Need help on optimizing Anki for long-term vocabulary retention

Hi just wanted to know If I could make any improvements on my anki deck options because I tend to forget cards that I’ve been doing for more than a year, 6-7 months ago I had put my desired retention up to 95% , but it was too much so I put it back to 90% (around a month ago).

I just do vocab cards like this

FRONT:

深謀 (word)

BACK:

しんぼう (reading)

thought-out plan (definition)

My deck options are

NEW CARDS AND LAPSES LEARNING STEPS: I left them empty cause I thought fsrs might do it better.

DISPLAY ORDER:

New card gather order:

Deck

New card sort order:

Card type, then order gathered

New/review order:

Show before reviews

Interday learning/review order:

Mix with reviews

Review sort order:

Ascending intervals

FSRS

Desired Retention:

0.90

FSRS parameters: 0.2532, 0.4902, 1.3207, 16.3767, 8.7124, 0.1557, 0.5130, 0.0013, 0.9152, 0.2671, 0.7276, 1.9673, 0.0025, 0.0975, 2.3655, 1.0000, 5.9999, 0.6279, 1.0564

I’ve already optimised my my fsrs. The settings I didn’t mention are untouched, hence why I felt no need to talk about. I only use the again and good buttons. Currently doing an average of 1,100 day for this deck. Not doing any new cards.

My retention rate

Total Pass Fail Retention

Today 1,053 100 91.3%

Yesterday 1,079 103 91.3%

Last week 7,185 685 91.3%

Last month 32,520 3,672 89.9%

All time 658,586 125,709 84%

Since I started using fsrs my retention rate has gone up in both mature and young cards especially mature ones which were at 0.70% retention rate before. Are roughly around 0.90 now.

Thanks for reading so far and any help is appreciated.

  1. Since you reduced your DR, have you rescheduled your cards (using the FSRS Helper add-on)? Or are they still scheduled whre they would be for 95% retention?

FSRS might do it better, but it also might not – and there’s no way to tell whether it will work better for you or won’t. FSRS doesn’t have a model for near-term memory, so blanking out your steps isn’t recommended for most learners.

  1. The question for you to ask yourself is – are you happy with the intervals you’re getting when New cards are introduced and after lapses?
  1. How long has it been since you optimized? You can (and probably should) re-optimize monthly.
  1. That’s an incredibly high workload for language learning. Are you satisfied with that workload?
  1. Are those cards that you last studied before you enabled FSRS? Or have they been rescheduled using FSRS?
1 Like

Hi thanks for the reply.

For N1. I’m not sure actually when I rescheduled, I didn’t bother noting it. The first time I used FSRS was also the time I put it up to 0.95%, (those 6-7 months ago) I know I rescheduled it, I’d like to think since I switched back to 0.90 I rescheduled it again, because the amount of cards since have halfed essentially (Also I didn’t use the FSRS helper add on and have been changing all my settings via ankidroid).

N2. I end up getting the new cards wrong or sometimes my already learned cards wrong whilst getting every other card right and then I see the cards I get wrong very quickly again leaving me not a big of a interval to really test my memory cause I’ll have seen that card just like 9 seconds ago (I’ll tend to average around 3 seconds on a card), so I’m not satisfied with it but I just kinda had in my mind the longer the interval the better it tests my memory so left it like that.

N3. Optimised it around a month ago and tried again today, but it says “the FSRS parameters currently appear to be optimal” so I haven’t fiddled with it anymore.

N4. If I could I would like to shorten it lol, but to no detriment of my retention rate If possible.

N5. They’ve been rescheduled, but I’m just noticing I’m struggling with long interval cards (because of the amount of cards it happens to).

Thanks for the reply :+1:

I think the only thing I can make improvements on is my lapses. Though I don’t know if the default would word for me because if I fail a card with the current way of letting fsrs handle it I mainly see it 3 times and still forget it so I don’t know whether I more steps than 3. And I just realised, I didn’t reply to your comment but mine instead for my previous response.

Also My fsrs simulator says I’ll reach 1,000 reviews in 94 days without doing new cards. So I feel like I’m doing something wrong.

Bringing your DR down from 95 to 90 was an important first step. But it is going to take some time for your too-heavy workload to lighten up, even if you’re not introducing more New cards. You’ve got a huge number of “Young” cards that need to grow to “Mature” – and that just takes time.

I’ve got my young/mature stats, if that helps.

■ New 5,461 26.46%

■ Learning 9 0.04%

■ Relearning 5 0.02%

■ Young 6,251 30.29%

■ Mature 8,909 43.17%

■ Suspended 1 0%

■ Buried 0 0%

Total 20,636

I think I mistook your retention information above for the “All” cards Stats > Retention table, and I was drawing your Young/Mature numbers from that. Sorry! [Except for things like FSRS parameters or card templates, which we need as text, most things you provide when asking for help – tables, graphs, settings – are better in their original form, as a screenshot.]

Your Young/Mature cards are more balanced than I thought, but it will still take time to bring your workload down.

1 Like

Thanks, I’ll provide a screenshot next time.

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.