Korean BERT Model base weight - v1

This model is the Korean implementation of bert model. The license is as follows. It is used as a base for various purposes. Pretrained models and Tokenizers related to KOBERT comply with the apache2.0 license according to the license of the dependency package.

Tokenizer

from transformers import AutoTokenizer
tokenizer = AutoTokenizer.from_pretrained('danielpark/bert-korean-tokenizer-v1')
tokenizer.encode("SKT의 KOBERT의 TOKENIZER 베이스를 활용합니다.")

Bert

import torch
from transformers import BertModel
model = BertModel.from_pretrained('danielpark/kobert-base-v1')
text = "SKT의 KOBERT를 base로 사용합니다."
inputs = tokenizer.batch_encode_plus([text])
out = model(input_ids = torch.tensor(inputs['input_ids']),
              attention_mask = torch.tensor(inputs['attention_mask']))
out.pooler_output.shape
torch.Size([1, 768])

References

[1] documentation https://sktelecom.github.io/project/kobert/ <BR> [2] github https://github.com/SKTBrain/KoBERT