Model Overview

PuddleJumper-Platypus2-13B-QLoRA-0.80-epoch is a 13 billion parameter language model created by merging two distinct models: totally-not-an-llm/PuddleJumper-13b and Weyaxi/Platypus2-13B-QLoRA-0.80-epoch. This merging approach aims to combine the strengths of both base models.

Performance Benchmarks

The model's performance has been evaluated on the Hugging Face Open LLM Leaderboard, achieving an average score of 47.85. Key benchmark results include:

• ARC (25-shot): 54.52
• HellaSwag (10-shot): 79.36
• MMLU (5-shot): 55.15
• TruthfulQA (0-shot): 54.32
• Winogrande (5-shot): 71.11
• DROP (3-shot): 20.49

Notably, the GSM8K (5-shot) score is 0.0, indicating limitations in complex mathematical reasoning tasks. The model generally performs well on common sense reasoning (HellaSwag, Winogrande) and general knowledge (MMLU, ARC).

Use Cases

This model is suitable for applications requiring:

• General text generation and understanding.
• Tasks benefiting from a broad range of capabilities derived from its merged components.
• Scenarios where a 13B parameter model offers a balance between performance and computational resources.