Skip to content

abstract_dataloader.ext

ADL specification extensions and non-generic utilities.

Unlike abstract_dataloader.generic, these implementations "extend" the ADL spec by imposing a particular conceptual framework on various functionality.

Warning

This module and its submodules are not automatically imported; you will need to explicitly import them: 

from abstract_dataloader.ext import torch

Info

Not all extension modules are included in the test suite or CI, and these moduels are generally held to a lower standard than the core abstract_dataloader.

Extension Modules

  • augment

    A protocol for specifying data augmentations.

  • graph

    A programming model for composing a DAG of callables into a single transform.

  • lightning

    A lightning datamodule wrapper for ADL datasets and pipelines.

  • objective

    A programming model for training objectives and multi-objective learning.

  • sample

    Dataset sampling utilities, including a low-discrepancy subset sampler.

  • torch

    Pytorch-specific utilities, including a collate function and a torch.nn.Module pipeline.

  • types

    Type-related utilities which are not part of the core ADL spec.