You can download the dataset from https://github.com/tianyi-lab/HallusionBench and use their official implementation to evaluate our model.
