microsoft / Semi-supervised-learning

A Unified Semi-Supervised Learning Codebase (NeurIPS'22)
https://usb.readthedocs.io
MIT License
1.37k stars 182 forks source link

Assertion on pseudolabel #215

Closed mioam closed 3 months ago

mioam commented 7 months ago

When I use "pseudolabel" which does not require a strong_transform, the assertion appears:

AssertionError: alg pseudolabel requires strong augmentation

I wonder if there is something wrong in https://github.com/microsoft/Semi-supervised-learning/blob/6cf697e6968f9f77040a9e0162306436a91a00dc/semilearn/datasets/cv_datasets/datasetbase.py#L59, which should be

assert self.alg in ['fullysupervised', 'supervised', 'pseudolabel', 'vat', 'pimodel', 'meanteacher', 'mixmatch',], f"alg {self.alg} requires strong augmentation"

Maybe "refixmatch" also requires a strong augmentation.

Hhhhhhao commented 5 months ago

Thanks for pointing that out. Would you like to open a PR for making this change?

github-actions[bot] commented 3 months ago

Stale issue message