donut_rus / tokenizer_config.json
Akajackson's picture
Training in progress, epoch 0
f355a72
raw
history blame
314 Bytes
{
"eos_token": "</s>",
"model_max_length": 512,
"pad_token": "<pad>",
"processor_class": "DonutProcessor",
"separate_vocabs": false,
"source_lang": "ru",
"sp_model_kwargs": {},
"special_tokens_map_file": null,
"target_lang": "en",
"tokenizer_class": "MarianTokenizer",
"unk_token": "<unk>"
}