|
--- |
|
language: ja |
|
license: cc-by-sa-4.0 |
|
library_name: transformers |
|
datasets: |
|
- cc100 |
|
- mc4 |
|
- oscar |
|
- wikipedia |
|
- izumi-lab/cc100-ja |
|
- izumi-lab/mc4-ja-filter-ja-normal |
|
- izumi-lab/oscar2301-ja-filter-ja-normal |
|
- izumi-lab/wikipedia-ja-20230720 |
|
- izumi-lab/wikinews-ja-20230728 |
|
|
|
widget: |
|
- text: 東京大学で[MASK]の研究をしています。 |
|
|
|
--- |
|
|
|
# DeBERTa V2 small Japanese |
|
|
|
This is a [DeBERTaV2](https://github.com/microsoft/DeBERTa) model pretrained on Japanese texts. |
|
The codes for the pretraining are available at [retarfi/language-pretraining](https://github.com/retarfi/language-pretraining/releases/tag/v2.2.1). |
|
|
|
|
|
## How to use |
|
|
|
You can use this model for masked language modeling as follows: |
|
|
|
```python |
|
from transformers import AutoTokenizer, AutoModelForMaskedLM |
|
tokenizer = AutoTokenizer.from_pretrained("izumi-lab/deberta-v2-small-japanese") |
|
model = AutoModelForMaskedLM.from_pretrained("izumi-lab/deberta-v2-small-japanese") |
|
... |
|
``` |
|
|
|
|
|
## Tokenization |
|
|
|
The model uses a sentencepiece-based tokenizer, the vocabulary was trained on the Japanese Wikipedia using [sentencepiece](https://github.com/google/sentencepiece). |
|
|
|
|
|
## Training Data |
|
|
|
We used the following corpora for pre-training: |
|
|
|
- [Japanese portion of CC-100](https://huggingface.co/datasets/izumi-lab/cc100-ja) |
|
- [Japanese portion of mC4](https://huggingface.co/datasets/izumi-lab/mc4-ja-filter-ja-normal) |
|
- [Japanese portion of OSCAR2301](izumi-lab/oscar2301-ja-filter-ja-normal) |
|
- [Japanese Wikipedia as of July 20, 2023](https://huggingface.co/datasets/izumi-lab/wikipedia-ja-20230720) |
|
- [Japanese Wikinews as of July 28, 2023](https://huggingface.co/datasets/izumi-lab/wikinews-ja-20230728) |
|
|
|
|
|
## Training Parameters |
|
|
|
- learning_rate: 6e-4 |
|
- total_train_batch_size: 2,016 |
|
- max_seq_length: 128 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-06 |
|
- lr_scheduler_type: linear schedule with warmup |
|
- training_steps: 1,000,000 |
|
- warmup_steps: 100,000 |
|
- precision: BF16 |
|
|
|
|
|
## Fine-tuning on General NLU tasks |
|
|
|
We evaluate our model with the average of five seeds. |
|
|
|
|
|
| Model | JSTS | JNLI | JCommonsenseQA | |
|
|---------------------------------------------------------------------------|------------------|-----------|----------------| |
|
| | Pearson/Spearman | acc | acc | |
|
| **DeBERTaV2 small** | **0.890/0.846** | **0.880** | **0.737** | |
|
| [UTokyo BERT small](https://huggingface.co/izumi-lab/bert-small-japanese) | 0.889/0.841 | 0.841 | 0.715 | |
|
|
|
|
|
## Citation |
|
|
|
Citation will be updated. |
|
Please check when you would cite. |
|
|
|
``` |
|
@article{Suzuki-etal-2023-ipm, |
|
title = {Constructing and analyzing domain-specific language model for financial text mining} |
|
author = {Masahiro Suzuki and Hiroki Sakaji and Masanori Hirano and Kiyoshi Izumi}, |
|
journal = {Information Processing \& Management}, |
|
volume = {60}, |
|
number = {2}, |
|
pages = {103194}, |
|
year = {2023}, |
|
doi = {10.1016/j.ipm.2022.103194} |
|
} |
|
``` |
|
|
|
|
|
## Licenses |
|
|
|
The pretrained models are distributed under the terms of the [Creative Commons Attribution-ShareAlike 4.0](https://creativecommons.org/licenses/by-sa/4.0/). |
|
|
|
|
|
## Acknowledgments |
|
|
|
This work was supported in part by JSPS KAKENHI Grant Number JP21K12010, and the JST-Mirai Program Grant Number JPMJMI20B1, Japan. |
|
|
