How to use yumenghuang/backward_tokenizer with Transformers:
# Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("yumenghuang/backward_tokenizer", dtype="auto")
