fix gpus default for Trainer.add_argparse_args (#6898)

(cherry picked from commit 9c9e2a0)

Author: awaelchli

CHANGELOG.md

@@ -5,10 +5,12 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 
-## [UnReleased] - 2021-MM-DD
+## [1.3.0] - 2021-MM-DD
 
 ### Added
 
+- Added utils for NaN/Inf detection for gradients and parameters ([#6834](https://github.com/PyTorchLightning/pytorch-lightning/pull/6834/))
+
 
 - Added more explicit exception message when trying to execute `trainer.test()` or `trainer.validate()` with `fast_dev_run=True` ([#6667](https://github.com/PyTorchLightning/pytorch-lightning/pull/6667))
 
@@ -81,8 +83,10 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 - Added support for `precision=64`, enabling training with double precision ([#6595](https://github.com/PyTorchLightning/pytorch-lightning/pull/6595))
 
+
 - Added support for DDP communication hooks ([#6736](https://github.com/PyTorchLightning/pytorch-lightning/issues/6736))
 
+
 - Added `artifact_location` argument to `MLFlowLogger` which will be passed to the `MlflowClient.create_experiment` call ([#6677](https://github.com/PyTorchLightning/pytorch-lightning/pull/6677))
 
 
@@ -111,6 +115,9 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 ### Deprecated
 
+- Deprecated `TrainerTrainingTricksMixin` in favor of a separate utilities module for NaN/Inf detection for gradients and parameters ([#6834](https://github.com/PyTorchLightning/pytorch-lightning/pull/6834/))
+
+
 - `period` has been deprecated in favor of `every_n_val_epochs` in the `ModelCheckpoint` callback ([#6146](https://github.com/PyTorchLightning/pytorch-lightning/pull/6146))
 
 
@@ -221,19 +228,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Fixed `EarlyStopping` logic when `min_epochs` or `min_steps` requirement is not met ([#6705](https://github.com/PyTorchLightning/pytorch-lightning/pull/6705))
 
 
-## [1.2.8] - 2021-04-13
-
-
-### Changed
-
-
-### Removed
-
-
-### Fixed
-
-
-- Fixed `EarlyStopping` logic when `min_epochs` or `min_steps` requirement is not met ([#6705](https://github.com/PyTorchLightning/pytorch-lightning/pull/6705))
+- Fixed `--gpus` default for parser returned by `Trainer.add_argparse_args` ([#6898](https://github.com/PyTorchLightning/pytorch-lightning/pull/6898))
 
 
 ## [1.2.7] - 2021-04-06

pytorch_lightning/utilities/argparse.py

@@ -186,7 +186,6 @@ def add_argparse_args(cls, parent_parser: ArgumentParser) -> ArgumentParser:
 
         if arg == 'gpus' or arg == 'tpu_cores':
             use_type = _gpus_allowed_type
-            arg_default = _gpus_arg_default
 
         # hack for types in (int, float)
         if len(arg_types) == 2 and int in set(arg_types) and float in set(arg_types):
@@ -238,13 +237,6 @@ def _gpus_allowed_type(x) -> Union[int, str]:
         return int(x)
 
 
-def _gpus_arg_default(x) -> Union[int, str]:
-    if ',' in x:
-        return str(x)
-    else:
-        return int(x)
-
-
 def _int_or_float_type(x) -> Union[int, float]:
     if '.' in str(x):
         return float(x)

pytorch_lightning/utilities/device_parser.py

@@ -59,11 +59,6 @@ def parse_gpu_ids(gpus: Optional[Union[int, str, List[int]]]) -> Optional[List[i
     If no GPUs are available but the value of gpus variable indicates request for GPUs
     then a MisconfigurationException is raised.
     """
-
-    # nothing was passed into the GPUs argument
-    if callable(gpus):
-        return None
-
     # Check that gpus param is None, Int, String or List
     _check_data_type(gpus)
 
@@ -97,10 +92,6 @@ def parse_tpu_cores(tpu_cores: Union[int, str, List]) -> Optional[Union[List[int
     Returns:
         a list of tpu_cores to be used or ``None`` if no TPU cores were requested
     """
-
-    if callable(tpu_cores):
-        return None
-
     _check_data_type(tpu_cores)
 
     if isinstance(tpu_cores, str):

tests/loggers/test_wandb.py

@@ -13,7 +13,6 @@
 # limitations under the License.
 import os
 import pickle
-import types
 from argparse import ArgumentParser
 from unittest import mock
 
