trakad-embed-v2 / tokenizer_config.json
hakansabunis's picture
trakad-embed-v2 — Turkish academic SimCSE fine-tune (633K theses, MultipleNegativesRankingLoss + subject-aware hard negatives)
f1a23a6 verified
raw
history blame contribute delete
314 Bytes
{
"add_prefix_space": true,
"backend": "tokenizers",
"bos_token": "<s>",
"cls_token": "<s>",
"eos_token": "</s>",
"is_local": false,
"mask_token": "<mask>",
"model_max_length": 128,
"pad_token": "<pad>",
"sep_token": "</s>",
"tokenizer_class": "XLMRobertaTokenizer",
"unk_token": "<unk>"
}