fix wrong parameter name in plotting function and change value tensor…

… in pytorch sparse tensor to float type

tonic/functional/to_sparse_tensor.py

@@ -9,13 +9,13 @@ def get_indices_values(events, sensor_size, ordering, merge_polarities):
     Parameters:
         merge_polarities (bool): flag that decides whether to combine polarities into a single channel
                                 or split them into separate channels. If True, the number of channels for
-                                indices is 1, otherwise it's the number of different polarities. Regardless 
+                                indices is 1, otherwise it's the number of different polarities. Regardless
                                 of this flag, all values assigned to indices will be 1, which signify a spike.
 
     Returns:
         sparse tensor in TxCxWxH format, where T is timesteps, C is the number of channels for each polarity,
         and W and H are always the size of the sensor.
-    """ 
+    """
     assert "x" and "t" and "p" in ordering
     x_index = ordering.find("x")
     t_index = ordering.find("t")
@@ -30,8 +30,9 @@ def get_indices_values(events, sensor_size, ordering, merge_polarities):
     values = np.ones(events.shape[0])
 
     # prevents polarities used as indices that are not 0
-    if len(np.unique(events[:, p_index])) == 1: merge_polarities = True
-
+    if len(np.unique(events[:, p_index])) == 1:
+        merge_polarities = True
+
     if merge_polarities:  # the indices need to start at 0
         events[:, p_index] = 0
         n_channels = 1
@@ -40,35 +41,49 @@ def get_indices_values(events, sensor_size, ordering, merge_polarities):
         n_channels = len(np.unique(events[:, p_index]))
 
     max_time = int(max(events[:, t_index]) + 1)
-    
-    if "y" in ordering:   
+
+    if "y" in ordering:
         y_index = ordering.find("y")
         indices = events[:, [t_index, p_index, x_index, y_index]]
     else:
         indices = events[:, [t_index, p_index, x_index]]
 
     return indices, values, max_time, n_channels
 
+
 def to_sparse_tensor_pytorch(events, sensor_size, ordering, merge_polarities):
     try:
         import torch
     except ImportError:
-        raise ImportError('The sparse tensor transform needs PyTorch installed. Please install a stable version ' +
-                          'of PyTorch or alternatively install Tonic with optional PyTorch dependencies.')
-    indices, values, max_time, n_channels = get_indices_values(events, sensor_size, ordering, merge_polarities)
+        raise ImportError(
+            "The sparse tensor transform needs PyTorch installed. Please install a"
+            " stable version "
+            + "of PyTorch or alternatively install Tonic with optional PyTorch"
+            " dependencies."
+        )
+    indices, values, max_time, n_channels = get_indices_values(
+        events, sensor_size, ordering, merge_polarities
+    )
     indices = torch.LongTensor(indices).T
-    values = torch.LongTensor(values)
+    values = torch.FloatTensor(values)
     return torch.sparse.FloatTensor(
         indices, values, torch.Size([max_time, n_channels, *sensor_size])
     )
 
+
 def to_sparse_tensor_tensorflow(events, sensor_size, ordering, merge_polarities):
-    indices, values, max_time, n_channels = get_indices_values(events, sensor_size, ordering, merge_polarities)
+    indices, values, max_time, n_channels = get_indices_values(
+        events, sensor_size, ordering, merge_polarities
+    )
     try:
         import tensorflow
     except ImportError:
-        raise ImportError('The sparse tensor transform needs PyTorch installed. Please install a stable version ' +
-                          'of PyTorch or alternatively install Tonic with optional PyTorch dependencies.')
+        raise ImportError(
+            "The sparse tensor transform needs PyTorch installed. Please install a"
+            " stable version "
+            + "of PyTorch or alternatively install Tonic with optional PyTorch"
+            " dependencies."
+        )
     return tensorflow.sparse.SparseTensor(
         indices, values, torch.Size([max_time, n_channels, *sensor_size])
     )

tonic/utils.py

@@ -21,13 +21,14 @@ def plot_event_grid(events, ordering, axis_array=(1, 3), plot_frame_number=False
         import matplotlib.pyplot as plt
     except ImportError:
         raise ImportError(
-            "Please install the matplotlib package to plot events. This is an optional dependency."
+            "Please install the matplotlib package to plot events. This is an optional"
+            " dependency."
         )
 
     events = events.squeeze()
     events = np.array(events)
     transform = transforms.Compose(
-        [transforms.ToVoxelGrid(num_time_bins=np.product(axis_array))]
+        [transforms.ToVoxelGrid(n_time_bins=np.product(axis_array))]
     )
     x_index = ordering.find("x")
     y_index = ordering.find("y")
@@ -71,7 +72,8 @@ def pad_tensors(batch):
 
     if not isinstance(batch[0][0], torch.Tensor):
         print(
-            "tonic.utils.pad_tensors expects a PyTorch Tensor of events. Please use ToSparseTensor or similar transform to convert the events."
+            "tonic.utils.pad_tensors expects a PyTorch Tensor of events. Please use"
+            " ToSparseTensor or similar transform to convert the events."
         )
         return None, None
     max_length = max([sample.size()[0] for sample, target in batch])