app.py

"""Gradio chat app for math reasoning with Qwen2.5 + LoRA."""

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

MODEL_ID = "Qwen/Qwen2.5-1.5B-Instruct"
ADAPTER_ID = "arinbalyan/math-reasoning-lora"
DEVICE = "cuda" if torch.cuda.is_available() else "cpu"


def load_model():
    tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, padding_side="left")
    tokenizer.pad_token = tokenizer.eos_token

    base = AutoModelForCausalLM.from_pretrained(
        MODEL_ID,
        torch_dtype=torch.float16 if DEVICE == "cuda" else torch.float32,
        device_map="auto",
    )
    model = PeftModel.from_pretrained(base, ADAPTER_ID)
    model.eval()
    return model, tokenizer


model, tokenizer = load_model()


def format_prompt(question: str) -> str:
    return f"<|im_start|>user\n{question}<|im_end|>\n<|im_start|>assistant\n"


def generate(message: str, history: list) -> str:
    prompt = format_prompt(message)
    inputs = tokenizer(prompt, return_tensors="pt").to(DEVICE)

    with torch.no_grad():
        outputs = model.generate(
            **inputs,
            max_new_tokens=512,
            temperature=0.7,
            top_p=0.9,
            do_sample=True,
            pad_token_id=tokenizer.eos_token_id,
        )

    response = tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True)
    return response.strip()


demo = gr.ChatInterface(
    fn=generate,
    title="🧮 Math Reasoning Chat",
    description="Qwen2.5-1.5B fine-tuned on GSM8K for chain-of-thought math reasoning.",
    theme="soft",
)

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