switch node labels to node features (node attributes)

[kevinkle]

This changes #26 , our nodes will then be un-"labeled" but instead tagged with an attribute for what we think the kmer is

def benchmark_task_val(args, writer=None, feat='node-label'):
477     all_vals = []
478     graphs = load_data.read_graphfile(args.datadir, args.bmname, max_nodes=args.max_nodes)
479 
480     if feat == 'node-feat' and 'feat_dim' in graphs[0].graph:
481         print('Using node features')
482         input_dim = graphs[0].graph['feat_dim']
483     elif feat == 'node-label' and 'label' in graphs[0].node[0]:
484         print('Using node labels')
485         for G in graphs:
486             for u in G.nodes():
487                 G.node[u]['feat'] = np.array(G.node[u]['label'])

[kevinkle]

Should keep this in mind as pytorch geometric's implementation of diffpool uses the same setup