Walmart-the-bag/Misted-7B
Misted-7B is a 7 billion parameter language model developed by Walmart-the-bag, created by spherically merging (slerp) the OpenHermes-2-Mistral-7B and Mistral-7B-SlimOrca models. This model is designed for general-purpose instruction following, leveraging the strengths of its base models. It achieves an average score of 66.94 on the Open LLM Leaderboard, demonstrating capabilities across reasoning, common sense, and language understanding tasks.
Loading preview...
Misted-7B Model Summary
Misted-7B is a 7 billion parameter language model developed by Walmart-the-bag, built upon the Mistral architecture. This model is a product of a spherical linear interpolation (slerp) merge, combining the teknium/OpenHermes-2-Mistral-7B and Open-Orca/Mistral-7B-SlimOrca models with a merge parameter t of 0.32. This merging strategy aims to synthesize the beneficial characteristics of both base models into a single, cohesive unit.
Key Capabilities & Performance
Misted-7B demonstrates solid performance across various benchmarks, as evaluated on the Open LLM Leaderboard. Its average score of 66.94 indicates proficiency in diverse tasks. Specific benchmark results include:
- AI2 Reasoning Challenge (25-Shot): 63.65
- HellaSwag (10-Shot): 84.14
- MMLU (5-Shot): 63.94
- TruthfulQA (0-shot): 52.00
- Winogrande (5-shot): 78.30
- GSM8k (5-shot): 59.59
These scores highlight its ability in reasoning, common sense, factual recall, and mathematical problem-solving. The model uses an Alpaca-style prompt format for instruction following.
Good For
- General-purpose instruction following tasks.
- Applications requiring a balance of reasoning and common sense.
- Scenarios where a 7B parameter model offers a good trade-off between performance and computational resources.