We are planning to use cohere embeddings and cohere provides int8 embeddings which require 1/4th of float32. Does Pinecone support storing such int8 embeddings at lower storage and memory costs?
And will Pinecone Serverless provide any Product Quantization?