Update services/model_service.py

services/model_service.py

@@ -42,7 +42,7 @@ class ModelService:
                 settings.MODEL_NAME,
                 torch_dtype=torch.float16 if quantization_device == "cuda" else torch.float32,
                 device_map="auto" if quantization_device == "cuda" else None,
-                load_in_8bit=(quantization_bits == 8),
+           #     load_in_8bit=(quantization_bits == 8),
                 trust_remote_code=True
             )
             logger.info(f"Model {settings.MODEL_NAME} loaded successfully on {quantization_device}.")