docs: specify directives

.github/CONTRIBUTING.md

@@ -103,7 +103,7 @@ def my_func(param_a: int, param_b: Optional[float] = None) -> str:
         >>> my_func(1, 2)
         3
 
-    .. note:: If you want to add something.
+    .. hint:: If you want to add something.
     """
     p = param_b if param_b else 0
     return str(param_a + p)

docs/source/pages/implement.rst

@@ -257,7 +257,7 @@ and tests gets formatted in the following way:
   3. ``new_metric(...)``: essentially wraps the ``_update`` and ``_compute`` private functions into one public function that
      makes up the functional interface for the metric.
 
-  .. note::
+  .. hint::
      The `functional mean squared error <https://github.com/Lightning-AI/torchmetrics/blob/master/src/torchmetrics/functional/regression/mse.py>`_
      metric is a is a great example of how to divide the logic.
 
@@ -270,9 +270,9 @@ and tests gets formatted in the following way:
      ``_new_metric_compute(...)`` function in its ``compute``. No logic should really be implemented in the module interface.
      We do this to not have duplicate code to maintain.
 
-   .. note::
-     The module `MeanSquaredError <https://github.com/Lightning-AI/torchmetrics/blob/master/src/torchmetrics/regression/mse.py>`_
-     metric that corresponds to the above functional example showcases these steps.
+  .. note::
+    The module `MeanSquaredError <https://github.com/Lightning-AI/torchmetrics/blob/master/src/torchmetrics/regression/mse.py>`_
+    metric that corresponds to the above functional example showcases these steps.
 
 4. Remember to add binding to the different relevant ``__init__`` files.
 
@@ -291,7 +291,7 @@ and tests gets formatted in the following way:
      so that different combinations of inputs and parameters get tested.
   5. (optional) If your metric raises any exception, please add tests that showcase this.
 
-  .. note::
+  .. hint::
     The `test file for MSE <https://github.com/Lightning-AI/torchmetrics/blob/master/tests/unittests/regression/test_mean_error.py>`_
     metric shows how to implement such tests.
 

docs/source/pages/lightning.rst

@@ -13,7 +13,7 @@ TorchMetrics in PyTorch Lightning
 TorchMetrics was originally created as part of `PyTorch Lightning <https://github.com/Lightning-AI/pytorch-lightning>`_, a powerful deep learning research
 framework designed for scaling models without boilerplate.
 
-.. note::
+.. caution::
 
     TorchMetrics always offers compatibility with the last 2 major PyTorch Lightning versions, but we recommend always
     keeping both frameworks up-to-date for the best experience.
@@ -69,9 +69,9 @@ LightningModule `self.log <https://lightning.ai/docs/pytorch/stable/extensions/l
 method, Lightning will log the metric based on ``on_step`` and ``on_epoch`` flags present in ``self.log(...)``. If
 ``on_epoch`` is True, the logger automatically logs the end of epoch metric value by calling ``.compute()``.
 
-.. note::
+.. caution::
 
-    ``sync_dist``, ``sync_dist_group`` and ``reduce_fx`` flags from ``self.log(...)`` don't affect the metric logging
+    The ``sync_dist``, ``sync_dist_group`` and ``reduce_fx`` flags from ``self.log(...)`` don't affect the metric logging
     in any manner. The metric class contains its own distributed synchronization logic.
 
     This, however is only true for metrics that inherit the base class ``Metric``,
@@ -136,7 +136,7 @@ Note that logging metrics this way will require you to manually reset the metric
 In general, we recommend logging the metric object to make sure that metrics are correctly computed and reset.
 Additionally, we highly recommend that the two ways of logging are not mixed as it can lead to wrong results.
 
-.. note::
+.. hint::
 
     When using any Modular metric, calling ``self.metric(...)`` or ``self.metric.forward(...)`` serves the dual purpose
     of calling ``self.metric.update()`` on its input and simultaneously returning the metric value over the provided

docs/source/pages/overview.rst

@@ -61,13 +61,13 @@ This metrics API is independent of PyTorch Lightning. Metrics can directly be us
     It is highly recommended to re-initialize the metric per mode as
     shown in the examples above.
 
-.. note::
+.. caution::
 
     Metric states are **not** added to the models ``state_dict`` by default.
     To change this, after initializing the metric, the method ``.persistent(mode)`` can
     be used to enable (``mode=True``) or disable (``mode=False``) this behaviour.
 
