I measured softmax's precision cliff. In fp32, cross-entropy returns exactly zero at margin 16.
Jon Barron posted a paper today — "A Fix for Softmax and Cross-Entropy" — claiming that the standard implementations of softmax and cross-entropy in every major framework have two numerical bugs. Not "imprecision." Bugs, with specific, falsifiable numbers attached: a precision cliff at margin ≈5.5…
Read the full story at DEV Community — Machine Learning ↗
Timeline · 1 report
- 2026-10-23 23:33 · DEV Community — Machine Learning
I measured softmax's precision cliff. In fp32, cross-entropy returns exactly zero at margin 16.