Zenz V2.5 Xsmall

🚀 zenz-v2.5-small

A conditional language model based on the GPT - 2 architecture specialized for Japanese kana - kanji conversion tasks.

🚀 Quick Start

zenz-v2.5 is a conditional language model with a GPT - 2 architecture, specifically designed for kana - kanji conversion tasks. It is intended for use in the neural kana - kanji conversion system "Zenzai".

• Character - level and byte - level BPE tokenizer
• High performance in kana - kanji conversion tasks
• High performance in context - aware conversion

zenz-v2.5 is available in three model sizes:

• zenz-v2.5-medium: A large - scale model with 310M parameters.
• zenz-v2.5-small: A medium - scale model with 91M parameters.
• zenz-v2.5-xsmall: A small - scale model with 26M parameters.

📚 Documentation

Model Description

It uses the tokenizer of ku-nlp/gpt2-small-japanese-char provided under CC - BY - SA 4.0.

This model is provided under CC - BY - SA 4.0.

• Developed by: Keita Miwa (𝕏)
• Model Type: GPT - 2
• Language(s) (NLP): Japanese
• License: CC - BY - SA 4.0

Model Sources

This model is designed to be used in conjunction with Zenzai (AzooKeyKanaKanjiConverter).

• Repository: https://github.com/ensan-hcl/AzooKeyKanaKanjiConverter

Data Sources

This model was built using the zenz-v2.5-dataset.

📄 License

This model is provided under the CC - BY - SA 4.0 license.

🔧 Technical Details

Acknowledgements

During the construction of this model, we received computational resource support from Sakura Internet Inc. We are very grateful for their assistance.

We also utilized the following libraries, tools, and language resources to build this model:

• MeCab (https://taku910.github.io/mecab/)
• ipadic - NEologd (https://github.com/neologd/mecab-ipadic-neologd)
• torch (https://pypi.org/project/torch/)
• transformers (https://pypi.org/project/transformers/)
• datasets (https://pypi.org/project/datasets/)
• jaconv (https://pypi.org/project/jaconv/)
• llama.cpp (https://github.com/ggerganov/llama.cpp)
• llm.c (https://github.com/karpathy/llm.c)