for (labels, value) in rows:
    pass

result = np.fromiter(your_generator, dtype=[('labels', '|S10'), ('value', float)])

result = np.array([(('a','b','c'), 1.23),(('a','c','d'), 2.34)],
                  dtype=[('label', ("|S10", 3)), ('value', float)])

dtype=[('label',[('A','|S10'),('B','|S10'),('C','|S10')]),('value',float)]