app.py

import gradio as gr
from transformers import AutoTokenizer, AutoModelForCausalLM, TextIteratorStreamer
from threading import Thread
import torch

# Load the model and tokenizer
model_name = "Qwen/Qwen2.5-Coder-1.5B"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForCausalLM.from_pretrained(model_name, torch_dtype=torch.float32, low_cpu_mem_usage=True)

def generate_code(prompt, max_length):
    inputs = tokenizer(prompt, return_tensors="pt")
    streamer = TextIteratorStreamer(tokenizer)
    
    generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=int(max_length))
    thread = Thread(target=model.generate, kwargs=generation_kwargs)
    thread.start()

    generated_text = ""
    for new_text in streamer:
        generated_text += new_text
        yield generated_text

# Gradio interface
iface = gr.Interface(
    fn=generate_code,
    inputs=[
        gr.Textbox(lines=5, label="Enter your prompt"),
        gr.Slider(minimum=50, maximum=500, value=200, step=10, label="Max Length")
    ],
    outputs=gr.Code(language="python", label="Generated Code"),
    title="Qwen2.5-Coder-1.5B Code Generator",
    description="Enter a prompt to generate Python code using Qwen2.5-Coder-1.5B",
    examples=[
        ["Write a Python function to calculate the factorial of a number.", 200],
        ["Create a class representing a simple bank account with deposit and withdraw methods.", 300]
    ]
)

iface.launch()