Reproduce “TomasuLLM: Out-of-Order Speculative Execution for LLM Agents” 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

Jiangnan Yu, Ceyu Xu, Mengming Li, Shiyu Huang, Yiran Xia, Jian Weng, Hui Xue, Haohui Mai and 1 more

TomasuLLM overlaps coding-agent tool work by drafting later calls and running eligible candidates in private sandboxes, while validating actions, dependencies, observations, and effects before in-order commit. Unsupported calls remain serial. Against Serial Pi, it reports 1.31× and 1.35× geometric-mean speedups on two benchmarks, and 1.27× matched-progress mean on SWE-Marathon, at added compute.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.