Hugging Face AXCXEPT/EZO-Common-T2-2B-gemma-2-it is a 2.6 billion parameter language model based on the Gemma-2-2B-it architecture, developed by AXCXEPT. This model is enhanced with multiple tuning techniques to improve general performance, excelling particularly in Japanese language tasks while designed for global applicability. It leverages high-quality Japanese Wikipedia and FineWeb data for instruction tuning, making it suitable for diverse use cases worldwide.
Model Overview
AXCXEPT/EZO-Common-T2-2B-gemma-2-it is a 2.6 billion parameter language model built upon the Gemma-2-2B-it base. Developed by AXCXEPT, this model incorporates multiple tuning techniques to achieve enhanced general performance, with a notable strength in Japanese language tasks. Despite its focus on Japanese data, it is engineered to address diverse global needs.
Key Features and Training
- Base Model: Utilizes the Gemma-2-2B-it architecture.
- Enhanced Performance: Improved through various tuning methods for general applicability.
- Japanese Language Proficiency: Demonstrates strong performance in Japanese tasks.
- Global Design: Engineered to meet a wide range of international use cases.
- Training Data: Instruction data was created by extracting high-quality content from Japanese Wikipedia and FineWeb datasets.
- Training Method: Employs a plain instruction tuning method, learning from exemplary responses to improve understanding and generation across languages and contexts.
Usage and Licensing
This model is provided for research and development purposes only and is considered an experimental prototype. It is not intended for commercial use or deployment in mission-critical environments. Users are responsible for understanding the associated risks. The model is subject to the Gemma Terms of Use, as it is based on Gemma-2-2B-it.