mega-small / tokenizer_config.json
Frankenstein-Labs's picture
Upload MEGA model
a0883c1 verified
Raw History Blame Contribute Delete
260 Bytes
{
"tokenizer_class": "MegaTokenizer",
"unk_token": "<unk>",
"bos_token": "<s>",
"eos_token": "</s>",
"pad_token": "<pad>",
"modality_tokens": [
"<image>",
"<audio>",
"<video>"
],
"model_type": "unigram_or_bpe",
"vocab_size": 800
}