Exploring Pinecone's Sparse-Dense Index

Join Pinecone’s very own James Briggs and Sebastian Bruch for a workshop exploring the ins and outs of our new approach to sparse-dense vector support. Announced recently, our index now supports sparse-dense embeddings regardless of model used. This means you can leverage the latest LLMs to power your search applications and deliver the best results possible.

We’ll walk through examples using both SPLADE and BM25 to generate sparse embeddings, discuss the tradeoffs between the two, and demonstrate how you can easily fine-tune your results upon query. You’ll leave the session with valuable insights and the resources needed to take your search results to the next level.

[j_hull]

Hello! Really interested how to do that sparse-embedding in production (and have it run quick). Can't find any resource anywhere to be able to do this...any ideas?