File size: 1,446 Bytes
9435cbe e26d4af 9435cbe |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 |
import gradio as gr
from transformers import pipeline
# Load a free Hugging Face model (small + free to run)
generator = pipeline("text2text-generation", model="google/flan-t5-small")
# Agent function
def agentic_ai(user_input):
# Step 1: Analyze input
analysis_prompt = f"Analyze the intent of this input: {user_input}"
analysis = generator(analysis_prompt, max_length=50, do_sample=False)[0]['generated_text']
# Step 2: Decide what to do (simple rule-based agent)
if "summarize" in user_input.lower():
task_prompt = f"Summarize this text in 2 lines: {user_input}"
elif "question" in user_input.lower() or "?" in user_input:
task_prompt = f"Answer this question briefly: {user_input}"
else:
task_prompt = f"Generate a helpful response: {user_input}"
# Step 3: LLM Response
response = generator(task_prompt, max_length=80, do_sample=False)[0]['generated_text']
# Step 4: Return both analysis + final response
return f"🔎 Agent Analysis: {analysis}\n\n💡 Agent Response: {response}"
# Gradio UI
demo = gr.Interface(
fn=agentic_ai,
inputs=gr.Textbox(lines=3, placeholder="Type your text here..."),
outputs="text",
title="🤖 Mini Agentic LLM App",
description="Smallest free demo of an Agentic AI using NLP + LLM on Hugging Face & Gradio. Input few lines or paragraph with question and Click Submit"
)
if __name__ == "__main__":
demo.launch()
|