We present LLM-Blender, an ensembling framework designed to attain consistently superior performance by leveraging the diverse strengths of multiple open-source large language models (LLMs). Our framework consists of two modules: PairRanker and GenFuser, addressing the observation that optimal LLMs for different examples can significantly vary. PairRanker employs a specialized pairwise comparison method to distinguish subtle differences between candidate outputs. It jointly encodes the input text and a pair of candidates, using cross-attention encoders to determine the superior one. Our results demonstrate that PairRanker exhibits the highest correlation with ChatGPT-based ranking. Then, GenFuser aims to merge the top-ranked candidates, generating an improved output by capitalizing on their strengths and mitigating their weaknesses. To facilitate large-scale evaluation, we introduce a benchmark dataset, MixInstruct, which is a mixture of multiple instruction datasets featuring oracle pairwise comparisons. Our LLM-Blender significantly outperform individual LLMs and baseline methods across various metrics, establishing a substantial performance gap.
LLM-Blender: Ensembling Large Language Models with Pairwise Ranking and Generative Fusion
An ensembling framework combines multiple large language models by ranking and merging outputs to achieve superior performance across various metrics.
- Year
- 2023
- Venue
- arXiv 2023
- Authors
- 3
- Hosting
- Abstract onlyARXIV-DEFAULT
Cite
Notes
Only stored in your browser.
Attribution
- Abstract & full text
- arxiv.org/abs/2306.02561v3ARXIV-DEFAULT
- TL;DR
- Semantic Scholar