OpenChat
OpenChat's 7B community chat model — conditioned reinforcement learning from comparisons achieves GPT-3.5-level performance.
