Method _stack_cache_tensors

KVCOMM/llm/kvcomm_engine.py:698–699

(cache: DynamicCache)

    @staticmethod
    def _stack_cache_tensors(cache: DynamicCache) -> Tuple[torch.Tensor, torch.Tensor]:
        return torch.stack(cache.key_cache), torch.stack(cache.value_cache)

    @staticmethod
    def _placeholder_length(cache: DynamicCache) -> int: