Reproduce “Cost-Effective In-Context Learning for Entity Resolution: A Design Space Exploration” 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

Meihao Fan, Xiaoyue Han, Ju Fan, Chengliang Chai, Nan Tang, Guoliang Li, Xiaoyong Du

BATCHER reduces repeated prompting in entity resolution by grouping match questions and selecting labeled examples to cover multiple questions. Its selection first limits annotation examples, then limits demonstration tokens; it does not guarantee better answers. On Walmart-Amazon, GPT-3.5-turbo-0301 batch prompting raised F1 from 67.54 to 78.92 and lowered estimated API cost from $1.43 to $0.33. Beer was an accuracy exception.

Summary by alphaXiv.

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