硬件MLU370-X4(1块)

命令：

vllm serve /home/project2/DeepSeek-R1-Distill-Qwen-7B  --served-model-name deepseek-r1:7b --max-model-len 4096 --use-v2-block-manager --tensor-parallel-size 1 --max-num-seqs 32 --max-num-batched-tokens 4096 --enable-chunked-prefill --dtype half  --enforce_eager --gpu_memory_utilization 0.7 --port 6006

报错：

INFO 04-15 09:19:36 mlu_hijack.py:35] Run vLLM in unpaged mode, Apply MLU optimization

INFO 04-15 09:19:36 dump_info.py:65] Cannot get device info

Apply vllm_mlu success, running in performance version !

INFO 04-15 09:20:04 api_server.py:585] vLLM API server version 0.6.4.post1

ERROR 04-15 09:20:12 engine.py:366] Not support chunked_prefill in unpaged mode.

希望将上下文长度突破2048，有什么办法吗？