Evaluating Large Language Models with fmeval

TL;DR

fmeval is an open source library designed to evaluate large language models across various tasks and responsible AI dimensions, emphasizing simplicity, coverage, extensibility, and performance.

Contribution

This paper introduces fmeval, a comprehensive evaluation library for LLMs, detailing its design principles and implementation, and demonstrating its practical use case.

Findings

Effective evaluation of LLMs across multiple tasks

Facilitates responsible AI assessment

Supports model selection for specific applications

Abstract

fmeval is an open source library to evaluate large language models (LLMs) in a range of tasks. It helps practitioners evaluate their model for task performance and along multiple responsible AI dimensions. This paper presents the library and exposes its underlying design principles: simplicity, coverage, extensibility and performance. We then present how these were implemented in the scientific and engineering choices taken when developing fmeval. A case study demonstrates a typical use case for the library: picking a suitable model for a question answering task. We close by discussing limitations and further work in the development of the library. fmeval can be found at https://github.com/aws/fmeval.