tokenizer_config.json

{"Ngram_vocab_size": 32768, "model_max_length": 512}