Improving synthesis prediction of small molecules at scale with RetroChimera

The RetroChimera model has been published in Nature, detailing its architecture and extensive validation studies. It demonstrates the ability to recall rare reaction types and has succeeded in zero-shot transfer and fine-tuning on proprietary datasets. RetroChimera aims to accelerate the development of new medicinally relevant molecules and advanced materials by automatically proposing high-quality synthesis routes. The model combines two complementary strengths, allowing it to produce better predictions than either model alone.