DCR框架提出科研智能体决策基准 RD-Bench 与校准决策模型 DCRM
Decision-Centric Research: A Calibrated Decision Model and Benchmark for Scientific Agents
给做科研智能体的朋友：这篇把评估粒度拆到单次决策，0.6B 小模型靠状态信息打到 89.4%，比 DeepSeek-V4.1-Flash 还高不少，校准后的不确定性输出也能直接用。
论文提出 Decision-Centric Research（DCR）框架，把科研智能体在证据、行动和策略上的单个决策作为监督与评估的基本单元，共定义 24 种决策类型。基于其中 10 种构建了含 2,000 个规则生成案例的 Research Decision Benchmark（RD-Bench）。配套的 DCRM 模型在一次前向传播中同时输出答案和选项概率，0.6B 的 state-aware 版本在 held-out 研究领域划分下达到 89.4% Top-1 准确率，高于 question-only 版本的 27.8% 和 DeepSeek-V4.1-Flash 的 36.4%。校准后预测集平均含 1.515 个候选，在 97.1% 覆盖率下比 question-only 版本平均缩小 57.4%。
Decision-Centric Research: A Calibrated Decision Model and Benchmark for Scientific Agents
Scientific agents make consequential choices about evidence, next actions and research strategy, but choices embedded in generated text are difficult to evaluate independently. Decision-Centric Research (DCR) is introduced to make individual choices the common unit of supervision, prediction and evaluation through typed predictions over an explicit Research State. DCR defines 24 decision types, ten of which are instantiated in the Research Decision Benchmark (RD-Bench) of 2,000 rule-generated cases. The Decision-Centric Research Model (DCRM) shares a candidate-token readout across types to produce answers and option probabilities in one forward pass. Under a held-out research-domain split, a 0.6B DCRM (state-aware) achieves 89.4% Top-1 accuracy, compared with 27.8% for DCRM (question-only) trained with the same recipe and 36.4% for DeepSeek-V4.1-Flash. All 30 type-seed comparisons against DCRM (question-only) are significant after Holm correction. After held-out calibration, the resulting prediction sets contain a mean of 1.515 candidates at 97.1% empirical label coverage, reducing mean set size by 57.4% relative to DCRM (question-only) at similar coverage. These results establish a compact state-aware decision layer with accurate typed prediction and an inspectable uncertainty interface.