• CriticGPT is a new AI model from OpenAI designed to identify mistakes in code generated by ChatGPT. Based on the GPT-4 family of large language models, it analyzes code and points out potential errors, making it easier for humans to spot mistakes that might otherwise go unnoticed. CriticGPT was trained on a large number of inputs containing deliberately inserted mistakes. Its capabilities extend beyond just code review - it is able to identify errors made by ChatGPT in some other tasks as well. OpenAI plans to integrate CriticGPT-like models into its RLHF labeling pipeline to provide its trainers with AI assistance.

    Friday, June 28, 2024
  • OpenAI has trained a model based on GPT-4 that can catch GPT-4 mistakes and help humans spot mistakes during the RLHF process.