@@ -172,11 +171,10 @@ def wrapper_something():
     params.wrapper_something_wo_name = lambda: lambda: '1'
     params.wrapper_something = wrapper_something
 
-    assert isinstance(params.gpus, types.FunctionType)
     params = WandbLogger._convert_params(params)
     params = WandbLogger._flatten_dict(params)
     params = WandbLogger._sanitize_callable_params(params)
-    assert params["gpus"] == '_gpus_arg_default'
+    assert params["gpus"] == "None"
     assert params["something"] == "something"
     assert params["wrapper_something"] == "wrapper_something"
     assert params["wrapper_something_wo_name"] == "<lambda>"

tests/trainer/test_trainer_cli.py

@@ -23,6 +23,7 @@
 import tests.helpers.utils as tutils
 from pytorch_lightning import Trainer
 from pytorch_lightning.utilities import argparse
+from tests.helpers.runif import RunIf
 
 
 @mock.patch('argparse.ArgumentParser.parse_args')
@@ -45,7 +46,7 @@ def test_default_args(mock_argparse, tmpdir):
 
 
 @pytest.mark.parametrize('cli_args', [['--accumulate_grad_batches=22'], ['--weights_save_path=./'], []])
-def test_add_argparse_args_redefined(cli_args):
+def test_add_argparse_args_redefined(cli_args: list):
     """Redefines some default Trainer arguments via the cli and
     tests the Trainer initialization correctness.
     """
@@ -90,7 +91,7 @@ def test_get_init_arguments_and_types():
 
 
 @pytest.mark.parametrize('cli_args', [['--callbacks=1', '--logger'], ['--foo', '--bar=1']])
-def test_add_argparse_args_redefined_error(cli_args, monkeypatch):
+def test_add_argparse_args_redefined_error(cli_args: list, monkeypatch):
     """Asserts thar an error raised in case of passing not default cli arguments."""
 
     class _UnkArgError(Exception):
@@ -171,27 +172,26 @@ def test_argparse_args_parsing(cli_args, expected):
     assert Trainer.from_argparse_args(args)
 
 
-@pytest.mark.parametrize(['cli_args', 'expected_gpu'], [
-    pytest.param('--gpus 1', [0]),
-    pytest.param('--gpus 0,', [0]),
+@pytest.mark.parametrize(['cli_args', 'expected_parsed', 'expected_device_ids'], [
+    pytest.param('', None, None),
+    pytest.param('--gpus 1', 1, [0]),
+    pytest.param('--gpus 0,', '0,', [0]),
 ])
-@pytest.mark.skipif(not torch.cuda.is_available(), reason="test requires GPU machine")
-def test_argparse_args_parsing_gpus(cli_args, expected_gpu):
+@RunIf(min_gpus=1)
+def test_argparse_args_parsing_gpus(cli_args, expected_parsed, expected_device_ids):
     """Test multi type argument with bool."""
     cli_args = cli_args.split(' ') if cli_args else []
     with mock.patch("argparse._sys.argv", ["any.py"] + cli_args):
         parser = ArgumentParser(add_help=False)
         parser = Trainer.add_argparse_args(parent_parser=parser)
         args = Trainer.parse_argparser(parser)
 
+    assert args.gpus == expected_parsed
     trainer = Trainer.from_argparse_args(args)
-    assert trainer.data_parallel_device_ids == expected_gpu
+    assert trainer.data_parallel_device_ids == expected_device_ids
 
 
-@pytest.mark.skipif(
-    sys.version_info < (3, 7),
-    reason="signature inspection while mocking is not working in Python < 3.7 despite autospec"
-)
+@RunIf(min_python="3.7.0")
 @pytest.mark.parametrize(['cli_args', 'extra_args'], [
     pytest.param({}, {}),
     pytest.param({'logger': False}, {}),

tests/utilities/test_argparse_utils.py

@@ -1,4 +1,4 @@
-from pytorch_lightning.utilities.argparse import parse_args_from_docstring
+from pytorch_lightning.utilities.argparse import parse_args_from_docstring, _gpus_allowed_type, _int_or_float_type
 
 
 def test_parse_args_from_docstring_normal():
@@ -48,3 +48,13 @@ def test_parse_args_from_docstring_empty():
         """
     )
     assert len(args_help.keys()) == 0
+
+
+def test_gpus_allowed_type():
+    assert _gpus_allowed_type('1,2') == '1,2'
+    assert _gpus_allowed_type('1') == 1
+
+
+def test_int_or_float_type():
+    assert isinstance(_int_or_float_type('0.0'), float)
+    assert isinstance(_int_or_float_type('0'), int)