jianwen-wang

### What is the issue? When using the LLaMA 3 model with OpenAI’s /v1/chat/completions API, it was discovered that the request message can’t exceed 2k tokens, while the LLaMA 3...