Based on what was said in the previous thread, I guess that affects difficulty via “Historical retention” and any weirdness my graphs have may come from that?
Historical retention only affects things if your cards have incomplete review histories somehow, or if you use “Ignore cards reviewed before”.
Can FSRS do same-day reviews for Relearn?
Theoretically yes, but I have never seen it happen in my decks.
I’m getting this error on AnkiDroid 2.19 beta 6. I thought it shouldn’t happen anymore?
That happened to me too. I worked around it with these steps:
I probably remember that Evaluate works on the search query and not the preset when I am using it, but at other times I may forget that.
One of the motivations for the idea of [Idea] A tool to test parameters/presets? was that when I use a text file to keep track of presets and parameters, I use the triple-click mouse button to select the whole output of Evaluate: “Log loss: 0.4017, RMSE(bins): 3.55%. Smaller numbers indicate a better fit to your review history.”, and the explanation adds unnecessary noise.
Maybe it should be separated into its own paragraph?
Did you ever optimize the parameters in any of the earlier betas and didn’t undo that? If yes, you don’t have the FSRS 4.5 parameters saved in your collection. So, this error is expected.
Thank you, that helped.
I checked the graphs for the presets, and they all have the general shape of the harder movies2anki query (a lot of 0-5%, few 5-20%), but the easier ones have fewer difficult cards, and more of them are suspended there.
I got a similar error and crash when clicking “Compute minimum recommended retention” on this same deck. Not sure the fix you already made will fix this too.
Anki 24.10 (8d9251e7) (ao)
Python 3.9.18 Qt 6.6.2 PyQt 6.6.1
Platform: Windows-10-10.0.19045
Traceback (most recent call last):
File "aqt.taskman", line 144, in _on_closures_pending
File "aqt.taskman", line 88, in <lambda>
File "aqt.taskman", line 108, in wrapped_done
File "aqt.operations", line 259, in wrapped_done
File "concurrent.futures._base", line 439, in result
File "concurrent.futures._base", line 391, in __get_result
File "concurrent.futures.thread", line 58, in run
File "aqt.operations", line 243, in wrapped_op
File "aqt.main", line 1530, in backup
File "anki.collection", line 341, in create_backup
File "anki._backend_generated", line 172, in create_backup
File "anki._backend", line 161, in _run_command
pyo3_runtime.PanicException: called `Result::unwrap()` on an `Err` value: PoisonError { .. }
It will be fixed after merging: Feat/align FSRS-rs with PyTorch Implementation by L-M-Sherlock · Pull Request #3540 · ankitects/anki · GitHub
Thanks, logged on Right click hangs UI due to empty context menu · Issue #3550 · ankitects/anki · GitHub.
As this is an experimental feature and the wording/design may change, it makes sense to hold off on making it translatable immediately to avoid potentially wasting translator’s time.
Logged on Disallow setting all days to easy · Issue #3551 · ankitects/anki · GitHub
I still can not use AnkiDroid for my reviews as I get the 17 parameters error. I tried to optimize my parameters but they are already optimized. I run AnkiDroid 2.19.1 and Anki Desktop directly from the main branch.
On AnkiDroid, delete the parameters and press Optimize. Then sync and press Optimize on Anki Desktop.
Hi all,
rc2 is now available, and is mainly focused on an FSRS update. There are still some tasks remaining for the milestone, so there’ll be at least one more rc, and hopefully within a few days. Some of the tasks might end up bumped to the following release.
If you saved 19 params with an older Anki version, you’ll either need to reoptimize in AnkiDroid, or reset the params to default.
Thanks !! Solved !
I found a bug: sometimes the “Your desired retention is below optimal. Increasing it is recommended.” message doesn’t appear if desired retention is 0.01 less than the recommended value. However, it’s not consistent. I can reproduce it in some presets, but not in others.
Never mind, the preset was told to ignore cards before 2014.
After optimizing the FSRS parameters, it seems that they are not rounded to 4 decimal places in calculations, but are only displayed as rounded.
And if you change the value of the parameter in a text box and then restore it back, it will recalculate slightly different values on the card’s S and D. (This is a small difference of course.)
The old (legacy) version saved the last used settings, making importation easier. This new version resets everything, and I have to select everything again. I emphasize this request because by removing the legacy version (RC2), it has harmed those who are now forced to use the new version, and we have lost the conveniences of the previous version.
For example, the field separator is always set to ‘pipe’ by default, and I always use ‘comma.’ I am forced to change it every time. The previous version used to save my preferences.
I think I’ve seen this happening even before the beta/rc releases.
