Fig. 10From: Prediction of 8-state protein secondary structures by a novel deep learning architectureModel loss variation trained using TR12148. Loss on CB513 dataset stopped decreasing after 130 iterations and the model inclined to be stableBack to article page