roberta-base-japanese

nlp-waseda/roberta-base-japanese

Model description

This is a Japanese RoBERTa base model pretrained on Japanese Wikipedia and the Japanese portion of CC-100.

As of 2026-04-20, the license of this model has been changed from CC BY-SA 4.0 to Apache License 2.0, as Creative Commons licenses are generally not well-suited for machine learning models.

How to use

You can use this model for masked language modeling as follows:

from transformers import AutoTokenizer, AutoModelForMaskedLM
tokenizer = AutoTokenizer.from_pretrained("nlp-waseda/roberta-base-japanese")
model = AutoModelForMaskedLM.from_pretrained("nlp-waseda/roberta-base-japanese")

sentence = '早稲田 大学 で 自然 言語 処理 を [MASK] する 。' # input should be segmented into words by Juman++ in advance
encoding = tokenizer(sentence, return_tensors='pt')
...

You can fine-tune this model on downstream tasks.

Tokenization

The input text should be segmented into words by Juman++ in advance. Juman++ 2.0.0-rc3 was used for pretraining. Each word is tokenized into tokens by sentencepiece.

BertJapaneseTokenizer now supports automatic JumanppTokenizer and SentencepieceTokenizer. You can use this model without any data preprocessing.

Vocabulary

The vocabulary consists of 32000 tokens including words (JumanDIC) and subwords induced by the unigram language model of sentencepiece.

Training procedure

This model was trained on Japanese Wikipedia (as of 20210920) and the Japanese portion of CC-100. It took a week using eight NVIDIA A100 GPUs.

The following hyperparameters were used during pretraining:

• learning_rate: 1e-4
• per_device_train_batch_size: 256
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 2
• total_train_batch_size: 4096
• max_seq_length: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 700000
• warmup_steps: 10000
• mixed_precision_training: Native AMP

Performance on JGLUE

See the Baseline Scores of JGLUE.