from transformers import AutoTokenizer

# Load the correct base tokenizer
tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")

# Save tokenizer config properly
tokenizer.save_pretrained("./POSNEG-Dataset-API")