A decorator for registering corrector classes.

A decorator for registering corrector classes.

A decorator for registering model classes.

A decorator for registering model classes.

A decorator for registering predictor classes.

A decorator for registering predictor classes.

same as matlab rgb2ycbcr only_y: only return Y channel Input: uint8, [0, 255] float, [0, 1]

Distribute the inception network computation to all available TPUs. Args: input_tensor: The input images. Assumed to be within [0, 255]. in

Distribute the inception network computation to all available TPUs. Args: input_tensor: The input images. Assumed to be within [0, 255]. in

Compute the sample at time `t_end` by DPM-Solver, given the initial `x` at time `t_start`. =========================================

Scale the parameters of a module and return it.

Scale the parameters of a module and return it.

Scale the parameters of a module and return it.

Create the drift and diffusion functions for the reverse SDE/ODE.

Create the drift and diffusion functions for the reverse SDE/ODE.

Set logging threshold on current logger.

ShakeDrop eval pass. See https://arxiv.org/abs/1802.02375 Args: x: Input to apply ShakeDrop to. mask_prob: Mask probability. alpha_m

ShakeDrop training pass. See https://arxiv.org/abs/1802.02375 Args: x: Input to apply ShakeDrop to. mask_prob: Mask probability. alp

Shake-shake regularization in testing mode. Args: xa: Input, branch A. xb: Input, branch B. Returns: Mix of input branches.

Shake-shake regularization in training mode. Shake-shake regularization interpolates between inputs A and B with *different* random uniform (per-

A wrapper tha configures and returns the update function of correctors.

A wrapper tha configures and returns the update function of correctors.

A wrapper that configures and returns the update function of predictors.

A wrapper that configures and returns the update function of predictors.

split the large images from original_dataroot into small overlapped images with size (p_size)x(p_size), and save them into taget_dataroot; on

blur + bicubic downsampling Args: x: HxWxC image, [0, 1] k: hxw, double sf: down-scale factor Return: downsam

blur + bicubic downsampling Args: x: HxWxC image, [0, 1] k: hxw, double sf: down-scale factor Return: downsam

Running one step of training or evaluation. This function will undergo `jax.lax.scan` so that multiple steps can be pmapped and jit-compiled toge

Running one step of training or evaluation. This function will undergo `jax.lax.scan` so that multiple steps can be pmapped and jit-compiled toge

Converts a torch Tensor into an image Numpy array of BGR channel order Input: 4D(B,(3/1),H,W), 3D(C,H,W), or 2D(H,W), any range, RGB channel

Convert [0.001, 1] to [0, 999].

Convert [0, 999] to [0.001, 1].

Convert [0.001, 1] to [0, 1].

Runs the training pipeline. Args: config: Configuration to use. workdir: Working directory for checkpoints and TF summaries. If this

Runs the training pipeline. Args: config: Configuration to use. workdir: Working directory for checkpoints and TF summaries. If this

Update target parameters to be closer to those of source parameters using an exponential moving average. :param target_params: the targe

Update target parameters to be closer to those of source parameters using an exponential moving average. :param target_params: the target

One update of the predictor. Args: x: A PyTorch tensor representing the current state t: A Pytorch tensor representing the current ti

One update of the corrector. Args: x: A PyTorch tensor representing the current state t: A PyTorch tensor representing the current ti