switch to torch TestCase for build-in dataset (#822)

Author: zhangguanheng66

test/common/torchtext_test_case.py

@@ -1,5 +1,5 @@
 # -*- coding: utf-8 -*-
-from unittest import TestCase
+from torch.testing._internal.common_utils import TestCase
 import json
 import logging
 import os

test/data/test_builtin_datasets.py

@@ -6,7 +6,6 @@
 import torchtext.data as data
 from torchtext.datasets import AG_NEWS
 import torch
-from torch.testing import assert_allclose
 from ..common.torchtext_test_case import TorchtextTestCase
 
 
@@ -99,10 +98,10 @@ def test_text_classification(self):
         ag_news_train, ag_news_test = AG_NEWS(root=datadir, ngrams=3)
         self.assertEqual(len(ag_news_train), 120000)
         self.assertEqual(len(ag_news_test), 7600)
-        assert_allclose(ag_news_train[-1][1][:10],
-                        torch.tensor([3525, 319, 4053, 34, 5407, 3607, 70, 6798, 10599, 4053]).long())
-        assert_allclose(ag_news_test[-1][1][:10],
-                        torch.tensor([2351, 758, 96, 38581, 2351, 220, 5, 396, 3, 14786]).long())
+        self.assertEqual(ag_news_train[-1][1][:10],
+                         torch.tensor([3525, 319, 4053, 34, 5407, 3607, 70, 6798, 10599, 4053]).long())
+        self.assertEqual(ag_news_test[-1][1][:10],
+                         torch.tensor([2351, 758, 96, 38581, 2351, 220, 5, 396, 3, 14786]).long())
 
     def test_imdb(self):
         from torchtext.experimental.datasets import IMDB
@@ -111,14 +110,14 @@ def test_imdb(self):
         train_dataset, test_dataset = IMDB()
         self.assertEqual(len(train_dataset), 25000)
         self.assertEqual(len(test_dataset), 25000)
-        assert_allclose(train_dataset[0][1][:10],
-                        torch.tensor([13, 1568, 13, 246, 35468, 43, 64, 398, 1135, 92]).long())
-        assert_allclose(train_dataset[-1][1][:10],
-                        torch.tensor([2, 71, 4555, 194, 3328, 15144, 42, 227, 148, 8]).long())
-        assert_allclose(test_dataset[0][1][:10],
-                        torch.tensor([13, 125, 1051, 5, 246, 1652, 8, 277, 66, 20]).long())
-        assert_allclose(test_dataset[-1][1][:10],
-                        torch.tensor([13, 1035, 14, 21, 28, 2, 1051, 1275, 1008, 3]).long())
+        self.assertEqual(train_dataset[0][1][:10],
+                         torch.tensor([13, 1568, 13, 246, 35468, 43, 64, 398, 1135, 92]).long())
+        self.assertEqual(train_dataset[-1][1][:10],
+                         torch.tensor([2, 71, 4555, 194, 3328, 15144, 42, 227, 148, 8]).long())
+        self.assertEqual(test_dataset[0][1][:10],
+                         torch.tensor([13, 125, 1051, 5, 246, 1652, 8, 277, 66, 20]).long())
+        self.assertEqual(test_dataset[-1][1][:10],
+                         torch.tensor([13, 1035, 14, 21, 28, 2, 1051, 1275, 1008, 3]).long())
 
         # Test API with a vocab input object
         old_vocab = train_dataset.get_vocab()
@@ -164,14 +163,14 @@ def test_squad1(self):
         train_dataset, dev_dataset = SQuAD1()
         self.assertEqual(len(train_dataset), 87599)
         self.assertEqual(len(dev_dataset), 10570)
-        assert_allclose(train_dataset[100]['question'],
-                        torch.tensor([7, 24, 86, 52, 2, 373, 887, 18, 12797, 11090, 1356, 2, 1788, 3273, 16]).long())
-        assert_allclose(train_dataset[100]['ans_pos'][0],
-                        torch.tensor([72, 72]).long())
-        assert_allclose(dev_dataset[100]['question'],
-                        torch.tensor([42, 27, 669, 7438, 17, 2, 1950, 3273, 17252, 389, 16]).long())
-        assert_allclose(dev_dataset[100]['ans_pos'][0],
-                        torch.tensor([45, 48]).long())
+        self.assertEqual(train_dataset[100]['question'],
+                         torch.tensor([7, 24, 86, 52, 2, 373, 887, 18, 12797, 11090, 1356, 2, 1788, 3273, 16]).long())
+        self.assertEqual(train_dataset[100]['ans_pos'][0],
+                         torch.tensor([72, 72]).long())
+        self.assertEqual(dev_dataset[100]['question'],
+                         torch.tensor([42, 27, 669, 7438, 17, 2, 1950, 3273, 17252, 389, 16]).long())
+        self.assertEqual(dev_dataset[100]['ans_pos'][0],
+                         torch.tensor([45, 48]).long())
 
