[Test refactor 1/5] Per-folder tests reorganization (#15725)

* Per-folder tests reorganization

Co-authored-by: sgugger <sylvain.gugger@gmail.com>
Co-authored-by: Stas Bekman <stas@stason.org>

.github/workflows/self-scheduled.yml

@@ -16,6 +16,8 @@ env:
   MKL_NUM_THREADS: 16
   PYTEST_TIMEOUT: 600
   SIGOPT_API_TOKEN: ${{ secrets.SIGOPT_API_TOKEN }}
+  TF_FORCE_GPU_ALLOW_GROWTH: true
+  RUN_PT_TF_CROSS_TESTS: 1
 
 jobs:
   run_all_tests_torch_gpu:

conftest.py

@@ -22,7 +22,7 @@ from os.path import abspath, dirname, join
 
 # allow having multiple repository checkouts and not needing to remember to rerun
 # 'pip install -e .[dev]' when switching between checkouts and running tests.
-git_repo_path = abspath(join(dirname(dirname(__file__)), "src"))
+git_repo_path = abspath(join(dirname(__file__), "src"))
 sys.path.insert(1, git_repo_path)
 
 # silence FutureWarning warnings in tests since often we can't act on them until

src/transformers/commands/add_new_model.py

@@ -131,7 +131,7 @@ class AddNewModelCommand(BaseTransformersCLICommand):
 
             shutil.move(
                 f"{directory}/test_modeling_{lowercase_model_name}.py",
-                f"{path_to_transformer_root}/tests/test_modeling_{lowercase_model_name}.py",
+                f"{path_to_transformer_root}/tests/{lowercase_model_name}/test_modeling_{lowercase_model_name}.py",
             )
         else:
             os.remove(f"{directory}/modeling_{lowercase_model_name}.py")
@@ -148,7 +148,7 @@ class AddNewModelCommand(BaseTransformersCLICommand):
 
             shutil.move(
                 f"{directory}/test_modeling_tf_{lowercase_model_name}.py",
-                f"{path_to_transformer_root}/tests/test_modeling_tf_{lowercase_model_name}.py",
+                f"{path_to_transformer_root}/tests/{lowercase_model_name}/test_modeling_tf_{lowercase_model_name}.py",
             )
         else:
             os.remove(f"{directory}/modeling_tf_{lowercase_model_name}.py")
@@ -165,7 +165,7 @@ class AddNewModelCommand(BaseTransformersCLICommand):
 
             shutil.move(
                 f"{directory}/test_modeling_flax_{lowercase_model_name}.py",
-                f"{path_to_transformer_root}/tests/test_modeling_flax_{lowercase_model_name}.py",
+                f"{path_to_transformer_root}/tests/{lowercase_model_name}/test_modeling_flax_{lowercase_model_name}.py",
             )
         else:
             os.remove(f"{directory}/modeling_flax_{lowercase_model_name}.py")

src/transformers/commands/add_new_model_like.py

@@ -554,7 +554,7 @@ def get_model_files(model_type: str, frameworks: Optional[List[str]] = None) ->
     ]
     test_files = filter_framework_files(test_files, frameworks=frameworks)
     # Add the test directory
-    test_files = [REPO_PATH / "tests" / f for f in test_files]
+    test_files = [REPO_PATH / "tests" / module_name / f for f in test_files]
     # Filter by existing files
     test_files = [f for f in test_files if f.exists()]
 

src/transformers/testing_utils.py

@@ -689,6 +689,10 @@ def get_tests_dir(append_path=None):
     # this function caller's __file__
     caller__file__ = inspect.stack()[1][1]
     tests_dir = os.path.abspath(os.path.dirname(caller__file__))
+
+    while not tests_dir.endswith("tests"):
+        tests_dir = os.path.dirname(tests_dir)
+
     if append_path:
         return os.path.join(tests_dir, append_path)
     else:
@@ -1270,10 +1274,10 @@ def pytest_terminal_summary_main(tr, id):
     orig_tbstyle = config.option.tbstyle
     orig_reportchars = tr.reportchars
 
-    dir = "reports"
+    dir = f"reports/{id}"
     Path(dir).mkdir(parents=True, exist_ok=True)
     report_files = {
-        k: f"{dir}/{id}_{k}.txt"
+        k: f"{dir}/{k}.txt"
         for k in [
             "durations",
             "errors",

templates/adding_a_new_model/cookiecutter-template-{{cookiecutter.modelname}}/test_modeling_flax_{{cookiecutter.lowercase_modelname}}.py

