Optimus-7B: A Math-Optimized Language Model

Optimus-7B is a 7 billion parameter language model developed by Q-bert, built upon the Mistral-7B-v0.1 architecture. Its primary differentiation lies in its fine-tuning on the MetaMathQA dataset, which specifically enhances its capabilities in mathematical reasoning and problem-solving.

Key Capabilities & Performance

This model has been evaluated on the Open LLM Leaderboard and achieves a competitive average score of 69.09. Notable benchmark results include:

• ARC (25-shot): 65.44
• HellaSwag (10-shot): 85.41
• MMLU (5-shot): 63.61
• TruthfulQA (0-shot): 55.79
• Winogrande (5-shot): 78.77
• GSM8K (5-shot): 65.50 (specifically for mathematical word problems)

Use Cases

Optimus-7B is particularly well-suited for applications requiring strong quantitative and logical reasoning. Developers can leverage this model for:

• Mathematical problem-solving: Generating solutions or explanations for math-related queries.
• Data analysis and interpretation: Assisting with tasks that involve numerical understanding.
• Educational tools: Creating interactive math tutors or content generation for STEM fields.

The model supports the ChatML format for interaction.