Update 1 app.py

app.py

@@ -1,11 +1,36 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
+from smolagents import CodeAgent, DuckDuckGoSearchTool, FinalAnswerTool, VisitWebpageTool, HfApiModel
 
-"""
-For more information on `huggingface_hub` Inference API support, please check the docs: https://huggingface.co/docs/huggingface_hub/v0.22.2/en/guides/inference
-"""
+# Initialize HuggingFace client
 client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
+# Create the smolagents agent (without UserInputTool since we'll get input from Gradio)
+agent = CodeAgent(
+    tools=[DuckDuckGoSearchTool(), VisitWebpageTool(), FinalAnswerTool()],
+    model=HfApiModel(),
+    max_steps=8,
+    verbosity_level=1
+)
+
+# Function to perform web research with a provided query
+def research_with_query(query):
+    result = agent.run(f"""
+    Think step by step:
+
+    1. The user has asked about: "{query}"
+    2. Use the DuckDuckGoSearchTool to search the web for information about this query.
+    3. From the search results, identify 1-2 relevant webpage URLs that might contain detailed information.
+    4. Use the VisitWebpageTool to visit each identified webpage and extract its content.
+    5. Combine the information from the search results and webpage visits.
+    6. Create a comprehensive bullet point summary of all collected information.
+    7. Each bullet point should start with "• " and be on a new line.
+    8. Use the FinalAnswerTool to present your bullet-point summary as the final answer.
+
+    Make sure your bullet points are clear, well-organized, and directly relevant to the user's query.
+    Include the most important and factual information from your research.
+    """)
+    return result
 
 def respond(
     message,
@@ -14,39 +39,55 @@ def respond(
     max_tokens,
     temperature,
     top_p,
+    use_web_search,
 ):
-    messages = [{"role": "system", "content": system_message}]
-
-    for val in history:
-        if val[0]:
-            messages.append({"role": "user", "content": val[0]})
-        if val[1]:
-            messages.append({"role": "assistant", "content": val[1]})
+    # Check if web search is enabled and message starts with a research request
+    if use_web_search and message.strip().lower().startswith(("search:", "research:", "find info:")):
+        query = message.split(":", 1)[1].strip()
+        
+        yield "Searching the web for information about your query. This may take a moment..."
+        
+        try:
+            # Perform the web search and get bullet point summary
+            research_results = research_with_query(query)
+            
+            # Return the research results
+            yield f"Here's what I found about '{query}':\n\n{research_results}"
+        except Exception as e:
+            yield f"Sorry, I encountered an error while searching the web: {str(e)}"
+    else:
+        # Regular chat completion for normal messages
+        messages = [{"role": "system", "content": system_message}]
 
-    messages.append({"role": "user", "content": message})
+        for val in history:
+            if val[0]:
+                messages.append({"role": "user", "content": val[0]})
+            if val[1]:
+                messages.append({"role": "assistant", "content": val[1]})
 
-    response = ""
+        messages.append({"role": "user", "content": message})
 
-    for message in client.chat_completion(
-        messages,
-        max_tokens=max_tokens,
-        stream=True,
-        temperature=temperature,
-        top_p=top_p,
-    ):
-        token = message.choices[0].delta.content
+        response = ""
 
-        response += token
-        yield response
+        for message in client.chat_completion(
+            messages,
+            max_tokens=max_tokens,
+            stream=True,
+            temperature=temperature,
+            top_p=top_p,
+        ):
+            token = message.choices[0].delta.content
+            response += token
+            yield response
 
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
+# Create the Gradio interface
 demo = gr.ChatInterface(
     respond,
     additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
+        gr.Textbox(
+            value="You are a helpful assistant. When users ask you to search for information with 'search:', 'research:', or 'find info:', you will search the web for them.", 
+            label="System message"
+        ),
         gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
         gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
         gr.Slider(
@@ -56,9 +97,15 @@ demo = gr.ChatInterface(
             step=0.05,
             label="Top-p (nucleus sampling)",
         ),
+        gr.Checkbox(value=True, label="Enable web search (use 'search:', 'research:', or 'find info:' to search)")
+    ],
+    examples=[
+        ["search: latest developments in quantum computing"],
+        ["research: climate change impacts in 2023"],
+        ["find info: benefits of meditation"],
+        ["Hello! How are you today?"]
     ],
 )
 
-
 if __name__ == "__main__":
-    demo.launch()
+    demo.launch()