BaT: Towards Self-Evolving Medical Research Agent with Stage Rubrics
2026-08-17 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors developed a system called Benchmark-as-Teacher (BaT) to help AI agents improve themselves when working on complex medical imaging tasks. BaT breaks down workflows into separate stages and uses special methods to evaluate and train the AI more effectively, focusing on stages where it struggles. Their approach showed big improvements over previous AI models on a medical imaging benchmark. This means the AI can better learn from its own mistakes and get smarter over time in handling healthcare-related tasks.
long-horizon agentsmedical imaging workflowsbenchmarkingcurriculum reinforcement learningpolicy updateGRPOAutoMedBench-Litestage-level rubricself-improvementagent training
Authors
Junqi Liu, Yufan He, Yexiao He, Pengfei Guo, Dong Yang, Andriy Myronenko, Can Zhao, Hanrong Ye, Tianhao Qi, Yuyin Zhou, Daguang Xu, Yucheng Tang
Abstract
Long-horizon agents are beginning to automate complete workflows that produce code, reports, and research artifacts. Medical imaging workflows are multi-stage and data-sensitive, while expert trajectories remain scarce and difficult to share. Structured benchmarks can localize failures through stage-level rubrics, but standard post-training discards these diagnostics before the next training round. We present Benchmark-as-Teacher (BaT), a recursive self-improvement system for agent post-training. BaT contains two linked components: the asynchronous Stage Bank data pipeline and BiCuRL (Bilevel Curriculum Reinforcement Learning), its self-improving post-training method. Stage Bank synthesizes content-isolated training states outside the policy-update loop. BiCuRL uses a fixed held-out evaluation to select the next stage curriculum, verifies rollouts with task rubrics, updates the policy with GRPO, and returns the candidate checkpoint to evaluation. On AutoMedBench-Lite, BaT-4B and BaT-9B more than double the Overall scores of their Qwen Instruct baselines. BaT-9B Agent reaches 79.6 Overall, exceeding Claude Opus 4.6 with Claude Code at 77.5.