add function to reduce tensors (similar to reduction in torch.nn)

pytorch_lightning/metrics/functional/reduction.py

@@ -0,0 +1,29 @@
+import torch
+
+
+def reduce(to_reduce: torch.Tensor, reduction: str) -> torch.Tensor:
+    """
+    reduces a given tensor by a given reduction method
+    Parameters
+    ----------
+    to_reduce : torch.Tensor
+        the tensor, which shall be reduced
+    reduction : str
+        a string specifying the reduction method.
+        should be one of 'elementwise_mean' | 'none' | 'sum'
+    Returns
+    -------
+    torch.Tensor
+        reduced Tensor
+    Raises
+    ------
+    ValueError
+        if an invalid reduction parameter was given
+    """
+    if reduction == 'elementwise_mean':
+        return torch.mean(to_reduce)
+    if reduction == 'none':
+        return to_reduce
+    if reduction == 'sum':
+        return torch.sum(to_reduce)
+    raise ValueError('Reduction parameter unknown.')