{% extends "base.html" %} {% block title %}Start Interpretation ยท Sparsifire{% endblock %} {% block content %}
Explorer

Interpret Features

{% csrf_token %} {% if form.errors %}
Please correct the errors below: {{ form.errors }}
{% endif %}
{{ form.run }} {% if form.run.errors %}
{{ form.run.errors }}
{% endif %}
{{ form.n_features }}
{{ form.n_features.help_text }}
{{ form.ollama_model }}
{{ form.ollama_model.help_text }}
{{ form.k_positive }}
{{ form.k_positive.help_text }}
{{ form.k_negative }}
{{ form.k_negative.help_text }}
{{ form.temperature }}
{{ form.temperature.help_text }}
{{ form.system_prompt }}
Customize the persona to guide the interpretation style.

Cancel
Local LLM Required
Ensure Ollama is running: ollama serve
Methodology

This process uses the Auto-Interpretability method described by O'Neill et al. (2024).

It feeds the LLM with a contrastive set of Top-K activating documents versus Random non-activating documents to distill the semantic meaning of each latent feature.

{% endblock %}