@@ -20,8 +20,8 @@ import unittest
 from transformers import is_flax_available, {{cookiecutter.camelcase_modelname}}Config
 from transformers.testing_utils import require_flax, slow
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor
 
 if is_flax_available():
     import numpy as np
@@ -345,8 +345,8 @@ from transformers import (
 )
 from transformers.testing_utils import require_sentencepiece, require_flax, require_tokenizers, slow
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor
 
 
 if is_flax_available():

templates/adding_a_new_model/cookiecutter-template-{{cookiecutter.modelname}}/test_modeling_tf_{{cookiecutter.lowercase_modelname}}.py

@@ -20,8 +20,8 @@ import unittest
 from transformers import is_tf_available, {{cookiecutter.camelcase_modelname}}Config
 from transformers.testing_utils import require_tf, slow
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_tf_common import TFModelTesterMixin, floats_tensor, ids_tensor
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_tf_common import TFModelTesterMixin, floats_tensor, ids_tensor
 
 
 if is_tf_available():
@@ -360,8 +360,8 @@ from transformers import (
 )
 from transformers.testing_utils import require_sentencepiece, require_tf, require_tokenizers, slow
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_tf_common import TFModelTesterMixin, ids_tensor
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor
 
 
 if is_tf_available():

templates/adding_a_new_model/cookiecutter-template-{{cookiecutter.modelname}}/test_modeling_{{cookiecutter.lowercase_modelname}}.py

@@ -18,13 +18,13 @@
 {% if cookiecutter.is_encoder_decoder_model == "False" -%}
 import unittest
 
-from tests.test_modeling_common import floats_tensor
+from ..test_modeling_common import floats_tensor
 from transformers import is_torch_available
 from transformers.testing_utils import require_torch, slow, torch_device
 
 from transformers import {{cookiecutter.camelcase_modelname}}Config
-from .test_configuration_common import ConfigTester
-from .test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
 
 
 if is_torch_available():
@@ -489,9 +489,9 @@ from transformers import is_torch_available
 from transformers.file_utils import cached_property
 from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device
 
-from .test_configuration_common import ConfigTester
-from .test_generation_utils import GenerationTesterMixin
-from .test_modeling_common import ModelTesterMixin, ids_tensor
+from ..test_configuration_common import ConfigTester
+from ..test_generation_utils import GenerationTesterMixin
+from ..test_modeling_common import ModelTesterMixin, ids_tensor
 
 
 if is_torch_available():

tests/albert/test_modeling_albert.py

@@ -20,8 +20,8 @@ from transformers import AlbertConfig, is_torch_available
 from transformers.models.auto import get_values
 from transformers.testing_utils import require_torch, slow, torch_device
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
 
 
 if is_torch_available():

tests/albert/test_modeling_flax_albert.py

@@ -19,7 +19,7 @@ import numpy as np
 from transformers import AlbertConfig, is_flax_available
 from transformers.testing_utils import require_flax, slow
 
-from .test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor, random_attention_mask
+from ..test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor, random_attention_mask
 
 
 if is_flax_available():

tests/albert/test_modeling_tf_albert.py

@@ -20,8 +20,8 @@ from transformers import AlbertConfig, is_tf_available
 from transformers.models.auto import get_values
 from transformers.testing_utils import require_tf, slow
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_tf_common import TFModelTesterMixin, ids_tensor
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor
 
 
 if is_tf_available():

tests/albert/test_tokenization_albert.py

@@ -15,14 +15,15 @@
 
 import os
 import unittest
+from os.path import dirname
 
 from transformers import AlbertTokenizer, AlbertTokenizerFast
 from transformers.testing_utils import require_sentencepiece, require_tokenizers, slow
 
-from .test_tokenization_common import TokenizerTesterMixin
+from ..test_tokenization_common import TokenizerTesterMixin
 
 
-SAMPLE_VOCAB = os.path.join(os.path.dirname(os.path.abspath(__file__)), "fixtures/spiece.model")
+SAMPLE_VOCAB = os.path.join(dirname(dirname(os.path.abspath(__file__))), "fixtures/spiece.model")
 
 
 @require_sentencepiece

tests/auto/test_configuration_auto.py

@@ -27,12 +27,12 @@ from transformers.models.roberta.configuration_roberta import RobertaConfig
 from transformers.testing_utils import DUMMY_UNKNOWN_IDENTIFIER
 
