TSNBench Dataset

TSNBench is a benchmark for evaluating the ability of large language models (LLMs) to understand and reason about Time-Sensitive Networking (TSN). The dataset contains two complementary datasets. The first dataset is a multiple-choice question-answering (MCQA) covering TSN concepts. The second component consists of open-ended TSN reasoning tasks that require models to compute worst-case delay (WCD) values for TSN flows under Credit-Based Shaper (CBS) and Cyclic Queuing and Forwarding (CQF) mechanisms. These open-ended tasks include network topology, flow information, routing paths, and ground-truth WCD values.