Japanese Roberta Base

🚀 japanese-roberta-base

This repository offers a base-sized Japanese RoBERTa model. It was trained using code from the Github repository rinnakk/japanese-pretrained-models by rinna Co., Ltd.. The model aims to provide effective solutions for Japanese natural language processing tasks.

🚀 Quick Start

Loading the Model

from transformers import AutoTokenizer, AutoModelForMaskedLM

tokenizer = AutoTokenizer.from_pretrained("rinna/japanese-roberta-base", use_fast=False)
tokenizer.do_lower_case = True  # due to some bug of tokenizer config loading

model = AutoModelForMaskedLM.from_pretrained("rinna/japanese-roberta-base")

Using the Model for Masked Token Prediction

⚠️ Important Note

• Use [CLS]: To predict a masked token, add a [CLS] token before the sentence as it was used during model training.
• Use [MASK] after tokenization: Directly typing [MASK] in an input string and replacing a token with [MASK] after tokenization yield different results. It's better to use [MASK] after tokenization, but the Huggingface Inference API only supports typing [MASK] in the input string.
• Provide position_ids explicitly: When position_ids are not provided for a Roberta* model, Huggingface's transformers constructs it starting from padding_idx instead of 0, which doesn't work well with rinna/japanese-roberta-base. So, construct position_ids starting from 0 manually.

💡 Usage Tip

When using the model for masked token prediction, carefully follow the above notes to get accurate results.

Example

# original text
text = "4年に1度オリンピックは開かれる。"

# prepend [CLS]
text = "[CLS]" + text

# tokenize
tokens = tokenizer.tokenize(text)
print(tokens)  # output: ['[CLS]', '▁4', '年に', '1', '度', 'オリンピック', 'は', '開かれる', '。']

# mask a token
masked_idx = 5
tokens[masked_idx] = tokenizer.mask_token
print(tokens)  # output: ['[CLS]', '▁4', '年に', '1', '度', '[MASK]', 'は', '開かれる', '。']

# convert to ids
token_ids = tokenizer.convert_tokens_to_ids(tokens)
print(token_ids)  # output: [4, 1602, 44, 24, 368, 6, 11, 21583, 8]

# convert to tensor
import torch
token_tensor = torch.LongTensor([token_ids])

# provide position ids explicitly
position_ids = list(range(0, token_tensor.size(1)))
print(position_ids)  # output: [0, 1, 2, 3, 4, 5, 6, 7, 8]
position_id_tensor = torch.LongTensor([position_ids])

# get the top 10 predictions of the masked token
with torch.no_grad():
    outputs = model(input_ids=token_tensor, position_ids=position_id_tensor)
    predictions = outputs[0][0, masked_idx].topk(10)

for i, index_t in enumerate(predictions.indices):
    index = index_t.item()
    token = tokenizer.convert_ids_to_tokens([index])[0]
    print(i, token)

"""
0 総会
1 サミット
2 ワールドカップ
3 フェスティバル
4 大会
5 オリンピック
6 全国大会
7 党大会
8 イベント
9 世界選手権
"""

✨ Features

• Transformer-based: A 12-layer, 768-hidden-size transformer-based masked language model.
• Trained on large datasets: Trained on Japanese CC-100 and Japanese Wikipedia to optimize a masked language modelling objective.

📚 Documentation

Model architecture

A 12-layer, 768-hidden-size transformer-based masked language model.

Training

The model was trained on Japanese CC-100 and Japanese Wikipedia to optimize a masked language modelling objective on 8*V100 GPUs for around 15 days. It reaches ~3.9 perplexity on a dev set sampled from CC-100.

Tokenization

The model uses a sentencepiece-based tokenizer, and the vocabulary was trained on the Japanese Wikipedia using the official sentencepiece training script.

Release date

August 25, 2021

How to cite

@misc{rinna-japanese-roberta-base,
    title = {rinna/japanese-roberta-base},
    author = {Zhao, Tianyu and Sawada, Kei},
    url = {https://huggingface.co/rinna/japanese-roberta-base}
}

@inproceedings{sawada2024release,
    title = {Release of Pre-Trained Models for the {J}apanese Language},
    author = {Sawada, Kei and Zhao, Tianyu and Shing, Makoto and Mitsui, Kentaro and Kaga, Akio and Hono, Yukiya and Wakatsuki, Toshiaki and Mitsuda, Koh},
    booktitle = {Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024)},
    month = {5},
    year = {2024},
    pages = {13898--13905},
    url = {https://aclanthology.org/2024.lrec-main.1213},
    note = {\url{https://arxiv.org/abs/2404.01657}}
}

📄 License

The MIT license

📦 Additional Information