remove USE_CUDA (huggingface#7861)

.github/workflows/self-push.yml

@@ -59,7 +59,6 @@ jobs:
         TF_FORCE_GPU_ALLOW_GROWTH: "true"
         # TF_GPU_MEMORY_LIMIT: 4096
         OMP_NUM_THREADS: 1
-        USE_CUDA: yes
       run: |
         source .env/bin/activate
         python -m pytest -n 2 --dist=loadfile -s ./tests/
@@ -110,7 +109,6 @@ jobs:
           TF_FORCE_GPU_ALLOW_GROWTH: "true"
           # TF_GPU_MEMORY_LIMIT: 4096
           OMP_NUM_THREADS: 1
-          USE_CUDA: yes
         run: |
           source .env/bin/activate
           python -m pytest -n 2 --dist=loadfile -s ./tests/

.github/workflows/self-scheduled.yml

@@ -57,7 +57,6 @@ jobs:
         TF_FORCE_GPU_ALLOW_GROWTH: "true"
         OMP_NUM_THREADS: 1
         RUN_SLOW: yes
-        USE_CUDA: yes
       run: |
         source .env/bin/activate
         python -m pytest -n 1 --dist=loadfile -s ./tests/
@@ -67,7 +66,6 @@ jobs:
         TF_FORCE_GPU_ALLOW_GROWTH: "true"
         OMP_NUM_THREADS: 1
         RUN_SLOW: yes
-        USE_CUDA: yes
       run: |
         source .env/bin/activate
         pip install -r examples/requirements.txt
@@ -120,7 +118,6 @@ jobs:
           TF_FORCE_GPU_ALLOW_GROWTH: "true"
           OMP_NUM_THREADS: 1
           RUN_SLOW: yes
-          USE_CUDA: yes
         run: |
           source .env/bin/activate
           python -m pytest -n 1 --dist=loadfile -s ./tests/
@@ -130,7 +127,6 @@ jobs:
           TF_FORCE_GPU_ALLOW_GROWTH: "true"
           OMP_NUM_THREADS: 1
           RUN_SLOW: yes
-          USE_CUDA: yes
         run: |
           source .env/bin/activate
           pip install -r examples/requirements.txt

docs/source/testing.rst

@@ -22,12 +22,12 @@ How transformers are tested
 
    * `self-hosted (push) <https://github.com/huggingface/transformers/blob/master/.github/workflows/self-push.yml>`__: runs fast tests on GPU only on commits on ``master``. It only runs if a commit on ``master`` has updated the code in one of the following folders: ``src``, ``tests``, ``.github`` (to prevent running on added model cards, notebooks, etc.)
 
-   * `self-hosted runner <https://github.com/huggingface/transformers/blob/master/.github/workflows/self-scheduled.yml>`__: runs slow tests on ``tests`` and ``examples``:
+   * `self-hosted runner <https://github.com/huggingface/transformers/blob/master/.github/workflows/self-scheduled.yml>`__: runs normal and slow tests on GPU in ``tests`` and ``examples``:
 
    .. code-block:: bash
 
-    RUN_SLOW=1 USE_CUDA=1 pytest tests/
-    RUN_SLOW=1 USE_CUDA=1 pytest examples/
+    RUN_SLOW=1 pytest tests/
+    RUN_SLOW=1 pytest examples/
 
    The results can be observed `here <https://github.com/huggingface/transformers/actions>`__.
 
@@ -393,7 +393,7 @@ On a GPU-enabled setup, to test in CPU-only mode add ``CUDA_VISIBLE_DEVICES=""``
                 
     CUDA_VISIBLE_DEVICES="" pytest tests/test_logging.py
 
-or if you have multiple gpus, you can tell which one to use in this test session, e.g. to use only the second gpu if you have gpus ``0`` and ``1``, you can run:
+or if you have multiple gpus, you can specify which one is to be used by ``pytest``. For example, to use only the second gpu if you have gpus ``0`` and ``1``, you can run:
 
 .. code-block:: bash
                 

scripts/fsmt/tests-to-run.sh

@@ -2,5 +2,5 @@
 
 # these scripts need to be run before any changes to FSMT-related code - it should cover all bases
 
-USE_CUDA=0 RUN_SLOW=1 pytest --disable-warnings tests/test_tokenization_fsmt.py tests/test_configuration_auto.py tests/test_modeling_fsmt.py examples/seq2seq/test_fsmt_bleu_score.py
-USE_CUDA=1 RUN_SLOW=1 pytest --disable-warnings tests/test_tokenization_fsmt.py tests/test_configuration_auto.py tests/test_modeling_fsmt.py examples/seq2seq/test_fsmt_bleu_score.py
+CUDA_VISIBLE_DEVICES="" RUN_SLOW=1 pytest --disable-warnings tests/test_tokenization_fsmt.py tests/test_configuration_auto.py tests/test_modeling_fsmt.py examples/seq2seq/test_fsmt_bleu_score.py
+RUN_SLOW=1 pytest --disable-warnings tests/test_tokenization_fsmt.py tests/test_configuration_auto.py tests/test_modeling_fsmt.py examples/seq2seq/test_fsmt_bleu_score.py

src/transformers/testing_utils.py

@@ -187,8 +187,10 @@ def require_torch_tpu(test_case):
 
 
 if _torch_available:
-    # Set the USE_CUDA environment variable to select a GPU.
-    torch_device = "cuda" if parse_flag_from_env("USE_CUDA") else "cpu"
+    # Set env var CUDA_VISIBLE_DEVICES="" to force cpu-mode
+    import torch
+
+    torch_device = "cuda" if torch.cuda.is_available() else "cpu"
 else:
     torch_device = None
 
@@ -485,9 +487,9 @@ def tearDown(self):
 def mockenv(**kwargs):
     """this is a convenience wrapper, that allows this:
 
-    @mockenv(USE_CUDA=True, USE_TF=False)
+    @mockenv(RUN_SLOW=True, USE_TF=False)
     def test_something():
-        use_cuda = os.getenv("USE_CUDA", False)
+        run_slow = os.getenv("RUN_SLOW", False)
         use_tf = os.getenv("USE_TF", False)
     """
     return unittest.mock.patch.dict(os.environ, kwargs)

tests/test_skip_decorators.py

@@ -23,10 +23,10 @@
 # the following 4 should be run. But since we have different CI jobs running
 # different configs, all combinations should get covered
 #
-# USE_CUDA=1 RUN_SLOW=1 pytest -rA tests/test_skip_decorators.py
-# USE_CUDA=0 RUN_SLOW=1 pytest -rA tests/test_skip_decorators.py
-# USE_CUDA=0 RUN_SLOW=0 pytest -rA tests/test_skip_decorators.py
-# USE_CUDA=1 RUN_SLOW=0 pytest -rA tests/test_skip_decorators.py
+# RUN_SLOW=1 pytest -rA tests/test_skip_decorators.py
+# RUN_SLOW=1 CUDA_VISIBLE_DEVICES="" pytest -rA tests/test_skip_decorators.py
+# RUN_SLOW=0 pytest -rA tests/test_skip_decorators.py
+# RUN_SLOW=0 CUDA_VISIBLE_DEVICES="" pytest -rA tests/test_skip_decorators.py
 
 import os
 import unittest