Sonika Johri @sonikaj.bsky.social · Jan 8 Even if a quantum model trained today does not meet the baseline performance to beat a classical deep learning model, its parameters are still useful in jumpstarting the training of a future larger model! As an example, see the training below on MNIST:
1 likes 1 replies
?
Replies
Now, philosophically, in ML/AI, the parameters encapsulate the knowledge learned from the training data. They are the essence of the model's predictive power.