[https://nvbugs/5570575][fix] : Use less kv cache memory on SM120#9054

peaceh-nv merged 1 commit intoNVIDIA:release/1.1NVIDIA/TensorRT-LLM:release/1.1from

peaceh-nv:fix-rel-1.1-1peaceh-nv/TensorRT-LLM:fix-rel-1.1-1Copy head branch name to clipboard