This work identifies the Logit Conflation Problem, where a to-ken’s logit aggregates prompt-independent factors, including linguistic fluency and parametric associations, with prompt-relevance, and proposes SEAL-Sampling to isolate this component through attention-weighted attribution.
Pinlong Zhao, Huijun Tang, Pengfei Jiao et al.· Annual Meeting of the Associ...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.