A Large Language Model for Chemistry and Retrosynthesis Predictions

2025/07/02 by Yueqing Zhang, Zhang, Yueqing, Wentao Liu +23 · 2 voices
Computer Science · Materials Science · Physics and Astronomy · #Advanced Graph Neural Networks #Chemical Physics (physics.chem-ph) #Computational Drug Discovery Methods #FOS: Physical sciences #Machine Learning in Materials Science #physics.chem-ph

paper · pdf · doi:10.48550/arxiv.2507.01444

openalex publication_date 2025/07/02 · arxiv published 2025/07/02 · arxiv updated 2025/07/10 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Large language models (LLM) have achieved impressive progress across a broad range of general-purpose tasks, but their effectiveness in chemistry remains limited due to scarce domain-specific datasets and the demand for precise symbolic and structural reasoning. Here we introduce ECNU-ChemGPT(name after East China Normal University), a chemistry-specialized LLM engineered for deep chemical knowledge understanding and accurate retrosynthetic route planning. Our approach is distinguished by four key strategies: structured prompt-based knowledge distillation from authoritative chemistry textbooks to construct a high-quality question-answering dataset; domain-specific prompt engineering using curated chemical keywords, combined with LLMs APIs for data derivation and knowledge distillation; large-scale fine-tuning on a meticulously cleaned and enriched Pistachio reaction dataset to enhance retrosynthesis prediction accuracy; and integration of BrainGPT, a dynamic multi-model scheduling framework that enables task-specific invocation of multiple specialized models trained for diverse chemistry-related tasks. ECNU-ChemGPT exhibits superior performance on chemistry question-answering and retrosynthetic planning benchmarks, outperforming leading general-purpose models-including Deepseek-R1, Qwen-2.5, and GPT-4o. In retrosynthesis, it achieves a Top-1 accuracy of 68.3% on the USPTO₅0K dataset and successfully reconstructed 13 complete experimental pathways for real-world drug molecules from medicinal chemistry journals. These results underscore the effectiveness of domain-adapted fine-tuning combined with dynamic multi-model task scheduling, providing a scalable and robust solution for chemical knowledge question answering and retrosynthetic planning.

Citations

• DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning
• Contextual Biasing to Improve Domain-specific Custom Vocabulary Audio Transcription without Explicit Fine-Tuning of Whisper Model
• Qwen2 Technical Report
• ChatGLM: A Family of Large Language Models from GLM-130B to GLM-4 All Tools
• SynAsk: Unleashing the Power of Large Language Models in Organic Synthesis
• Evolutionary optimization of model merging recipes
• Developing ChemDFM as a large language foundation model for chemistry
• Small LLMs Are Weak Tool Learners: A Multi-LLM Agent
• EduChat: A Large-Scale Language Model-based Chatbot System for Intelligent Education
• Large language models in medicine
• Transfer learning enables predictions in network biology
• C-Eval: A Multi-Level Multi-Discipline Chinese Evaluation Suite for Foundation Models
• LLaMA: Open and Efficient Foundation Language Models
• Large Language Models Encode Clinical Knowledge
• Large language models encode clinical knowledge
• High-Resolution Image Synthesis with Latent Diffusion Models
• High-Resolution Image Synthesis with Latent Diffusion Models
• Measuring Massive Multitask Language Understanding
• Mapping the Space of Chemical Reactions Using Attention-Based Neural Networks
• Drug discovery with explainable artificial intelligence
• Language Models are Few-Shot Learners
• A Graph to Graphs Framework for Retrosynthesis Prediction
• Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
• "Found in Translation": Predicting Outcomes of Complex Organic Chemistry\n Reactions using Neural Sequence-to-Sequence Models
• Retrosynthetic reaction prediction using neural sequence-to-sequence models
• Neural‐Symbolic Machine Learning for Retrosynthesis and Reaction Prediction

Discussions

• Zhang et al. A Large Language Model for Chemistry and Retrosynthesis Predictions arxiv.org/pdf/2507.01444 [bsky, 3 points, 0 comments]
• #physics.chem-ph Origin | Interest | Match [bsky, 0 points, 0 comments]

Related

• Reasoning-Driven Retrosynthesis Prediction with Large Language Models via Reinforcement Learning
• Enhancing Chemical Reaction and Retrosynthesis Prediction with Large Language Model and Dual-task Learning
• BatGPT-Chem: A Foundation Large Model For Retrosynthesis Prediction
• LLM-Augmented Chemical Synthesis and Design Decision Programs
• Retro-Expert: Collaborative Reasoning for Interpretable Retrosynthesis