Question Regarding the Small File Size of Your .h5 Dataset

[fjwaaaa]

I hope this message finds you well. I recently noticed that the .h5 files you provided have a remarkably small footprint—only about 814 MB in total—and I am very impressed. I’m curious about how you achieved such efficient storage.

In my current project, I need to normalize the pixel values of each image in the .h5 files to the range [0, 1]. However, after performing the normalization and re-saving the files, their total size ballooned to approximately 5.6 GB. Upon researching, I discovered that specifying a compression parameter (e.g., compression="gzip") during saving can help reduce the file size.

Could you please let me know if you applied a similar compression strategy when creating your .h5 files? If not, I would greatly appreciate it if you could share the method you used to keep the file sizes so compact.

Thank you very much for your time and assistance.