[allovernow]

I might take you up on that. How does your dataset facilitate self supervised experimentation?

I've a good amount of experience playing with autoencoders but this is the first I've heard of contrastive learning.