Merge branch 'master' into tf_qa_pipelines_test

Signed-off-by: Morgan Funtowicz <morgan@huggingface.co>

.circleci/config.yml

@@ -12,9 +12,23 @@ jobs:
             - checkout
             - run: sudo pip install .[sklearn,tf-cpu,torch,testing]
             - run: sudo pip install codecov pytest-cov
-            - run: python -m pytest -n 8 --dist=loadfile -s -v ./tests/ --cov
+            - run: python -m pytest -n 4 --dist=loadfile -s -v ./tests/ --cov
             - run: codecov
-
+    run_all_tests_torch_and_tf:
+        working_directory: ~/transformers
+        docker:
+            - image: circleci/python:3.5
+        environment:
+            OMP_NUM_THREADS: 1
+            RUN_SLOW: yes
+            RUN_CUSTOM_TOKENIZERS: yes
+        resource_class: xlarge
+        parallelism: 1
+        steps:
+            - checkout
+            - run: sudo pip install .[mecab,sklearn,tf-cpu,torch,testing]
+            - run: python -m pytest -n 4 --dist=loadfile -s -v ./tests/
+            - no_output_timeout: 4h
     run_tests_torch:
         working_directory: ~/transformers
         docker:
@@ -27,7 +41,7 @@ jobs:
             - checkout
             - run: sudo pip install .[sklearn,torch,testing]
             - run: sudo pip install codecov pytest-cov
-            - run: python -m pytest -n 8 --dist=loadfile -s -v ./tests/ --cov
+            - run: python -m pytest -n 4 --dist=loadfile -s -v ./tests/ --cov
             - run: codecov
     run_tests_tf:
         working_directory: ~/transformers
@@ -41,7 +55,7 @@ jobs:
             - checkout
             - run: sudo pip install .[sklearn,tf-cpu,testing]
             - run: sudo pip install codecov pytest-cov
-            - run: python -m pytest -n 8 --dist=loadfile -s -v ./tests/ --cov
+            - run: python -m pytest -n 4 --dist=loadfile -s -v ./tests/ --cov
             - run: codecov
     run_tests_custom_tokenizers:
         working_directory: ~/transformers
@@ -65,7 +79,7 @@ jobs:
             - checkout
             - run: sudo pip install .[sklearn,torch,testing]
             - run: sudo pip install -r examples/requirements.txt
-            - run: python -m pytest -n 8 --dist=loadfile -s -v ./examples/
+            - run: python -m pytest -n 4 --dist=loadfile -s -v ./examples/
     deploy_doc:
         working_directory: ~/transformers
         docker:

.github/workflows/self-scheduled.yml

@@ -48,4 +48,4 @@ jobs:
       run: |
         source .env/bin/activate
         python -m pytest -n 1 --dist=loadfile -s -v ./tests/
-
+        

tests/test_modeling_auto.py

@@ -78,6 +78,8 @@ class AutoModelTest(unittest.TestCase):
             model, loading_info = AutoModelForPreTraining.from_pretrained(model_name, output_loading_info=True)
             self.assertIsNotNone(model)
             self.assertIsInstance(model, BertForPreTraining)
+            for value in loading_info.values():
+                self.assertEqual(len(value), 0)
 
     @slow
     def test_lmhead_model_from_pretrained(self):

tests/test_pipelines.py

@@ -2,9 +2,16 @@ import unittest
 from typing import Iterable, List, Optional
 
 from transformers import pipeline
-from transformers.pipelines import Pipeline
+from transformers.pipelines import (
+    FeatureExtractionPipeline,
+    FillMaskPipeline,
+    NerPipeline,
+    Pipeline,
+    QuestionAnsweringPipeline,
+    TextClassificationPipeline,
+)
 
-from .utils import require_tf, require_torch
+from .utils import require_tf, require_torch, slow
 
 
 QA_FINETUNED_MODELS = [
@@ -284,7 +291,6 @@ class MultiColumnInputTestCase(unittest.TestCase):
             self._test_multicolumn_pipeline(nlp, valid_samples, invalid_samples, mandatory_output_keys)
 
     @require_tf
-    @unittest.skip("This test is failing intermittently. Skipping it until we resolve.")
     def test_tf_question_answering(self):
         mandatory_output_keys = {"score", "answer", "start", "end"}
         valid_samples = [
@@ -304,3 +310,30 @@ class MultiColumnInputTestCase(unittest.TestCase):
         for tokenizer, model, config in TF_QA_FINETUNED_MODELS:
             nlp = pipeline(task="question-answering", model=model, config=config, tokenizer=tokenizer, framework="tf")
             self._test_multicolumn_pipeline(nlp, valid_samples, invalid_samples, mandatory_output_keys)
+
+
+class PipelineCommonTests(unittest.TestCase):
+
+    pipelines = (
+        NerPipeline,
+        FeatureExtractionPipeline,
+        QuestionAnsweringPipeline,
+        FillMaskPipeline,
+        TextClassificationPipeline,
+    )
+
+    @slow
+    @require_tf
+    def test_tf_defaults(self):
+        # Test that pipelines can be correctly loaded without any argument
+        for default_pipeline in self.pipelines:
+            with self.subTest(msg="Testing Torch defaults with PyTorch and {}".format(default_pipeline.task)):
+                default_pipeline(framework="tf")
+
+    @slow
+    @require_torch
+    def test_pt_defaults(self):
+        # Test that pipelines can be correctly loaded without any argument
+        for default_pipeline in self.pipelines:
+            with self.subTest(msg="Testing Torch defaults with PyTorch and {}".format(default_pipeline.task)):
+                default_pipeline(framework="pt")

tests/test_tokenization_bert_japanese.py

@@ -130,7 +130,6 @@ class BertJapaneseTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
         assert encoded_pair == [2] + text + [3] + text_2 + [3]
 
 
-@custom_tokenizers
 class BertJapaneseCharacterTokenizationTest(TokenizerTesterMixin, unittest.TestCase):
 
     tokenizer_class = BertJapaneseTokenizer