Reproduce “Decoupled Predictive Decision Making: A Shared Conditional Predictor for 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

Lang Mo, Xingxuan Zhang, Qi Wang, Mingchao Hao, Hao Yuan, Chunjing Gan, Wang Lu, Peng Cui

DPDM separates semantic representation from inference over labeled history: a frozen language model encodes cases, and one shared conditional predictor estimates targets from historical labels. Using Qwen3.5-27B, it achieves 82.33% mean accuracy versus 78.80% for few-shot prompting across 50 benchmarks with identical histories, labels, and queries. This supports an advantage in the evaluated setup, not every task or model pairing.

Summary by alphaXiv.

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