Hugging Face UTENA-7B-V3 Overview
UTENA-7B-V3 is a 7 billion parameter language model developed by AI-B, constructed through a merge of two distinct models: AI-B/UTENA-7B-UNA-V2 and AI-B/UTENA-7B-NSFW-V2. This merge was performed using the slerp method within mergekit, combining the strengths of its constituent models.
Key Capabilities & Performance
The model's performance has been evaluated on the Open LLM Leaderboard, achieving a competitive average score of 67.42. Specific benchmark results include:
- AI2 Reasoning Challenge (25-Shot): 65.96
- HellaSwag (10-Shot): 85.70
- MMLU (5-Shot): 64.72
- TruthfulQA (0-shot): 53.64
- Winogrande (5-shot): 80.27
- GSM8k (5-shot): 54.21
These scores indicate a balanced capability across reasoning, common sense, factual recall, and mathematical problem-solving. The model is also available in quantized GGUF format for optimized inference.
When to Use This Model
UTENA-7B-V3 is suitable for applications requiring a versatile 7B parameter model with solid performance across a range of general language understanding and generation tasks. Its balanced benchmark results suggest it can be a strong candidate for use cases where a single model needs to handle diverse cognitive demands.