Repo for Agent-RLHF
Recently, Agents based on language language models (LLMs) are one of the hottest topics in the research field now. However, these agents rely heavily on the powerful commercial LLMs like GPT-4, and relatively light-weight 13b/7b models still cannot play an important role in this scenario. Thus, we would like to build the first dataset for RLHF targeted at the agent applications, which can help to enhance the ability of open-sourced LLMs in the agent scenarios.
- AgentTuning
- UltraFeedback