{% extends 'base.html' %} {% block title %}Extraction Lab ยท Menhir{% endblock %} {% block body_class %}extraction-lab-page{% endblock %} {% block extra_head %} {% endblock %} {% block app_main %}
EXTRACTION WORKBENCH

Extraction Lab

Test entity extraction and proposition recall against prompt variants. Measure mention/proposition recall/precision against gold ground truth.

Pure read No graph writes. Read-only extraction tracing against production client pipeline.

Previous Episodes (Context)

Optional prior episodes that will be considered during extraction (respecting context_episode_count limit).

Gold Ground Truth

Expected extraction results (for scoring).

Experiment arms

Every enabled arm receives the same message, episodes, and gold criteria.

Ready.
Results will appear here in comparable columns.
{% endblock %} {% block extra_scripts %} {% endblock %}