Description: Try out ChatProtect, a tool that automatically detects and revises hallucinations from large language models. Visit our website to explore examples or interact with the tool in a live demo. Enjoy protected chatting!
chat (6479) chatgpt (1172) revision (488) mitigation (171) gpt-4 (108) gpt-3 (78) language models (14) vicuna (7) contradictions (6) hallucinations (5)
ChatProtect detects and removes hallucinations of LLMs when you chat with them.
Try out some examples or your own live chat: Biography Scientific History QA Sportsī¸ Math Adversarial Breakout Live Chat
ChatProtect is a project by the Secure, Reliable, and Intelligent Systems Lab at ETH Zürich.