Use HuggingFacePipeline instead of HuggingFaceEndpoint for local models

from langchain_huggingface import HuggingFacePipeline
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline

Load the model and tokenizer from local path

model_path = "unsloth/Llama-3.3-70B-Instruct-bnb-4bit"
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(
model_path,
torch_dtype=torch.bfloat16,
quantization_config=quantization_config,
device_map="auto",
trust_remote_code=True
)

Can I use this unsloth/Llama-4-Scout-17B-16E-Instruct-GGUF same as unsloth/llama3.3?