DMax: Aggressive Parallel Decoding for dLLMs

arxiv:2604.08302

Original source · Paper JSON

LLMs: architectures, context, training · 2026-04-09

Cites (4) · Cited by (0)

Complete cleaned Markdown · Source Markdown including bibliography

Read in parts

1. Part 1
2. Part 2
3. Part 3

Contents

• DMax: Aggressive Parallel Decoding for dLLMs
• Abstract
• 1 Introduction
• 2 Preliminaries
• 3 Methodology
• 3.1 On-Policy Uniform Training
• 3.2 Soft Parallel Decoding
• 4 Experiments
• 4.1 Experimental Setup
• 4.2 Experimental Results
• 5 Ablation Study
• 6 Related Work
• 7 Conclusion