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 sample

Info

This module is not included in the test suite or CI, and is generally held to a lower standard than the core abstract_dataloader.

The current extension modules are:

  • augment: A protocol for specifying data augmentations.
  • graph: A programming model based on composing a DAG of callables into a single transform.
  • lightning: A lightning datamodule wrapper for ADL datasets and pipelines.
  • objective: Standardized learning objectives, and a programming model for multi-objective learning.
  • sample: Dataset sampling utilities, including a low-discrepancy subset sampler.
  • types: Type-related utilities which are not part of the core ADL spec.