open-spaced-repetition/FSRS-Anki-20k

Introduction

FSRS-Anki-20k is a dataset of 20k collections from Anki for FSRS project. It is a random sample of collections with 5000+ revlog entries, so it should contain a mix of older (still active) users, and newer users. Entries are pre-sorted in (cid, id) order.

There are two versions of the dataset: ./revlogs and ./dataset. The ./revlogs version contains the raw revlog entries, while the ./dataset version contains the dataset preprocessed by ./revlogs2dataset.py. The size of the raw revlog entries is ~50GB, and the size of the preprocessed dataset is ~20GB.

Data Format

Revlogs

Please see the protocol buffer definition in stats.proto. For the fields that are not self-explanatory, please refer to the Anki Database Structure.

Dataset

The columns of the dataset are as follows:

• card_id: the unique identifier for each flashcard.
• review_th: the ordinal number of the review among all reviews done by the user.
• delta_t: the number of days since the last review of this flashcard. -1 if this is the first review.
• rating: the rating given by the user for this review, 1: again, 2: hard, 3: good, 4: easy. Where only 'again' indicates a failed recall, and all other scores indicate a successful recall.

Preprocess

1. read the revlog entries from the .revlog file.
2. remove the revlog entries generated by reviews in filtered decks when the user disables the option "Reschedule cards based on my answers in this deck".
3. remove the revlog entries generated by manual (re)scheduling like Forget and Set Due Date.
4. keep the revlog entries from the latest learning start sequence for each flashcard.
5. calculate the time between reviews of each flashcard (delta_t), review order (review_th), and encode card_id numerically.
6. save the dataset to a .csv file.

License

This dataset is released under the FSRS-Anki-20k License.

Related Projects

SRS Benchmark