         # Test API with a vocab input object
         old_vocab = train_dataset.get_vocab()
@@ -185,14 +184,14 @@ def test_squad2(self):
         train_dataset, dev_dataset = SQuAD2()
         self.assertEqual(len(train_dataset), 130319)
         self.assertEqual(len(dev_dataset), 11873)
-        assert_allclose(train_dataset[200]['question'],
-                        torch.tensor([84, 50, 1421, 12, 5439, 4569, 17, 30, 2, 15202, 4754, 1421, 16]).long())
-        assert_allclose(train_dataset[200]['ans_pos'][0],
-                        torch.tensor([9, 9]).long())
-        assert_allclose(dev_dataset[200]['question'],
-                        torch.tensor([41, 29, 2, 66, 17016, 30, 0, 1955, 16]).long())
-        assert_allclose(dev_dataset[200]['ans_pos'][0],
-                        torch.tensor([40, 46]).long())
+        self.assertEqual(train_dataset[200]['question'],
+                         torch.tensor([84, 50, 1421, 12, 5439, 4569, 17, 30, 2, 15202, 4754, 1421, 16]).long())
+        self.assertEqual(train_dataset[200]['ans_pos'][0],
+                         torch.tensor([9, 9]).long())
+        self.assertEqual(dev_dataset[200]['question'],
+                         torch.tensor([41, 29, 2, 66, 17016, 30, 0, 1955, 16]).long())
+        self.assertEqual(dev_dataset[200]['ans_pos'][0],
+                         torch.tensor([40, 46]).long())
 
         # Test API with a vocab input object
         old_vocab = train_dataset.get_vocab()

test/data/test_field.py

@@ -2,7 +2,6 @@
 from collections import Counter
 import os
 
-from numpy.testing import assert_allclose
 import torch
 import torchtext.data as data
 import pytest
@@ -376,9 +375,9 @@ def test_numerical_features_no_vocab(self):
         test_float_data = ["1.1", "0.1", "3.91", "0.2", "10.2"]
 
         numericalized_int = int_field.numericalize(test_int_data)
-        assert_allclose(numericalized_int.data.numpy(), [1, 0, 1, 3, 19])
+        self.assertEqual(numericalized_int.data, [1, 0, 1, 3, 19])
         numericalized_float = float_field.numericalize(test_float_data)
-        assert_allclose(numericalized_float.data.numpy(), [1.1, 0.1, 3.91, 0.2, 10.2])
+        self.assertEqual(numericalized_float.data, [1.1, 0.1, 3.91, 0.2, 10.2])
 
         # Test with postprocessing applied
         int_field = data.Field(sequential=False, use_vocab=False,
@@ -396,9 +395,9 @@ def test_numerical_features_no_vocab(self):
         test_float_data = ["1.1", "0.1", "3.91", "0.2", "10.2"]
 
         numericalized_int = int_field.numericalize(test_int_data)
-        assert_allclose(numericalized_int.data.numpy(), [2, 1, 2, 4, 20])
+        self.assertEqual(numericalized_int.data, [2, 1, 2, 4, 20])
         numericalized_float = float_field.numericalize(test_float_data)
-        assert_allclose(numericalized_float.data.numpy(), [0.55, 0.05, 1.955, 0.1, 5.1])
+        self.assertEqual(numericalized_float.data, [0.55, 0.05, 1.955, 0.1, 5.1])
 
     def test_errors(self):
         # Test that passing a non-tuple (of data and length) to numericalize

test/data/test_metrics.py

@@ -1,5 +1,4 @@
 from torchtext.data.metrics import bleu_score
-from torch.testing import assert_allclose
 from ..common.torchtext_test_case import TorchtextTestCase
 
 
@@ -19,19 +18,19 @@ def test_bleu_score(self):
         # Partial match
         candidate = [['My', 'full', 'pytorch', 'test']]
         refs = [[['My', 'full', 'pytorch', 'test', '!'], ['Different']]]
-        assert_allclose(bleu_score(candidate, refs), 0.7788007)
+        self.assertEqual(bleu_score(candidate, refs), 0.7788007)
 
         # Bigrams and unigrams only
         candidate = [['My', 'pytorch', 'test']]
         refs = [[['My', 'full', 'pytorch', 'test'], ['Different']]]
-        assert_allclose(bleu_score(candidate, refs, max_n=2,
-                                   weights=[0.5, 0.5]), 0.5066641)
+        self.assertEqual(bleu_score(candidate, refs, max_n=2,
+                                    weights=[0.5, 0.5]), 0.5066641)
 
         # Multi-sentence corpus
         candidate = [['My', 'full', 'pytorch', 'test'], ['Another', 'Sentence']]
         refs = [[['My', 'full', 'pytorch', 'test'], ['Completely', 'Different']],
                 [['No', 'Match']]]
-        assert_allclose(bleu_score(candidate, refs), 0.8408964)
+        self.assertEqual(bleu_score(candidate, refs), 0.8408964)
 
         # Empty input
         candidate = [[]]
@@ -52,13 +51,13 @@ def test_bleu_score(self):
 
         # The comments below give the code used to get each hardcoded bleu score
         # nltk.translate.bleu_score.corpus_bleu(refs, candidate)
-        assert_allclose(bleu_score(candidate, refs), 0.4573199)
+        self.assertEqual(bleu_score(candidate, refs), 0.4573199)
         # nltk.translate.bleu_score.corpus_bleu(refs, candidate, weights=[0.33]*3)
-        assert_allclose(bleu_score(candidate, refs, 3,
-                        weights=[0.33, 0.33, 0.33]), 0.4901113)
+        self.assertEqual(bleu_score(candidate, refs, 3,
+                         weights=[0.33, 0.33, 0.33]), 0.4901113)
         # nltk.translate.bleu_score.corpus_bleu(refs, candidate, weights=[0.5]*2)
-        assert_allclose(bleu_score(candidate, refs, 2,
-                        weights=[0.5, 0.5]), 0.5119535)
+        self.assertEqual(bleu_score(candidate, refs, 2,
+                         weights=[0.5, 0.5]), 0.5119535)
         # nltk.translate.bleu_score.corpus_bleu(refs, candidate, weights=[1])
-        assert_allclose(bleu_score(candidate, refs, 1,
-                        weights=[1]), 0.5515605)
+        self.assertEqual(bleu_score(candidate, refs, 1,
+                         weights=[1]), 0.5515605)

test/test_build.py

@@ -3,7 +3,6 @@
 import os
 from collections import Counter
 
-import numpy as np
 import torch
 import torchtext.data
 
@@ -130,16 +129,16 @@ def test_vectors_get_vecs(self):
         self.assertEqual(vec.vectors.shape[0], len(vec))
 
         tokens = ['chip', 'baby', 'Beautiful']
-        token_vecs = vec.get_vecs_by_tokens(tokens).numpy()
+        token_vecs = vec.get_vecs_by_tokens(tokens)
         self.assertEqual(token_vecs.shape[0], len(tokens))
         self.assertEqual(token_vecs.shape[1], vec.dim)
-        torch.testing.assert_allclose(vec[tokens[0]].numpy(), token_vecs[0])
-        torch.testing.assert_allclose(vec[tokens[1]].numpy(), token_vecs[1])
-        torch.testing.assert_allclose(vec['<unk>'].numpy(), token_vecs[2])
+        self.assertEqual(vec[tokens[0]], token_vecs[0])
+        self.assertEqual(vec[tokens[1]], token_vecs[1])
+        self.assertEqual(vec['<unk>'], token_vecs[2])
 
