AQ-MedAI/ClinAlign-30B-A3B

ClinAlign

ClinAlign is a clinician-grounded healthcare alignment framework that scales from instance rubrics to a reusable library of clinical principles, enabling robust preference alignment and inference-time self-revision for medical LLMs.

<p align="center"> 📃 <a href="https://arxiv.org/abs/2602.09653" target="_blank">Paper</a> </p>

<p align="center"> <img src="assets/logo.png" alt="ClinAlign" width="520"/> </p>

🔥 Highlights

• —Clinician-verified rubrics (HealthRubrics): a physician-validated preference dataset built by having clinicians revise and finalize LLM-drafted, checkable rubrics.
• —Reusable principle library (HealthPrinciples): distilled clinician consensus as 119 broadly reusable principles organized by clinical dimensions (urgency / uncertainty / expertise / task type).
• —Scalable supervision: principles can be converted into per-question rubrics for new, unlabeled medical queries—scaling training data without per-instance clinician authoring.
• —Inference-time alignment tool: retrieve matched principles → generate rubric references → guide iterative self-revision at test time.

🥇 Results

🤗 Models: <a href="https://huggingface.co/AQ-MedAI/ClinAligh-4B" target="blank">ClinAlign-4B</a> • <a href="https://huggingface.co/AQ-MedAI/ClinAligh-30B-A3B" target="blank">ClinAlign-30B-A3B</a>

<p align="center"> <img src="assets/result.png" alt="ClinAlign results" width="820"/> </p>

Acknowledgement

• —verl
• —simple-evals