WalledEval: A Comprehensive Safety Evaluation Toolkit for Large Language Models

TL;DR

WalledEval is a comprehensive toolkit for evaluating the safety of large language models across multiple benchmarks, including safety, robustness, and cultural appropriateness, with new tools and datasets for thorough assessment.

Contribution

It introduces WalledEval, a versatile safety evaluation framework supporting diverse models, benchmarks, and custom mutations, along with new safety datasets and a content moderation tool.

Findings

Supports over 35 safety benchmarks across multiple safety domains.

Includes custom mutators for robustness testing against text mutations.

Provides new datasets for cultural safety and exaggerated safety assessment.

Abstract

WalledEval is a comprehensive AI safety testing toolkit designed to evaluate large language models (LLMs). It accommodates a diverse range of models, including both open-weight and API-based ones, and features over 35 safety benchmarks covering areas such as multilingual safety, exaggerated safety, and prompt injections. The framework supports both LLM and judge benchmarking and incorporates custom mutators to test safety against various text-style mutations, such as future tense and paraphrasing. Additionally, WalledEval introduces WalledGuard, a new, small, and performant content moderation tool, and two datasets: SGXSTest and HIXSTest, which serve as benchmarks for assessing the exaggerated safety of LLMs and judges in cultural contexts. We make WalledEval publicly available at https://github.com/walledai/walledeval.