Model Overview

luqmanxyz/Maya_Hermes-2.5-Mistral-7B is a 7 billion parameter language model derived from the OpenHermes-2.5-Mistral-7B architecture. It has been fine-tuned using Direct Preference Optimization (DPO) on the argilla/distilabel-intel-orca-dpo-pairs dataset, aiming to enhance its instruction-following and response quality.

Key Capabilities & Performance

This model demonstrates competitive performance on the Open LLM Leaderboard, with an average score of 68.60. Specific benchmark results include:

• AI2 Reasoning Challenge (25-Shot): 66.30
• HellaSwag (10-Shot): 85.07
• MMLU (5-Shot): 63.23
• TruthfulQA (0-shot): 55.89
• Winogrande (5-shot): 78.85
• GSM8k (5-shot): 62.24

These scores indicate its proficiency in reasoning, common sense, language understanding, and mathematical problem-solving.

When to Use This Model

This model is a strong candidate for general-purpose applications requiring a capable 7B parameter model. Its DPO fine-tuning suggests improved alignment and response quality, making it suitable for tasks such as:

• Instruction following
• Question answering
• Text generation
• Reasoning-based tasks

Its balanced performance across various benchmarks makes it a versatile choice for developers seeking a robust and efficient language model.