D

Deberta V2 Base Japanese

Developed by ku-nlp
A Japanese DeBERTa V2 base model pretrained on Japanese Wikipedia, CC-100, and OSCAR corpora, suitable for masked language modeling and downstream task fine-tuning.
Downloads 38.93k
Release Time : 1/5/2023

Model Overview

This is a DeBERTa V2 model pretrained on large-scale Japanese corpora, primarily designed for Japanese masked language modeling tasks, but can also be fine-tuned for various natural language understanding tasks.

Model Features

High-Quality Japanese Pretraining
Pretrained on high-quality Japanese corpora including Japanese Wikipedia, CC-100, and OSCAR, covering a wide range of Japanese linguistic features.
Professional Tokenization
Input text requires professional tokenization via Juman++ to ensure accurate understanding of Japanese text by the model.
Multi-Task Adaptability
In addition to masked language modeling, it can be fine-tuned for various natural language understanding tasks such as text classification and question answering.

Model Capabilities

Japanese Text Understanding
Masked Language Modeling
Natural Language Processing Task Fine-tuning

Use Cases

Natural Language Understanding
Text Classification
Can be used for Japanese text classification tasks such as sentiment analysis and topic classification.
Achieved 0.970 accuracy on the MARC-ja task
Semantic Similarity Calculation
Can be used to calculate semantic similarity between Japanese text pairs.
Achieved Pearson correlation coefficient of 0.922 on the JSTS task
Question Answering System
Can be used to build Japanese question answering systems.
Achieved F1 score of 0.951 on the JSQuAD task
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
© 2025AIbase