docs: undefined to_sparse() function

[puigde]

In the code snippet below This second example is about loading the dataset as a GNN with random data and classes:

import random
from torch_geometric.utils import from_networkx
import numpy as np

def to_torch_randomdata(dataset, features, classes):
    for d in dataset:
        G = d["graph"]
        for n in G.nodes():
            G.nodes[n]['features'] = list(np.random.rand(features).astype(np.float32))
            G.nodes[n]['classes'] = random.randint(0,classes - 1)
        gnn = from_networkx(G)
        gnn = to_sparse(gnn)
        yield {"name": d["name"], "gnn": gnn}
        
dataset = to_torch_randomdata(load_dataset("<dataset_folder>"), 10, 2)

The function to_sparse is undefined.