Semantic/Hybrid Search in the Browser

RestartKernel 7 hours ago | parent | next [-]

This is very good. Super snappy and gives me what I'm looking for. Great write-up too.

How much more would it cost to use a multilingual tokeniser? I realise this is somewhat pointless for an English blog, but services like Uber Eats have something like that going on (server-side of course, and they may just be translating queries).

	▲	bartdegoede an hour ago \| parent [-]
		It’s gets much larger (there’s the potion-multilingual-128M model, distilled from bge-m3 which has like 100 languages), so it gets to ~120m range for the lookup table (4x smaller if you’d do int8 quantization, so you could get to to 30-35mb).

▲

rdksu 7 hours ago | parent | prev | next [-]

This is a very exciting field to make stuff in! I also experimented with a completely client side wallpaper search xD,a while back.

In case some one is interested here is it : https://web-inky-ten-60.vercel.app/

▲

docheinestages 7 hours ago | parent [-]

I think it's good etiquette to ask the user if they want to download the model or any assets larger than the average page size, in case they have limited bandwidth.

	▲	rdksu 6 hours ago \| parent [-]
		Hi ! This was made mostly for me to use. I didnt think much about the UX back then. Will keep it in mind. Thanks for the feedback.

▲

docheinestages 7 hours ago | parent | prev [-]

Every browser should have an embedding API to avoid having to worry about the plumbing and data transfer.

	▲	bartdegoede an hour ago \| parent [-]
		Chrome ships with one (https://news.ycombinator.com/item?id=43973502) but AFAICT it’s only used for history, there’s no api for it. There’s also an included LLM? https://www.thatprivacyguy.com/blog/chrome-silent-nano-insta...