Support for Online Learning? about fsrs-rs

[L-M-Sherlock]

There are two ways for auto-optimization: Online Learning and Periodic Retraining. I prefer the latter because it is easier to implement and more stable.

[L-M-Sherlock]

To implement it, we need to integrate the scheduler into Anki at first. By the way, online learning doesn't mean that it must have experience replay. The main feature of online learning is it could optimize its current weights by the new reviews on the fly.

[user1823]

To implement it, we need to integrate the scheduler into Anki at first.

Got it.

By the way, online learning doesn't mean that it must have experience replay.

I know that. My logic was that if you had implemented online learning, you would also have implemented experience replay to make the results better.