This is really cool. Any plans to release the dataset?

We include the dataset pipeline in the codebase so far, might release dataset.