2 papers
cs.LG2026
Certifying Model Upgrades with Slice-Wise Non-Regression and Incumbent Fallback
Shengwei Zhang, Tao Wu, Fei Qian
An updated model can improve an aggregate metric while degrading a slice that matters to a downstream user. We study checkpoint selection subject to non-regression tolerances relat…
cs.CL2026
When Edit Localization Amplifies Relative Selection Bias: Gradient Geometry, Target Mismatch, and Importance Weighting
Shengwei Zhang, Haoda Dai, Yifei Li +1
Human corrections identify editable spans, but the examples receiving corrections may come from a selective feedback channel. We analyze this interaction at a fixed model checkpoin…