my_model/config/kbvqa_config.py

import os

# Model and Tokenizer Settings
KBVQA_MODEL_NAME = "m7mdal7aj/fine_tunned_llama_2_merged"
QUANTIZATION = '4bit'
MAX_CONTEXT_WINDOW = 4000
ADD_EOS_TOKEN = False
TRUST_REMOTE = False
USE_FAST = True
LOW_CPU_MEM_USAGE = True

# Access Token
HUGGINGFACE_TOKEN = os.getenv("HUGGINGFACE_TOKEN")