Model Overview

The MaziyarPanahi/TheTop-5x7B-Instruct-S4-v0.1 is a 7 billion parameter instruction-tuned language model developed by MaziyarPanahi. This model was constructed using mergekit, a toolkit designed for merging pre-trained language models, specifically combining top 7B models and applying a SLERP (Spherical Linear Interpolation) merge to other 7B models. This approach allows for the creation of a robust model by leveraging the strengths of multiple base models.

Key Capabilities & Performance

The model demonstrates competitive performance across a range of benchmarks, as evaluated on the Open LLM Leaderboard:

• Average Score: 74.94
• AI2 Reasoning Challenge (25-Shot): 72.18
• HellaSwag (10-Shot): 88.29
• MMLU (5-Shot): 65.03
• TruthfulQA (0-shot): 65.56
• Winogrande (5-shot): 85.16
• GSM8k (5-shot): 73.39

These scores indicate strong general reasoning, common sense, and mathematical capabilities for a model of its size.

Good for

• General-purpose conversational AI: Its instruction-tuned nature and broad benchmark performance suggest suitability for various dialogue applications.
• Reasoning tasks: Scores on ARC and MMLU indicate good performance in logical and knowledge-based reasoning.
• Academic and research applications: The detailed evaluation results provide a solid basis for further research and fine-tuning.