core-3/kuno-royale-7B
core-3/kuno-royale-7B is a 7 billion parameter language model created by core-3, formed by merging SanjiWatsuki/Kunoichi-DPO-v2-7B and eren23/ogno-monarch-jaskier-merge-7b using LazyMergekit. This model demonstrates competitive performance across various benchmarks, including MMLU, HellaSwag, and GSM8K, making it suitable for general-purpose text generation and reasoning tasks. With a 4096-token context length, it offers a balanced option for developers seeking a capable 7B model.
Loading preview...
Model Overview
kuno-royale-7B is a 7 billion parameter language model developed by core-3. It is a merged model, combining the strengths of two base models: SanjiWatsuki/Kunoichi-DPO-v2-7B and eren23/ogno-monarch-jaskier-merge-7b. The merge was performed using LazyMergekit, employing an slerp merge method with specific parameter weighting for self-attention and MLP layers.
Performance Highlights
kuno-royale-7B exhibits strong benchmark performance, achieving an average score of 74.74 across a suite of evaluations. Key scores include:
- MMLU: 65.13
- HellaSwag: 88.20
- GSM8K: 69.90
These scores position it competitively against its constituent models and other 7B-class models like SanjiWatsuki/Kunoichi-DPO-v2-7B, often slightly outperforming them in several categories. However, the developers note that a newer version, core-3/kuno-royale-v2-7b, is likely to offer improved performance.
Intended Use Cases
This model is suitable for a range of general-purpose natural language processing tasks, including:
- Text generation
- Question answering
- Reasoning tasks
- Conversational AI
Its balanced performance across various benchmarks suggests its utility in applications requiring robust language understanding and generation capabilities.