Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

On the Rejection Criterion for Proxy-based Test-time Alignment

About

Recent works proposed test-time alignment methods that rely on a small aligned model as a proxy that guides the generation of a larger base (unaligned) model. The implicit reward approach skews the large model distribution, whereas the nudging approach defers the generation of the next token to the small aligned model when the large base one is unconfident about its outcome. In this work, we first show that both approaches can be reduced to sampling from similar graphical models, where they differ only in the definition of a rejection criterion (or distribution). Moreover, we argue that the confidence criterion is ill-motivated due to linguistic phenomena like ambiguous phrasing. We propose a novel rejection criterion based on a conservative confidence bet. Experimentally, our novel approach outperforms previous work on several datasets.

Ayoub Hammal, Pierre Zweigenbaum, Caio Corro• 2026

Related benchmarks

TaskDatasetResultRank
Math ReasoningGSM8K
Accuracy82.4
254
Commonsense ReasoningARC Challenge
Accuracy90.5
243
Commonsense ReasoningCommonsenseQA
Accuracy (pass@1)79.3
108
Mathematical ReasoningMATH 500
Accuracy (pass@1)61.8
88
Math ReasoningSVAMP
Accuracy90.3
85
Showing 5 of 5 rows

Other info

Follow for update