Model Overview
Weyaxi/Nous-Hermes-Platypus2-13B-QLoRA-0.80-epoch is a 13 billion parameter language model created by merging two distinct models: NousResearch/Nous-Hermes-Llama2-13b and Platypus2-13B-QLoRA-0.80-epoch. This fusion aims to combine the strengths of its base models, resulting in a versatile instruction-tuned LLM.
Key Capabilities & Performance
This model has been evaluated on the Open LLM Leaderboard, showcasing its general-purpose capabilities across a range of tasks. Notable benchmark scores include:
- Avg. Score: 52.89
- ARC (25-shot): 59.9
- HellaSwag (10-shot): 83.29
- MMLU (5-shot): 56.69
- TruthfulQA (0-shot): 51.08
- Winogrande (5-shot): 75.22
While it performs well in common sense reasoning (HellaSwag, Winogrande) and general knowledge (MMLU), its performance on complex mathematical reasoning (GSM8K: 1.44) and reading comprehension (DROP: 42.65) indicates areas for potential improvement or specific use case considerations.
When to Use This Model
This model is a strong candidate for applications requiring a balanced instruction-tuned LLM with a 13 billion parameter count. Its performance profile suggests suitability for:
- General text generation and understanding
- Common sense reasoning tasks
- Instruction following in various domains
Developers should consider its specific benchmark scores when evaluating its fit for tasks demanding high accuracy in complex math or deep reading comprehension.