Assistant Eval Runner

Pick cases in the active dataset tab (or none to run the whole dataset), choose a model, run — results land in Phoenix. The Help tab explains the workflow.

{% for group in datasets %} {% empty %}

No eval cases registered.

{% endfor %}

Recent runs

{% for run in runs %} {% empty %} {% endfor %}
Dataset Experiment Cases Model Runs Status Started Finished Result
{{ run.dataset }} {{ run.experiment_name|default:"" }} {% if run.case_ids %}{{ run.case_ids|join:", " }}{% else %}(all){% endif %} {{ run.model }}{% if run.git_label %} ({{ run.git_label }}){% endif %}{% if run.prompt_overrides %} [prompts: {{ run.prompt_overrides|join:", " }}]{% endif %} {{ run.runs }} {% if run.status == "done" %}finished{% elif run.status == "failed" %}crashed{% else %}{{ run.status }}{% endif %} {{ run.started_at|date:"M d, H:i:s"|default:"" }} {{ run.finished_at|date:"M d, H:i:s"|default:"" }} {% if run.phoenix_link %} Phoenix {% elif run.status == "failed" %} {{ run.error }} {% endif %}
No runs yet.