Pith. sign in

Title resolution pending

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it

fields

cs.LG 1

years

2026 1

verdicts

UNVERDICTED 1

representative citing papers

Pando: Do Interpretability Methods Work When Models Won't Explain Themselves?

cs.LG · 2026-04-13 · unverdicted · novelty 6.0

Pando shows that when models give no or misleading explanations, gradient-based attribution and relevance patching improve accuracy in predicting held-out model decisions by 3-5 percentage points over black-box methods, while logit lens, sparse autoencoders, and circuit tracing provide no reliable 3

citing papers explorer

Showing 1 of 1 citing paper.

  • Pando: Do Interpretability Methods Work When Models Won't Explain Themselves? cs.LG · 2026-04-13 · unverdicted · none · ref 17

    Pando shows that when models give no or misleading explanations, gradient-based attribution and relevance patching improve accuracy in predicting held-out model decisions by 3-5 percentage points over black-box methods, while logit lens, sparse autoencoders, and circuit tracing provide no reliable 3