Register Tokens Let Diffusion LLMs Reason Across Cleared Context, Gains Up to 19.5 Pts

sprocket-lab · hf · 2026-09-17

Approach

Masked diffusion LLMs (dLLMs) generate text via iterative denoising with bidirectional attention. Cross-chunk reasoning normally requires keeping earlier output in context. This work instead uses a few register tokens: fixed-position tokens whose continuous hidden states are trained to carry reasoning progress.

Training and results

Related event: Register Tokens Let Diffusion LLMs Keep Reasoning After Clearing Context(2 posts)→

Original post →

More from Research

Research channel →