A collection of datasets for AI safety and alignment research, including jailbreak robustness evaluation, alignment assessment, toxicity detection, ha