vocab/grok_1/__init__.py

"""

## reference
https://github.com/xai-org/grok-1/blob/main/run.py

        vocab_size=128 * 1024,
        pad_token=0,
        eos_token=2,
"""


import os
import sentencepiece
from tokenizer import sptokenizer_patch

CURRENT_DIR = os.path.dirname(os.path.abspath(__file__))
MODEL_FILE = os.path.join(CURRENT_DIR, "tokenizer.model")

tokenizer = sentencepiece.SentencePieceProcessor(model_file=MODEL_FILE)

# print(tokenizer.decode([1,2,3], skip_special_tokens=True))