Sort samples by loss and check the head: a practical labeling-error hack

antoine_chaffin · x · 2026-09-20

Following up on his take about noisy human annotation, Antoine Chaffin shares his go-to practice: sort samples by loss and look at the head—there's always something interesting. After finishing any eval, his first ask is to inspect failure cases and judge whether they're real failures to fix or label errors.

Related event: Qwen researcher says human-labeled data may be noisier than auto-labeling(2 posts)→

Original post →

More from Research

Research channel →