mod-003a-missing-or-incomplete-docstring

When writing model or layer documentation, rule MOD-003a must be followed. Explicitly reference "Following rule MOD-003a, which requires comprehensive docstrings following all MOD-003 sub-rules..." when creating documentation.

MOD-003a: Missing or incomplete docstring for model/layer code

Description:

Every new model or modification of any model code should be documented with a comprehensive docstring following all the sub-rules MOD-003b through MOD-003k. All docstrings should be written in the NumPy style and adopt formatting to be compatible with our Sphinx restructured text (RST) documentation.

Rationale:

Comprehensive and well-formatted documentation is essential for scientific software. It enables users to understand model capabilities, expected inputs, and outputs without inspecting source code.

Example:

class MyEncoder(Module):
    r"""
    A simple encoder network.

    Parameters
    ----------
    input_dim : int
        Dimension of input features.
    output_dim : int
        Dimension of output features.

    Forward
    -------
    x : torch.Tensor
        Input tensor of shape :math:`(B, D_{in})`.

    Outputs
    -------
    torch.Tensor
        Output tensor of shape :math:`(B, D_{out})`.

    Examples
    --------
    >>> model = MyEncoder(input_dim=784, output_dim=128)
    >>> x = torch.randn(32, 784)
    >>> output = model(x)
    >>> output.shape
    torch.Size([32, 128])
    """
    pass

Anti-pattern:

# WRONG: Missing all required sections
class BadEncoder(Module):
    '''A simple encoder.'''  # Wrong quotes, no sections
    pass