 
-sys.path.append(str(Path(__file__).parent.parent / "utils"))
+sys.path.append(str(Path(__file__).parent.parent.parent / "utils"))
 
 from test_module.custom_configuration import CustomConfig  # noqa E402
 
 
-SAMPLE_ROBERTA_CONFIG = os.path.join(os.path.dirname(os.path.abspath(__file__)), "fixtures/dummy-config.json")
+SAMPLE_ROBERTA_CONFIG = os.path.join(os.path.dirname(os.path.abspath(__file__)), "../fixtures/dummy-config.json")
 
 
 class AutoConfigTest(unittest.TestCase):

tests/auto/test_feature_extraction_auto.py

@@ -31,17 +31,17 @@ from transformers import (
 from transformers.testing_utils import DUMMY_UNKNOWN_IDENTIFIER
 
 
-sys.path.append(str(Path(__file__).parent.parent / "utils"))
+sys.path.append(str(Path(__file__).parent.parent.parent / "utils"))
 
 from test_module.custom_configuration import CustomConfig  # noqa E402
 from test_module.custom_feature_extraction import CustomFeatureExtractor  # noqa E402
 
 
-SAMPLE_FEATURE_EXTRACTION_CONFIG_DIR = os.path.join(os.path.dirname(os.path.abspath(__file__)), "fixtures")
+SAMPLE_FEATURE_EXTRACTION_CONFIG_DIR = os.path.join(os.path.dirname(os.path.abspath(__file__)), "../fixtures")
 SAMPLE_FEATURE_EXTRACTION_CONFIG = os.path.join(
-    os.path.dirname(os.path.abspath(__file__)), "fixtures/dummy_feature_extractor_config.json"
+    os.path.dirname(os.path.abspath(__file__)), "../fixtures/dummy_feature_extractor_config.json"
 )
-SAMPLE_CONFIG = os.path.join(os.path.dirname(os.path.abspath(__file__)), "fixtures/dummy-config.json")
+SAMPLE_CONFIG = os.path.join(os.path.dirname(os.path.abspath(__file__)), "../fixtures/dummy-config.json")
 
 
 class AutoFeatureExtractorTest(unittest.TestCase):

tests/auto/test_modeling_auto.py

@@ -29,10 +29,10 @@ from transformers.testing_utils import (
     slow,
 )
 
-from .test_modeling_bert import BertModelTester
+from ..bert.test_modeling_bert import BertModelTester
 
 
-sys.path.append(str(Path(__file__).parent.parent / "utils"))
+sys.path.append(str(Path(__file__).parent.parent.parent / "utils"))
 
 from test_module.custom_configuration import CustomConfig  # noqa E402
 

tests/auto/test_modeling_tf_auto.py

@@ -26,7 +26,7 @@ from transformers.testing_utils import (
     slow,
 )
 
-from .test_modeling_bert import BertModelTester
+from ..bert.test_modeling_bert import BertModelTester
 
 
 if is_tf_available():

tests/auto/test_processor_auto.py

@@ -41,7 +41,7 @@ from transformers.testing_utils import PASS, USER, is_staging_test
 from transformers.tokenization_utils import TOKENIZER_CONFIG_FILE
 
 
-sys.path.append(str(Path(__file__).parent.parent / "utils"))
+sys.path.append(str(Path(__file__).parent.parent.parent / "utils"))
 
 from test_module.custom_configuration import CustomConfig  # noqa E402
 from test_module.custom_feature_extraction import CustomFeatureExtractor  # noqa E402
@@ -50,11 +50,11 @@ from test_module.custom_tokenization import CustomTokenizer  # noqa E402
 
 
 SAMPLE_PROCESSOR_CONFIG = os.path.join(
-    os.path.dirname(os.path.abspath(__file__)), "fixtures/dummy_feature_extractor_config.json"
+    os.path.dirname(os.path.abspath(__file__)), "../fixtures/dummy_feature_extractor_config.json"
 )
-SAMPLE_VOCAB = os.path.join(os.path.dirname(os.path.abspath(__file__)), "fixtures/vocab.json")
+SAMPLE_VOCAB = os.path.join(os.path.dirname(os.path.abspath(__file__)), "../fixtures/vocab.json")
 
-SAMPLE_PROCESSOR_CONFIG_DIR = os.path.join(os.path.dirname(os.path.abspath(__file__)), "fixtures")
+SAMPLE_PROCESSOR_CONFIG_DIR = os.path.join(os.path.dirname(os.path.abspath(__file__)), "../fixtures")
 
 
 class AutoFeatureExtractorTest(unittest.TestCase):

tests/bart/test_modeling_bart.py

@@ -25,9 +25,9 @@ from transformers import BartConfig, is_torch_available
 from transformers.file_utils import cached_property
 from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device
 
