This run verifies experiment #{{ experiment.id }}:
{{ experiment.flag }}. See what the change did.
{% endif %}
{% if is_demo and not inline_assets %}
{% set far = report.findings_json | selectattr("rule_id", "equalto", "H03") | rejectattr("status", "equalto", "ok") | list %}
This is the demo run that ships with kneebench: a real benchmark of
{{ report.summary.model }} on {{ report.summary.gpu }}{% if far %}, measured from far away over the internet,
which is why the network findings stand out{% endif %}. Every number and chart below works the same way for
your own runs.
Benchmark your server. Delete this run any time; kneebench demo brings it back.
{% endif %}
{% if not inline_assets %}{% set slo = report.summary.slo %}
What if…judge this run against other targets{% if has_estimates %}, or estimate another request size{% endif %}. Uses the stored measurements; nothing is re-run.
{% if has_estimates %}
{% endif %}
{% endif %}
{% include "report_body.html" %}
{% endblock %}
{% block scripts %}
{% if inline_assets %}{% else %}{% endif %}
{% endblock %}