Introducing the RAG Lab: Test and compare different search configurations to find the optimal setup for your use case.
New RAG strategies to maximize the potential of your information at retrieval time:
1. Context expansion: Add 1 or more text blocks before and after your found information for greater context. Go to docs here
2. Metadata enhancement: Incorporate parameters like origin, title, summary, labels, and fields. Go to docs here
As of October 7th, ChatGPT 3.5 will no longer be available in our LLM selection due to its deprecation on OpenAI's platform. We recommend transitioning to GPT-4 mini for those wishing to continue using OpenAI models. We're grateful for the contributions of ChatGPT 3.5 and look forward to embracing newer technologies.
New tools in the CLI and Python SDK now support moving/copying resources between KBs, ideal for managing development and production environments. Go the docs here
Important API change: page_number and page_size parameters are now deprecated in /search and /find endpoints. Instead, use the new top_k parameter (similar to /ask), which is equivalent to setting page_number=0 and page_size=top_k. For more results, simply increase the top_k value.