{% extends "study/base.html" %} {% block title %}FSRS optimizer{% endblock %} {% block content %} {% include "study/settings_tabs.html" %}

FSRS optimizer

Fits the 19 FSRS-5 weights to your review history by minimising log-loss (Adam, finite-difference gradients, weights clipped to the published bounds). Needs at least {{ min_reviews }} scored reviews (reviews on a later day than the previous one); you have {{ scored }}.

{% csrf_token %}
{% if params %}
{% csrf_token %}
{% endif %}
{% if params %}
{{ params.review_count }}reviews fitted
{{ params.log_loss_before|floatformat:4 }} → {{ params.log_loss_after|floatformat:4 }}log-loss
{{ rmse_before|floatformat:2 }}% → {{ rmse_after|floatformat:2 }}%RMSE (bins)
{{ params.optimized_at|date:"Y-m-d H:i" }}optimized
{% endif %} {% for i, w, d, lo, hi in rows %}{% endfor %}
wCurrentDefaultBounds
w{{ i }}{{ w|floatformat:4 }}{{ d|floatformat:4 }}{{ lo }} – {{ hi }}

Command line: python manage.py optimize_fsrs --user demo. Algorithm notes: docs/fsrs.md.

{% endblock %}