perf(glm): write MLA projections directly in token-major layout - #1710

perf(glm): write MLA projections directly in token-major layout#1710

shihaobai merged 2 commits into

mainModelTC/LightLLM:mainfrom

bsh/glm-prefill-mla-token-major-bmmModelTC/LightLLM:bsh/glm-prefill-mla-token-major-bmmCopy head branch name to clipboard