MolLean checks chemical reasoning against an explicit molecular state
Original title: MolLean: A Process-Verifiable Language for Chemical Reasoning
On 8,800 examples and 12 tasks, Qwen3.6-27B whole-trace refinement beat direct chain-of-thought on every task.
Researchers introduce MolLean, a process-verifiable language that grounds chemical reasoning in an explicit molecular state. A layered Lean-based kernel checks state transitions and assertions, while a typed witness protocol separates native structural checks from external measurements. MolLean-MCP feeds rejected actions back without changing state. With Qwen3.6-27B, whole-trace refinement beat direct chain-of-thought on all 12 tasks; stepwise MolLean-MCP feedback did better on reaction tasks.
- Evaluation covered 8,800 matched examples across 12 tasks.
- Tasks span generation, editing, optimization, and reactions.
- Whole-trace refinement outperformed direct chain-of-thought on all 12 tasks.
- MolLean-MCP reached better performance on reaction tasks.
- Rejected actions leave the molecular state unchanged.
- Products
- MolLeanMolLean-MCP
- AI models
- Qwen3.6-27B
From the original article
1 min read at huggingface.co
Chemical chain-of-thought can contain hallucinated structural and property claims. We thereby introduce MolLean, a process-verifiable language that grounds chemical reasoning in an explicit molecular state. A layered Lean-based kernel checks state transitions and assertions over molecular readouts, from primitive graph edits to compound transformations.
By Jiatong Li, Yatao Bian · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.