Creating an LSA from Infogain-weighted model raises "can't create more dimensions than there are documents" error

Creating an LSA with Information Gain weighted model raises "can't create more dimensions than there are documents".
All other weighting methods work.
Tested with Python 2.7.6
Tryed both Pattern 2.6 from PyPI and the latest version from Git.

Example:

from pattern.vector import Model, Document, LSA

docs = [
    {'attr A': 2, 'attr B': 1},
    {'attr A': 1, 'Attr B': 3},
    {'Attr C': 1, 'Attr D': 1},
]
m = Model(weight='ig')

for d in docs:
    m.append(Document(d))

m.lsa = LSA(m)

Start by running the provided Python example with Model(weight='ig'), Document, and LSA to reproduce the error. Inspect the pattern.vector weighting and LSA entry points; done means the Infogain-weighted model creates its LSA without the dimensions error while other weighting methods continue to work.