We empirically demonstrate that popular CLIP adaptation approaches, such as Adapters, Prompt Learning, and Test-Time Adaptation, substantially degrade the calibration capabilities of the zero-shot baseline in the presence of distributional drift. We identify the increase in logit ranges as the underlying cause of miscalibration of CLIP adaptation methods, contrasting with previous work on calibrating fully-supervised models. Motivated by these observations, we present a simple and model-agnostic solution to mitigate miscalibration, by scaling the logit range of each sample to its zero-shot prediction logits
ECCV 2024 Paper: is currently pursuing his Ph.D. in developing reliable deep learning models. Earlier, he completed his master’s thesis on accelerating MRI reconstruction. He has published 25+ research articles in renowned venues.
Recorded on Nov 19, 2024
SOCIAL SHARE CARD GENERATOR