File size: 1,864 Bytes
7a3dd67 e0da0b4 abaf9f1 0dad39b 32a90bc 86179ff 32a90bc 0dad39b abaf9f1 16f48ef 86179ff abaf9f1 86179ff abaf9f1 32a90bc 0dad39b abaf9f1 32a90bc abaf9f1 2ebc95a |
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 |
from transformers import AutoTokenizer, AutoModelForCausalLM, LlamaConfig
from config.config import settings
from sentence_transformers import SentenceTransformer
import torch
import logging
logger = logging.getLogger(__name__)
class ModelService:
_instance = None
def __new__(cls):
if cls._instance is None:
cls._instance = super().__new__(cls)
cls._instance._initialized = False
return cls._instance
def __init__(self):
if not self._initialized:
self._initialized = True
self._load_models()
def _load_models(self):
try:
# Load tokenizer
self.tokenizer = AutoTokenizer.from_pretrained(settings.MODEL_NAME)
# Load model configuration
config = LlamaConfig.from_pretrained(settings.MODEL_NAME)
# Check quantization type and adjust accordingly
if config.get('quantization_config', {}).get('type', '') == 'compressed-tensors':
logger.warning("Quantization type 'compressed-tensors' is not supported. Switching to 'bitsandbytes_8bit'.")
config.quantization_config['type'] = 'bitsandbytes_8bit'
# Load model with the updated configuration
self.model = AutoModelForCausalLM.from_pretrained(
settings.MODEL_NAME,
config=config,
torch_dtype=torch.float16 if settings.DEVICE == "cuda" else torch.float32,
device_map="auto" if settings.DEVICE == "cuda" else None
)
# Load sentence embedder
self.embedder = SentenceTransformer(settings.EMBEDDER_MODEL)
except Exception as e:
logger.error(f"Error loading models: {e}")
raise
def get_models(self):
return self.tokenizer, self.model, self.embedder |