3 ms·

I dont agree with this, either we know what the pretext task was, or we fine tune with labels, but in either case, it's no different to analyze than supervised

I dont agree with this, either we know what the pretext task was, or we fine tune with labels, but in either case, it's no different to analyze than supervised models. I do agree that more parameters and a bigger training set makes the models more opaque (or at least more expensive to understand), but the self supervision is not the reason.

visarga 6y ago
Or instead of analyzing the model we could analyze the training data for biases. Most bias comes from data.
- swayson 6y ago
  Its bias all the way down... Code, Data, Model, Use-Case, ...