tokenizer_config.json

{"tokenizer_class": "T5TokenizerFast", "model_max_length": 512, "tokenizer_file": "merges.txt"}