MIX4_ja-en_fugumt / tokenizer_config.json
twieland's picture
Training in progress, step 5000
3f2946a
{
"eos_token": "</s>",
"model_max_length": 512,
"name_or_path": "staka/fugumt-ja-en",
"pad_token": "<pad>",
"separate_vocabs": false,
"source_lang": "ja",
"sp_model_kwargs": {},
"special_tokens_map_file": "C:\\Users\\Trevor/.cache\\huggingface\\hub\\models--staka--fugumt-ja-en\\snapshots\\8cb8ff81a8625a626c6f0f19cc5082c6181f223a\\special_tokens_map.json",
"target_lang": "en",
"tokenizer_class": "MarianTokenizer",
"unk_token": "<unk>"
}