OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Kocharelectra

KoCharELECTRA is a public Python project providing Base and Small Korean ELECTRA checkpoints with a custom syllable-level tokenizer for use through Hugging Face Transformers.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Kocharelectra — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use KoCharELECTRA when a Korean NLP workflow calls for a character-level ELECTRA checkpoint and can accommodate the project's custom tokenizer.

Who it is for

  • Developers encoding Korean text with character-level tokenization.
  • Researchers and practitioners adapting Korean pretrained models for classification, named-entity recognition or sentiment analysis.

Documented uses

  • Tokenize Korean text into syllables, spaces and punctuation.
  • Load a Base or Small Korean ELECTRA model and produce text representations.
  • Fine-tune a discriminator checkpoint for a Korean NLP task.
Before you choose

Requirements & limits

The repository is public and licensed under Apache-2.0. Its README says the model checkpoints are hosted for Transformers loading.

  • The tokenizer is character-level, not wordpiece-level, and must be loaded from the repository's custom implementation.
  • The documented tokenizer setup requires Transformers 2.9.1 or later.
  • Both checkpoints have a stated maximum sequence length of 512 and vocabulary size of 11,568.
  • Hanja was removed during preprocessing and is not represented in the vocabulary.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Load a KoCharELECTRA checkpoint and tokenizer

Prepare a Base or Small checkpoint and its matching custom tokenizer through Hugging Face Transformers.

Before you start

  • Python environment with Transformers version 2.9.1 or later.
  • The repository’s tokenization_kocharelectra.py file.

Steps

  1. Import ElectraModel from transformers and KoCharElectraTokenizer from tokenization_kocharelectra.

  2. Load both components with the same checkpoint identifier: monologg/kocharelectra-base-discriminator or monologg/kocharelectra-small-discriminator.

Expected outcome

A model and its matching character-level tokenizer are loaded for further text processing or task-specific work.

Guide limitations

  • Use KoCharElectraTokenizer rather than the standard ElectraTokenizer.
  • The README says ElectraModel does not return pooled_output.
  • Hanja characters were excluded during preprocessing and are not in the vocabulary.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
monologg/KoCharELECTRA · Official homepage title · retained in its source language
How we review sourcesSuggest a correction