{% extends "base.html" %} {% load static %} {% block title %}Home — PRISM△{% endblock %} {% block extra_head %} {% endblock %} {% block content %}

PRISMΔ

Illuminating the "black box" of embeddings.

Load Dataset
1
Ingest
Dense Embeddings
2
Train SAE
Sparse Features
3
Interpret
Semantic Labels
Reference: "Disentangling Dense Embeddings with Sparse Autoencoders" (O'Neill et al., 2024)
{% endblock %} {% block extra_js %} {% endblock %}