app.py

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

# Initialize model and tokenizer
MODEL_NAME = "TheBloke/deepseek-coder-1.3b-instruct-GPTQ"
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, use_fast=True)

model = AutoModelForCausalLM.from_pretrained(
    MODEL_NAME,
    device_map="cpu",  # Optimized for CPU
    quantization_config=BaseQuantizeConfig(),  # Required for GPTQ models
    torch_dtype=torch.float32,  # Better CPU compatibility
    low_cpu_mem_usage=True
)

def generate_text(prompt, max_length=100, temperature=0.7):
    inputs = tokenizer(prompt, return_tensors="pt").to("cpu")
    
    with torch.no_grad():
        outputs = model.generate(
            **inputs,
            max_length=max_length,
            temperature=temperature,
            pad_token_id=tokenizer.eos_token_id
        )
    
    return tokenizer.decode(outputs[0], skip_special_tokens=True)

# Gradio UI
with gr.Blocks(theme="soft") as demo:
    gr.Markdown("# 🧠 DeepSeek Coder 1.3B Text Generator\nOptimized for CPU execution on HuggingFace Spaces")
    
    with gr.Row():
        with gr.Column():
            prompt = gr.Textbox(
                label="Input Prompt",
                placeholder="Enter your programming/code-related question...",
                lines=5
            )
            max_length = gr.Slider(50, 500, value=150, label="Max Output Length")
            temperature = gr.Slider(0.1, 1.0, value=0.7, label="Creativity Level")
            submit = gr.Button("Generate Code", variant="primary")
        
        output = gr.Textbox(label="Generated Output", lines=10)
    
    submit.click(
        fn=generate_text,
        inputs=[prompt, max_length, temperature],
        outputs=output
    )
    
    gr.Examples(
        examples=[
            ["Write a Python function to calculate Fibonacci numbers"],
            ["Explain the difference between list and tuples in Python"],
            ["Create a simple Flask API endpoint for user registration"]
        ],
        fn=generate_text,
        inputs=[prompt, max_length, temperature],
        outputs=output,
        cache_examples=False  # Save memory
    )

if __name__ == "__main__":
    demo.launch()