Work

Representation Alignment as a Bottleneck in LLM-Based Retrosynthesis Planning

Year: 2026

Type: preprint

Abstract: While LLMs show promise in general reasoning, symbolic planning in chemistry remains a bottleneck. Direct ''SMILES-to-PDDL'' attempts fail because they force models to juggle chemical analysis and planning-language structuring simultaneously. We hypothesize that this failure stems from a lack of intermediate abstractions rather than insufficient model capacity. By decomposing retrosynthesis into molecule mapping, reaction mapping, and PDDL generation, we achieve high success rates where end-to-end approaches fail. This provides evidence that a primary bottleneck lies in representation alignment rather than raw model capacity. Our structural analysis demonstrates that intermediate representations are essential in retrosynthesis planning, highlighting the importance of representation-centric design in future systems.

Source: arXiv (Cornell University)

DOI: 10.48550/arxiv.2609.35571

Authors: Yoo, Hyunwoo, Huang, Cassie, Shin, Haebin, Zhang, Li, Rosen, Gail L.

Institutions: Drexel University, University of Michigan

Topic: AI-based Problem Solving and Planning

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: green

Funder: National Science Foundation

Award: III: Small: Learning Multi-scale Sequence Features for Predicting Gene to Microbiome Function

OpenAlex ID: W7214783281

Cites:

Cited by:

Related to: