jordonpeter01/AI-App-Factory-3

1from huggingface_hub import snapshot_download2from transformers import AutoModelForCausalLM, AutoTokenizer3from flask import Flask, request, jsonify4import torch5 6app = Flask(__name__)7 8# Carregue o modelo e o tokenizador9model_id = "gpt2"10device = torch.device("cuda" if torch.cuda.is_available() else "cpu")11model = AutoModelForCausalLM.from_pretrained(model_id).to(device)12tokenizer = AutoTokenizer.from_pretrained(model_id)13 14@app.route("/generate", methods=["POST"])15def generate():16    data = request.get_json()17    prompt = data.get("prompt", "")18 19    # Tokenize a entrada20    inputs = tokenizer.encode(prompt, return_tensors="pt").to(device)21 22    # Gere a resposta usando o modelo23    outputs = model.generate(inputs, max_length=100, temperature=0.7)24 25    # Decodifique a resposta26    response = tokenizer.decode(outputs[0], skip_special_tokens=True)27 28    return jsonify(response)29 30if __name__ == "__main__":31    app.run(debug=True)32