| import os | |
| from transformers import LlamaTokenizer | |
| CURRENT_DIR = os.path.dirname(os.path.abspath(__file__)) | |
| TOKENIZER_DIR = os.path.join(CURRENT_DIR, "BELLE-LLaMA-EXT-7B") | |
| tokenizer = LlamaTokenizer.from_pretrained(TOKENIZER_DIR) | |
| # print(tokenizer.tokenize("大道发生")) |