AIbase
EN
Home
Bert Base Japanese Basic Char V2
B

Bert Base Japanese Basic Char V2

Developed by hiroshi-matsuda-rit
This is a Japanese BERT pre-trained model based on character-level tokenization and whole word masking techniques, requiring no dependency on `fugashi` or `unidic_lite` toolkits.
Large Language Model
Transformers
 Japanese#Japanese Character-level BERT#Dependency-free Tokenization#Whole Word Masking
Downloads 14
Release Time : 3/2/2022

Model Overview

This model is a Japanese BERT pre-trained model that employs character-level tokenization and whole word masking techniques, suitable for various Japanese natural language processing tasks.

Model Features

Character-level Tokenization
Uses character-level tokenization, eliminating the need for external tokenization toolkits.
Whole Word Masking
Employs whole word masking during pre-training to enhance the model's understanding of Japanese text.
Lightweight Dependencies
No dependency on `fugashi` or `unidic_lite` toolkits, simplifying deployment and usage.

Model Capabilities

Text Classification
Named Entity Recognition
Question Answering
Text Generation

Use Cases

Natural Language Processing
Japanese Text Classification
Used for classifying Japanese text, such as sentiment analysis, topic classification, etc.
Japanese Named Entity Recognition
Used to identify named entities in Japanese text, such as person names, locations, organization names, etc.
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
EnglishįŽ€äŊ“中文įšéĢ”中文ãĢãģんご
Š 2025AIbase