Reproduce “Enabling Flexible Multi-LLM Integration for Scalable Knowledge Aggregation” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Zhenglun Kong, Zheng Zhan, Shiyue Hou, Yifan Gong, Xin Meng, Pengwei Sui, Peiyan Dong, Xuan Shen and 5 more

Fusion-X transfers selected specialist knowledge into one target language model by scoring each source for an input, thresholding and weighting the selected probability distributions, and training the target to match them. On Llama-2-7B Big-Bench Hard, it raises average exact-match accuracy from 39.59 to 41.70 and reduces below-target tasks from 10 to five, without eliminating regressions.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.