SpliNet / tokenizer_metadata.json
Angshul's picture
Upload repaired SpliNet 2B-token pretrained model
4ec5e47 verified
Raw History Blame Contribute Delete
489 Bytes
{
"source": "allenai/c4",
"config": "en",
"split": "train",
"model_type": "unigram",
"vocab_size": 32000,
"lowercase": true,
"normalization": "NFKC + SentencePiece nmt_nfkc",
"training_text_bytes": 536874807,
"training_documents": 247312,
"special_token_ids": {
"<unk>": 0,
"<s>": 1,
"</s>": 2,
"<pad>": 3,
"<cls>": 4,
"<sep>": 5,
"<mask>": 6
},
"sha256_spiece_model": "119ec6b2af9cbbc56f297bd606b69f79f6e3130a34ee56122ee813a58d15bb9d"
}