90% is the default setting. For language learning, 80% ~ 90% is OK.
2 Likes
Hi, I also use 1d in the learning steps. If I’m using the helper addon with auto reschedule card after review then I should be good?
The reason people were typically using interday learning cards in the past was to avoid the card’s ease being reduced when you forget during the initial acquisition stage. FSRS copes with this better, so they are not as useful. While it will cope if you use longer learning steps, I’d suggest you just let FSRS schedule the cards once they grow into the 1+ day range.
5 Likes
大佬。我又来打扰了。我想问一个问题了
我刚把fsrs的版本从v3.20.0升级到v4.5.6
卡片评价的再次复习间隔相较v3.20.0，至少翻了3倍。
如图
同一张卡的相邻2次“简单”评价，在v3.20.0和v4.5.6版本间，间隔日期差距好大。
这是正常现象吗？ 
很正常，v3 对简单的处理非常粗暴，没有利用复习数据。v4对简单进行了优化，会学习你点简单的那些复习的实际记忆情况。
1 Like
Does FSRS take card history even if you manually reset aka “forgot” the card?
1 Like
FSRS just ignores the revlogs earlier than your manual reset.
1 Like
叶大您好！看到GitHub上的Anki电脑端Beta版本似乎是支持了FSRS直接在Anki内的配置，想问问如果想要最近入坑FSRS的话应该等FSRS内置进Anki出正式版还是直接用fsrs4anki里的配置方法来配置Anki？感谢！ 
本人同时大多数时间使用Anki iOS端（？）感觉如果iOS正式版FSRS还需要一段时间才出来就用手动方法吧
直接用 fsrs4anki 里的配置方法更适合你的情况。Anki iOS 支持 custom scheduling 中的 FSRS。
1 Like
The beta version of AnkiMobile supports integrated FSRS. If you’re using the desktop beta, I recommend you apply for the AnkiMobile beta as well.
1 Like
Could you give me a link to attend the beta testing of AnkiMobile?
I’ve sent you a link. If other people wish to join beta testing, please message me on tenderapp: Beta Testing - AnkiMobile Manual
I just calculated my model weights using a backup. I got these:
0.5076, 75.4008, 100.0000, 100.0000, 3.5036, 2.5230, 0.5681, 0.0002, 2.2480, 0.1003, 1.3617, 2.0646, 0.1004, 0.3152, 1.9241, 0.7750, 1.9470
The 100 look odd. Where can I check the documentation for each number? thank you
Here is the document:
Thank you. I don’t find obvious what each number respresent. There are 17 numbers and I don’t see that many items explained. I still don’t understand which the third and four numbers represent.
I think having the list using the name will be much clearer. [So(G), Do(G), etc] or a list in bullet points enumarating the order and labeling the value.
It’s pretty hard to name them. The first four parameters are the initial stabilities for four ratings.
Sorry if this is already mentioned somewhere, I couldn’t find it. When switching to the builtin FSRS, do I just delete everything in Custom Scheduling?
Yes. It’s mentioned in
2 Likes
Hello, it’s me again. I have a question regarding the new cards. When I press the “good” button a few times, it calculates the learning steps correctly as per the settings. However, after I have gone through all the steps, Anki schedules the next review for the card quite far in the future (several weeks, a month). I find this quite concerning because I often forget what I had for breakfast in the evening, so for me, this is a relatively long interval.
Do I need to pay attention to tweaking any part of the algorithm (or Anki), or should I just trust that this is how the algorithm works?
