File size: 2,415 Bytes
22ca508 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 |
"""
Cargador de modelos GPT para uso local
"""
import torch
from transformers import AutoTokenizer, AutoModelForCausalLM
import logging
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
class ModelLoader:
def __init__(self):
self.model = None
self.tokenizer = None
# Optimización para Apple Silicon
if torch.backends.mps.is_available():
self.device = "mps"
elif torch.cuda.is_available():
self.device = "cuda"
else:
self.device = "cpu"
def load_model(self, model_name="microsoft/DialoGPT-medium"):
"""
Carga un modelo GPT desde Hugging Face
Args:
model_name (str): Nombre del modelo en Hugging Face Hub
"""
try:
logger.info(f"Cargando modelo: {model_name}")
logger.info(f"Usando dispositivo: {self.device}")
# Cargar tokenizer
self.tokenizer = AutoTokenizer.from_pretrained(model_name)
# Agregar pad_token si no existe
if self.tokenizer.pad_token is None:
self.tokenizer.pad_token = self.tokenizer.eos_token
# Cargar modelo
self.model = AutoModelForCausalLM.from_pretrained(
model_name,
torch_dtype=torch.float16 if self.device == "cuda" else torch.float32,
device_map="auto" if self.device == "cuda" else None
)
if self.device == "cpu":
self.model = self.model.to(self.device)
logger.info("Modelo cargado exitosamente")
return True
except Exception as e:
logger.error(f"Error al cargar el modelo: {str(e)}")
return False
def get_model_info(self):
"""Retorna información del modelo cargado"""
if self.model is None:
return {"status": "No hay modelo cargado"}
return {
"status": "Modelo cargado",
"device": self.device,
"model_type": type(self.model).__name__,
"vocab_size": self.tokenizer.vocab_size if self.tokenizer else "N/A"
}
def is_loaded(self):
"""Verifica si hay un modelo cargado"""
return self.model is not None and self.tokenizer is not None
|