Nice read. I agree that for a lot of hobby use cases you can just load the embeddings from parquet and compute the similarities in-memory. To find similarity b

Nice read. I agree that for a lot of hobby use cases you can just load the embeddings from parquet and compute the similarities in-memory. To find similarity between my blogposts [1] I wanted to experiment with a local vector database and found ChromaDB fairly easy to use (similar to SQLite just a file on your machine). [1] https://staticnotes.org/posts/how-recommendations-work/ https://staticnotes.org/posts/how-recommendations-work/