Update app.py

app.py

@@ -1,32 +1,92 @@
 import gradio as gr
 from huggingface_hub import InferenceClient
+import json
+import re
 
-"""
-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
-"""
 client = InferenceClient("HuggingFaceH4/zephyr-7b-beta")
 
+SYSTEM_PROMPT = """You are an AI agent planner that helps break down tasks into clear, actionable steps. For each task, you will:
+1. Analyze the task and break it down into specific sub-tasks
+2. Create a structured plan with numbered steps
+3. Include any relevant considerations or potential challenges
+4. Format the response as a JSON with the following structure:
+{
+    "task_analysis": "Brief analysis of the main task",
+    "steps": [
+        {
+            "step_number": 1,
+            "description": "Step description",
+            "estimated_time": "Time estimate",
+            "considerations": ["List of considerations"]
+        }
+    ],
+    "potential_challenges": ["List of potential challenges"],
+    "resources_needed": ["List of required resources"]
+}
+
+Keep your responses focused and practical."""
+
+def parse_json_response(response_text):
+    """Extract JSON from the response text."""
+    try:
+        # Find JSON pattern in the text
+        json_match = re.search(r'\{.*\}', response_text, re.DOTALL)
+        if json_match:
+            json_str = json_match.group()
+            return json.loads(json_str)
+        return None
+    except json.JSONDecodeError:
+        return None
+
+def format_plan(plan_json):
+    """Format the JSON plan into a readable markdown string."""
+    if not plan_json:
+        return "Error: Could not parse the plan. Please try again."
+    
+    output = []
+    output.append("# Task Analysis")
+    output.append(plan_json.get("task_analysis", ""))
+    output.append("\n## Detailed Steps")
+    
+    for step in plan_json.get("steps", []):
+        output.append(f"\n### Step {step.get('step_number')}: {step.get('description')}")
+        output.append(f"- Estimated time: {step.get('estimated_time')}")
+        if step.get('considerations'):
+            output.append("\nConsiderations:")
+            for consideration in step['considerations']:
+                output.append(f"- {consideration}")
+    
+    if plan_json.get("potential_challenges"):
+        output.append("\n## Potential Challenges")
+        for challenge in plan_json["potential_challenges"]:
+            output.append(f"- {challenge}")
+    
+    if plan_json.get("resources_needed"):
+        output.append("\n## Required Resources")
+        for resource in plan_json["resources_needed"]:
+            output.append(f"- {resource}")
+    
+    return "\n".join(output)
 
 def respond(
     message,
     history: list[tuple[str, str]],
-    system_message,
-    max_tokens,
-    temperature,
-    top_p,
+    system_message=SYSTEM_PROMPT,
+    max_tokens=1024,
+    temperature=0.7,
+    top_p=0.95,
 ):
     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]})
+    for user_msg, assistant_msg in history:
+        if user_msg:
+            messages.append({"role": "user", "content": user_msg})
+        if assistant_msg:
+            messages.append({"role": "assistant", "content": assistant_msg})
 
     messages.append({"role": "user", "content": message})
 
     response = ""
-
     for message in client.chat_completion(
         messages,
         max_tokens=max_tokens,
@@ -35,19 +95,21 @@ def respond(
         top_p=top_p,
     ):
         token = message.choices[0].delta.content
-
         response += token
-        yield response
+        
+        # Try to parse and format JSON as it comes in
+        plan_json = parse_json_response(response)
+        if plan_json:
+            formatted_response = format_plan(plan_json)
+            yield formatted_response
+        else:
+            yield response
 
-
-"""
-For information on how to customize the ChatInterface, peruse the gradio docs: https://www.gradio.app/docs/chatinterface
-"""
 demo = gr.ChatInterface(
     respond,
     additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=512, step=1, label="Max new tokens"),
+        gr.Textbox(value=SYSTEM_PROMPT, label="System message", lines=5),
+        gr.Slider(minimum=1, maximum=2048, value=1024, step=1, label="Max new tokens"),
         gr.Slider(minimum=0.1, maximum=4.0, value=0.7, step=0.1, label="Temperature"),
         gr.Slider(
             minimum=0.1,
@@ -57,8 +119,10 @@ demo = gr.ChatInterface(
             label="Top-p (nucleus sampling)",
         ),
     ],
+    title="AI Agent Planner",
+    description="I help break down tasks into clear, actionable steps. Describe your task, and I'll create a detailed plan.",
+    theme=gr.themes.Soft(),
 )
 
-
 if __name__ == "__main__":
     demo.launch()