Dominik Weckmüller
Dominik Weckmüller
Folks, it's here! 🥳 I added webgpu support in the new branch and it's fast! There was a simple problem in the old code where I would call `Promise.all()` for...
Finally managed to come up with the correct GitHub Action. - You can now find the WebGPU app here: https://do-me.github.io/SemanticFinder/webgpu/ - Normal page uses wasm: https://do-me.github.io/SemanticFinder/
According to https://huggingface.co/spaces/Xenova/webgpu-embedding-benchmark/ you usually get even better speed-ups when processing in batches. At the moment, the naive logic in SemanticFinder just processes one single chunk at a time which...
It's easy, you simply didn't clone/check out the webgpu branch yet (if I see correctly). If you add the branch to your repo it will work.
> According to https://huggingface.co/spaces/Xenova/webgpu-embedding-benchmark/ you usually get even better speed-ups when processing in batches. At the moment, the naive logic in SemanticFinder just processes one single chunk at a time...
As WebGPU has been supported for quite a while now under https://do-me.github.io/SemanticFinder/webgpu/ I'll close this issue. My considerations about batch processing should probably best be continued in https://github.com/do-me/SemanticFinder/issues/49
Sure! What algorithm are you referring to particularly? For dimensionality reduction, I was considering t-SNE, PCA and UMAP. They all have pros and cons with PCA being great and widely...
fyi: there is also https://github.com/tantaraio/voy, Rust-based wasm DB as alternative to Orama. However seems a little dead? Seeing all these projects I think we created a pretty solid "vector DB"...
Here's another hot candidate for crazy speed improvements on the indexing side: static models with model2vec: https://github.com/MinishLab/model2vec/issues/75. Curious how to run this in JS.
Fyi: lancedb seems like the best file-based vector DB out there (https://github.com/lancedb/lancedb), similar to sqlite-vec but with more functionality (full-text search etc.). Seems superior to voy and is also written...