Shrinks learning rate by a specified factor. :param optimizer: optimizer whose learning rate must be shrunk. :param shrink_factor: facto

Saves model checkpoint.

Set random seed for reproducible experiments Inputs: seed number

Set random seed for reproducible experiments Inputs: seed number

Initialize the weights

Prunes heads of the model. heads_to_prune: dict of {layer_num: list of heads to prune in this layer} See base class PreTrainedModel

Computes top-k accuracy, from predicted and true labels. :param scores: scores from the model :param targets: true labels :param k:

Clips gradients computed during backpropagation to avoid explosion of gradients. :param optimizer: optimizer with the gradients to be clippe

r""" Returns: Example: ```python # Assumption: *get_visual_embeddings(image)* gets the visual embeddings of the ima

r""" labels (`torch.LongTensor` of shape `(batch_size, total_sequence_length)`, *optional*): Labels for computing the masked langu

r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): Labels for computing the multiple choice classification los

r""" labels (`torch.LongTensor` of shape `(batch_size, total_sequence_length)`, *optional*): Labels for computing the sequence cla

r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): Labels for computing the sequence classification/regression

r""" region_to_phrase_position (`torch.LongTensor` of shape `(batch_size, total_sequence_length)`, *optional*): The positions depi

Saves model checkpoint. :param data_name: base name of processed dataset :param epoch: epoch number :param epochs_since_improvement: