FSRS for Rust, including Optimizer and Scheduler
Implement modern spaced repetition algorithms, simulation, and parameter optimization in Rust.
fsrs-rs is a Rust implementation of the Free Spaced Repetition Scheduler (FSRS), offering scheduling capabilities, simulations, and parameter optimization. It suits developers and data-driven learning system architects who need native Rust spaced repetition logic rather than heavy Python or LibTorch stacks.
The crate handles review state calculations, data migration from SM-2 style data, and parameter optimization from review logs using scalar and SIMD Rust kernels. It also includes single-user Cost ADR policy simulation and training for FSRS-6 and FSRS-7 models.
It suits Rust developers building custom spaced repetition platforms, memory trainers, or educational apps who require direct control over optimization and scheduling. It does not suit teams looking for a ready-to-use end-user application or those working outside the Rust ecosystem without utilizing bindings.
Compute custom parameters directly from your card review history logs.
Convert existing legacy ease factors and intervals into modern FSRS memory states.
Train and simulate single-user cost-conditioned desired-retention policies using CPU/Rayon.
Access core functionality via bindings for C, Python, Node.js, Dart, and PHP.
Free under the BSD-3-Clause license. There are no paid hosted options mentioned in the repository, so the only cost is your own development and compute time.
880 open remote AI-training roles, updated November 15, 2026; most pay $20–$175 an hour. Zekai lists them free; the application goes to the employer’s platform.
Use rs-fsrs if you only need to schedule cards. Use fsrs-rs if your application also requires parameter optimization.
The crate uses scalar and SIMD Rust kernels for inference and analytic gradients for training, removing any Burn or libtorch dependency.
See Zekai first in Google