docker拉取sglang“harbor.sourcefind.cn:5443/dcu/admin/base/sglang:0.5.10rc0-ubuntu22.04-dtk26.04-py3.10-20260410镜像后。
运行蚂蚁·安诊儿医疗大模型量化版(链接“https://www.modelscope.cn/models/MedAIBase/AntAngelMed-INT4”)
python3 -m sglang.launch\_server
--model-path ./MedAIBase/AntAngelMed-INT4
--host 0.0.0.0 --port 30012
--trust-remote-code
--attention-backend fa3
--mem-fraction-static 0.9
--tp-size 1
显示缺少“No module named 'aiter'”,但是aiter不支持海光的卡
RuntimeError: Get GPU arch from rocminfo failed "Unknown GPU architecture: gfx928. Supported architectures: ['native', 'gfx90a', 'gfx908', 'gfx940', 'gfx941', 'gfx942', 'gfx945', 'gfx1100', 'gfx950']