Sampling with crossdock dataset

When I sampling with the crossdock dataset (sample_diffusion_with_pocket.py), I encountered many problems (Mostly because certain values do not exist in the dataset). I suspect that the inconsistency in the data preprocessing process and training configuration is the cause. Is there a model trained of crossdock? Also, could you provide the crossdock data preprocessing code?