Update app.py

app.py

@@ -8,6 +8,71 @@ from smolagents import CodeAgent, DuckDuckGoSearchTool, OpenAIServerModel, Visit
 # --- Constants ---
 DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 
+class AttachmentDownloadTool(Tool):
+    name = "attachment-downloader"
+    description = "Downloads the file associated with the given task_id. If it does not exist, return None. input: task_id。output: attachment files or None"
+    inputs = {
+        "task_id": {
+            "type": "str",
+            "description": "task_id that needs to download attachment files."
+        }
+    }
+    output_type = File
+
+    def forward(self, task_id):
+        download_url = f"{api_url}/files/"
+        
+        try:
+            response = requests.get(download_url + task_id, stream=True, timeout=15)
+            if response.status_code != 200:
+                return None
+            file_obj = io.BytesIO(response.content)
+            file_obj.seek(0)
+            return file_obj
+        except Exception as e:
+            return None
+
+class ImageCaptionTool(Tool):
+    name = "image-captioner"
+    description = "Identify the content of the input image and describe it in natural language. Input: image. Output: description text."
+    inputs = {
+        "image": {
+            "type": "image",
+            "description": "Images that need to be identified and described"
+        }
+    }
+    output_type = str
+
+    def setup(self):
+        self.model = OpenAIServerModel(
+            model_id="Qwen/Qwen2.5-VL-32B-Instruct",
+            api_base="https://api.siliconflow.cn/v1/",
+            api_key=os.getenv('MODEL_TOKEN'),
+        )
+
+    def forward(self, image):
+        prompt = "Please describe the content of this picture in detail."
+        return self.model(prompt, images=[image])
+
+class AudioToTextTool(Tool):
+    name = "audio-to-text"
+    description = "Convert the input audio content to text. Input: audio. Output: text."
+    inputs = {
+        "audio": {
+            "type": "audio",
+            "description": "The audio file that needs to be transcribed"
+        }
+    }
+    output_type = str
+
+    def setup(self):
+        # 使用 HuggingFace Hub 上的 Whisper 大模型
+        self.model = HfApiModel(model_id="openai/whisper-large-v3")  # 或其他支持音频转写的模型
+
+    def forward(self, audio):
+        prompt = "Please transcribe this audio content into text."
+        return self.model(prompt, audios=[audio])
+
 # --- Basic Agent Definition ---
 # ----- THIS IS WERE YOU CAN BUILD WHAT YOU WANT ------
 # class BasicAgent:
@@ -21,17 +86,37 @@ DEFAULT_API_URL = "https://agents-course-unit4-scoring.hf.space"
 class BasicAgent:
     def __init__(self):
 
-        self.model = OpenAIServerModel(
-            model_id="Qwen/Qwen2.5-VL-32B-Instruct",
+        self.think_model = OpenAIServerModel(
+            model_id="THUDM/GLM-Z1-32B-0414",
             api_base="https://api.siliconflow.cn/v1/",
             api_key=os.getenv('MODEL_TOKEN'),
         )
+        self.base_model = OpenAIServerModel(
+            model_id="THUDM/GLM-4-32B-0414",
+            api_base="https://api.siliconflow.cn/v1/",
+            api_key=os.getenv('MODEL_TOKEN'),
+        )
+        # self.vision_model = OpenAIServerModel(
+        #     model_id="Qwen/Qwen2.5-VL-32B-Instruct",
+        #     api_base="https://api.siliconflow.cn/v1/",
+        #     api_key=os.getenv('MODEL_TOKEN'),
+        # )
 
-        self.tools = [DuckDuckGoSearchTool(), VisitWebpageTool()]
+        self.tools = [AttachmentDownloadTool, ImageCaptionTool, AudioToTextTool]
+
+        web_agent = ToolCallingAgent(
+            tools=[DuckDuckGoSearchTool(), VisitWebpageTool()],
+            model=self.base_model,
+            max_steps=10,
+            name="web_search_agent",
+            description="Runs web searches for you.",
+        )
         
         self.agent = CodeAgent(
             tools=self.tools,
-            model=self.model,
+            model=self.think_model,
+            managed_agents=[web_agent,],
+            additional_authorized_imports=["time", "numpy", "pandas"],
             max_steps=20
         )
         print("BasicAgent initialized.")