-from .test_configuration_common import ConfigTester
-from .test_generation_utils import GenerationTesterMixin
-from .test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor
+from ..generation.test_generation_utils import GenerationTesterMixin
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor
 
 
 if is_torch_available():

tests/bart/test_modeling_flax_bart.py

@@ -20,8 +20,8 @@ import timeout_decorator  # noqa
 from transformers import BartConfig, is_flax_available
 from transformers.testing_utils import require_flax, slow
 
-from .test_generation_flax_utils import FlaxGenerationTesterMixin
-from .test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor
+from ..generation.test_generation_flax_utils import FlaxGenerationTesterMixin
+from ..test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor
 
 
 if is_flax_available():

tests/bart/test_modeling_tf_bart.py

@@ -21,9 +21,9 @@ from transformers import BartConfig, BartTokenizer, is_tf_available
 from transformers.file_utils import cached_property
 from transformers.testing_utils import require_tf, slow
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_tf_common import TFModelTesterMixin, ids_tensor
-from .test_modeling_tf_core import TFCoreModelTesterMixin
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor
+from ..utils.test_modeling_tf_core import TFCoreModelTesterMixin
 
 
 if is_tf_available():

tests/bart/test_tokenization_bart.py

@@ -20,7 +20,7 @@ from transformers.file_utils import cached_property
 from transformers.models.roberta.tokenization_roberta import VOCAB_FILES_NAMES
 from transformers.testing_utils import require_tokenizers, require_torch
 
-from .test_tokenization_common import TokenizerTesterMixin, filter_roberta_detectors
+from ..test_tokenization_common import TokenizerTesterMixin, filter_roberta_detectors
 
 
 @require_tokenizers

tests/barthez/test_tokenization_barthez.py

@@ -18,7 +18,7 @@ import unittest
 from transformers import BarthezTokenizer, BarthezTokenizerFast, BatchEncoding
 from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow
 
-from .test_tokenization_common import TokenizerTesterMixin
+from ..test_tokenization_common import TokenizerTesterMixin
 
 
 @require_tokenizers

tests/bartpho/test_tokenization_bartpho.py

@@ -15,13 +15,14 @@
 
 import os
 import unittest
+from os.path import dirname
 
 from transformers.models.bartpho.tokenization_bartpho import VOCAB_FILES_NAMES, BartphoTokenizer
 
-from .test_tokenization_common import TokenizerTesterMixin
+from ..test_tokenization_common import TokenizerTesterMixin
 
 
-SAMPLE_VOCAB = os.path.join(os.path.dirname(os.path.abspath(__file__)), "fixtures/test_sentencepiece_bpe.model")
+SAMPLE_VOCAB = os.path.join(dirname(dirname(os.path.abspath(__file__))), "fixtures/test_sentencepiece_bpe.model")
 
 
 class BartphoTokenizerTest(TokenizerTesterMixin, unittest.TestCase):

tests/beit/test_feature_extraction_beit.py

@@ -22,7 +22,7 @@ from datasets import load_dataset
 from transformers.file_utils import is_torch_available, is_vision_available
 from transformers.testing_utils import require_torch, require_vision
 
-from .test_feature_extraction_common import FeatureExtractionSavingTestMixin, prepare_image_inputs
+from ..test_feature_extraction_common import FeatureExtractionSavingTestMixin, prepare_image_inputs
 
 
 if is_torch_available():

tests/beit/test_modeling_beit.py

@@ -25,8 +25,8 @@ from transformers.file_utils import cached_property, is_torch_available, is_visi
 from transformers.models.auto import get_values
 from transformers.testing_utils import require_torch, require_vision, slow, torch_device
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor
 
 
 if is_torch_available():

tests/beit/test_modeling_flax_beit.py

@@ -21,8 +21,8 @@ from transformers import BeitConfig
 from transformers.file_utils import cached_property, is_flax_available, is_vision_available
 from transformers.testing_utils import require_flax, require_vision, slow
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_flax_common import FlaxModelTesterMixin, floats_tensor, ids_tensor
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_flax_common import FlaxModelTesterMixin, floats_tensor, ids_tensor
 
 
 if is_flax_available():

tests/bert/test_modeling_bert.py

@@ -20,9 +20,9 @@ from transformers import BertConfig, is_torch_available
 from transformers.models.auto import get_values
 from transformers.testing_utils import require_torch, require_torch_gpu, slow, torch_device
 
