Method get_seq_length

src/utils/cache.py:282–298

返回缓存状态的序列长度 Args: layer_idx: 可选的层索引 Returns: int: 序列长度

(self, layer_idx=0)

        return keys_to_return, values_to_return

    def get_seq_length(self, layer_idx=0) -> int:
        """
        返回缓存状态的序列长度

        Args:
            layer_idx: 可选的层索引

        Returns:
            int: 序列长度
        """
        is_empty_layer = (
            len(self._quantized_key_cache) == 0  # no cache in any layer
            or len(self._quantized_key_cache) <= layer_idx  # skipped `layer_idx` and hasn't run a layer with cache after it
            or not self._quantized_key_cache[layer_idx].numel()  # the layer has no cache
        )
        layer_seq_length = self._quantized_key_cache[layer_idx].shape[-2] if not is_empty_layer else 0
        return layer_seq_length

    def clear_query(self):
        """