Tabnine, the Tel Aviv-based AI coding platform company, has announced significant enhancements to its coding assistant. At the heart of these enhancements is the integration of retrieval augmented generation techniques, which enable the platform to deliver finely-tuned suggestions by drawing on specific code and engineering patterns found within a team’s codebase or integrated development environment. RAG is typically used in enterprise applications to reduce hallucinations generated by large language models. By injecting context and factual information into the prompt, RAG ensures that the response from LLMs is grounded.
Read more here.