[fix] [nvbug/5252057] Fix kv cache reuse on PyTorch multimodal#4025

symphonylyh merged 1 commit intoNVIDIA:mainNVIDIA/TensorRT-LLM:mainfrom

yechank-nvidia:kv_cache_multimodalyechank-nvidia/TensorRT-LLM:kv_cache_multimodalCopy head branch name to clipboard