9902aaa77d03a2a5d1ccc3f8e260dfee2be86819,python/tests/test_embeddings.py,,test_normalize,#,123

Before Change

    normed = norm_weights(wv.weights)
    norms = np.linalg.norm(normed, axis=1)
    for norm in norms:
        assert norm == 0 or np.allclose(norm, 1, rtol=1e-4)


def test_vocab_truncation():
    model = random_model()

After Change

def test_normalize():
    wv = random_model()(keep_unused=True)
    normed = norm_weights(wv.weights)
    gold_norms = np.zeros_like(wv.weights)
    for i in range(len(gold_norms)):
        norm = np.sqrt(np.sum(np.square(wv.weights[i])))
        gold_norms[i] = wv.weights[i] if norm == 0.0 else wv.weights[i] / norm
    np.testing.assert_allclose(normed, gold_norms)


def test_vocab_truncation():

Instances