JMingo/gemma-4-E4B-it-Japanese

VISIONConcurrent Unit Cost:1Model Size:7.9BQuant:FP8Context Size:32kTool Calling:SupportedPublished:May 16, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

JMingo/gemma-4-E4B-it-Japanese is a 7.9 billion parameter instruction-tuned language model, based on Google's Gemma-4-E4B-it architecture, specifically optimized for Japanese language processing. This model achieves a smaller size by removing multimodal features and focuses exclusively on Japanese by eliminating non-Japanese tokens from its vocabulary. It is designed for applications requiring efficient and focused Japanese text generation and understanding.

Loading preview...

Overview

JMingo/gemma-4-E4B-it-Japanese is a specialized version of Google's Gemma-4-E4B-it model, fine-tuned for the Japanese language. This 7.9 billion parameter model has been engineered to process and generate Japanese text efficiently by making two key modifications:

Key Differentiators

  • Japanese-Centric Vocabulary: Non-Japanese tokens have been deliberately removed from the model's vocabulary. This forces the model to "think" exclusively in Japanese, potentially leading to more coherent and contextually appropriate Japanese outputs.
  • Optimized Size: Multimodal capabilities, such as audio and vision features, have been stripped away. This reduction in complexity results in a smaller model size, which can be beneficial for deployment in resource-constrained environments or for faster inference.

Use Cases

This model is particularly well-suited for applications where:

  • High-quality Japanese text generation is paramount.
  • Efficient processing of Japanese language is required.
  • A smaller model footprint is advantageous, without sacrificing Japanese language proficiency.