In-Browser Semantic AI Search - free and open source!

[@Supabase]

Ready to store your embeddings on the server? 👉 database.new

[@kirilkirchev285]

Cool & useful as always. Such videos really boosts my knowledge over new technologies and applying them in my apps. Thanks Supabase team!

[@dshukertjr]

This is awesome! I always wanted to create something cool with transformers.js

[@kevinkkirimii]

Dog is food 😊

[@catoshimeowmoto]

Reminds me of the "Not a hot dog 🌭" from Silicon valley tv series 😂

[@imdevbutok]

Dependendo das circunstâncias tudo é comida.

[@takeshikriang]

Super cool❤

[@ctbwx]

I don't think "sleeping" is what the model was referring to for the "bed" result on the "fun" query... 😏

[@Supabase]

Oh, you mean like jumping on the bed? Yah, that's fun too 😉

[@StephenRayner]

Haha

[@StephenRayner]

Hmm 🤔 thoughts on using a generated column for the vector?

[@allanandliftedhands2669]

Does this add any additional costs from supabase or its just the cost for the llm one will be using. Also are the dependency kotlin/android compatible?

[@Supabase]

This specific example is a static / client-side only application and doesn't incur any cost except for the electricity powering the computer it is running on. If you wanted to store the embeddings in a server environment somewhere, then you would need a Postgres instance for example, which you can get on the free tier with Supabase 👉 database.new

[@eliglanz]

I wonder tho- any way to essentially restrict the semantic search not to return something far fetched if the available options are not a 'real' match? Because it essentially performs ranking in order to return values. And it will return 3 items, even tho they have no semantic sense, since they're the closest available...

[@andy111007]

any plans to have python version of it?

[@intebuddy]

Everywhere I go, I see his face (claude).

[@eliglanz]

If you sell enough databases, you could buy yourself a boat..