ethicalabs/Kurtis-E1.1-Qwen3-4B Overview
ethicalabs/Kurtis-E1.1-Qwen3-4B is a 4 billion parameter language model based on the Qwen3 architecture, fine-tuned by ethicalabs using the Flower federated learning framework. This model is designed for general-purpose language understanding and generation, with a notable performance across a wide range of academic and professional subjects.
Key Capabilities & Performance
The model's performance was evaluated using the LM Evaluation Harness on a Mac Mini M4 Pro, demonstrating a solid understanding across various domains. Key evaluation results on the MMLU benchmark include:
- Overall MMLU Score: 0.6849
- Social Sciences: Achieved a strong 0.7813, with high scores in areas like high school government (0.8756) and psychology (0.8679).
- Humanities: Scored 0.5951, showing proficiency in subjects such as high school European history (0.7879) and international law (0.7686).
- STEM: Recorded a 0.6943, with notable results in high school biology (0.8742) and computer security (0.7800).
Good For
- General Knowledge Applications: Its balanced performance across MMLU categories suggests suitability for tasks requiring broad factual recall and understanding.
- Educational Content Generation: Excels in social sciences and humanities, making it useful for generating content related to these fields.
- Reasoning Tasks: The MMLU scores indicate a capability for reasoning across diverse subjects.