If you find this tutorial helpful, I'd be grateful if you could give it a star.
中文版请见：FSRS4Anki 使用指北
- Step 1: Enabling the FSRS Scheduler
- Step 2: Personalizing FSRS
- Configuring Different Parameters for Different Decks
- FAQ
To get started with FSRS, you'll need to follow a two-step process.
- First, you'll need to enable the FSRS scheduler in your Anki application.
- Next, you'll need to personalize FSRS to suit your learning patterns.
Let's now discuss both of these steps in detail.
Go to Tools > Preferences > Review > Enable V3 Scheduler.
- Go to the following page and copy all of the code. /~https://github.com/open-spaced-repetition/fsrs4anki/blob/main/fsrs4anki_scheduler.js
- In Anki, open the deck options of any deck (it doesn’t matter which deck). Find the Advanced Settings column, and paste the code you copied into the Custom Scheduling field:
- Ensure that the learning and re-learning steps are shorter than 1 day in any deck you want to use with FSRS. Other settings, such as “Graduating interval” and “Easy interval”, don’t matter. For more details about which Anki settings matter and which are obsolete, see the FAQs.
After you perform the above steps, the FSRS4Anki Scheduler should ideally be active. If you want to confirm this, you can change this part of the code:
const display_memory_state = false;to:
const display_memory_state = true;Then open any deck for review and you'll see the following message:
This shows that the FSRS scheduler is running normally. If you don’t see D, S and R, and only see “FSRS enabled”, it means that the card is in the “learning” or “relearning” stage, not in the “review” stage.
You can then change the code back and the message will no longer display.
Personalizing FSRS for your learning needs involves a two-step process.
- First, you'll need to train the FSRS parameters for your collection using the FSRS optimizer, tailoring the algorithm to your learning patterns.
- Next, you'll need to choose the desired retention rate and maximum interval.
Let's now discuss both of these steps in detail.
For most users, it is advisable to use one of the following two methods (Google Colab and Hugging Face) for training the parameters. Advanced users can explore other options mentioned here.
Note that the FSRS optimizer requires a minimum of 2,000 reviews to produce accurate results. If you don't have enough data, you can skip this step and use the default parameters instead, which are already entered into the scheduler code.
Method 1: Training using Google Colab
Open the optimizer's notebook. You don't need to configure the coding environment yourself and you can use Google's machines for free (you'll need to have a Google account):
After the Colab website opens, switch to the folder tab. Once the Optimizer connects to Google's machines, you can right-click to upload your deck file/collection file exported from Anki.
When exporting these files, make sure to select "Include scheduling information" and "Support older Anki versions". You don't need to include media.
A note on Privacy
The decks that you upload to the optimizer can't be accessed by the author of FSRS. This can be verified by anyone who understands code because the code of the optimizer is open-source.
Google may have access to the uploaded data. But, the risk is similar to uploading the data to your personal Google Drive folder.
If you are too worried about privacy, you still have two options.
- Advanced users can run the script locally using the options mentioned here.
- Other users can export their collection with blanked-out fields. To do this, go through the following steps:
- Take a backup by going to
File → Create Backupjust in case anything goes wrong. - Go to
Browse > Notes > Find and Replace. - Type
(.|\n)*in the "Find" field and keep the "Replace With" field empty. - Check (✓) the "Treat input as regular expression" option. Uncheck "Selected notes only" if you want to apply this to all notes.
- Export your collection using the steps mentioned above.
- Restore the contents of your notes by going to
Edit → Undo Find and Replace.
- Take a backup by going to
- After uploading the file, replace the
collection-2022-09-18@13-21-58.colpkgwith the name of your uploaded file. - Replace
Asia/Shanghaiwith your timezone. The notebook contains a link to the list of time zones. - Also, replace the value of
next_day_starts_at. To find this value, Go toTools > Preferences > Review > Next day starts atin your Anki.
Then, run the optimizer by either pressing Ctrl+F9 or going to Runtime > Run all.
Wait for the code to finish running. Then, go to section 2.2 (Result), where the optimized parameters will be available. Copy these parameters.
Method 2: Training using Hugging Face
Simply upload your exported decks to this website and it will optimise it for you.
https://huggingface.co/spaces/open-spaced-repetition/fsrs4anki_app
After training the parameters by either of the methods above, replace the parameters in the FSRS code that you copied earlier.
Even after you start using FSRS, you should re-train the parameters once in every two months. However, it depends on how old your collection is. Users with relatively newer collections might want to re-optimize monthly. Re-optimization will ensure that FSRS works well with your current patterns of learning.
Now, you need to choose your requestRetention, which denotes the retention rate (i.e. the fraction of the cards recalled successfully) that FSRS will try to achieve.
As an aid in deciding this value, you can view your past retention rate in Anki stats. For example, if your retention rate in the past was 90%, you can set 0.90 as your requestRetention.
