AnkiWeb
- Rating
- 1 (π 1 Β· π 0)
- Updated
- 2025-07-27
- Anki versions
- 25.07.1~
- Description language
- en
AnkiWeb addon 1202210378
Provides an adaptive FSRS v5/v6 retention scheduler minimizing expected review workload until card retirement, now deprecated and intended for lifelong learning.
Open on AnkiWeb GitHub Ask about alternatives
stale
| Min Anki | Max Anki | Updated |
|---|---|---|
| 25.02 | 25.07.1+ | 2025-07-27 |
Loadingβ¦
Reorders Anki review cards by expected long-term knowledge gain, displays each card's gain, and supports FSRS 4.5β6.
Overrides FSRS desired retention per card using ordered search, note-field threshold, or grade rules, requiring Anki builds with dynamic retention reviewer hooks.
Supports FSRS rescheduling, postponing, advancing, breaks, load balancing, easy days, sibling dispersal, flattening, stats, and misuse remedies for Anki reviews.
Transfers scheduling data, review history, and FSRS memory state between Anki cards, with configurable shortcuts and options to delete the old card or change deck.
Assigns FSRS parameters to cards using ordered search rules, keeping deck presets when no dynamic rule matches.
Predicts Anki study time, finish time, pacing, retention, fatigue, and distractions using offline machine learning, customizable progress metrics, and fully adjustable UI styling.
This addon implements an adaptive retention scheduler that replaces the default fixed desired retention strategy used in Anki's FSRS. It supports both FSRS v5 and v6.
Note: This addon is now deprecated. If you're interested in other scheduling strategies, I would recommend trying No Scheduler.
By default, FSRS uses a fixed desired retention and optimizes the knowledge-workload ratio over time.
This addon introduces a different optimization objective: minimizing the expected total review workload until a card is retired, where retirement is defined as reaching a stability of 36,500 (approximately 100 years).
Formally, the scheduler solves the following recursive problem:
$$
w(\text{card}) = \begin{cases}
\min_{r}\mathbb{E}_{\text{card'}}[w(\text{card}')] + 1 & \text{if } \text{stability} < 36500 \
0 & \text{otherwise}
\end{cases}
$$
where:
The above problem can be efficiently solved by dynamic programming.
This addon is designed for users with lifelong learning goals, where minimizing long-term workload is more important than short-term efficiency.