Introduce recent changes of FSRS4Anki, and want to collect some feedback

Anki Scheduling
1

Continuing the discussion from New features of FSRS4Anki from v3.0.0 to v3.6.0:

During the past one month, FSRS4Anki only added three features. It is stable to me. So I want to collect some feedback for it. And here are the changes between v3.6.0 and v3.9.2:

Scheduler

  • Fix: set the good interval when the learning step only contains one step.

Optimizer

  • Feat: calculate the stability and difficulty for each review history and save it in prediction.tsv.
  • Fix: use the true repetition expectation from revlog to optimize suggested retention.
  • Feat: evaluate the model with cross-entropy loss and compare the model before/after training.

Simulator

  • Feat: count retention expectation of all cards.

Helper

  • Fix: reschedule each card only once.

FSRS4Anki v3.9.2 has been released at:

By the way, I am waiting for the V3 scheduler at Ankimobile and Ankidroid to support the new custom scheduling feature, which allows FSRS4Anki to work on phones.

2

do you have any news that those apps will have the custom scheduler option. Or is it something you wish they had.

3

They have already supported V3 Scheduler. It is reasonable to support the new feature of V3, right?

4

at least for ankidroid a feature request was already sent: [Feature request] Add support for the custom study feature recently implemented into Anki (2.1.55) · Issue #12620 · ankidroid/Anki-Android · GitHub

So they are at least aware that there’s some demand for it.

5

And it’s already in the current AnkiMobile beta.

6

@dae I have a question: If the scheduler save state variables in cards’ custom data field than sync them to a phone device and do some reviews there, will the custom data be lost? Or just keep it the same?

7

yes but from the replies and the commit history. It dosen’t look like they are working on it

8

@L.M.Sherlock I believe older v3 versions & v2 should leave any custom study data alone and not change it, but I have not tested to make sure.