Update README.md

README.md

@@ -6,4 +6,10 @@ tags: []
 
 This is a transformers-compatible `llava-critic-7b` model converted from [lmms-lab/llava-critic-7b](https://huggingface.co/lmms-lab/llava-critic-7b/tree/main) using [convert_llava_onevision_weights_to_hf.py](https://github.com/huggingface/transformers/blob/main/src/transformers/models/llava_onevision/convert_llava_onevision_weights_to_hf.py).
 
-However, there may be some **precision problems** to be fixed. **See [#34467](https://github.com/huggingface/transformers/issues/34467) for details.**
+However, there may be some **precision problems** to be fixed. **See [#34467](https://github.com/huggingface/transformers/issues/34467) for details.**
+
+## Requirements for vLLM
+
+The latest vLLM (0.6.3.post1) has an sever bug when serving models, See [vllm #9848](https://github.com/vllm-project/vllm/issues/9848) for details.
+
+I recommend using `vllm==0.6.2` to avoid this issue.