LRSampler

class jlpy.pytorch.base.linear_regression.base_impl.LRSampler

Bases: Sampler[list[int]]

A custom torch batched Sampler - LRSampler.

• Custom Sampler:

  • Controlling how to shuffle.

  • Return the size for iteration.

• Custom batched sampler:

  • Control how to shuffle.

  • Control batch size.

  • Control the drop_last.

  • Return the size for batched iteration.

• Sampler __iter__():

  • Return a list of indices.

  • Using the list of indices to get elements from Dataset class.

  • Sampler will call __getitems__() first, then fall back to call __getitem__().

• Dataset class:

  • Sampler class only need to get the len of the dataset.

  • According to the len of the dataset, Sampler will procee all the rest itself.

  • Dataloader will iterate the Sampler and get the samples from Dataset.

Methods

__init__

Construct a class instance.

__init__(dsize: int, batch_size: int = 32, shuffle: bool = False, drop_last: bool = True) → None

Construct a class instance.