protein-ligand affinity prediction

 **PDBBindv2016**  | Binding Affinity Regression | Benchmark Evaluation (Effectiveness) | Each sample in the PDBBind v2016 dataset is a complex, but we extracted the sequence data with substantial information loss to yield a protein-ligand sequence pair. We maintained the same split setting used in a previous study, where the refined set (excluding the core set) is treated as  training (train.csv) and validation (valid.csv) sets, while the core set (complexes with the highest resolution) is treated as the test set (test.csv).