Changing initialization of max_batch_size

Previous behavior: if None or greater than the dataset length, it is set to the dataset length
Current behavior: If None, it is set to the dataset length + 1, if the dataset has a length. Otherwise, it is set to 1

bs_scheduler/batch_size_schedulers.py

@@ -35,11 +35,13 @@ def __init__(self, dataloader: DataLoader, batch_size_manager: Optional[BatchSiz
         assert max_batch_size is None or isinstance(max_batch_size, int)
         assert isinstance(min_batch_size, int)
         if max_batch_size is None:
-            max_batch_size = len(self.dataloader.dataset)
+            if dataloader is not None and hasattr(dataloader, 'dataset') and hasattr(dataloader.dataset, '__len__'):
+                max_batch_size = len(dataloader.dataset) + 1
+            else:
+                max_batch_size = 1
         else:
             if max_batch_size < 0:
                 raise ValueError(f"Maximum batch size must be greater than 0, but is {max_batch_size}.")
-            max_batch_size = min(len(self.dataloader.dataset), max_batch_size)
         self.max_batch_size: int = max_batch_size
 
         if min_batch_size < 0:
@@ -173,8 +175,8 @@ class LambdaBS(BSScheduler):
         batch_size_manager (Optional[BatchSizeManager]): If not None, a custom class which manages the batch size,
             which provides a getter and setter for the batch size. Default: None.
         max_batch_size (Optional[int]): Upper limit for the batch size so that a batch of size max_batch_size fits
-            in the memory. If None or greater than the lenght of the dataset wrapped by the dataloader, max_batch_size
-            is set to `len(self.dataloader.dataset)`. Default: None.
+            in the memory. If None, max_batch_size is set to `len(self.dataloader.dataset) if available else 0 + 1`.
+            Default: None.
         min_batch_size (int): Lower limit for the batch size which must be greater than 0. Default: 1.
         verbose (bool): If ``True``, prints a message to stdout for each update. Default: ``False``.
 
@@ -237,8 +239,8 @@ class MultiplicativeBS(BSScheduler):
         batch_size_manager (Optional[BatchSizeManager]): If not None, a custom class which manages the batch size,
             which provides a getter and setter for the batch size. Default: None.
         max_batch_size (Optional[int]): Upper limit for the batch size so that a batch of size max_batch_size fits
-            in the memory. If None or greater than the lenght of the dataset wrapped by the dataloader, max_batch_size
-            is set to `len(self.dataloader.dataset)`. Default: None.
+            in the memory. If None, max_batch_size is set to `len(self.dataloader.dataset) if available else 0 + 1`.
+            Default: None.
         min_batch_size (int): Lower limit for the batch size which must be greater than 0. Default: 1.
         verbose (bool): If ``True``, prints a message to stdout for each update. Default: ``False``.
 
@@ -299,8 +301,8 @@ class StepBS(BSScheduler):
         batch_size_manager (Optional[BatchSizeManager]): If not None, a custom class which manages the batch size,
             which provides a getter and setter for the batch size. Default: None.
         max_batch_size (Optional[int]): Upper limit for the batch size so that a batch of size max_batch_size fits
-            in the memory. If None or greater than the lenght of the dataset wrapped by the dataloader, max_batch_size
-            is set to `len(self.dataloader.dataset)`. Default: None.
+            in the memory. If None, max_batch_size is set to `len(self.dataloader.dataset) if available else 0 + 1`.
+            Default: None.
         min_batch_size (int): Lower limit for the batch size which must be greater than 0. Default: 1.
         verbose (bool): If ``True``, prints a message to stdout for each update. Default: ``False``.
 
