Anki 24.10 Beta

Anki Beta Testing
66

A user has updated to Anki 24.10 beta and HyperTTS sarted to throw errors:
name '__file__' is not defined"

2024-09-30_hypertts_file
2024-09-30_hypertts_file1914×1025 135 KB

seems to be triggered when an external dependency tries to do: import pkg_resources

2 Likes
67

Happens with Chinese Support 3 as well:

import pkg_resources
File "pkg_resources", line 74, in <module>
NameError: name '__file__' is not defined
68

Fixed in polish graphs of simulator, true retention table and forgetting curve by L-M-Sherlock ¡ Pull Request #3448 ¡ ankitects/anki ¡ GitHub

1 Like
69

When I start Anki with add-ons and click “Rename” of a deck, nothing seems to happen. Only when I try to enter the deck or its Options, “Processing…” appears. It also appears even if I don’t try to rename a deck. When AnkiMorphs was enabled, something about unseen morphs has appeared after “Processing…” at least once. Only after the processing ends, do all the deck rename dialogs appear.

71

“Compute minimum recommended retention” cannot be canceled (in the middle of an iteration?).

72

[apparently the error about parameters happened with 24.06]

73

The limit for Maximum reviews/day in the FSRS simulator is only 1000, while the limit under daily limits is 9999. Also can we get a new beta maybe since so much has been fixed already

image
image969×456 45.2 KB

74

Has this issue been resolved. I have seen it yet to be added as an issue on GitHub

1 Like
75

When optimizing the deck using the new Anki with FSR, I noticed that even after optimization, the program keeps the old Anki parameters and states that the current parameters are optimal. However, if you reset the values to default before optimizing, Anki includes the new parameters correctly, and the optimization works as expected.

image

image

76

It’s still not really “as expected” - FSRS-5 is supposed to work with same-day reviews, but it doesn’t right now, and the last 2 parameters never change. LMSherlock already fixed it, so now we’re just waiting for the next beta.

77

Does that mean you’re better off optimizing with Google Colab and putting the numbers into your deck presets manually?

78

The issue isn’t necessarily with the optimization of the two new parameters, but rather that it seems when the deck is already optimized in the current version, it doesn’t migrate to the new version because it’s considered “already optimized.”

79

I can’t reproduce that. My presets that used FSRS-4.5 parameters can be optimized.

80

Are there any news about this :confused:

image
image914×133 10 KB

81

Right now <1d intervals are not supported. If you leave the (re)learning steps field empty, only intervals equal to or greater than 1 day will be given to you.

82

It seems Sherlock has taken a couple of days off. I was wondering if he would take this thing into consideration.

83

What’s happening is that FSRS is adding two zeroes to the old parameters and then calculating the RMSE to decide which parameters are better. In your case, the RMSE of the extended old parameters is better than the RMSE of the new parameters, which is the reason Anki is not updating the parameters.

84

Depends on how far has he gotten to it but I’d rather want this not get merged. It’ll delay the stable version.

85

I believe that the ball is in dae’s court right now.

But IMO, letting FSRS schedule <1d intervals even when the user has set the learning steps can lead to some confusion. Plus, we are still not convinced that FSRS can accurately calculate <1d intervals. So, this is unlikely to land up in the 24.10 release.

1 Like
86

It is compatible with manual learning steps, no?