Japanese Finance Sentiment ModernBERT

This is a Japanese ModernBERT model fine-tuned on financial texts in the Japanese language.

Model Architecture

The model architecture follows the same structure as ModernBERT.

Training Data

We used textual information from the Economy Watchers Survey published by the Financial Services Agency of Japan. We are training the model using data from 2001 to 2009.

Training

The training code is publicly available at the following notebook: Training Notebook

Usage

from transformers import AutoTokenizer, AutoModelForSequenceClassification
tokenizer = AutoTokenizer.from_pretrained('takaito/modernbert-ja-310m-eco-watcher')
model = AutoModelForSequenceClassification.from_pretrained('takaito/modernbert-ja-310m-eco-watcher')

License

MIT License

Citation

@article{xxxx,
  title = {xxxx}
  author = {Kaito Takano and Kei Nakagawa},
  journal = {xxxx},
  volume = {xx},
  number = {x},
  pages = {xx-xx},
  year = {20xx},
  doi = {xxxx}
}
Downloads last month
34
Safetensors
Model size
0.3B params
Tensor type
F32
ยท
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for takaito/modernbert-ja-310m-eco-watcher

Finetuned
(8)
this model