You can set a higher requestRetention but keep in mind that as you increase the requestRetention above 0.90, the review load (reviews/day) will increase very rapidly. For the same reason, it is not advisable to use a requestRetention greater than 0.97.
After deciding the value of requestRetention, put this into the scheduler code. At the same time, decide the value of maximumInterval, which is the maximum interval any card is allowed to attain. The value in the FSRS scheduler code overrides the value set in Anki's deck options.
After performing the above steps, you are ready to start using FSRS. Just start reviewing and FSRS will do its work.
After setting up FSRS in your Anki, you can install the FSRS4Anki Helper add-on and use it to reschedule your existing cards. This is a one-time measure to reschedule the cards that were previously scheduled according to Anki's built-in algorithm. The add-on also offers many other useful features. Read more about the add-on here: /~https://github.com/open-spaced-repetition/fsrs4anki-helper
You can also generate different parameters for different decks and configure them separately in the code. In the default configuration, deckParams already contains three groups of parameters.
The group "global config for FSRS4Anki" is global parameters.
The group "MainDeck1" are the parameters applied to the deck "MainDeck1" and its sub-decks.
Similarly, the third group is the parameters applied to the deck "MainDeck2::SubDeck::SubSubDeck" and its sub-decks. You can replace these with the decks you want to configure. If you need more, feel free to copy and add them.
const deckParams = [
{
// Default parameters of FSRS4Anki for global
"deckName": "global config for FSRS4Anki",
"w": [0.4, 0.6, 2.4, 5.8, 4.93, 0.94, 0.86, 0.01, 1.49, 0.14, 0.94, 2.18, 0.05, 0.34, 1.26, 0.29, 2.61],
// The above parameters can be optimized via FSRS4Anki optimizer.
// For details about the parameters, please see: /~https://github.com/open-spaced-repetition/fsrs4anki/wiki/The-Algorithm
// User's custom parameters for global
"requestRetention": 0.9, // recommended setting: 0.75 - 0.95
"maximumInterval": 36500,
// FSRS only modifies the long-term scheduling. So (re)learning steps in deck options work as usual.
// I recommend setting steps shorter than 1 day.
},
{
// Example 1: User's custom parameters for this deck and its sub-decks.
"deckName": "MainDeck1",
"w": [0.6, 0.9, 2.9, 6.8, 4.72, 1.02, 1, 0.04, 1.49, 0.17, 1.02, 2.15, 0.07, 0.35, 1.17, 0.32, 2.53],
"requestRetention": 0.9,
"maximumInterval": 36500,
},
{
// Example 2: User's custom parameters for this deck and its sub-decks.
// Don't omit any keys.
"deckName": "MainDeck2::SubDeck::SubSubDeck",
"w": [0.6, 0.9, 2.9, 6.8, 4.72, 1.02, 1, 0.04, 1.49, 0.17, 1.02, 2.15, 0.07, 0.35, 1.17, 0.32, 2.53],
"requestRetention": 0.9,
"maximumInterval": 36500,
}
];If there are some decks you don't want to use FSRS with, you can add their names to the skip_decks list.
const skip_decks = ["MainDeck3", "MainDeck4::SubDeck"];Q0: Why should the (re)learning steps be kept low with FSRS? What is considered as long interval by FSRS (from faq it seems maybe 1 day)?
A0: Due to limitations of Anki's custom scheduling, the FSRS scheduler can't determine whether the reviews in the learning or relearning steps occur on the same day or on different days. So, the scheduler assumes that all those reviews occur on the same day. This means that even if the last learning step is greater than 1 day, the first interval set by the FSRS scheduler can still be equal to 1 day. To prevent this unexpected behaviour, it is recommend to not use learning or relearning steps of more than or equal to 1 day.
If you still want to use learning (or relearning) steps greater than one day, it is recommended to reschedule your cards daily using the FSRS helper add-on. The helper add-on can read the entire review history of the card and, thus, provide more accurate next intervals in such cases.
"Auto reschedule the card after each review" allows you to set learning steps longer than 1d (it's not recommended otherwise), but then the intervals displayed above the answer buttons won't match real intervals. So pick your poison - disable this option and use learning steps no longer than or equal to 1d, or enable it but then your displayed intervals won't match real intervals. To enable this feature, you need to install the helper add-on: https://ankiweb.net/shared/info/759844606
Q1: Will AnkiDroid kill this (since it still doesn't use v3 scheduling)?
A1: AnkiDroid currently supports the v3 scheduler, which can be enabled through an advanced setting. However, it doesn't support custom scheduling yet, which is required for FSRS to work. AnkiDroid is expected to support FSRS in v2.17. For more details, please see: ankidroid/Anki-Android#12620
Till then, you can enable the "auto-reschedule after sync" option in the FSRS helper add-on. This way, when you sync your reviews from AnkiDroid to Desktop, they would be automatically rescheduled according to the FSRS algorithm. For best results, it is recommended to sync the reviews daily.
