I only used it because they said the “easy days” might be activated with that option. I also prefer using the “FSRS helper,” but I didn’t understand the point of that setting. It seemed like “garbage” to me, so I restored the backup.
If your FSRS parameters change after optimization or you change Desired retention, the intervals of your cards will take this into account only at the next review or immediately if you turn on the function.
Reschedule Cards on Change
1 Like
The 24.10 Beta 3 has a new feature: “Let FSRS control short term schedule (experimental)”.
How do I turn this on? Do I have only to remove all learning and re-learning steps from the deck preset?
It is already turned on. Observe your learning steps after your current user-defined learning steps have been worked through for a given card. You can perhaps use only one (or 2, or more, or not at all as you like) learning step(s) and it will take the helms from then on.
I don’t get it. Will it only add additional learning steps to already defined?
So how does it work if I have empty learning steps?
No, it can do everything. If you wish to leave it empty, though it is not recommended if FSRS 4 beginning weights more than one day, it can still do scheduling.
In my case it works best, when you at least use one learning step there.
You can perhaps use only one (or 2, or more, or not at all as you like) learning step(s) and it will take the helms from then on.
It looks like it’s not working right now.
Huh… 
But it is for me. Do you mind sharing your learning steps
Yep, fixed in beta 3
Can confirm, I have the same issue
It’s a bit complicated
The intended behavior is that it only works if you leave the (re)learning steps field(s) empty. If they are not empty, your user-specified steps are used instead.
The current behavior is that you can get a <1d interval anyway, even if you have specified your own steps.
2 Likes
I don’t know if this is a bug or the default behavior of everyday learning cards (I don’t remember them behaving this way).
A backlog of Interday (learn, relearn) cards, say 20 cards seems to override the “Again” interval.
- The first card of those 20, is only shown after all those 20 cards have been cycled through and the cycle repeats.
- It ignores how short the again interval is (e.g. 1 second) and thus shows the same card after, say, 5 minutes after all cards have been cycled through.
- It does not advance through the queue depending on the “again” interval like it used to.
- This behavior does not occur with normal fresh new–>learn cards. They follow the interval correctly.
Here for more context:
Here is the card info for one of those interday cards I believe. (There was a 0-second interval shown, which is an odd occurrence. I believe this was around the time I tried to change the learning steps using the console. Anki seems to register 1s in the console as 0 in the reviewer, but that is a different problem)
Behavior is also seen elsewhere.
Here is another interday card. Here the 0-second interval is not present. (I think, I then raised the interval in the console by 1 second, making it 2s instead of 1s odd enough and then it returned to being 1s in the reviewer).
1 Like
It was caused by the update check connections to 162.55.238.74 being blocked.
1 Like
How is the quantity in the ‘reviews’ field measured? I’ve noticed that when I click ‘Optimize’ for the same deck, sometimes it shows 0 (zero), and other times it shows the total number of reviews completed. However, this seems to happen randomly. I’m not sure if returning 0 has any practical effect, but I find this alternation a bit strange. Could you explain why this happens?
Could you reproduce it in all presets? I guess the optimization is too fast in the preset with a small amount of reviews, so the number doesn’t update to the real amount before the optimization is complete.
It was supposed to be fixed in Fix FSRS progress update issues by abdnh · Pull Request #3420 · ankitects/anki · GitHub. Looks like the fix is not working properly.
It’s happening in the same preset. If I keep clicking ‘Optimize’ multiple times, sometimes it shows zero, and other times it shows the actual number of reviews. This behavior seems random
It’s a visual bug.
17 vs. 19 parameters mismatch on MacOS 24.10 vs. iOS 24.08:
- It seems like when I optimize FSRS params, now we have two extra params (assumingly for the intraday learning steps). So optimizing leads to 19 FSRS params as opposed to previously 17.
- Now, when I want to study a deck with 19 FSRS params on mobile (iOS), I get the error in the screenshot. I’ve got the latest iOS version (24.08).
- If I go to the settings and just delete the last two params (so I have 17 params again), everything works again on iOS.
- This is probably expected and fine since the iOS version is not expected to “know” the intraday learning steps yet. Just sharing it here so devs are aware of the current mismatch.
