vocab/gpt_neox_chinese_v1/README.md

added vocab (size: 54634) with 22 dummy tokens (new size: 54656)
Vocab size: 54634

训练数据

https://github.com/huggingface/transformers/blob/main/src/transformers/models/gpt_neox_japanese/tokenization_gpt_neox_japanese.py