[AMD][MI35X] Bump Qwen3.5 MXFP4 MI355X SGLang AgentX to v0.5.18 and retune all-reduce, prefill, and CUDA graph - #2737

[AMD][MI35X] Bump Qwen3.5 MXFP4 MI355X SGLang AgentX to v0.5.18 and retune all-reduce, prefill, and CUDA graph#2737

yichiche wants to merge 8 commits into

mainSemiAnalysisAI/InferenceX:mainfrom

amd/qwen3.5-fp4-mi355x-sglang-agentic-v0.5.18SemiAnalysisAI/InferenceX:amd/qwen3.5-fp4-mi355x-sglang-agentic-v0.5.18Copy head branch name to clipboard