X-Ragent: Tool-Augmented RL Grounds Chest X-Ray Reasoning, +8.6 on ChestAgentBench
CarnegieMellonU · hf · 2026-10-14
Researchers from ICT CAS, CMU, and collaborators propose X-Ragent, a post-training framework for tool-augmented RL on chest X-ray reasoning that decouples learned auxiliary-evidence seeking from validity-gated quantitative measurement acquisition, preventing inapplicable evidence from entering diagnostic reasoning.
They build Rad-CoT, a 7.7K-sample corpus combining causally ordered tool-augmented trajectories with report-style diagnostic reasoning, initialize Qwen3-VL-8B via agentic SFT, then optimize with GRPO using an accuracy-gated reward that applies process-level terms for measurement consistency and tool cost only to correct trajectories—preventing plausible process behavior from compensating for wrong diagnoses.
X-Ragent achieves the strongest results across three CXR benchmarks, including an 8.6-point improvement on the fully held-out ChestAgentBench.
More from Research
- Why OCR models mix scripts: "cruel" and 残酷 have nearly identical embeddings — VikParuchuri · 2026-10-16
- OCR models keep writing the right word in the wrong alphabet — and frontier models do it too — VikParuchuri · 2026-10-16
- Tsinghua professor's ApexIntelligence raises $50M to build self-evolving foundation models — DeryaTR_ · 2026-10-16
- Anders Sandberg on fluid computation: error budgets tighten quadratically over time — anderssandberg · 2026-10-16
- How the fluid computation works: force field advects particle to 0/1 boxes — anderssandberg · 2026-10-16
- FlashAttention creator Tri Dao joins Radical Numerics as scientific advisor — exnx · 2026-10-16