Center cropping implementation from ADM. https://github.com/openai/guided-diffusion/blob/8fb3ad9197f16bbc40620447b2742e13458d2831/guided_diff

From https://stackoverflow.com/questions/9860711/cucomplex-h-and-exp and https://forums.developer.nvidia.com/t/complex-number-exponential-function/246

Calculate KID score based on accumulated extracted features from the two distributions. Returns a tuple of mean and standard deviation of KID

Calculate PRDC score based on accumulated extracted features from the two distributions. Implementation inspired by `Fid Score`_

Compute coefficient of x1

Compute coefficient of x1

Special purposed function for computing numerical stabled d_alpha_t / alpha_t

Special purposed function for computing numerical stabled d_alpha_t / alpha_t

Compute the drift term of the SDE

Compute coefficient of x0

Compute coefficient of x0

Take a list of samples (as dictionary) and create a batch, preserving the keys. If `tensors` is True, `ndarray` objects are combined into ten

Forward method of inception net. Copy of the forward method from this file: https://github.com/toshas/torch-fidelity/blob/master/torc

xz: (batch, dim, seqlen)

xz: (batch, dim, seqlen)

xz: (batch, dim, seqlen)

r"""Pass the input through the encoder layer. Args: hidden_states: the sequence to the encoder layer (required). resi

"position_ids" is just to be compatible with Transformer generation. We don't use it. num_last_tokens: if > 0, only return the logits

Wrapper function: transfrom velocity prediction model to denoiser Args: velocity: [batch_dim, ...] shaped tensor; velocity model o

Wrapper function: transfrom score prediction model to velocity Args: score: [batch_dim, ...] shaped tensor; score model output

Set the logits for none top-k values to -inf. Done in-place.