GaRAGe: A Benchmark with Grounding Annotations for RAG Evaluation

TL;DR

GaRAGe is a comprehensive benchmark with detailed grounding annotations designed to evaluate how well large language models can identify and utilize relevant information in long-form question answering across diverse topics and complexities.

Contribution

This paper introduces GaRAGe, a novel benchmark with human-curated annotations for fine-grained evaluation of LLM grounding capabilities in RAG tasks.

Findings

Models often over-summarize rather than ground answers on relevant passages.

Models have a low true positive rate in deflecting when no relevant info is available.

Performance drops significantly with time-sensitive questions and sparse sources.

Abstract

We present GaRAGe, a large RAG benchmark with human-curated long-form answers and annotations of each grounding passage, allowing a fine-grained evaluation of whether LLMs can identify relevant grounding when generating RAG answers. Our benchmark contains 2366 questions of diverse complexity, dynamism, and topics, and includes over 35K annotated passages retrieved from both private document sets and the Web, to reflect real-world RAG use cases. This makes it an ideal test bed to evaluate an LLM's ability to identify only the relevant information necessary to compose a response, or provide a deflective response when there is insufficient information. Evaluations of multiple state-of-the-art LLMs on GaRAGe show that the models tend to over-summarise rather than (a) ground their answers strictly on the annotated relevant passages (reaching at most a Relevance-Aware Factuality Score of 60%), or (b) deflect when no relevant grounding is available (reaching at most 31% true positive rate in deflections). The F1 in attribution to relevant sources is at most 58.9%, and we show that performance is particularly reduced when answering time-sensitive questions and when having to draw knowledge from sparser private grounding sources.