N = 5
T_train, T_test = 10, 10
train, test = [], []
for i in xrange(T_train):
x = np.random.choice([0, 1], size=(N,), p=[1./3, 2./3])
y = np.random.choice([0, 1], size=(N,), p=[1./3, 2./3])
train.append((x, y))
for i in xrange(T_test):
x = np.random.choice([0, 1], size=(N,), p=[1./3, 2./3])
y = np.random.choice([0, 1], size=(N,), p=[1./3, 2./3])
test.append((x, y))
model = RRBM(N, train)
# check partition function
assert np.allclose(model.Z(model.W, model.b, train[0][0]), model.enum_Z(model.W, model.b, train[0][0]), atol = 0.01)
gW_fd, gb_fd = fd(model)
gW, gb = model.g(model.W, model.b)
# check gradient
assert np.allclose(gW, gW_fd, atol=0.01)
assert np.allclose(gb, gb_fd, atol=0.01)
# check conversion
W2, b2 = model.vec2theta(model.theta2vec(model.W, model.b))
assert (W2 == model.W).all()
assert (b2 == model.b).all()
else:
lst.append(str((l1, l2)))
return map(lambda x: eval(x), np.random.choice(lst, size=N, replace=False))
# tag to tag projection experiments
# number of tag sizes to get for the learning curve
test_size = 3000
sizes = map(lambda x : 10*x, xrange(1, 101))
# number of experiments to run with each size
experiments = 100
# model
model = RRBM(lex_col.N, 64, 0, [], C=0.1)
# main loop over different experiments (parallelize)
for experiment in xrange(experiments):
test = []
test_keys = sample_from_d(d, test_size, lex_col[lang1], lex_col[lang2])
test_keys_set = set(test)
for (l1, l2) in test_keys:
embedding = E_dict[l2] if l2 in E_dict else zero
test.append((lex_col[lang1][l1], lex_col[lang2][l2], embedding))
entry1 = lex_col[lang1].pp(l1)
entry2 = lex_col[lang2].pp(l2)
#if "num=PL" in entry2[1]:
# print l2, entry2[1]
