May I ask how long it takes for you to iterate a pre training task and what GPU you will be using

[sansanyuchen]

May I ask how long it takes for you to iterate a pre training task and what GPU you will be using

[Luffy03]

Hi, many thanks for your attention to our work!
We use 8*H800 for pre-tranining. For the pre-training duration, it depends on the data and model size. Please refer to the details in our repo Large-Scale-Medical. For example, for model_base in SSL, it requires about one week for pre-training 160K CT (but you need to cache the data).