Update modeling_deberta.py

modeling_deberta.py

@@ -1251,20 +1251,24 @@ class DebertaV2ForCausalLM(DebertaV2ForMaskedLM):
             ],
             dim=-1
         )
-        attention_mask = torch.cat(
-            [
-                attention_mask,
-                torch.full((batch_size, self.n_masks + 1), attention_mask[0, -1], device=attention_mask.device),
-            ],
-            dim=-1
-        )
-        position_ids = torch.cat(
-            [
-                position_ids,
-                torch.arange(0, self.n_masks + 1, device=position_ids.device).unsqueeze(0) + position_ids[:, -1:],
-            ],
-            dim=-1
-        )
+
+        if attention_mask is not None:
+            attention_mask = torch.cat(
+                [
+                    attention_mask,
+                    torch.full((batch_size, self.n_masks + 1), attention_mask[0, -1], device=attention_mask.device),
+                ],
+                dim=-1
+            )
+
+        if position_ids is not None:
+            position_ids = torch.cat(
+                [
+                    position_ids,
+                    torch.arange(0, self.n_masks + 1, device=position_ids.device).unsqueeze(0) + position_ids[:, -1:],
+                ],
+                dim=-1
+            )
 
         outputs = super().forward(
             input_ids,