-        token_one_vec = vec.get_vecs_by_tokens(tokens[0], lower_case_backup=True).numpy()
+        token_one_vec = vec.get_vecs_by_tokens(tokens[0], lower_case_backup=True)
         self.assertEqual(token_one_vec.shape[0], vec.dim)
-        torch.testing.assert_allclose(vec[tokens[0].lower()].numpy(), token_one_vec)
+        self.assertEqual(vec[tokens[0].lower()], token_one_vec)
 
     def test_download_charngram_vectors(self):
         c = Counter({'hello': 4, 'world': 3, 'ᑌᑎIᑕOᗪᕮ_Tᕮ᙭T': 5, 'freq_too_low': 2})
@@ -157,7 +156,7 @@ def test_download_charngram_vectors(self):
             expected_stoi = {x: index for index, x in enumerate(expected_itos)}
             self.assertEqual(v.itos, expected_itos)
             self.assertEqual(dict(v.stoi), expected_stoi)
-            vectors = v.vectors.numpy()
+            vectors = v.vectors
 
             # The first 5 entries in each vector.
             expected_charngram = {
@@ -167,11 +166,11 @@ def test_download_charngram_vectors(self):
             }
 
             for word in expected_charngram:
-                torch.testing.assert_allclose(
+                self.assertEqual(
                     vectors[v.stoi[word], :5], expected_charngram[word])
 
-            torch.testing.assert_allclose(vectors[v.stoi['<unk>']], np.zeros(100))
-            torch.testing.assert_allclose(vectors[v.stoi['OOV token']], np.zeros(100))
+            self.assertEqual(vectors[v.stoi['<unk>']], torch.zeros(100))
+            self.assertEqual(vectors[v.stoi['OOV token']], torch.zeros(100))
 
     def test_download_custom_vectors(self):
         c = Counter({'hello': 4, 'world': 3, 'ᑌᑎIᑕOᗪᕮ_Tᕮ᙭T': 5, 'freq_too_low': 2})
@@ -187,7 +186,7 @@ def test_download_custom_vectors(self):
 
             self.assertEqual(v.itos, ['<unk>', '<pad>', '<bos>',
                                       'ᑌᑎIᑕOᗪᕮ_Tᕮ᙭T', 'hello', 'world'])
-            vectors = v.vectors.numpy()
+            vectors = v.vectors
 
             # The first 5 entries in each vector.
             expected_fasttext_simple_en = {
@@ -196,10 +195,10 @@ def test_download_custom_vectors(self):
             }
 
             for word in expected_fasttext_simple_en:
-                torch.testing.assert_allclose(
+                self.assertEqual(
                     vectors[v.stoi[word], :5], expected_fasttext_simple_en[word])
 
-            torch.testing.assert_allclose(vectors[v.stoi['<unk>']], np.zeros(300))
+            self.assertEqual(vectors[v.stoi['<unk>']], torch.zeros(300))
 
     def test_download_fasttext_vectors(self):
         c = Counter({'hello': 4, 'world': 3, 'ᑌᑎIᑕOᗪᕮ_Tᕮ᙭T': 5, 'freq_too_low': 2})
@@ -219,7 +218,7 @@ def test_download_fasttext_vectors(self):
             expected_stoi = {x: index for index, x in enumerate(expected_itos)}
             self.assertEqual(v.itos, expected_itos)
             self.assertEqual(dict(v.stoi), expected_stoi)
-            vectors = v.vectors.numpy()
+            vectors = v.vectors
 
             # The first 5 entries in each vector.
             expected_fasttext_simple_en = {
@@ -228,11 +227,11 @@ def test_download_fasttext_vectors(self):
             }
 
             for word in expected_fasttext_simple_en:
-                torch.testing.assert_allclose(
+                self.assertEqual(
                     vectors[v.stoi[word], :5], expected_fasttext_simple_en[word])
 
-            torch.testing.assert_allclose(vectors[v.stoi['<unk>']], np.zeros(300))
-            torch.testing.assert_allclose(vectors[v.stoi['OOV token']], np.zeros(300))
+            self.assertEqual(vectors[v.stoi['<unk>']], torch.zeros(300))
+            self.assertEqual(vectors[v.stoi['OOV token']], torch.zeros(300))
 
     def test_download_glove_vectors(self):
         c = Counter({'hello': 4, 'world': 3, 'ᑌᑎIᑕOᗪᕮ_Tᕮ᙭T': 5, 'freq_too_low': 2})
@@ -253,7 +252,7 @@ def test_download_glove_vectors(self):
             self.assertEqual(v.itos, expected_itos)
             self.assertEqual(dict(v.stoi), expected_stoi)
 
