method · first seen 12 Oct 2026
training-free decoding
Improving model outputs at inference time through decoding rules without any additional training.
Mentions per week
PapersCodeDiscussion
Earliest sources
- Intent-Triggered Anchoring for Multi-Token Format Units in Parallel Decoding of Diffusion Language ModelsPapers · 12 Oct 2026 · Junhyeok Lee, Kyu Sung Choi
Latest sources
- Intent-Triggered Anchoring for Multi-Token Format Units in Parallel Decoding of Diffusion Language ModelsPapers · 12 Oct 2026 · Junhyeok Lee, Kyu Sung Choi