collections for PL back-compatibility (#750)

.github/workflows/docs-check.yml

@@ -76,7 +76,7 @@ jobs:
  working-directory: ./docs
  run: |
  make clean
- make html --debug --jobs 2 SPHINXOPTS="-W --keep-going" -b linkcheck
+ make html --debug SPHINXOPTS="-W --keep-going" -b linkcheck
 
  - name: Upload built docs
  uses: actions/upload-artifact@v2

CHANGELOG.md

@@ -31,7 +31,6 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Metrics having third party dependencies removed from global import ([#463](https://github.com/PyTorchLightning/metrics/pull/463))
 - Untokenized for `BLEUScore` input stay consistent with all the other text metrics ([#640](https://github.com/PyTorchLightning/metrics/pull/640))
 - Arguments reordered for `TER`, `BLEUScore`, `SacreBLEUScore`, `CHRFScore` now expect input order as predictions first and target second ([#696](https://github.com/PyTorchLightning/metrics/pull/696))
-- Renamed `torchmetrics.collections` to `torchmetrics.metrics_collections` to avoid clashing with system's `collections` package ([#695](https://github.com/PyTorchLightning/metrics/pull/695))
 - Changed dtype of metric state from `torch.float` to `torch.long` in `ConfusionMatrix` to accommodate larger values ([#708](https://github.com/PyTorchLightning/metrics/issues/708))
 - Unify `preds`, `target` input argument's naming across all text metrics ([#723](https://github.com/PyTorchLightning/metrics/issues/723), [#727](https://github.com/PyTorchLightning/metrics/issues/727))
  * `bert`, `bleu`, `chrf`, `sacre_bleu`, `wip`, `wil`, `cer`, `ter`, `wer`, `mer`, `rouge`, `squad`

tests/bases/test_collections.py

@@ -20,7 +20,7 @@
 from tests.helpers.testers import DummyMetricDiff, DummyMetricSum
 from torchmetrics import Metric
 from torchmetrics.classification import Accuracy
-from torchmetrics.metric_collections import MetricCollection
+from torchmetrics.collections import MetricCollection
 
 seed_all(42)
 

torchmetrics/__init__.py

@@ -55,6 +55,7 @@
  Specificity,
  StatScores,
 )
+from torchmetrics.collections import MetricCollection # noqa: E402
 from torchmetrics.image import ( # noqa: E402
  PSNR,
  SSIM,
@@ -63,7 +64,6 @@
  StructuralSimilarityIndexMeasure,
 )
 from torchmetrics.metric import Metric # noqa: E402
-from torchmetrics.metric_collections import MetricCollection # noqa: E402
 from torchmetrics.regression import ( # noqa: E402
  CosineSimilarity,
  ExplainedVariance,

torchmetrics/metric_collections.py → torchmetrics/collections.py

@@ -12,7 +12,6 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
-from collections import OrderedDict
 from copy import deepcopy
 from typing import Any, Dict, Hashable, Iterable, Optional, Sequence, Tuple, Union
 
@@ -22,6 +21,9 @@
 from torchmetrics.metric import Metric
 from torchmetrics.utilities import rank_zero_warn
 
+# this is just a bypass for this module name collision with build-in one
+from torchmetrics.utilities.imports import OrderedDict
+
 
 class MetricCollection(nn.ModuleDict):
  """MetricCollection class can be used to chain metrics that have the same call pattern into one single class.

torchmetrics/metric.py

@@ -15,10 +15,9 @@
 import inspect
 import operator as op
 from abc import ABC, abstractmethod
-from collections.abc import Sequence
 from contextlib import contextmanager
 from copy import deepcopy
-from typing import Any, Callable, Dict, Generator, List, Optional, Union
+from typing import Any, Callable, Dict, Generator, List, Optional, Sequence, Union
 
 import torch
 from torch import Tensor

torchmetrics/utilities/imports.py

@@ -13,6 +13,7 @@
 # limitations under the License.
 """Import utilities."""
 import operator
+from collections import OrderedDict # noqa: F401
 from importlib import import_module
 from importlib.util import find_spec
 from typing import Callable, Optional