Tiger Gemma 12B v3 Overview

The TheDrummer/Tiger-Gemma-12B-v3 is a 12 billion parameter model built upon the Gemma 3 architecture, offering a substantial 32768 token context window. This iteration focuses on refining the model's output and behavior, moving towards a more neutral and less overtly positive tone, especially when engaging with challenging subjects. It is also highlighted for its potential vision capabilities.

Key Capabilities & Features

• Neutral Tone: Engineered to provide more balanced and less positive responses, particularly for sensitive or difficult topics.
• Enhanced Steerability: Offers improved control over output, allowing for better navigation of complex themes.
• Paragraph-Oriented Output: Prioritizes generating responses in paragraph format rather than relying heavily on markdown.
• Vision Capable: The model is noted to possess vision capabilities, suggesting multimodal applications.

Good For

• Applications requiring a neutral and objective tone in generated text.
• Use cases where fine-grained control and steerability over content, especially for nuanced subjects, is crucial.
• Scenarios benefiting from paragraph-style responses over structured markdown.
• Potential multimodal applications leveraging its vision capabilities.