diff --git a/src/diffusers/models/attention.py b/src/diffusers/models/attention.py
index 4c970d062d..e9454a467a 100644
--- a/src/diffusers/models/attention.py
+++ b/src/diffusers/models/attention.py
@@ -221,11 +221,15 @@ class Transformer2DModel(ModelMixin, ConfigMixin):
         # 3. Output
         if self.is_input_continuous:
             if not self.use_linear_projection:
-                hidden_states = hidden_states.reshape(batch, height, weight, inner_dim).permute(0, 3, 1, 2)
+                hidden_states = (
+                    hidden_states.reshape(batch, height, weight, inner_dim).permute(0, 3, 1, 2).contiguous()
+                )
                 hidden_states = self.proj_out(hidden_states)
             else:
                 hidden_states = self.proj_out(hidden_states)
-                hidden_states = hidden_states.reshape(batch, height, weight, inner_dim).permute(0, 3, 1, 2)
+                hidden_states = (
+                    hidden_states.reshape(batch, height, weight, inner_dim).permute(0, 3, 1, 2).contiguous()
+                )
 
             output = hidden_states + residual
         elif self.is_input_vectorized: