Giskard is an open-source testing and evaluation framework for AI models, including Large Language Models (LLMs), generative AI, and Retrieval-Augmented Generation (RAG) applications. It offers tools for identifying vulnerabilities, generating test cases, and evaluating model performance across various domains such as tabular data, natural language processing, and computer vision. Giskard provides both a Python library for data scientists to test AI models in their development environment and an enterprise collaborative hub for AI quality, security, and compliance management. The framework aims to help AI practitioners detect and prevent biases, performance issues, and security vulnerabilities in AI systems, while ensuring compliance with AI regulations like the EU AI Act.