-from .test_configuration_common import ConfigTester
-from .test_generation_utils import GenerationTesterMixin
-from .test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
+from ..generation.test_generation_utils import GenerationTesterMixin
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
 
 
 if is_torch_available():

tests/bert/test_modeling_flax_bert.py

@@ -19,7 +19,7 @@ import numpy as np
 from transformers import BertConfig, is_flax_available
 from transformers.testing_utils import require_flax, slow
 
-from .test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor, random_attention_mask
+from ..test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor, random_attention_mask
 
 
 if is_flax_available():

tests/bert/test_modeling_tf_bert.py

@@ -20,9 +20,9 @@ from transformers import BertConfig, is_tf_available
 from transformers.models.auto import get_values
 from transformers.testing_utils import require_tf, slow
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_tf_common import TFModelTesterMixin, floats_tensor, ids_tensor
-from .test_modeling_tf_core import TFCoreModelTesterMixin
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_tf_common import TFModelTesterMixin, floats_tensor, ids_tensor
+from ..utils.test_modeling_tf_core import TFCoreModelTesterMixin
 
 
 if is_tf_available():

tests/bert/test_tokenization_bert.py

@@ -29,7 +29,7 @@ from transformers.models.bert.tokenization_bert import (
 )
 from transformers.testing_utils import require_tokenizers, slow
 
-from .test_tokenization_common import TokenizerTesterMixin, filter_non_english
+from ..test_tokenization_common import TokenizerTesterMixin, filter_non_english
 
 
 @require_tokenizers

tests/bert_generation/test_modeling_bert_generation.py

@@ -19,9 +19,9 @@ import unittest
 from transformers import BertGenerationConfig, is_torch_available
 from transformers.testing_utils import require_torch, slow, torch_device
 
-from .test_configuration_common import ConfigTester
-from .test_generation_utils import GenerationTesterMixin
-from .test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
+from ..generation.test_generation_utils import GenerationTesterMixin
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
 
 
 if is_torch_available():

tests/bert_generation/test_tokenization_bert_generation.py

@@ -15,17 +15,18 @@
 
 import os
 import unittest
+from os.path import dirname
 
 from transformers import BertGenerationTokenizer
 from transformers.file_utils import cached_property
 from transformers.testing_utils import require_sentencepiece, require_torch, slow
 
-from .test_tokenization_common import TokenizerTesterMixin
+from ..test_tokenization_common import TokenizerTesterMixin
 
 
 SPIECE_UNDERLINE = "▁"
 
-SAMPLE_VOCAB = os.path.join(os.path.dirname(os.path.abspath(__file__)), "fixtures/test_sentencepiece.model")
+SAMPLE_VOCAB = os.path.join(dirname(dirname(os.path.abspath(__file__))), "fixtures/test_sentencepiece.model")
 
 
 @require_sentencepiece

tests/bert_japanese/test_tokenization_bert_japanese.py

@@ -29,7 +29,7 @@ from transformers.models.bert_japanese.tokenization_bert_japanese import (
 )
 from transformers.testing_utils import custom_tokenizers
 
-from .test_tokenization_common import TokenizerTesterMixin
+from ..test_tokenization_common import TokenizerTesterMixin
 
 
 @custom_tokenizers

tests/bertweet/test_tokenization_bertweet.py

@@ -18,7 +18,7 @@ import unittest
 
 from transformers.models.bertweet.tokenization_bertweet import VOCAB_FILES_NAMES, BertweetTokenizer
 
-from .test_tokenization_common import TokenizerTesterMixin
+from ..test_tokenization_common import TokenizerTesterMixin
 
 
 class BertweetTokenizationTest(TokenizerTesterMixin, unittest.TestCase):

tests/big_bird/test_modeling_big_bird.py

@@ -17,14 +17,13 @@
 
 import unittest
 
-from tests.test_modeling_common import floats_tensor
 from transformers import BigBirdConfig, is_torch_available
 from transformers.models.auto import get_values
 from transformers.models.big_bird.tokenization_big_bird import BigBirdTokenizer
 from transformers.testing_utils import require_torch, slow, torch_device
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
 
 
 if is_torch_available():

tests/big_bird/test_modeling_flax_big_bird.py

@@ -19,7 +19,7 @@ import numpy as np
 from transformers import BigBirdConfig, is_flax_available
 from transformers.testing_utils import require_flax, slow
 
