1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
|
--- a/tests/test_evaluator.py 2023-05-14 11:01:54.449768849 +0200
+++ b/tests/test_evaluator.py 2023-05-14 11:06:15.182738125 +0200
@@ -16,6 +16,7 @@
from time import sleep
from unittest import TestCase, mock
+from unittest import skip
from datasets import ClassLabel, Dataset, Features, Sequence, Value
from PIL import Image
@@ -128,6 +128,7 @@
return [{"text": "Lorem ipsum"} for _ in inputs]
+@skip("require network")
class TestEvaluator(TestCase):
def setUp(self):
self.data = Dataset.from_dict({"label": [1, 0], "text": ["great movie", "horrible movie"]})
@@ -230,6 +230,7 @@
)
+@skip("require network")
class TestTextClassificationEvaluator(TestCase):
def setUp(self):
self.data = Dataset.from_dict({"label": [1, 0], "text": ["great movie", "horrible movie"]})
@@ -394,6 +394,7 @@
self.assertAlmostEqual(results["latency_in_seconds"], results["total_time_in_seconds"] / len(data), 5)
+@skip("require network")
class TestTextClassificationEvaluatorTwoColumns(TestCase):
def setUp(self):
self.data = Dataset.from_dict(
@@ -452,6 +452,7 @@
self.assertEqual(results["accuracy"], 1.0)
+@skip("require network")
class TestImageClassificationEvaluator(TestCase):
def setUp(self):
self.data = Dataset.from_dict(
@@ -534,6 +535,7 @@
self.assertEqual(results["accuracy"], 0)
+@skip("require network")
class TestQuestionAnsweringEvaluator(TestCase):
def setUp(self):
self.data = Dataset.from_dict(
@@ -716,6 +716,7 @@
)
self.assertEqual(results["overall_accuracy"], 0.5)
+ @skip("require network")
def test_class_init(self):
evaluator = TokenClassificationEvaluator()
self.assertEqual(evaluator.task, "token-classification")
@@ -735,6 +736,7 @@
)
self.assertEqual(results["overall_accuracy"], 2 / 3)
+ @skip("require network")
def test_overwrite_default_metric(self):
accuracy = load("seqeval")
results = self.evaluator.compute(
@@ -750,6 +752,7 @@
)
self.assertEqual(results["overall_accuracy"], 1.0)
+ @skip("require network")
def test_data_loading(self):
# Test passing in dataset by name with data_split
data = self.evaluator.load_data("evaluate/conll2003-ci", split="validation[:1]")
@@ -863,6 +866,7 @@
self.pipe = DummyTextGenerationPipeline(num_return_sequences=4)
self.evaluator = evaluator("text-generation")
+ @skip("require network")
def test_class_init(self):
evaluator = TextGenerationEvaluator()
self.assertEqual(evaluator.task, "text-generation")
@@ -877,6 +877,7 @@
results = self.evaluator.compute(data=self.data)
self.assertIsInstance(results["unique_words"], int)
+ @skip("require nltk")
def test_overwrite_default_metric(self):
word_length = load("word_length")
results = self.evaluator.compute(
@@ -906,6 +910,7 @@
self.assertEqual(processed_predictions, {"data": ["A", "B", "C", "D"]})
+@skip("require network")
class TestText2TextGenerationEvaluator(TestCase):
def setUp(self):
self.data = Dataset.from_dict(
@@ -979,6 +984,7 @@
self.assertEqual(results["bleu"], 0)
+@skip("require network")
class TestAutomaticSpeechRecognitionEvaluator(TestCase):
def setUp(self):
self.data = Dataset.from_dict(
--- a/tests/test_trainer_evaluator_parity.py 2023-05-14 17:50:29.224525549 +0200
+++ b/tests/test_trainer_evaluator_parity.py 2023-05-14 17:37:40.947501195 +0200
@@ -269,6 +269,7 @@
self.assertEqual(transformers_results["eval_HasAns_f1"], evaluator_results["HasAns_f1"])
self.assertEqual(transformers_results["eval_NoAns_f1"], evaluator_results["NoAns_f1"])
+ @unittest.skip('require eval_results.json')
def test_token_classification_parity(self):
model_name = "hf-internal-testing/tiny-bert-for-token-classification"
n_samples = 500
--- a/tests/test_load.py 2023-05-20 15:45:58.855473557 +0200
+++ b/tests/test_load.py 2023-05-20 15:50:41.620071500 +0200
