Projects the params to be within lp_bound (according to an lp) of the identity map. First thing we do is clip the params to be

Projects the params to be within lp_bound (according to an lp) of the identity map. First thing we do is clip the params to be

Get the distribution q(x_t | x_0). :param x_start: the [N x C x ...] tensor of noiseless inputs. :param t: the number of dif

Restore the parameters stored with the `store` method. Useful to validate the model with EMA parameters without affecting the original op

Runs an attack with targets randomly selected from all classes besides the correct one. The attack should be a function from (inputs, labels)

Returns a tensor of an entity, which may or may not already be a tensor

Assigns a random name to this minibatch and saves the examples and labels in two separate files: <random_name>.examples.npy a

Sets the environment variable 'MISTER_ED_GPU'. Defaults to using gpu if cuda is available ARGS: manual : bool - we set the 'MISTE

Create a list of timesteps to use from an original diffusion process, given the number of timesteps we want to take from equally-sized portio

Computes the norm used in "Spatially Transformed Adversarial Examples"

Save the current parameters for restoring later. Args: parameters: Iterable of `torch.nn.Parameter`; the parameters to be temp

Modifies the constant scaling that we keep to weight f_adv vs D(.) in our loss function. IF the attack was successful

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

Update the reweighting using losses from a model. Sub-classes should override this method to update the reweighting using lo

Update the reweighting using losses from a model. Call this method from each rank with a batch of timesteps and the correspo

Fused `upsample_2d()` followed by `tf.nn.conv2d()`. Padding is performed only once at the beginning, not between the operations. The f

Prints out validation values interim for use in iterative techniques ARGS: new_examples: Variable (NxCxHxW) - [0.0, 1.0] images t

Ported from JAX.

Get a numpy array of weights, one per diffusion step. The weights needn't be normalized, but must be positive.

End time of the SDE.