-.. note::
+.. important::
 
     Due to specialized logic around metric states, we in general do **not** recommend
     that metrics are initialized inside other metrics (nested metrics), as this can lead
@@ -306,7 +306,7 @@ This pattern is implemented for the following operators (with ``a`` being metric
 * Positive Value (``pos(a)``)
 * Indexing (``a[0]``)
 
-.. note::
+.. caution::
 
     Some of these operations are only fully supported from Pytorch v1.4 and onwards, explicitly we found:
     ``add``, ``mul``, ``rmatmul``, ``rsub``, ``rmod``
@@ -381,7 +381,7 @@ inside your LightningModule. In most cases we just have to replace ``self.log``
             # remember to reset metrics at the end of the epoch
             self.valid_metrics.reset()
 
-.. note::
+.. important::
 
     `MetricCollection` as default assumes that all the metrics in the collection
     have the same call signature. If this is not the case, input that should be

src/torchmetrics/audio/dnsmos.py

@@ -54,11 +54,13 @@ class DeepNoiseSuppressionMeanOpinionScore(Metric):
     - ``dnsmos`` (:class:`~torch.Tensor`): float tensor of DNSMOS values reduced across the batch
         with shape ``(...,4)`` indicating [p808_mos, mos_sig, mos_bak, mos_ovr] in the last dim.
 
-    .. note:: using this metric requires you to have ``librosa``, ``onnxruntime`` and ``requests`` installed.
+    .. hint::
+        Using this metric requires you to have ``librosa``, ``onnxruntime`` and ``requests`` installed.
         Install as ``pip install torchmetrics['audio']`` or alternatively `pip install librosa onnxruntime-gpu requests`
         (if you do not have GPU enabled machine install `onnxruntime` instead of `onnxruntime-gpu`)
 
-    .. note:: the ``forward`` and ``compute`` methods in this class return a reduced DNSMOS value
+    .. caution::
+        The ``forward`` and ``compute`` methods in this class return a reduced DNSMOS value
         for a batch. To obtain the DNSMOS value for each sample, you may use the functional counterpart in
         :func:`~torchmetrics.functional.audio.dnsmos.deep_noise_suppression_mean_opinion_score`.
 

src/torchmetrics/audio/nisqa.py

@@ -43,10 +43,12 @@ class NonIntrusiveSpeechQualityAssessment(Metric):
     - ``nisqa`` (:class:`~torch.Tensor`): float tensor reduced across the batch with shape ``(5,)`` corresponding to
       overall MOS, noisiness, discontinuity, coloration and loudness in that order
 
-    .. note:: Using this metric requires you to have ``librosa`` and ``requests`` installed. Install as
+    .. hint::
+        Using this metric requires you to have ``librosa`` and ``requests`` installed. Install as
         ``pip install librosa requests``.
 
-    .. note:: The ``forward`` and ``compute`` methods in this class return values reduced across the batch. To obtain
+    .. caution::
+        The ``forward`` and ``compute`` methods in this class return values reduced across the batch. To obtain
         values for each sample, you may use the functional counterpart
         :func:`~torchmetrics.functional.audio.nisqa.non_intrusive_speech_quality_assessment`.
 

src/torchmetrics/audio/pesq.py

@@ -45,12 +45,14 @@ class PerceptualEvaluationSpeechQuality(Metric):
 
     - ``pesq`` (:class:`~torch.Tensor`): float tensor of PESQ value reduced across the batch
 
-    .. note:: using this metrics requires you to have ``pesq`` install. Either install as ``pip install
+    .. hint::
+        Using this metrics requires you to have ``pesq`` install. Either install as ``pip install
         torchmetrics[audio]`` or ``pip install pesq``. ``pesq`` will compile with your currently
         installed version of numpy, meaning that if you upgrade numpy at some point in the future you will
         most likely have to reinstall ``pesq``.
 
-    .. note:: the ``forward`` and ``compute`` methods in this class return a single (reduced) PESQ value
+    .. caution::
+        The ``forward`` and ``compute`` methods in this class return a single (reduced) PESQ value
         for a batch. To obtain a PESQ value for each sample, you may use the functional counterpart in
         :func:`~torchmetrics.functional.audio.pesq.perceptual_evaluation_speech_quality`.
 

src/torchmetrics/audio/srmr.py

@@ -48,11 +48,12 @@ class SpeechReverberationModulationEnergyRatio(Metric):
 
     - ``srmr`` (:class:`~torch.Tensor`): float scaler tensor
 
-    .. note:: using this metrics requires you to have ``gammatone`` and ``torchaudio`` installed.
+    .. hint::
+        Using this metrics requires you to have ``gammatone`` and ``torchaudio`` installed.
         Either install as ``pip install torchmetrics[audio]`` or ``pip install torchaudio``
         and ``pip install git+https://github.com/detly/gammatone``.
 
