Model Overview
Ensemble5-Platypus2-13B-QLora-0.80-epoch is a 13 billion parameter language model developed by Weyaxi. This model is a strategic merge of two distinct base models: yeontaek/llama-2-13B-ensemble-v5 and Platypus2-13B-QLoRA-0.80-epoch. The merging process aims to combine the strengths of its constituent models, offering a versatile tool for various natural language processing tasks.
Key Capabilities & Performance
Evaluated on the Hugging Face Open LLM Leaderboard, Ensemble5-Platypus2-13B-QLora-0.80-epoch achieves an overall average score of 49.5. Specific benchmark results highlight its performance across different domains:
- ARC (25-shot): 59.73
- HellaSwag (10-shot): 82.66
- MMLU (5-shot): 56.94
- TruthfulQA (0-shot): 52.92
- Winogrande (5-shot): 74.43
- GSM8K (5-shot): 1.9
- DROP (3-shot): 17.93
These scores indicate a balanced capability in common sense reasoning, reading comprehension, and general knowledge, while showing areas for improvement in complex mathematical reasoning (GSM8K).
Intended Use Cases
This model is suitable for applications requiring a general-purpose language model with 13 billion parameters. Its performance profile suggests it can be effectively used for:
- Text generation and completion
- Question answering (especially common sense and factual)
- Summarization
- Content creation
Developers should consider its specific benchmark scores when aligning it with tasks that demand high accuracy in areas like HellaSwag or ARC.