Evaluating LLM Input Comprehension and Guardrail Robustness through Noise-Based Attacks Data

Data for the research project, including datasets used, the spikee noise plugin, and result files from running spikee against the target LLMs.