tokenizer_config.json

{
  "clean_up_tokenization_spaces": true,
  "eos_token": "</s>",
  "model_max_length": 512,
  "pad_token": "<pad>",
  "return_tensors": "pt",
  "separate_vocabs": false,
  "source_lang": "en",
  "sp_model_kwargs": {},
  "target_lang": "fr",
  "tokenizer_class": "MarianTokenizer",
  "unk_token": "<unk>"
}