326 B
326 B
metax-c500-vllm
- 支持
gpt-oss:将vllm目录覆盖到镜像中的/opt/conda/lib/python3.10/site-packages/vllm。运行gpt-oss时需指定VLLM_ATTENTION_BACKEND=TRITON_ATTN_VLLM_V1 - 将
code_generator.py覆盖到镜像中的/opt/conda/lib/python3.10/site-packages/triton/compiler/code_generator.py