GSS BUFFER#

This module contains a version of the reservoir buffer that is specifically designed for the GSS model.

Classes#

class utils.gss_buffer.Buffer(buffer_size, device, minibatch_size, model=None)[source]#

Bases: object

Memory buffer for the GSS model. The buffer supports only examples and labels tensors.

add_data(examples, labels=None)[source]#

Adds the data to the memory buffer according to the reservoir strategy.

Parameters:

  • examples – tensor containing the images

  • labels – tensor containing the labels

  • logits – tensor containing the outputs of the network

  • task_labels – tensor containing the task labels

drop_cache()[source]#
empty()[source]#

Set all the tensors to None.

functional_reservoir(x, y, batch_c, bigX=None, bigY=None, indices=None)[source]#
get_all_data(transform=None)[source]#

Return all the items in the memory buffer.

Parameters:

transform (<module 'torchvision.transforms' from '/opt/hostedtoolcache/Python/3.10.14/x64/lib/python3.10/site-packages/torchvision/transforms/__init__.py'>) – the transformation to be applied (data augmentation)

Returns:

a tuple with all the items in the memory buffer

Return type:

Tuple

get_data(size, transform=None, give_index=False, random=False, device=None)[source]#

Random samples a batch of size items.

Parameters:

  • size (int) – the number of requested items

  • transform (<module 'torchvision.transforms' from '/opt/hostedtoolcache/Python/3.10.14/x64/lib/python3.10/site-packages/torchvision/transforms/__init__.py'>) – the transformation to be applied (data augmentation)

Returns:

a tuple with the requested items

Return type:

Tuple

get_grad_score(batch_x, batch_y, X, Y, indices)[source]#
init_tensors(examples, labels)[source]#

Initializes just the required tensors.

Parameters:

  • examples (Tensor) – tensor containing the images

  • labels (Tensor) – tensor containing the labels

  • logits – tensor containing the outputs of the network

  • task_labels – tensor containing the task labels

is_empty()[source]#

Returns true if the buffer is empty, false otherwise.

Return type:

bool

reset_fathom()[source]#