Adaptive Multi-Step Lookahead Decoding for Diffusion Language Models
Masked diffusion language models (DLMs) enable parallel text generation by iteratively refining masked tokens, offering a promising alternative to autoregressive decoding. Recent lookahead-based decoding methods improve the accuracy--efficiency trade-off by exploring future decoding states before committing token updates. However, existing approaches mainly rely on shallow one-step lookahead, which optimizes immediate information gain but can be suboptimal for longer-horizon decoding trajectories. Meanwhile, we find that a naive extension for deeper lookahead is also ineffective, as fixed-dept
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- PossiblePossibly related (embedding) · 56%DiffusionGemma: 4x faster text generation →
- PossiblePossibly related (embedding) · 54%What if context compression is a diffusion noise function? Proposal + honest results from untrained-model experiments [R] →
- PossiblePossibly related (embedding) · 50%New Server Hopes to Break Through AI’s “Memory Wall” →
- LinkedLinked via arxiv author · 85%Yingqian Cui →
“Adaptive Multi-Step Lookahead Decoding for Diffusion Language Models”
- LinkedLinked via arxiv author · 85%Xinwei Deng →
“Adaptive Multi-Step Lookahead Decoding for Diffusion Language Models”
- LinkedLinked via arxiv author · 85%Lantao Mei →
“Adaptive Multi-Step Lookahead Decoding for Diffusion Language Models”
- LinkedLinked via arxiv author · 85%Chang Liu →
“Adaptive Multi-Step Lookahead Decoding for Diffusion Language Models”
- LinkedLinked via arxiv author · 85%Charu C. Aggarwal →
“Adaptive Multi-Step Lookahead Decoding for Diffusion Language Models”