WebGLM is an official implementation of a KDD 2023 paper that enhances large language models with web search capabilities for question answering. The system is built on ChatGLM and trained using Reinforcement Learning from Human Feedback (RLHF). It retrieves relevant web information to augment the LLM’s responses, improving factual accuracy and coverage. Models are available in 10B and 2B parameter versions on HuggingFace.