@@ -61,6 +61,7 @@
hf_modules_cache=self.hf_modules_cache,
)
+ @pytest.mark.skip("require network")
def test_HubEvaluationModuleFactory_with_internal_import(self):
# "squad_v2" requires additional imports (internal)
factory = HubEvaluationModuleFactory(
@@ -72,6 +73,7 @@
module_factory_result = factory.get_module()
assert importlib.import_module(module_factory_result.module_path) is not None
+ @pytest.mark.skip("require network")
def test_HubEvaluationModuleFactory_with_external_import(self):
# "bleu" requires additional imports (external from github)
factory = HubEvaluationModuleFactory(
@@ -83,6 +85,7 @@
module_factory_result = factory.get_module()
assert importlib.import_module(module_factory_result.module_path) is not None
+ @pytest.mark.skip("require network")
def test_HubEvaluationModuleFactoryWithScript(self):
factory = HubEvaluationModuleFactory(
SAMPLE_METRIC_IDENTIFIER,
@@ -115,6 +118,7 @@
module_factory_result = factory.get_module()
assert importlib.import_module(module_factory_result.module_path) is not None
+ @pytest.mark.skip("require network")
def test_cache_with_remote_canonical_module(self):
metric = "accuracy"
evaluation_module_factory(
@@ -127,6 +131,7 @@
metric, download_config=self.download_config, dynamic_modules_path=self.dynamic_modules_path
)
+ @pytest.mark.skip("require network")
def test_cache_with_remote_community_module(self):
metric = "lvwerra/test"
evaluation_module_factory(
--- a/tests/test_metric.py 2023-05-20 15:54:32.558477445 +0200
+++ b/tests/test_metric.py 2023-05-20 15:55:40.775415987 +0200
@@ -316,6 +316,7 @@
self.assertDictEqual(expected_results[1], results[1])
del results
+ @pytest.mark.skip('')
def test_distributed_metrics(self):
with tempfile.TemporaryDirectory() as tmp_dir:
(preds_0, refs_0), (preds_1, refs_1) = DummyMetric.distributed_predictions_and_references()
@@ -736,6 +736,7 @@
self.assertDictEqual(dummy_result_1, combined_evaluation.compute(predictions=preds, references=refs))
+ @pytest.mark.skip('require network')
def test_modules_from_string(self):
expected_result = {"accuracy": 0.5, "recall": 0.5, "precision": 1.0}
predictions = [0, 1]
--- a/tests/test_metric_common.py 2023-05-20 15:57:02.399146066 +0200
+++ b/tests/test_metric_common.py 2023-05-20 15:59:25.167947472 +0200
@@ -99,6 +99,7 @@
evaluation_module_name = None
evaluation_module_type = None
+ @pytest.mark.skip('require network')
def test_load(self, evaluation_module_name, evaluation_module_type):
doctest.ELLIPSIS_MARKER = "[...]"
evaluation_module = importlib.import_module(
--- a/tests/test_trainer_evaluator_parity.py 2023-05-20 16:00:55.986549706 +0200
+++ b/tests/test_trainer_evaluator_parity.py 2023-05-20 16:02:51.808766855 +0200
@@ -4,6 +4,7 @@
import subprocess
import tempfile
import unittest
+import pytest
import numpy as np
import torch
@@ -33,6 +33,7 @@
def tearDown(self):
shutil.rmtree(self.dir_path, ignore_errors=True)
+ @pytest.mark.skip('require network')
def test_text_classification_parity(self):
model_name = "philschmid/tiny-bert-sst2-distilled"
@@ -121,6 +122,7 @@
self.assertEqual(transformers_results["eval_accuracy"], evaluator_results["accuracy"])
+ @pytest.mark.skip('require network')
def test_image_classification_parity(self):
# we can not compare to the Pytorch transformers example, that uses custom preprocessing on the images
model_name = "douwekiela/resnet-18-finetuned-dogfood"
@@ -179,6 +181,7 @@
self.assertEqual(transformers_results["eval_accuracy"], evaluator_results["accuracy"])
+ @pytest.mark.skip('require network')
def test_question_answering_parity(self):
model_name_v1 = "anas-awadalla/bert-tiny-finetuned-squad"
model_name_v2 = "mrm8488/bert-tiny-finetuned-squadv2"
|
