Hi, I am learning the scGCN method, and want to draw the UMAP figure as in your article (Figure 4a). After running the 'train.py', I got serveral excel files, but how can I get the low-represented vectors for cells computed by GCN model to draw the UMAP? Thanks.