app.py

import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
from flask import Flask, request, jsonify
from threading import Thread

app = Flask(__name__)

tokenizer = AutoTokenizer.from_pretrained("HuggingFaceH4/zephyr-7b-beta")
model = AutoModelForCausalLM.from_pretrained(
    "HuggingFaceH4/zephyr-7b-beta",
    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
    device_map="auto"
)

@app.route("/api/chat", methods=["POST"])
def chat():
    data = request.get_json()
    question = data.get("question", "")

    prompt = f"Eres BITER, un mentor experto en negocios. Siempre respondes en español con consejos breves y útiles.\nUsuario: {question}\nBITER:"
    
    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
    outputs = model.generate(**inputs, max_new_tokens=200)
    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    respuesta_final = response.split("BITER:")[-1].strip()

    return jsonify({"choices": [{"message": {"content": respuesta_final}}]})

def run():
    app.run(host='0.0.0.0', port=7860)

Thread(target=run).start()