Scaling RAG and Embedding Computations with Ray and Pinecone
Discover how to efficiently scale retrieval augmented generation (RAG) and embedding computations in this 29-minute conference talk. Learn about the challenges of developing RAG-based LLM applications and the data-intensive processes involved. Explore the powerful combination of Ray Data, an open-source distributed machine learning data processing library, and Pinecone, a leading vector database. Understand how these tools enable the generation of one billion embeddings in less than a day on a limited budget. Gain insights into the fundamentals of Ray Data and Pinecone, and learn how machine learning practitioners can leverage them for RAG embedding computation. Presented by Cheng Su, Manager of Data Team at Anyscale, and Roy Miara, Engineering Manager of Generative Search at Pinecone, this talk provides valuable knowledge for those looking to optimize their LLM applications and embedding processes.