Fig. 2From: DLM-DTI: a dual language model for the prediction of drug-target interaction with hint-based learningThe process flow of DLM-DTI. The drug and target sequences feed into their respective encoders. The encoded sequences are then merged, and the probability of bindings is computed using the interaction prediction head. DLM-DTI only utilizes the class token (CLS) of each encoded sequence because the class token preserves the abstract meaning of the entire sequence. The features of target sequences are computed using a teacher-student-based architecture, specifically employing a hint-based learning strategyBack to article page