llama_huggingface.py

from langchain_huggingface import ChatHuggingFace, HuggingFaceEndpoint
import os
from dotenv import load_dotenv
load_dotenv()


llm=''
chat_model=''
HF_TOKEN=os.getenv("HF_TOKEN")


def init_llama_chatmodel(repo_id):
    global llm
    global chat_model
    llm = HuggingFaceEndpoint(
        repo_id=repo_id,
        task="text-generation"
    )
    chat_model = ChatHuggingFace(llm=llm)
    return chat_model

# def get_llama_endpoint(repo_id):
#     llm = HuggingFaceEndpoint(
#         repo_id=repo_id,
#         task="text-generation",
#         max_new_tokens=512,
#         do_sample=False,
#         repetition_penalty=1.03,
#     )
#     return llm

# def get_llama_chatmodel(llm):
#     chat_model = ChatHuggingFace(llm=llm)
#     return chat_model