Training Chemical Plausibility-Aware Large Language Models for Single-Step Retrosynthesis

Researchers introduced Top-K prompting and the C3LM (Chemistry Constraint-Consistent Language Model), trained on the CREED-CCV-2+USPTO-XL dataset of ~45.6 million verified reactions, achieving state-of-the-art performance on the OOD URSA-expert-2026 benchmark for single-step retrosynthesis. The study, released on arXiv (2608.18940v1), shows that LLMs and conventional models explore complementary reaction spaces, motivating ensemble-based retrosynthesis systems.

arXiv:2608.18940v1 Announce Type: cross Abstract: Single-step retrosynthesis is a central component of computer-aided synthesis planning, yet its intrinsically one-to-many nature is poorly captured by single-answer evaluation and benchmarking protocols. To address this, we introduce Top-K prompting as a robust training and inference paradigm to better capture diverse, plausible reaction predictions. We compile CREED-CCV-2+USPTO-XL, an ultra-large-scale dataset of ~45.6 million verified reactions to train the C3LM Chemistry Constraint-Consistent Language Model . By integrating fine-tuning with ChemCensor-based and novelty-oriented rewards, our model achieves state-of-the-art performance on the OOD URSA-expert-2026 benchmark. Further analysis of reaction uniqueness shows that LLMs and conventional models explore complementary reaction spaces, motivating ensemble-based retrosynthesis systems. Overall, our results establish Top-K, plausibility-aware training as a practical new direction for robust future LLM-based synthesis planning.