app.py

import torch
import gradio as gr
from transformers import pipeline
import openai
import os
from dotenv import load_dotenv

load_dotenv()
openai.api_key = os.getenv("OPENAI_API_KEY")

# Use a FREE, NON-MICROSOFT model
model_name = "TinyLlama/TinyLlama-1.1B-Chat-v1.0"

# Create pipeline
generator = pipeline(
    "text-generation",
    model=model_name,
    torch_dtype=torch.bfloat16
    #device_map="cpu"   # change to "cuda" if GPU
)

def explain_topic(topic):
    prompt = f"<|system|>Explain the following topic in a simple way.\n<|user|>{topic}\n<|assistant|>"

    result = generator(
        prompt,
        max_new_tokens=200,
        temperature=0.5,
        do_sample=True
    )

    text = result[0]["generated_text"]

    # Clean output
    if "<|assistant|>" in text:
        text = text.split("<|assistant|>")[-1].strip()

    return text
gr.close_all()

# Build Gradio Interface
app = gr.Interface(
    fn=explain_topic,
    inputs=[gr.Textbox(label="Enter topic to explain", lines=5)],
    outputs=[gr.Textbox(label="Explanation", lines=8)],
    title="AI Explainer (3mtt & Hackathon)",
    description="Uses Hugging Face Transformers pipeline to explain any topic."
)

app.launch(debug=True)