Forward function. Args: x: Input feature, tensor size (B, C, D, H, W).

Forward function.

Forward function.

Forward inference. Returns predictions of validation, testing. Args: inputs (torch.Tensor): batch input tensor collated by

Forward training. Returns dict of losses of training. Args: inputs (torch.Tensor): batch input tensor collated by

Generator Path Regularization. Path regularization is proposed in StyleGAN2, which can help the improve the continuity of the latent space. M

Location of cropped area in original image

Get sampler for normal metrics. Directly returns the dataloader. Args: model (nn.Module): Model to evaluate. dataload

Get sampler for generative metrics. Returns a dummy iterator, whose return value of each iteration is a dict containing batch size and

Get the device of a module. Args: module (nn.Module): A module contains the parameters. Returns: torch.device: The device of

Convert a grayscale image to BGR image. Args: img (ndarray): The input image. Returns: ndarray: The converted BGR image.

Convert a grayscale image to RGB image. Args: img (ndarray): The input image. Returns: ndarray: The converted RGB image.

This the function to fetch gt label. Returns: LabelData: gt label.

Convert an image from the src colorspace to dst colorspace. Args: img (ndarray): The input image. src (str): The source colorspac

Numpy array to tensor. HWC to CHW Args: imgs (list[ndarray] | ndarray): Input images. bgr2rgb (bool): Whether to change bgr to rg

Initialize the weights in backbone. Args: pretrained (str, optional): Path to pre-trained weights. Defaults to No

Does a linear interpolation of two parameters/ buffers. Args: a (torch.Tensor): Interpolation start point, refer to orig state.

Create a masked version of a given loss function. To use this decorator, the loss function must have the signature like `loss_func(pred, targ