However, if you only use AnkiDroid, you're out of luck.
Q2: Would a "dormant" deck not actively being used affect the results?
A2: The cards in the "dormant" deck would have long intervals. So FSRS4Anki will predict you will forget them with a high possibility. If you remember them, FSRS4Anki will give longer intervals, but not be linear to the delay.
Q3: Does the algorithm change the way the card ease changes? If I just do good and hard will I be stuck in ease hell?
A3: FSRS4Anki substitutes the card ease with the difficulty variable. The lower the difficulty, the higher the factor. The ease hell is solved by mean reversion of difficulty. If you press good continuously, the difficulty will converge to
Q4: How can I see if it's working?
A4: You can use the AnkiWebView Inspector add-on. Open the inspector before you review the cards. Then you will get into debug mode and see the custom scheduling code in the inspector. For more details, please see: How does the scheduler work?
Q5: Does it work on existing cards, using their past review history? Or does it only work with information that is yet to be created in the future?
A5: The scheduler doesn't use the review history, but the optimizer does. The scheduler uses the ease factor and interval of cards to set memory states for existing cards. And you can use FSRS4Anki Helper to reschedule the existing cards based on the entire review history.
Q6: Once I started using this FSRS algorithm on my existing deck, if I ever have to go back to using Anki's built-in algorithm for the same deck, would that still be possible?
A6: The ease factor is modified by the v3 scheduler as usual, and FSRS4Anki doesn’t intervene in it. FSRS4Anki only modifies the interval of cards. So you can go back without any problem.
Q7: Once FSRS is running, which Anki settings will become irrelevant (in the sense that changing them won't affect scheduling anymore), and which won't?
A7: FSRS only modifies the long-term scheduling. So Learning steps and Relearning steps work as usual. And I recommend not setting a step of more than one day. For example, if your current steps are 10m 1h 1d 2d, you had better remove the 1d 2d from the steps.
In the latest version of FSRS4Anki, maximum interval has been supported. You can modify them in fsrs4anki_scheduler.js.
The graduating interval, easy interval, new interval, starting ease, interval modifier, easy bonus, and hard interval become irrelevant.
Q8: How to set different parameters for specific decks?
A8:
Step 1: Add code to the front template of the cards (if the version of Anki that you are using is above 2.1.62, you can skip this step)
Copy the following code <div id=deck deck_name="{{Deck}}"></div> and paste it in the front template of cards.
Step 2: Optimize the parameters for your specific deck
Step 3: Set different parameters for specific decks and their sub-decks
Q9: Can I use (addon name), which modifies Anki's algorithm and changes the math behind it, together with FSRS?
A9: No. Any algorithm that affects Anki's built-in scheduler will either become useless once you start using FSRS, or will actively interfere with FSRS and cause problems.
Q10: FSRS is giving huge intervals to recently learnt cards. How can I fix this?
A10: For many users, the default scheduler in Anki (SM-2) tends to show new cards at unnecessarily short intervals. So, when users switch to FSRS, they tend to feel that the intervals given to new cards are too large. But these larger intervals match the target retention (configured in the scheduler code) better. By using these larger intervals, FSRS can prevent many of the unnecessary reviews that happen when using SM-2. So, it is advisable to try using these larger first intervals for a few days and see how it goes.
If you still want to decrease the intervals, you can increase the requestRetention in the scheduler. But note that this will decrease all the intervals, not just the first intervals.
Q11: How can I grade the card to make FSRS more effective?
A11: The grade should be chosen based only on how easy it was to answer the card, not how long you want to wait until you see it again.
For example, if you habitually avoid the easy button because it shows long intervals, you can end up in a negative cycle: You'd be making the "easy" situations even rarer and easy grade's intervals longer and longer.
This means you should ignore the intervals shown above the answer buttons and instead focus on how well you recall the information.
If you still want to see a deck sooner rather than later, for example because you have an exam coming up, you can use the Advance function of the Helper add-on. Advance is the preferable method because it doesn't skew the grading history of the cards.
Q12: I only use "Again" and "Good", will FSRS work fine?
A12: Yes. FSRS is about equally accurate for people who rarely use "Hard" and "Easy" and for people who use all 4 buttons a lot. However, this is not the final conclusion, and as we gather more data, this conclusion may change.
You haven't seen the answer for your question? Here are many questions asked by others: /~https://github.com/open-spaced-repetition/fsrs4anki/issues?q=is%3Aissue+label%3Aquestion+
If you still have trouble, please open a new issue to provide the details about that.
/~https://github.com/open-spaced-repetition/fsrs4anki/issues/new/choose