/EfficientTrain

1.5−3.0× lossless training or pre-training speedup. An off-the-shelf, easy-to-implement algorithm for the efficient training of foundation visual backbones.

Primary LanguagePythonMIT LicenseMIT

Issues