{% extends "base.html" %} {% block title %}Run queue · {{ profile.name }}{% endblock %} {% block content %}
Build a list of specific runs -- any mix of models, quantizations, engines, backends, power states, even router runs -- then start the whole queue. Lemonade holds one model at a time, so the queue runs sequentially in the background, exactly like Sweep (unlike Sweep, which is one model's pulled quantizations × a Cartesian product of axes, this queue is whatever specific combinations you add). Only combinations this profile can actually run are selectable.
{% if running_batch %}A sweep/queue is already running ({{ running_batch.completed_count }}/{{ running_batch.total_count }} done) -- view progress before starting another.
{% endif %} {% if capabilities.error %}Couldn't reach this instance ({{ capabilities.error }}).
{% elif not models_by_base %}No models pulled yet -- pull one first.
{% else %} {% set base_names = models_by_base.keys() | list %} {% set first_variants = models_by_base[base_names[0]] %}