avalanche.benchmarks.classic.ExMLCIFAR10

class avalanche.benchmarks.classic.ExMLCIFAR10(scenario='split', run_id=0)[source]

ExML scenario on CIFAR10.

The pretrained models and class splits are taken from https://arxiv.org/abs/2112.06511

__init__(scenario='split', run_id=0)[source]

Init.

Parameters:
  • scenario – If ‘split’, use a class-incremental scenario with 5 experiences (2 classes each). If ‘joint’, use a single experience with all the classes. This should be used only as a baseline since it is not a continual scenario.

  • run_id – an integer in [0, 4]. Each run uses a different set of expert models and data splits.

Methods

__init__([scenario, run_id])

Init.

Attributes

streams