-    .. note::
+    .. attention::
         This implementation is experimental, and might not be consistent with the matlab
         implementation `SRMRToolbox`_, especially the fast implementation.
         The slow versions, a) fast=False, norm=False, max_cf=128, b) fast=False, norm=True, max_cf=30, have

src/torchmetrics/audio/stoi.py

@@ -50,7 +50,8 @@ class ShortTimeObjectiveIntelligibility(Metric):
 
     - ``stoi`` (:class:`~torch.Tensor`): float scalar tensor
 
-    .. note:: using this metrics requires you to have ``pystoi`` install. Either install as ``pip install
+    .. hint::
+        Using this metrics requires you to have ``pystoi`` install. Either install as ``pip install
         torchmetrics[audio]`` or ``pip install pystoi``.
 
     Args:

src/torchmetrics/classification/calibration_error.py

@@ -214,7 +214,7 @@ class MulticlassCalibrationError(Metric):
     - ``target`` (:class:`~torch.Tensor`): An int tensor of shape ``(N, ...)`` containing ground truth labels, and
       therefore only contain values in the [0, n_classes-1] range (except if `ignore_index` is specified).
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:

src/torchmetrics/classification/cohen_kappa.py

@@ -50,7 +50,7 @@ class labels.
       Additionally, we convert to int tensor with thresholding using the value in ``threshold``.
     - ``target`` (:class:`~torch.Tensor`): An int tensor of shape ``(N, ...)``.
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:
@@ -175,7 +175,7 @@ class labels.
       convert probabilities/logits into an int tensor.
     - ``target`` (:class:`~torch.Tensor`): An int tensor of shape ``(N, ...)``.
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:

src/torchmetrics/classification/dice.py

@@ -76,7 +76,7 @@ class Dice(Metric):
             - ``'samples'``: Calculate the metric for each sample, and average the metrics
               across samples (with equal weights for each sample).
 
-            .. note::
+            .. hint::
                What is considered a sample in the multi-dimensional multi-class case
                depends on the value of ``mdmc_average``.
 

src/torchmetrics/classification/hinge.py

@@ -55,7 +55,7 @@ class BinaryHingeLoss(Metric):
       ground truth labels, and therefore only contain {0,1} values (except if `ignore_index` is specified). The value
       1 always encodes the positive class.
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:
@@ -189,7 +189,7 @@ class MulticlassHingeLoss(Metric):
       ground truth labels, and therefore only contain values in the [0, n_classes-1] range (except if `ignore_index`
       is specified).
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:

src/torchmetrics/classification/jaccard.py

@@ -52,7 +52,7 @@ class BinaryJaccardIndex(BinaryConfusionMatrix):
       Additionally, we convert to int tensor with thresholding using the value in ``threshold``.
     - ``target`` (:class:`~torch.Tensor`): An int tensor of shape ``(N, ...)``.
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:
@@ -170,7 +170,7 @@ class MulticlassJaccardIndex(MulticlassConfusionMatrix):
       probabilities/logits into an int tensor.
     - ``target`` (:class:`~torch.Tensor`): An int tensor of shape ``(N, ...)``.
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:
@@ -307,7 +307,7 @@ class MultilabelJaccardIndex(MultilabelConfusionMatrix):
       sigmoid per element. Additionally, we convert to int tensor with thresholding using the value in ``threshold``.
     - ``target`` (:class:`~torch.Tensor`): An int tensor of shape ``(N, C, ...)``
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:

src/torchmetrics/classification/matthews_corrcoef.py

@@ -48,7 +48,7 @@ class BinaryMatthewsCorrCoef(BinaryConfusionMatrix):
       per element. Additionally, we convert to int tensor with thresholding using the value in ``threshold``.
     - ``target`` (:class:`~torch.Tensor`): An int tensor of shape ``(N, ...)``
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:
@@ -156,7 +156,7 @@ class MulticlassMatthewsCorrCoef(MulticlassConfusionMatrix):
       probabilities/logits into an int tensor.
     - ``target`` (:class:`~torch.Tensor`): An int tensor of shape ``(N, ...)``
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:
@@ -268,7 +268,7 @@ class MultilabelMatthewsCorrCoef(MultilabelConfusionMatrix):
       per element. Additionally, we convert to int tensor with thresholding using the value in ``threshold``.
     - ``target`` (:class:`~torch.Tensor`): An int tensor of shape ``(N, C, ...)``
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:

