As it is heavily vibe coded, I think member of technical staff at antropic has no clue....

Next Prompt: remove t@t and force commit.

refulgentis 16 hours ago

Bloviating nonsense, 3rd time I’ve seen something like this in HN since TurboQuant came out. You don’t need float32, never did. Source: I’ve been writing on device embedding code for 4 years.

spoaceman7777 16 hours ago

Well. That is insane. O_O Fantastic job!

cute_boi 15 hours ago

Another vibe coded slop where they can't even spend time on Readme or documentation around code...

esafak 18 hours ago

lancedb and duckdb integrations would be great...

zuzululu 17 hours ago

what could i use this for as part of my agentic workflow? codebase indexing? docs ?

kyxsc 17 hours ago

notes/docs/wiki is a great use case

myshapeprotocol 12 hours ago

[dead]

anthropic-dario 6 hours ago

[dead]

tracespect 14 hours ago

[flagged]