MedClip - Pretraining CLIP on medical data

This team is getting pretty big, but it’s also an ambitious project so let’s allow more than 10 here, actually. It’ll be very important that you guys organize well :slight_smile: We can provide you also with multiple TPU VMs I think :slight_smile:

4 Likes

In order to get access to the data, you need to create an account on https://physionet.org and apply for credentialed access
The credentialing process is a bit painful, you need to complete this course: CITI Program Course Instructions
I can’t remember how long it takes for them to approve the application, I did this about a year ago. I can confirm that I can download the dataset as a credentialed user

3 Likes

Is there a place on the team? I recently worked on pre-training a language model using a part of MIMIC-3 clinical notes (PhysioNet Data Repository)

1 Like

adding you @Vasudev

Actually giving you guys 2 TPU VMs directly since there is so much interest! I’ve split the team randomely into two, but shouldn’t really change anything :slight_smile: Feel free to continue organizing as before - you’ll just have acces to 2 TPU VMs tomorrow :slight_smile:

1 Like

Sorry folks, Due to a medical emergency at home, I’ll have to withdraw my name from the team. I’m sure the project will turn out to be amazing.

Thanks again, for your time.