-from .test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor, random_attention_mask
+from ..test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor, random_attention_mask
 
 
 if is_flax_available():

tests/big_bird/test_tokenization_big_bird.py

@@ -15,17 +15,18 @@
 
 import os
 import unittest
+from os.path import dirname
 
 from transformers import BigBirdTokenizer, BigBirdTokenizerFast
 from transformers.file_utils import cached_property
 from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow
 
-from .test_tokenization_common import TokenizerTesterMixin
+from ..test_tokenization_common import TokenizerTesterMixin
 
 
 SPIECE_UNDERLINE = "▁"
 
-SAMPLE_VOCAB = os.path.join(os.path.dirname(os.path.abspath(__file__)), "fixtures/test_sentencepiece.model")
+SAMPLE_VOCAB = os.path.join(dirname(dirname(os.path.abspath(__file__))), "fixtures/test_sentencepiece.model")
 
 
 @require_sentencepiece

tests/bigbird_pegasus/test_modeling_bigbird_pegasus.py

@@ -22,9 +22,9 @@ import unittest
 from transformers import BigBirdPegasusConfig, is_torch_available
 from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device
 
-from .test_configuration_common import ConfigTester
-from .test_generation_utils import GenerationTesterMixin
-from .test_modeling_common import ModelTesterMixin, ids_tensor
+from ..generation.test_generation_utils import GenerationTesterMixin
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import ModelTesterMixin, ids_tensor
 
 
 if is_torch_available():

tests/blenderbot/test_modeling_blenderbot.py

@@ -21,9 +21,9 @@ from transformers import BlenderbotConfig, is_torch_available
 from transformers.file_utils import cached_property
 from transformers.testing_utils import require_sentencepiece, require_tokenizers, require_torch, slow, torch_device
 
-from .test_configuration_common import ConfigTester
-from .test_generation_utils import GenerationTesterMixin
-from .test_modeling_common import ModelTesterMixin, ids_tensor
+from ..generation.test_generation_utils import GenerationTesterMixin
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import ModelTesterMixin, ids_tensor
 
 
 if is_torch_available():

tests/blenderbot/test_modeling_flax_blenderbot.py

@@ -20,8 +20,8 @@ import timeout_decorator  # noqa
 from transformers import BlenderbotConfig, is_flax_available
 from transformers.testing_utils import jax_device, require_flax, slow
 
-from .test_generation_flax_utils import FlaxGenerationTesterMixin
-from .test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor
+from ..generation.test_generation_flax_utils import FlaxGenerationTesterMixin
+from ..test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor
 
 
 if is_flax_available():

tests/blenderbot/test_modeling_tf_blenderbot.py

@@ -20,8 +20,8 @@ from transformers import BlenderbotConfig, BlenderbotTokenizer, is_tf_available
 from transformers.file_utils import cached_property
 from transformers.testing_utils import require_tf, require_tokenizers, slow
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_tf_common import TFModelTesterMixin, ids_tensor
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor
 
 
 if is_tf_available():

tests/blenderbot_small/test_modeling_blenderbot_small.py

@@ -21,9 +21,9 @@ from transformers import BlenderbotSmallConfig, is_torch_available
 from transformers.file_utils import cached_property
 from transformers.testing_utils import require_torch, slow, torch_device
 
-from .test_configuration_common import ConfigTester
-from .test_generation_utils import GenerationTesterMixin
-from .test_modeling_common import ModelTesterMixin, ids_tensor
+from ..generation.test_generation_utils import GenerationTesterMixin
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import ModelTesterMixin, ids_tensor
 
 
 if is_torch_available():

tests/blenderbot_small/test_modeling_flax_blenderbot_small.py

@@ -20,8 +20,8 @@ import timeout_decorator  # noqa
 from transformers import BlenderbotSmallConfig, is_flax_available
 from transformers.testing_utils import require_flax, slow
 
-from .test_generation_flax_utils import FlaxGenerationTesterMixin
-from .test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor
+from ..generation.test_generation_flax_utils import FlaxGenerationTesterMixin
+from ..test_modeling_flax_common import FlaxModelTesterMixin, ids_tensor
 
 
 if is_flax_available():

tests/blenderbot_small/test_modeling_tf_blenderbot_small.py

@@ -20,8 +20,8 @@ from transformers import BlenderbotSmallConfig, BlenderbotSmallTokenizer, is_tf_
 from transformers.file_utils import cached_property
 from transformers.testing_utils import require_tf, require_tokenizers, slow
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_tf_common import TFModelTesterMixin, ids_tensor
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor
 
 
 if is_tf_available():

tests/blenderbot_small/test_tokenization_blenderbot_small.py

@@ -23,7 +23,7 @@ from transformers.models.blenderbot_small.tokenization_blenderbot_small import (
     BlenderbotSmallTokenizer,
 )
 
