Lang Mo, Xingxuan Zhang, Qi Wang, Mingchao Hao, Hao Yuan, Chunjing Gan, Wang Lu, Peng Cui (Stable AI) · Conditional Prediction Agents
Decoupled Predictive Decision Making (DPDM)
DPDM frames agent decisions as conditional prediction, pairing a language model that encodes current and historical records with a frozen LimiX-2 predictor that needs no parameter updates at deployment. With Qwen3.5-27B as the encoder, mean accuracy across 50 benchmarks rises from 78.80% with few-shot prompting to 84.92% under the evaluated history budgets. The authors also report gains of 34.26% on ToolPRMBench's BFCL and 28.87% on ToolSandbox, and say accuracy improves as history grows.
Open the source ↗Similar Research & Safety releases
- Harness-S: Learning from Sibling Harnesses for Generalizable Agent Training
Kangning Zhang, Wenxiang Jiao, Bowen Fang, Shuai Shao, Jiachen Zhu, Rong Shan, Shijian Wang, Hao Wang (Shanghai Jiao Tong University) · Sibling Harness Training · Week 46, 2026
- CheMeleon
MIT (Zalte, Burns, Abreu, Green, with BASF researchers) · Molecular Foundation Model · Week 46, 2026
- Towards Foundation Decision Models: Drex
Nace.AI (Dhankhar et al.) · Open Decision Models · Week 46, 2026
- Random Is Almost Enough? Stable and Low-Cost Benchmark Compression via BestRand
George Ling, Shanshan Zhong, Lijin Yang, Hao Yang, Nanxiang Li, Zhongzhan Huang (Carnegie Mellon University) · Benchmark Compression · Week 46, 2026
- Agile Deliberation
Google Research (Otilia Stretcu, Leijie Wang) · Interactive Classifier Framework · Week 46, 2026
- AI-PPG-ACC-HTN
Google Research · Smartwatch hypertension AI · Week 46, 2026
Published by Quanta AI Labs, an AI agency in India. Browse every week in AI News Weekly.