app.py

from flask import Flask, request, jsonify
from transformers import AutoTokenizer, AutoModelForCausalLM
from huggingface_hub import login
import torch
import os

app = Flask(__name__)

# ✅ Securely fetch HF Token from environment (invisible to users)
hf_token = os.getenv("HF_TOKEN")
if not hf_token:
    raise ValueError("HF_TOKEN is not set in environment variables!")

# 🔐 Authenticate
login(token=hf_token)

# 🔄 Load model from Hugging Face
model_id = "dexcommunity/dex"
print("🔄 Loading model...")

# ⚠️ FIXED this line: fast tokenizer disabled to avoid tokenizer.json crash
tokenizer = AutoTokenizer.from_pretrained(model_id, token=hf_token, use_fast=False)

model = AutoModelForCausalLM.from_pretrained(model_id, token=hf_token)

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)
print("✅ Model loaded!")

@app.route('/chat', methods=['POST'])
def chat():
    try:
        data = request.get_json()
        msg = data.get("message", "")
        if not msg:
            return jsonify({"error": "No message sent"}), 400

        prompt = f"User: {msg}\nDex:"
        inputs = tokenizer(prompt, return_tensors="pt").to(device)
        outputs = model.generate(
            inputs.input_ids,
            max_length=256,
            do_sample=True,
            top_k=50,
            top_p=0.95,
            temperature=0.7,
            pad_token_id=tokenizer.eos_token_id
        )
        text = tokenizer.decode(outputs[0], skip_special_tokens=True)
        reply = text.split("Dex:")[-1].strip()
        return jsonify({"reply": reply})
    except Exception as e:
        return jsonify({"error": str(e)}), 500

if __name__ == "__main__":
    app.run(host='0.0.0.0', port=7860)