IAAIWaveFront Decoding
WaveFront Decoding: Parallelized Self-Speculative Decoding for Looped Language Models
Training-free self-speculative decoding that arranges token states at different positions and depths into a diagonal wavefront, so drafting new positions at shallow depth and verifying earlier ones at full depth happen in the same batched call of the shared block.
Inside the method
Diagonal wavefront: shallow drafts ‖ mid-depth ‖ full-depth verification, one batched call
Simplified conceptual schematic. Consult the paper for the complete architecture.
- Recurrence family
- Full-stack recurrence
- Depth control
- Mixed depths per call; full depth for acceptance
- KV / state strategy
- Optional cross-recurrence KV sharing
Reading note
Across six Spec-Bench categories, 2.42× speedup on Ouro-2.6B and 3.54× on Huginn-3.5B over autoregressive decoding, rising to 4.81× on Huginn with cross-recurrence KV sharing.
Sources checked 2026-10-28. This catalog does not imply independent reproduction.
Related works
Cite this work
@misc{wavefrontdecoding2026,
title = {WaveFront Decoding: Parallelized Self-Speculative Decoding for Looped Language Models},
author = {Hyeongju Ha and Jae-Joon Kim},
year = {2026},
eprint = {2609.23033},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.23033}
}