 The M.T. Albert model is a novel approach to address the challenges associated with predicting pharmacological properties of small molecules. It uses a large-scale pre-training process to extract contextual information from smile strings, leverages multitask learning to share information across multiple downstream tasks, and utilizes smiles enumeration to increase data diversity. This enables the model to achieve superior performance compared to other state of the art methods on most of the 60 data sets. Furthermore, the model employs attention mechanisms to focus on smiles character features essential to target properties, providing valuable insights into the model's decision-making process. This article was authored by Xiaochan Zhang, Chengkuan Wu, Jiaqai Yi, and others.