app.py

# app.py
from flask import Flask, send_file, request, jsonify
from transformers import AutoModelForCausalLM, AutoTokenizer
import torch
from functools import lru_cache
import os

app = Flask(__name__)

@lru_cache(maxsize=1)
def load_model():
    """Load model and tokenizer with caching"""
    tokenizer = AutoTokenizer.from_pretrained("amd/AMD-OLMo-1B")
    model = AutoModelForCausalLM.from_pretrained(
        "amd/AMD-OLMo-1B",
        torch_dtype=torch.float16,
        device_map="auto"
    )
    return model, tokenizer

def generate_response(prompt, model, tokenizer):
    """Generate response from the model"""
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    with torch.no_grad():
        outputs = model.generate(
            **inputs,
            max_length=200,
            num_return_sequences=1,
            temperature=0.7,
            top_p=0.9,
            repetition_penalty=1.2,
            pad_token_id=tokenizer.eos_token_id
        )
        response = tokenizer.decode(outputs[0], skip_special_tokens=True)
        return response.replace(prompt, "").strip()

@app.route('/')
def home():
    return send_file('index.html')

@app.route('/message', methods=['POST'])
def message():
    try:
        data = request.json
        user_message = data.get('message', '')
        
        if not user_message:
            return jsonify({"response": "عذراً، لم أفهم رسالتك"})
        
        model, tokenizer = load_model()
        response = generate_response(user_message, model, tokenizer)
        
        return jsonify({"response": response})
    
    except Exception as e:
        return jsonify({"response": f"عذراً، حدث خطأ: {str(e)}"})

if __name__ == '__main__':
    app.run(debug=True)