Torch 数据加载器工具

collate_sparse_matrix_batch(batch)

用于从稀疏张量创建批次的整理函数。

这是整理各种长度的稀疏矩阵所必需的。

参数

名称	类型	描述	默认值
batch	list[Tensor]	要整理成批次的张量列表。	必需

返回值

类型	描述
Tensor	整理成 CSR（压缩稀疏行）格式的张量。

源代码位于 bionemo/scdl/util/torch_dataloader_utils.py

def collate_sparse_matrix_batch(batch: list[torch.Tensor]) -> torch.Tensor:
    """Collate function to create a batch out of sparse tensors.

    This is necessary to collate sparse matrices of various lengths.

    Args:
        batch: A list of Tensors to collate into a batch.

    Returns:
        The tensors collated into a CSR (Compressed Sparse Row) Format.
    """
    batch_rows = torch.cumsum(
        torch.tensor([0] + [sparse_representation.shape[1] for sparse_representation in batch]), dim=0
    )
    batch_cols = torch.cat([sparse_representation[1] for sparse_representation in batch]).to(torch.int32)
    batch_values = torch.cat([sparse_representation[0] for sparse_representation in batch])
    if len(batch_cols) == 0:
        max_pointer = 0
    else:
        max_pointer = int(batch_cols.max().item() + 1)
    batch_sparse_tensor = torch.sparse_csr_tensor(batch_rows, batch_cols, batch_values, size=(len(batch), max_pointer))
    return batch_sparse_tensor