-from .test_tokenization_common import TokenizerTesterMixin
+from ..test_tokenization_common import TokenizerTesterMixin
 
 
 class BlenderbotSmallTokenizerTest(TokenizerTesterMixin, unittest.TestCase):

tests/byt5/test_tokenization_byt5.py

@@ -24,7 +24,7 @@ from typing import Tuple
 from transformers import AddedToken, BatchEncoding, ByT5Tokenizer
 from transformers.file_utils import cached_property, is_tf_available, is_torch_available
 
-from .test_tokenization_common import TokenizerTesterMixin
+from ..test_tokenization_common import TokenizerTesterMixin
 
 
 if is_torch_available():

tests/camembert/test_tokenization_camembert.py

@@ -15,16 +15,17 @@
 
 import os
 import unittest
+from os.path import dirname
 
 from transformers import CamembertTokenizer, CamembertTokenizerFast
 from transformers.file_utils import is_torch_available
 from transformers.testing_utils import require_sentencepiece, require_tokenizers, slow
 
-from .test_tokenization_common import TokenizerTesterMixin
+from ..test_tokenization_common import TokenizerTesterMixin
 
 
-SAMPLE_VOCAB = os.path.join(os.path.dirname(os.path.abspath(__file__)), "fixtures/test_sentencepiece.model")
-SAMPLE_BPE_VOCAB = os.path.join(os.path.dirname(os.path.abspath(__file__)), "fixtures/test_sentencepiece_bpe.model")
+SAMPLE_VOCAB = os.path.join(dirname(dirname(os.path.abspath(__file__))), "fixtures/test_sentencepiece.model")
+SAMPLE_BPE_VOCAB = os.path.join(dirname(dirname(os.path.abspath(__file__))), "fixtures/test_sentencepiece_bpe.model")
 
 FRAMEWORK = "pt" if is_torch_available() else "tf"
 

tests/canine/test_modeling_canine.py

@@ -21,8 +21,8 @@ from typing import List, Tuple
 from transformers import CanineConfig, is_torch_available
 from transformers.testing_utils import require_torch, slow, torch_device
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_common import ModelTesterMixin, _config_zero_init, global_rng, ids_tensor, random_attention_mask
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import ModelTesterMixin, _config_zero_init, global_rng, ids_tensor, random_attention_mask
 
 
 if is_torch_available():

tests/canine/test_tokenization_canine.py

@@ -24,7 +24,7 @@ from transformers.file_utils import cached_property
 from transformers.testing_utils import require_tokenizers, require_torch
 from transformers.tokenization_utils import AddedToken
 
-from .test_tokenization_common import TokenizerTesterMixin
+from ..test_tokenization_common import TokenizerTesterMixin
 
 
 class CanineTokenizationTest(TokenizerTesterMixin, unittest.TestCase):

tests/clip/test_feature_extraction_clip.py

@@ -21,7 +21,7 @@ import numpy as np
 from transformers.file_utils import is_torch_available, is_vision_available
 from transformers.testing_utils import require_torch, require_vision
 
-from .test_feature_extraction_common import FeatureExtractionSavingTestMixin
+from ..test_feature_extraction_common import FeatureExtractionSavingTestMixin
 
 
 if is_torch_available():

tests/clip/test_modeling_clip.py

@@ -36,8 +36,14 @@ from transformers.testing_utils import (
     torch_device,
 )
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_common import ModelTesterMixin, _config_zero_init, floats_tensor, ids_tensor, random_attention_mask
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import (
+    ModelTesterMixin,
+    _config_zero_init,
+    floats_tensor,
+    ids_tensor,
+    random_attention_mask,
+)
 
 
 if is_torch_available():

tests/clip/test_modeling_flax_clip.py

@@ -8,7 +8,7 @@ import transformers
 from transformers import CLIPConfig, CLIPTextConfig, CLIPVisionConfig, is_flax_available, is_torch_available
 from transformers.testing_utils import is_pt_flax_cross_test, require_flax, slow
 
