Model Overview

nlpguy/Lelantos-low-tune is a 7 billion parameter language model developed by nlpguy. It was created using the task arithmetic merge method via mergekit, combining openaccess-ai-collective/DPOpenHermes-7B-v2 as its base with SanjiWatsuki/Lelantos-7B.

Key Capabilities & Performance

This model exhibits robust performance across a range of benchmarks, as evaluated on the Open LLM Leaderboard. It achieved an average score of 70.82, with notable results including:

• AI2 Reasoning Challenge (25-Shot): 67.06
• HellaSwag (10-Shot): 86.06
• MMLU (5-Shot): 64.11
• TruthfulQA (0-shot): 61.33
• Winogrande (5-shot): 79.56
• GSM8k (5-shot): 66.79

These scores indicate strong general reasoning, common sense, and language understanding abilities, making it a versatile choice for various applications.

Merge Details

The model's architecture leverages a specific configuration where layers from the base model and SanjiWatsuki/Lelantos-7B were merged with a weight of 0.5 across all 32 layers, building upon openaccess-ai-collective/DPOpenHermes-7B-v2.