src/torchmetrics/classification/precision_fixed_recall.py

@@ -60,7 +60,7 @@ class BinaryPrecisionAtFixedRecall(BinaryPrecisionRecallCurve):
       ground truth labels, and therefore only contain {0,1} values (except if `ignore_index` is specified). The value
       1 always encodes the positive class.
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:
@@ -193,7 +193,7 @@ class MulticlassPrecisionAtFixedRecall(MulticlassPrecisionRecallCurve):
       ground truth labels, and therefore only contain values in the [0, n_classes-1] range (except if `ignore_index`
       is specified).
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns a tuple of either 2 tensors or 2 lists containing:
@@ -338,7 +338,7 @@ class MultilabelPrecisionAtFixedRecall(MultilabelPrecisionRecallCurve):
       ground truth labels, and therefore only contain {0,1} values (except if `ignore_index` is specified). The value
       1 always encodes the positive class.
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns a tuple of either 2 tensors or 2 lists containing:

src/torchmetrics/classification/precision_recall_curve.py

@@ -67,7 +67,7 @@ class BinaryPrecisionRecallCurve(Metric):
       ground truth labels, and therefore only contain {0,1} values (except if `ignore_index` is specified). The value
       1 always encodes the positive class.
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:
@@ -244,7 +244,7 @@ class MulticlassPrecisionRecallCurve(Metric):
       ground truth labels, and therefore only contain values in the [0, n_classes-1] range (except if `ignore_index`
       is specified).
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:
@@ -441,7 +441,7 @@ class MultilabelPrecisionRecallCurve(Metric):
     - ``target`` (:class:`~torch.Tensor`): An int tensor of shape ``(N, C, ...)``. Target should be a tensor containing
       ground truth labels, and therefore only contain {0,1} values (except if `ignore_index` is specified).
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following a tuple of either 3 tensors or

src/torchmetrics/classification/ranking.py

@@ -51,7 +51,7 @@ class MultilabelCoverageError(Metric):
     - ``target`` (:class:`~torch.Tensor`): An int tensor of shape ``(N, C, ...)``. Target should be a tensor
       containing ground truth labels, and therefore only contain {0,1} values (except if `ignore_index` is specified).
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:
@@ -171,7 +171,7 @@ class MultilabelRankingAveragePrecision(Metric):
     - ``target`` (:class:`~torch.Tensor`): An int tensor of shape ``(N, C, ...)``. Target should be a tensor
       containing ground truth labels, and therefore only contain {0,1} values (except if `ignore_index` is specified).
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:
@@ -291,7 +291,7 @@ class MultilabelRankingLoss(Metric):
     - ``target`` (:class:`~torch.Tensor`): An int tensor of shape ``(N, C, ...)``. Target should be a tensor
       containing ground truth labels, and therefore only contain {0,1} values (except if `ignore_index` is specified).
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:

src/torchmetrics/classification/recall_fixed_precision.py

@@ -59,7 +59,7 @@ class BinaryRecallAtFixedPrecision(BinaryPrecisionRecallCurve):
       ground truth labels, and therefore only contain {0,1} values (except if `ignore_index` is specified). The value
       1 always encodes the positive class.
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns the following output:
@@ -194,7 +194,7 @@ class MulticlassRecallAtFixedPrecision(MulticlassPrecisionRecallCurve):
       ground truth labels, and therefore only contain values in the [0, n_classes-1] range (except if `ignore_index`
       is specified).
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns a tuple of either 2 tensors or 2 lists containing:
@@ -337,7 +337,7 @@ class MultilabelRecallAtFixedPrecision(MultilabelPrecisionRecallCurve):
       ground truth labels, and therefore only contain {0,1} values (except if `ignore_index` is specified). The value
       1 always encodes the positive class.
 
-    .. note::
+    .. tip::
        Additional dimension ``...`` will be flattened into the batch dimension.
 
     As output to ``forward`` and ``compute`` the metric returns a tuple of either 2 tensors or 2 lists containing: