Preprint
Aug 2026
Respect Your Zero-Shot Uncertainty: Conservative Calibration for Test-Time-Adapted Vision-Language Models
It is shown that TTA can increase confidence and reduce entropy even when the top-1 prediction and its correctness remain unchanged, a failure mode the authors term prediction-preserving sharpening, and proposed Zero-Shot-Anchored Entropy Calibration (ZAEC), a label-free post-hoc method that uses zero-shot entropy as a...
Jing-Yan Jiang, Yaru Sun, Xiao Chen et al.
· 0 citations