3 papers
cs.CV2026
Learning Dynamic Evidence Routes for Vision Transformer Probing
Zice Wang, Zhenyu Zhang
Probing frozen vision transformers typically uses permutation-invariant aggregation (GAP or ), treating patch tokens as an unstructured set. Content-dependent probe…
cs.CL2026
Framing Effects in Independent-Agent Large Language Models: A Cross-Family Behavioral Analysis
Zice Wang, Zhenyu Zhang
In many real-world applications, large language models (LLMs) operate as independent agents without interaction, thereby limiting coordination. In this setting, we examine how prom…
cs.LG2026
Spectral Overfitting in Noisy Linear Probing of Pretrained Representations
Zice Wang, Zhenyu Zhang
Frozen pretrained features are often treated as a safe interface for downstream learning: only a small linear readout is trained, while the backbone is fixed. We show that this rea…