chemoinformatics

This paper investigates the impact of prompt engineering on drug toxicity prediction using large language models, finding that natural variance in LLM outputs outweighs prompt fine-tuning, while chemoinformatic feature extraction improves model performance.