tokenization_tinybuddy.py

from transformers import GPT2Tokenizer

class TinyBuddyTokenizer(GPT2Tokenizer):
    """Slow GPT-2 tokenizer wrapper. Bypasses tokenizer.json entirely."""
    vocab_files_names = {"vocab_file": "vocab.json", "merges_file": "merges.txt"}

    def __init__(self, vocab_file, merges_file, **kwargs):
        super().__init__(vocab_file=vocab_file, merges_file=merges_file, **kwargs)