def test_LdaCgsSeq_SeedTypes(self): """ Test for issue #74 issues. """ from tempfile import NamedTemporaryFile import os c = random_corpus(1000, 50, 6, 100) tmp = NamedTemporaryFile(delete=False, suffix='.npz') try: m0 = LdaCgsSeq(c, 'document', K=10) m0.train(n_iterations=20) m0.save(tmp.name) m1 = LdaCgsSeq.load(tmp.name) self.assertTrue(type(m0.seed) == type(m1.seed)) print("seed types:", type(m0._mtrand_state[0]), type(m1._mtrand_state[0])) self.assertTrue( type(m0._mtrand_state[0]) == type(m1._mtrand_state[0])) self.assertTrue( type(m0._mtrand_state[1]) == type(m1._mtrand_state[1])) self.assertTrue( type(m0._mtrand_state[2]) == type(m1._mtrand_state[2])) self.assertTrue( type(m0._mtrand_state[3]) == type(m1._mtrand_state[3])) self.assertTrue( type(m0._mtrand_state[4]) == type(m1._mtrand_state[4])) finally: try: os.remove(tmp.name) except WindowsError: pass
def test_LdaCgsSeq_SeedTypes(self): """ Test for issue #74 issues. """ from tempfile import NamedTemporaryFile import os c = random_corpus(1000, 50, 6, 100) tmp = NamedTemporaryFile(delete=False, suffix='.npz') try: m0 = LdaCgsSeq(c, 'document', K=10) m0.train(n_iterations=20) m0.save(tmp.name) m1 = LdaCgsSeq.load(tmp.name) self.assertTrue(type(m0.seed) == type(m1.seed)) print("seed types:", type(m0._mtrand_state[0]), type(m1._mtrand_state[0])) self.assertTrue(type(m0._mtrand_state[0]) == type(m1._mtrand_state[0])) self.assertTrue(type(m0._mtrand_state[1]) == type(m1._mtrand_state[1])) self.assertTrue(type(m0._mtrand_state[2]) == type(m1._mtrand_state[2])) self.assertTrue(type(m0._mtrand_state[3]) == type(m1._mtrand_state[3])) self.assertTrue(type(m0._mtrand_state[4]) == type(m1._mtrand_state[4])) finally: try: os.remove(tmp.name) except WindowsError: pass
def test_LdaCgsSeq_IO(self): from tempfile import NamedTemporaryFile import os c = random_corpus(1000, 50, 6, 100) tmp = NamedTemporaryFile(delete=False, suffix='.npz') try: m0 = LdaCgsSeq(c, 'document', K=10) m0.train(n_iterations=20) m0.save(tmp.name) m1 = LdaCgsSeq.load(tmp.name) self.assertTrue(m0.context_type == m1.context_type) self.assertTrue(m0.K == m1.K) self.assertTrue((m0.alpha == m1.alpha).all()) self.assertTrue((m0.beta == m1.beta).all()) self.assertTrue(m0.log_probs == m1.log_probs) for i in range(max(len(m0.corpus), len(m1.corpus))): self.assertTrue(m0.corpus[i].all() == m1.corpus[i].all()) self.assertTrue(m0.V == m1.V) self.assertTrue(m0.iteration == m1.iteration) for i in range(max(len(m0.Z), len(m1.Z))): self.assertTrue(m0.Z[i].all() == m1.Z[i].all()) self.assertTrue(m0.top_doc.all() == m1.top_doc.all()) self.assertTrue(m0.word_top.all() == m1.word_top.all()) self.assertTrue(m0.inv_top_sums.all() == m1.inv_top_sums.all()) self.assertTrue(m0.seed == m1.seed) self.assertTrue(m0._mtrand_state[0] == m1._mtrand_state[0]) self.assertTrue((m0._mtrand_state[1] == m1._mtrand_state[1]).all()) for s1,s2 in zip(m0._mtrand_state[2:], m1._mtrand_state[2:]): self.assertTrue(s1 == s2) m0 = LdaCgsSeq(c, 'document', K=10) m0.train(n_iterations=20) m0.save(tmp.name) m1 = LdaCgsSeq.load(tmp.name) self.assertTrue(not hasattr(m1, 'log_prob')) finally: try: os.remove(tmp.name) except WindowsError: pass
def test_LdaCgsSeq_IO(self): from tempfile import NamedTemporaryFile import os c = random_corpus(1000, 50, 6, 100) tmp = NamedTemporaryFile(delete=False, suffix='.npz') try: m0 = LdaCgsSeq(c, 'document', K=10) m0.train(n_iterations=20) m0.save(tmp.name) m1 = LdaCgsSeq.load(tmp.name) self.assertTrue(m0.context_type == m1.context_type) self.assertTrue(m0.K == m1.K) self.assertTrue((m0.alpha == m1.alpha).all()) self.assertTrue((m0.beta == m1.beta).all()) self.assertTrue(m0.log_probs == m1.log_probs) for i in range(max(len(m0.corpus), len(m1.corpus))): self.assertTrue(m0.corpus[i].all() == m1.corpus[i].all()) self.assertTrue(m0.V == m1.V) self.assertTrue(m0.iteration == m1.iteration) for i in range(max(len(m0.Z), len(m1.Z))): self.assertTrue(m0.Z[i].all() == m1.Z[i].all()) self.assertTrue(m0.top_doc.all() == m1.top_doc.all()) self.assertTrue(m0.word_top.all() == m1.word_top.all()) self.assertTrue(m0.inv_top_sums.all() == m1.inv_top_sums.all()) self.assertTrue(m0.seed == m1.seed) self.assertTrue(m0._mtrand_state[0] == m1._mtrand_state[0]) self.assertTrue((m0._mtrand_state[1] == m1._mtrand_state[1]).all()) for s1, s2 in zip(m0._mtrand_state[2:], m1._mtrand_state[2:]): self.assertTrue(s1 == s2) m0 = LdaCgsSeq(c, 'document', K=10) m0.train(n_iterations=20) m0.save(tmp.name) m1 = LdaCgsSeq.load(tmp.name) self.assertTrue(not hasattr(m1, 'log_prob')) finally: try: os.remove(tmp.name) except WindowsError: pass