app.py

import gradio as gr
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM

# Model and Tokenizer Setup
model_name = "unsloth/gemma-3-4b-it-unsloth-bnb-4bit"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(
    model_name,
    load_in_4bit=True,
    device_map="auto",
    torch_dtype=torch.bfloat16, #important for speed.
)

def generate_response(prompt):
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    outputs = model.generate(**inputs, max_new_tokens=256) # Adjust max_new_tokens as needed
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    return response

# Gradio Interface
iface = gr.Interface(
    fn=generate_response,
    inputs=gr.Textbox(lines=5, placeholder="Enter your prompt here..."),
    outputs=gr.Textbox(),
    title="Gemma 3-4B Inference",
    description="Run the unsloth/gemma-3-4b-it-unsloth-bnb-4bit model.",
)

if __name__ == "__main__":
    iface.launch(server_name="0.0.0.0", server_port=7860) #important for spaces.