| import os | |
| from transformers import AutoTokenizer, BloomTokenizerFast | |
| CURRENT_DIR = os.path.dirname(os.path.abspath(__file__)) | |
| TOKENIZER_DIR = os.path.join(CURRENT_DIR, "tokenizer") | |
| tokenizer = AutoTokenizer.from_pretrained(TOKENIZER_DIR, trust_remote_code=True) | |
| # vocab_size = len(tokenizer.get_vocab()) | |
| # vocab_size = tokenizer.vocab_size | |