[CLAP] Replace hard-coded batch size to enable dynamic ONNX export

src/transformers/models/clap/modeling_clap.py

@@ -89,24 +89,14 @@ def window_partition(hidden_states, window_size):
     return windows
 
 
-# Adapted from https://github.com/LAION-AI/CLAP/blob/6ad05a971ba0622f6acee8c41993e0d02bbed639/src/open_clip/htsat.py#L263
 def window_reverse(windows, window_size, height, width):
     """
-    Args:
-        windows (`torch.FloatTensor` of shape `(num_windows * batch_size, window_size, window_size, num_channels)`):
-            Input windows
-        window_size (`int`):
-            Window size
-        height (`int`):
-            Height of the resized audio
-        width (`int`):
-            Width of the resized audio
+    Merges windows to produce higher resolution features.
     """
-    batch_size = int(windows.shape[0] / (height * width / window_size / window_size))
-
-    hidden_states = windows.view(batch_size, height // window_size, width // window_size, window_size, window_size, -1)
-    hidden_states = hidden_states.permute(0, 1, 3, 2, 4, 5).contiguous().view(batch_size, height, width, -1)
-    return hidden_states
+    num_channels = windows.shape[-1]
+    windows = windows.view(-1, height // window_size, width // window_size, window_size, window_size, num_channels)
+    windows = windows.permute(0, 1, 3, 2, 4, 5).contiguous().view(-1, height, width, num_channels)
+    return windows
 
 
 # Copied from transformers.models.roberta.modeling_roberta.create_position_ids_from_input_ids