MatthieuJ/Jason1903_SLERP
MatthieuJ/Jason1903_SLERP is a 7 billion parameter language model created by MatthieuJ, formed by merging yam-peleg/Experiment26-7B and chihoonlee10/T3Q-Mistral-Orca-Math-DPO using the SLERP method. This model is designed to combine the strengths of its base models, achieving an average score of 76.77 on the Open LLM Leaderboard. It demonstrates strong performance across various reasoning and language understanding tasks, including a 70.74 on GSM8k and 73.12 on ARC, making it suitable for general-purpose applications requiring robust reasoning capabilities.
Loading preview...
Model Overview
MatthieuJ/Jason1903_SLERP is a 7 billion parameter language model developed by MatthieuJ. It is a product of merging two distinct models, yam-peleg/Experiment26-7B and chihoonlee10/T3Q-Mistral-Orca-Math-DPO, utilizing the SLERP (Spherical Linear Interpolation) merge method via mergekit.
Key Capabilities
This merged model aims to leverage the strengths of its components, demonstrating solid performance across a range of benchmarks. According to the Open LLM Leaderboard Evaluation Results, Jason1903_SLERP achieves an average score of 76.77.
- Reasoning: Scores 73.12 on the AI2 Reasoning Challenge (25-Shot) and 70.74 on GSM8k (5-shot), indicating proficiency in complex problem-solving and mathematical reasoning.
- Language Understanding: Achieves 89.13 on HellaSwag (10-Shot) and 85.08 on Winogrande (5-shot), showcasing strong common sense and contextual understanding.
- Knowledge & Truthfulness: Records 64.43 on MMLU (5-Shot) and 78.13 on TruthfulQA (0-shot).
Good For
Jason1903_SLERP is well-suited for applications requiring a balanced performance across general language understanding, reasoning, and mathematical tasks. Its 7B parameter size makes it a viable option for scenarios where computational resources are a consideration, while still delivering competitive benchmark results.