Sample gaussian tensor from a Gaussian distribution parametrized with mu and logsigma Args: mu: mean of the Gaussian logsigma: st

Save model in checkpoints Args: model (nn.Module): model args: command line arguments Returns: None: This function d

TRAINING Args: model (MnistDPL): network dataset (BaseDataset): dataset Kandinksy _loss (ADDMNIST_DPL): loss function

Where each tensor is batched

Create sinusoidal timestep embeddings. :param t: a 1-D Tensor of N indices, one per batch element. These ma

This function performs a hyper-parameter tuning of the model using a WandB sweep. Args: args: parsed command line arguments

Wandb log epoch metrics: accuracy on labels, accuracy on concepts, learning rate, test loss Args: kwargs: key-value dictionary with metric

Wandb log loss step Args: i (int): iteration number epoch (int): iteration number loss: loss value losses (default

Addmnist concept match loss Args: out_dict: output dictionary args: command line arguments Returns: loss: loss value

Addmnist cumulative loss Args: out_dict: output dictionary args: command line arguments Returns: loss: loss value

Addmnist rec class Args: out_dict: output dictionary args: command line arguments Returns: loss: loss value

Kandinsky cumulative loss Args: out_dict: output dictionary args: command line arguments Returns: loss: loss value

Xor cumulative loss Args: out_dict: output dictionary args: command line arguments Returns: loss: loss value

SDDOIA cumulative loss Args: out_dict: output dictionary args: command line arguments Returns: loss: loss value

Xor cumulative loss Args: out_dict: output dictionary args: command line arguments Returns: loss: loss value

Args: index (int): Index Returns: tuple: (sample, target) where target is class_index of the target class.

Retrieves the sample at the specified index. Args: idx: Index of the sample to retrieve. Returns: T

For program termination on cluster Args: signum (int): signal number frame: frame Returns: None: This function does

Init method Args: self: instance Returns: None: This function does not return a value.

Initialization method Args: self: instance loss: loss value nr_classes (int, default=19): number of outpu

Initialize method Args: self: instance loss: loss function nr_classes: number of classes Returns