How does the F1 score of LLM-as-a-judge evaluation compare to exact match for multi-hop HotPotQA when using it

Abstract

Abstract: Extractive reading comprehension question answering (QA) datasets are typically evaluated using Exact Match (EM) and F1-score, but these metrics often fail to fully capture model performance. With the success of large language models (LLMs), they have been employed in various tasks, including serving as judges (LLM-as-a-judge). In this paper, we reassess the performance of QA models using LLM-as-a-judge across four reading comprehension QA datasets. We examine different families of LLMs and various answer types to evaluate the effectiveness of LLM-as-a-judge in these tasks. Our results show th

Research Question

How does the F1 score of LLM-as-a-judge evaluation compare to exact match for multi-hop HotPotQA when using iterative retrieval with reranking versus 128K-token context windows, across varying numbers of adversarial distractors?

Verification Level

Descriptive public verification status only; aggregate claim counts are public, but individual claim records are not exposed here.

Quality Tier

Descriptive public triage only; this tier does not alter current publication or DOI behavior.

Quality Dimensions

Automated triage signals derived from public fields; not human peer review or independent validation.

Correction Record

Public corrections are additive records. Current status does not claim the synthesis is error-free.

Provenance