mahmoodlab/UNI

Access to the pretraining data

Opened this issue · 0 comments

Hi Authors,

Thanks for the great work. I would like to know whether the 100k WSI (100M patches) you used to train the UNI model has been released. I know you have provided the links to download the datasets from their sources in the "Data Availability" section, but the data collection and pre-processing are also difficult. If possible, can you provide the pre-processed training data?

Best,
Xiaohan