API Reference

Engines

The following classes are part of the torch_ttt.engine module:

ttt_engine.TTTEngine	Original image rotation-based test-time training approach.
ttt_pp_engine.TTTPPEngine	TTT++ approach: feature alignment-based + SimCLR loss.
masked_ttt_engine.MaskedTTTEngine	Masked token prediction-based test-time training engine.
actmad_engine.ActMADEngine	ActMAD approach: multi-level pixel-wise feature alignment.
tent_engine.TentEngine	TENT: Fully test-time adaptation by entropy minimization.
eata_engine.EataEngine	EATA: Efficient Test-Time Adaptation without Forgetting.
memo_engine.MemoEngine	MEMO: Test-Time Robustness via Augmentation.
deyo_engine.DeYOEngine	DeYO: Destroy Your Object – Test-Time Adaptation with PLPD.