-from .test_modeling_flax_common import FlaxModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
+from ..test_modeling_flax_common import FlaxModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
 
 
 if is_flax_available():

tests/clip/test_modeling_tf_clip.py

@@ -26,8 +26,8 @@ from transformers import CLIPConfig, CLIPTextConfig, CLIPVisionConfig
 from transformers.file_utils import is_tf_available, is_vision_available
 from transformers.testing_utils import is_pt_tf_cross_test, require_tf, require_vision, slow
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_tf_common import TFModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_tf_common import TFModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
 
 
 if is_tf_available():

tests/clip/test_tokenization_clip.py

@@ -22,7 +22,7 @@ from transformers import CLIPTokenizer, CLIPTokenizerFast
 from transformers.models.clip.tokenization_clip import VOCAB_FILES_NAMES
 from transformers.testing_utils import require_ftfy, require_tokenizers
 
-from .test_tokenization_common import TokenizerTesterMixin
+from ..test_tokenization_common import TokenizerTesterMixin
 
 
 @require_tokenizers

tests/convbert/test_modeling_convbert.py

@@ -17,13 +17,12 @@ import os
 import tempfile
 import unittest
 
-from tests.test_modeling_common import floats_tensor
 from transformers import ConvBertConfig, is_torch_available
 from transformers.models.auto import get_values
 from transformers.testing_utils import require_torch, require_torch_gpu, slow, torch_device
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor, random_attention_mask
 
 
 if is_torch_available():

tests/convbert/test_modeling_tf_convbert.py

@@ -19,8 +19,8 @@ import unittest
 from transformers import ConvBertConfig, is_tf_available
 from transformers.testing_utils import require_tf, slow
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_tf_common import TFModelTesterMixin, ids_tensor
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor
 
 
 if is_tf_available():

tests/convnext/test_feature_extraction_convnext.py

@@ -21,7 +21,7 @@ import numpy as np
 from transformers.file_utils import is_torch_available, is_vision_available
 from transformers.testing_utils import require_torch, require_vision
 
-from .test_feature_extraction_common import FeatureExtractionSavingTestMixin, prepare_image_inputs
+from ..test_feature_extraction_common import FeatureExtractionSavingTestMixin, prepare_image_inputs
 
 
 if is_torch_available():

tests/convnext/test_modeling_convnext.py

@@ -23,8 +23,8 @@ from transformers import ConvNextConfig
 from transformers.file_utils import cached_property, is_torch_available, is_vision_available
 from transformers.testing_utils import require_torch, require_vision, slow, torch_device
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import ModelTesterMixin, floats_tensor, ids_tensor
 
 
 if is_torch_available():

tests/cpm/test_tokenization_cpm.py

@@ -16,7 +16,7 @@
 from transformers.models.cpm.tokenization_cpm import CpmTokenizer
 from transformers.testing_utils import custom_tokenizers
 
-from .test_modeling_xlnet import XLNetModelTest
+from ..xlnet.test_modeling_xlnet import XLNetModelTest
 
 
 @custom_tokenizers

tests/ctrl/test_modeling_ctrl.py

@@ -18,9 +18,9 @@ import unittest
 from transformers import CTRLConfig, is_torch_available
 from transformers.testing_utils import require_torch, slow, torch_device
 
-from .test_configuration_common import ConfigTester
-from .test_generation_utils import GenerationTesterMixin
-from .test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
+from ..generation.test_generation_utils import GenerationTesterMixin
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_common import ModelTesterMixin, ids_tensor, random_attention_mask
 
 
 if is_torch_available():

tests/ctrl/test_modeling_tf_ctrl.py

@@ -19,8 +19,8 @@ import unittest
 from transformers import CTRLConfig, is_tf_available
 from transformers.testing_utils import require_tf, slow
 
-from .test_configuration_common import ConfigTester
-from .test_modeling_tf_common import TFModelTesterMixin, ids_tensor
+from ..test_configuration_common import ConfigTester
+from ..test_modeling_tf_common import TFModelTesterMixin, ids_tensor
 
 
 if is_tf_available():

tests/ctrl/test_tokenization_ctrl.py

@@ -19,7 +19,7 @@ import unittest
 
 from transformers.models.ctrl.tokenization_ctrl import VOCAB_FILES_NAMES, CTRLTokenizer
 
-from .test_tokenization_common import TokenizerTesterMixin
+from ..test_tokenization_common import TokenizerTesterMixin
 
 
 class CTRLTokenizationTest(TokenizerTesterMixin, unittest.TestCase):