Handle None joint_attention_kwargs in JointTransformerBlock

Author: guiyrt

src/diffusers/models/attention.py

@@ -192,7 +192,7 @@ def forward(
         hidden_states: torch.FloatTensor,
         encoder_hidden_states: torch.FloatTensor,
         temb: torch.FloatTensor,
-        joint_attention_kwargs: Dict[str, Any] = {},
+        joint_attention_kwargs: Optional[Dict[str, Any]] = None,
     ):
         if self.use_dual_attention:
             norm_hidden_states, gate_msa, shift_mlp, scale_mlp, gate_mlp, norm_hidden_states2, gate_msa2 = self.norm1(
@@ -208,6 +208,10 @@ def forward(
                 encoder_hidden_states, emb=temb
             )
 
+        # Empty dict if None is passed
+        if joint_attention_kwargs is None:
+            joint_attention_kwargs = {}
+
         # Attention.
         attn_output, context_attn_output = self.attn(
             hidden_states=norm_hidden_states,