@@ -346,8 +348,8 @@ class MultiStepBS(BSScheduler):
         batch_size_manager (Optional[BatchSizeManager]): If not None, a custom class which manages the batch size,
             which provides a getter and setter for the batch size. Default: None.
         max_batch_size (Optional[int]): Upper limit for the batch size so that a batch of size max_batch_size fits
-            in the memory. If None or greater than the lenght of the dataset wrapped by the dataloader, max_batch_size
-            is set to `len(self.dataloader.dataset)`. Default: None.
+            in the memory. If None, max_batch_size is set to `len(self.dataloader.dataset) if available else 0 + 1`.
+            Default: None.
         min_batch_size (int): Lower limit for the batch size which must be greater than 0. Default: 1.
         verbose (bool): If ``True``, prints a message to stdout for each update. Default: ``False``.
 
@@ -410,8 +412,8 @@ class ConstantBS(BSScheduler):
         batch_size_manager (Optional[BatchSizeManager]): If not None, a custom class which manages the batch size,
             which provides a getter and setter for the batch size. Default: None.
         max_batch_size (Optional[int]): Upper limit for the batch size so that a batch of size max_batch_size fits
-            in the memory. If None or greater than the lenght of the dataset wrapped by the dataloader, max_batch_size
-            is set to `len(self.dataloader.dataset)`. Default: None.
+            in the memory. If None, max_batch_size is set to `len(self.dataloader.dataset) if available else 0 + 1`.
+            Default: None.
         min_batch_size (int): Lower limit for the batch size which must be greater than 0. Default: 1.
         verbose (bool): If ``True``, prints a message to stdout for each update. Default: ``False``.
 
@@ -475,8 +477,8 @@ class LinearBS(BSScheduler):
         batch_size_manager (Optional[BatchSizeManager]): If not None, a custom class which manages the batch size,
             which provides a getter and setter for the batch size. Default: None.
         max_batch_size (Optional[int]): Upper limit for the batch size so that a batch of size max_batch_size fits
-            in the memory. If None or greater than the lenght of the dataset wrapped by the dataloader, max_batch_size
-            is set to `len(self.dataloader.dataset)`. Default: None.
+            in the memory. If None, max_batch_size is set to `len(self.dataloader.dataset) if available else 0 + 1`.
+            Default: None.
         min_batch_size (int): Lower limit for the batch size which must be greater than 0. Default: 1.
         verbose (bool): If ``True``, prints a message to stdout for each update. Default: ``False``.
 
@@ -534,8 +536,8 @@ class ExponentialBS(BSScheduler):
         batch_size_manager (Optional[BatchSizeManager]): If not None, a custom class which manages the batch size,
             which provides a getter and setter for the batch size. Default: None.
         max_batch_size (Optional[int]): Upper limit for the batch size so that a batch of size max_batch_size fits
-            in the memory. If None or greater than the lenght of the dataset wrapped by the dataloader, max_batch_size
-            is set to `len(self.dataloader.dataset)`. Default: None.
+            in the memory. If None, max_batch_size is set to `len(self.dataloader.dataset) if available else 0 + 1`.
+            Default: None.
         min_batch_size (int): Lower limit for the batch size which must be greater than 0. Default: 1.
         verbose (bool): If ``True``, prints a message to stdout for each update. Default: ``False``.
 
@@ -720,8 +722,8 @@ class PolynomialBS(BSScheduler):
         batch_size_manager (Optional[BatchSizeManager]): If not None, a custom class which manages the batch size,
             which provides a getter and setter for the batch size. Default: None.
         max_batch_size (Optional[int]): Upper limit for the batch size so that a batch of size max_batch_size fits
-            in the memory. If None or greater than the lenght of the dataset wrapped by the dataloader, max_batch_size
-            is set to `len(self.dataloader.dataset)`. Default: None.
+            in the memory. If None, max_batch_size is set to `len(self.dataloader.dataset) if available else 0 + 1`.
+            Default: None.
         min_batch_size (int): Lower limit for the batch size which must be greater than 0. Default: 1.
         verbose (bool): If ``True``, prints a message to stdout for each update. Default: ``False``.
 
@@ -777,8 +779,8 @@ class CosineAnnealingBS(BSScheduler):
         batch_size_manager (Optional[BatchSizeManager]): If not None, a custom class which manages the batch size,
             which provides a getter and setter for the batch size. Default: None.
         max_batch_size (Optional[int]): Upper limit for the batch size so that a batch of size max_batch_size fits
-            in the memory. If None or greater than the lenght of the dataset wrapped by the dataloader, max_batch_size
-            is set to `len(self.dataloader.dataset)`. Default: None.
+            in the memory. If None, max_batch_size is set to `len(self.dataloader.dataset) if available else 0 + 1`.
+            Default: None.
         min_batch_size (int): Lower limit for the batch size which must be greater than 0. Default: 1.
         verbose (bool): If ``True``, prints a message to stdout for each update. Default: ``False``.
 
@@ -971,8 +973,8 @@ class IncreaseBSOnPlateau(BSScheduler):
         batch_size_manager (Optional[BatchSizeManager]): If not None, a custom class which manages the batch size,
             which provides a getter and setter for the batch size. Default: None.
         max_batch_size (Optional[int]): Upper limit for the batch size so that a batch of size max_batch_size fits
-            in the memory. If None or greater than the lenght of the dataset wrapped by the dataloader, max_batch_size
-            is set to `len(self.dataloader.dataset)`. Default: None.
+            in the memory. If None, max_batch_size is set to `len(self.dataloader.dataset) if available else 0 + 1`.
+            Default: None.
         min_batch_size (int): Lower limit for the batch size which must be greater than 0. Default: 1.
         verbose (bool): If ``True``, prints a message to stdout for each update. Default: ``False``.
 
@@ -1138,8 +1140,8 @@ class CyclicBS(BSScheduler):
         max_batch_size (Optional[int]): Upper batch size boundary in the cycle. Functionally, it defines the cycle
             amplitude (upper_batch_size_bound - base_batch_size). The batch size at any cycle is the sum of
             base_batch_size and some scaling of the amplitude; therefore, upper_batch_size_bound may not actually be
-            reached depending on scaling function. If None or greater than the lenght of the dataset wrapped by the
-            dataloader, max_batch_size is set to `len(self.dataloader.dataset)`. Default: None.
+            reached depending on scaling function. If None, max_batch_size is set to
+            `len(self.dataloader.dataset) if available else 0 + 1`. Default: None.
         min_batch_size (int): Lower limit for the batch size which must be greater than 0. Default: 1.
         verbose (bool): If ``True``, prints a message to stdout for each update. Default: ``False``.
 
@@ -1274,8 +1276,8 @@ class CosineAnnealingBSWithWarmRestarts(BSScheduler):
         batch_size_manager (Optional[BatchSizeManager]): If not None, a custom class which manages the batch size,
             which provides a getter and setter for the batch size. Default: None.
         max_batch_size (Optional[int]): Upper limit for the batch size so that a batch of size max_batch_size fits
-            in the memory. If None or greater than the lenght of the dataset wrapped by the dataloader, max_batch_size
-            is set to `len(self.dataloader.dataset)`. Default: None.
+            in the memory. If None, max_batch_size is set to `len(self.dataloader.dataset) if available else 0 + 1`.
+            Default: None.
         min_batch_size (int): Lower limit for the batch size which must be greater than 0. Default: 1.
         verbose (bool): If ``True``, prints a message to stdout for each update. Default: ``False``.
 
@@ -1359,8 +1361,8 @@ class OneCycleBS(BSScheduler):
         batch_size_manager (Optional[BatchSizeManager]): If not None, a custom class which manages the batch size,
             which provides a getter and setter for the batch size. Default: None.
         max_batch_size (Optional[int]): Upper limit for the batch size so that a batch of size max_batch_size fits
-            in the memory. If None or greater than the lenght of the dataset wrapped by the dataloader, max_batch_size
-            is set to `len(self.dataloader.dataset)`. Default: None.
+            in the memory. If None, max_batch_size is set to `len(self.dataloader.dataset) if available else 0 + 1`.
+            Default: None.
         min_batch_size (int): Lower limit for the batch size which must be greater than 0. Default: 1.
         verbose (bool): If ``True``, prints a message to stdout for each update. Default: ``False``.