tokenizer_config.json

{
    "tokenizer": "SmilesTokenizerSeparateTaskToken",
    "path_to_vocabulary": "data/vocabularies/guacamol.txt",
    "max_molecule_length": 128
  }