shadowml/DareBeagle-7B
DareBeagle-7B is a 7 billion parameter language model created by shadowml, formed by merging mlabonne/NeuralBeagle14-7B and mlabonne/NeuralDaredevil-7B using a slerp merge method. This model demonstrates strong general reasoning capabilities, achieving an average score of 74.58 on the Open LLM Leaderboard across various benchmarks. It is suitable for tasks requiring robust understanding and generation, with a context length of 4096 tokens.
Loading preview...
DareBeagle-7B Overview
DareBeagle-7B is a 7 billion parameter language model developed by shadowml. It is a product of a strategic merge between two base models: mlabonne/NeuralBeagle14-7B and mlabonne/NeuralDaredevil-7B. This merge was executed using the slerp (spherical linear interpolation) method, specifically configured to blend the self-attention and MLP layers of the constituent models.
Key Capabilities & Performance
DareBeagle-7B exhibits strong performance across a range of benchmarks, as evaluated on the Open LLM Leaderboard. Its average score is 74.58, indicating solid general-purpose reasoning and language understanding. Specific benchmark results include:
- AI2 Reasoning Challenge (25-Shot): 71.67
- HellaSwag (10-Shot): 88.01
- MMLU (5-Shot): 65.03
- TruthfulQA (0-shot): 68.98
- Winogrande (5-shot): 82.32
- GSM8k (5-shot): 71.49
Good For
- General-purpose text generation: Its balanced performance across various benchmarks makes it suitable for a wide array of language tasks.
- Reasoning tasks: Demonstrated proficiency in AI2 Reasoning Challenge and GSM8k suggests capabilities in logical inference and problem-solving.
- Applications requiring robust understanding: High scores on HellaSwag and Winogrande indicate strong common sense and contextual understanding.