Integrate the Reward, Don't Sample It: Window Likelihood for Interruptions and Backchannels in Full-Duplex Speech Models
Paper recorded by Signals 4 on 2026-10-26 in cs.CL. Abstract reproduced from arXiv; link to the original below.
Published 2026-10-26 on arXiv · recorded by Signals 4 on 2026-10-27
Category: cs.CL · 自然语言处理 · first seen 2026-10-27
Abstract
A full-duplex spoken dialogue model must yield when the user cuts in and hold when the user merely backchannels. Sequence-level RL with a timing reward (GRPO) is the usual recipe; we test its whole-rollout form on four open models and find two failures. A binary window reward gives zero advantage on exactly the windows that most need learning (a zero-gradient trap), and the one score per rollout i
Read on arXiv →
Cite this page: Integrate the Reward, Don't Sample It: Window Likelihood for Interruptions and Backchannels in Full-Duplex Speech Models: the #150 most recent of 660 cs.CL papers we have recorded (as of 2026-10-26). Source: Signals 4 (Signals API) — https://data.jiangzhang.ca/signals4/t/papers/integrate-the-reward-don-t-sample-it-window-likelihood-for-interruptions-and-bac.html
Free to quote with attribution to “Signals 4 (Signals API)”. Machine-readable:
papers.json
Get 4 AI signals a day by email — free.
Get 4 AI signals a day by email — free