update

app.py

@@ -23,7 +23,7 @@ MAX_INPUT_TOKENS = int(os.getenv("MAX_INPUT_TOKENS", "32_000"))
 
 @spaces.GPU()
 @torch.inference_mode()
-def generate(message: dict, history: list[dict], system_prompt: str = "", max_new_tokens: int = 512, enable_thinking: bool = True) -> Iterator[str]:
+def generate(message: str | dict, history: list[dict], system_prompt: str = "", max_new_tokens: int = 512, enable_thinking: bool = True) -> Iterator[str]:
     # Build messages for Qwen3 (text-only format)
     messages = []
     if system_prompt:
@@ -47,12 +47,16 @@ def generate(message: dict, history: list[dict], system_prompt: str = "", max_ne
             content = item["content"]
             if isinstance(content, str):
                 messages.append({"role": "user", "content": content})
-            else:
-                # For now, just use the text part (Qwen3-14B is text-only)
-                messages.append({"role": "user", "content": message.get("text", "")})
+            elif isinstance(content, dict):
+                # Handle dict format
+                messages.append({"role": "user", "content": content.get("text", "")})
     
     # Add current user message
-    current_message = message.get("text", "")
+    # Handle both string and dict message formats
+    if isinstance(message, str):
+        current_message = message
+    else:
+        current_message = message.get("text", "")
     messages.append({"role": "user", "content": current_message})
 
     # Apply chat template with enable_thinking parameter