Function all_gather

modelscope/utils/torch_utils.py:323–357

Run all_gather on arbitrary picklable data (not necessarily tensors). Args: data: any picklable object group: a torch process group. By default, will use a group which contains all ranks on gloo backend. Returns: list[data]: list of data gathered from

(data, group=None)

def all_gather(data, group=None):
    """
    Run all_gather on arbitrary picklable data (not necessarily tensors).
    Args:
        data: any picklable object
        group: a torch process group. By default, will use a group which
            contains all ranks on gloo backend.
    Returns:
        list[data]: list of data gathered from each rank
    """
    if get_world_size() == 1:
        return [data]
    if group is None:
        group = _get_global_gloo_group()
    if dist.get_world_size(group) == 1:
        return [data]

    tensor = _serialize_to_tensor(data, group)

    size_list, tensor = _pad_to_largest_tensor(tensor, group)
    max_size = max(size_list)

    # receiving Tensor from all ranks
    tensor_list = [
        torch.empty((max_size, ), dtype=torch.uint8, device=tensor.device)
        for _ in size_list
    ]
    dist.all_gather(tensor_list, tensor, group=group)

    data_list = []
    for size, tensor in zip(size_list, tensor_list):
        buffer = tensor.cpu().numpy().tobytes()[:size]
        data_list.append(pickle.loads(buffer))

    return data_list


def is_on_same_device(model: torch.nn.Module) -> bool: