LogitDynamics detects errors in ViT predictions by training a linear probe on layerwise logit values and top-class instability statistics extracted via auxiliary heads.
Learning multiple layers of features from tiny images
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2representative citing papers
Across 3,938 CIFAR-10 runs, CosineAnnealingWarmRestarts leads on mean accuracy while scheduler preference is strongly architecture-dependent; CyclicLR helps some mobile/conv nets but not overall.
citing papers explorer
-
LogitDynamics: Reliable ViT Error Detection from Layerwise Logit Trajectories
LogitDynamics detects errors in ViT predictions by training a linear probe on layerwise logit values and top-class instability statistics extracted via auxiliary heads.
-
Systematic Evaluation of Learning Rate Scheduling Strategies Across Heterogeneous Architectures
Across 3,938 CIFAR-10 runs, CosineAnnealingWarmRestarts leads on mean accuracy while scheduler preference is strongly architecture-dependent; CyclicLR helps some mobile/conv nets but not overall.