ARES: An Automated Evaluation Framework for Retrieval-Augmented Generation Systems
By Jon Saad-Falcon, Omar Khattab, Christopher Potts, Matei A. Zaharia. North American Chapter of the Association for Computational Linguistics, 2023
325 citations recorded by the source corpus.
Evaluating retrieval-augmented generation (RAG) systems traditionally relies on hand annotations for input queries, passages to retrieve, and responses to generate. We introduce ARES, an Automated RAG Evaluation System, for evaluating RAG systems along the dimensions of context relevance, answer faithfulness, and answer relevance. By creating its own synthetic training data, ARES finetunes lightweight LM judges to assess the quality of individual RAG components. To mitigate potential prediction errors, ARES utilizes a small set of human-annotated datapoints for prediction-powered inference (PPI). Across eight different knowledge-intensive tasks in KILT, SuperGLUE, and AIS, ARES accurately evaluates RAG systems while using only a few hundred human annotations during evaluation. Furthermore, ARES judges remain effective across domain shifts, proving accurate even after changing the type of queries and/or documents used in the evaluated RAG systems. We make our code and datasets publicly available on Github.
View Paper: https://www.semanticscholar.org/paper/4df2b1e7d54fe5ad81dc2ed6774b93ef7891b3c8
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
- Hao Yu
- Aoran Gan
- Kai Zhang
- Shiwei Tong
- Qi Liu
- Zhaofeng Liu
- Yunfan Gao
- Yun Xiong
Citations in the corpus
- Evaluation of Retrieval-Augmented Generation: A Survey (2024, background): “The interplay between the retrieval and generation components means that the entire system’s performance cannot be fully understood by evaluating each component in isolation [41,13].”
- Retrieval-Augmented Generation for AI-Generated Content: A Survey (2024, unknown): “Another three benchmarks, RAGAS [362], ARES [363] and TruLens [364], consider three different aspects: Faithfulness, Answer Relevance, and Context Relevance, respectively.”
- Retrieval-Augmented Generation for Large Language Models: A Survey (2023, methodology): “Concurrently, state-of-the-art automated tools like RA-GAS [Es et al. , 2023], ARES [Saad-Falcon et al. , 2023], and TruLens 8 employ LLMs to adjudicate the quality scores.”
- BERGEN: A Benchmarking Library for Retrieval-Augmented Generation (2024, unknown): “There exist numerous works using LLMs as evaluators [35,70,94].”