Fig. 5: Error distributions of autoregressive prediction results of battery operating features across different phases (Early, Mid, Late) in a charging cycle.
From: Early prediction of lithium-ion battery degradation with a generative pre-trained transformer
T, I, and V represent temperature, current, and voltage, respectively. The error distributions of Long Short-Term Memory network (LSTM) and Transformer are depicted as violin plots in the pink area, while the error distributions of Generative Pre-trained Transformer (GPT-Small) model are shown in the deep blue area, with the median indicated by the black line and the interquartile range (IQR) indicated by the red line. Source data are provided as a Source Data file.