-            vectors = v.vectors.numpy()
+            vectors = v.vectors
 
             # The first 5 entries in each vector.
             expected_twitter = {
@@ -262,11 +261,11 @@ def test_download_glove_vectors(self):
             }
 
             for word in expected_twitter:
-                torch.testing.assert_allclose(
+                self.assertEqual(
                     vectors[v.stoi[word], :5], expected_twitter[word])
 
-            torch.testing.assert_allclose(vectors[v.stoi['<unk>']], np.zeros(25))
-            torch.testing.assert_allclose(vectors[v.stoi['OOV token']], np.zeros(25))
+            self.assertEqual(vectors[v.stoi['<unk>']], torch.zeros(25))
+            self.assertEqual(vectors[v.stoi['OOV token']], torch.zeros(25))
 
     def test_extend(self):
         c = Counter({'hello': 4, 'world': 3, 'ᑌᑎIᑕOᗪᕮ_Tᕮ᙭T': 5, 'freq_too_low': 2})
@@ -281,7 +280,7 @@ def test_extend(self):
 
             self.assertEqual(v.itos[:6], ['<unk>', '<pad>', '<bos>',
                                           'ᑌᑎIᑕOᗪᕮ_Tᕮ᙭T', 'hello', 'world'])
-            vectors = v.vectors.numpy()
+            vectors = v.vectors
 
             # The first 5 entries in each vector.
             expected_fasttext_simple_en = {
@@ -290,10 +289,10 @@ def test_extend(self):
             }
 
             for word in expected_fasttext_simple_en:
-                torch.testing.assert_allclose(
+                self.assertEqual(
                     vectors[v.stoi[word], :5], expected_fasttext_simple_en[word])
 
-            torch.testing.assert_allclose(vectors[v.stoi['<unk>']], np.zeros(300))
+            self.assertEqual(vectors[v.stoi['<unk>']], torch.zeros(300))
 
     def test_vectors_custom_cache(self):
         c = Counter({'hello': 4, 'world': 3, 'ᑌᑎIᑕOᗪᕮ_Tᕮ᙭T': 5, 'freq_too_low': 2})
@@ -312,7 +311,7 @@ def test_vectors_custom_cache(self):
 
             self.assertEqual(v.itos, ['<unk>', '<pad>', '<bos>',
                                       'ᑌᑎIᑕOᗪᕮ_Tᕮ᙭T', 'hello', 'world'])
-            vectors = v.vectors.numpy()
+            vectors = v.vectors
 
             # The first 5 entries in each vector.
             expected_fasttext_simple_en = {
@@ -321,7 +320,7 @@ def test_vectors_custom_cache(self):
             }
 
             for word in expected_fasttext_simple_en:
-                torch.testing.assert_allclose(
+                self.assertEqual(
                     vectors[v.stoi[word], :5], expected_fasttext_simple_en[word])
 
-            torch.testing.assert_allclose(vectors[v.stoi['<unk>']], np.zeros(300))
+            self.assertEqual(vectors[v.stoi['<unk>']], torch.zeros(300))

test/test_vocab.py

@@ -5,7 +5,6 @@
 
 
 import numpy as np
-from numpy.testing import assert_allclose
 import torch
 from torchtext import vocab
 
@@ -89,7 +88,7 @@ def test_vocab_set_vectors(self):
         expected_vectors = np.array([[0.0, 0.0], [0.0, 0.0], [0.0, 0.0],
                                      [0.0, 0.0], [0.1, 0.2], [0.5, 0.6],
                                      [0.3, 0.4]])
-        assert_allclose(v.vectors.numpy(), expected_vectors)
+        self.assertEqual(v.vectors, expected_vectors)
 
     def test_errors(self):
         c = Counter({'hello': 4, 'world': 3, 'ᑌᑎIᑕOᗪᕮ_Tᕮ᙭T': 5, 'freq_too_low': 2})