Gunulhona/Gemma-3-4B

Overview

Gunulhona/Gemma-3-4B is a 4.3 billion parameter instruction-tuned language model, developed by merging multiple pre-trained Gemma-based models. This model was created using the Karcher Mean merge method, with huihui-ai/gemma-3-4b-it-abliterated serving as the base model.

Key Capabilities

This merged model integrates diverse functionalities from its constituent models, aiming for a broad range of applications. The merge includes models specialized in:

• Graph Database Interaction: neo4j/text-to-cypher-Gemma-3-4B-Instruct-2025.04.0 contributes to text-to-Cypher query generation.
• Medical Applications: google/medgemma-4b-it and google/medgemma-1.5-4b-it enhance its performance in medical contexts.
• Translation: google/translategemma-4b-it provides robust translation capabilities.
• Document Generation: ZySec-AI/gemma-3-4b-document-writer supports document writing tasks.

Good For

This model is suitable for use cases requiring a combination of specialized skills, particularly in areas like:

• Generating Cypher queries from natural language.
• Assisting with medical text processing or information retrieval.
• Performing multilingual translation tasks.
• Aiding in the creation of various document types.
• General instruction-following tasks where a compact, multi-purpose model is beneficial.