Reproduce “LoopFill: Pipelined Multi-Token Speculative Decoding for Looped Transformers” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Siqi Zhu, Suozhi Huang, Yi Zhao, Jiaxuan You
LoopFill accelerates looped-model decoding by combining retrieved continuations and shallow-loop predictions in one draft tree. These guesses start later token computations early, while full-depth verification selects tokens and preserves the autoregressive output distribution. Under greedy decoding, decode-latency speedup averages 5.85× across seven benchmarks for Ouro-2.6B, versus 4.19× for LoopAhead; it does not measure end-to-end throughput.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.