Update app.py

app.py

@@ -17,6 +17,9 @@ from llama_cpp_agent.llm_output_settings import (
 )
 from llama_cpp_agent.tools import WebSearchTool
 from llama_cpp_agent.prompt_templates import web_search_system_prompt, research_system_prompt
+from langchain import HuggingFaceHub
+from llama_cpp_agent.llm_output_settings import LlmStructuredOutputSettings, LlmStructuredOutputType
+
 
 # UI related imports and definitions
 css = """
@@ -92,6 +95,26 @@ examples = [
     ["filetype:pdf intitle:python"]
 ]
 
+
+class HuggingFaceHubWrapper:
+    def __init__(self, repo_id, model_kwargs, huggingfacehub_api_token):
+        self.model = HuggingFaceHub(
+            repo_id=repo_id,
+            model_kwargs=model_kwargs,
+            huggingfacehub_api_token=huggingfacehub_api_token
+        )
+
+    def get_provider_default_settings(self):
+        return LlmStructuredOutputSettings(
+            output_type=LlmStructuredOutputType.JSON,
+            include_system_prompt=False,
+            include_user_prompt=False,
+            include_assistant_prompt=False,
+        )
+
+    def __call__(self, *args, **kwargs):
+        return self.model(*args, **kwargs)
+
 # Utility functions
 def get_server_time():
     utc_time = datetime.now(timezone.utc)
@@ -117,7 +140,7 @@ class CitingSources(BaseModel):
 
 # Model function
 def get_model(temperature, top_p, repetition_penalty):
-    return HuggingFaceHub(
+    return HuggingFaceHubWrapper(
         repo_id="mistralai/Mistral-7B-Instruct-v0.3",
         model_kwargs={
             "temperature": temperature,
@@ -127,6 +150,7 @@ def get_model(temperature, top_p, repetition_penalty):
         },
         huggingfacehub_api_token=huggingface_token
     )
+
 def get_messages_formatter_type(model_name):
     model_name = model_name.lower()
     if any(keyword in model_name for keyword in ["meta", "aya"]):