tokenizer_config.json

{
  "model_max_length": 512,
  "padding": "max_length",
  "truncation": true
}