On: Dall-E 2

[j7ake]

The problem is that with DNA sequences you dont have good training sets.

With text and images you can leverage “ground truth” data (verified by humans) to train your model.

The DNA sequences I would look for methods that don’t require good ground truth data.