abdelstark's picture
Initial publish: landing + slides + sample compare report + PCA assets
acb2bb6 verified
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<meta name="viewport" content="width=device-width, initial-scale=1.0">
<title>latent-inspector</title>
<style>
:root {
color-scheme: dark;
--bg: #0d1117;
--panel: #161b22;
--panel-2: #11161d;
--text: #e6edf3;
--muted: #8b949e;
--accent: #79c0ff;
--accent-2: #58a6ff;
--border: #30363d;
}
* { box-sizing: border-box; }
body {
margin: 0;
min-height: 100vh;
font-family: "Segoe UI", system-ui, sans-serif;
color: var(--text);
background: radial-gradient(circle at top, #182032 0%, var(--bg) 45%);
}
main {
max-width: 1040px;
margin: 0 auto;
padding: 4rem 1.5rem 5rem;
}
.eyebrow {
margin: 0;
color: var(--muted);
text-transform: uppercase;
letter-spacing: 0.12em;
font-size: 0.78rem;
}
h1 {
margin: 0.35rem 0 0;
font-size: clamp(2.3rem, 5vw, 4rem);
line-height: 1.05;
}
.lede {
max-width: 760px;
margin: 1rem 0 0;
color: var(--muted);
font-size: 1.08rem;
line-height: 1.6;
}
.grid {
display: grid;
grid-template-columns: repeat(auto-fit, minmax(300px, 1fr));
gap: 1rem;
margin-top: 2rem;
}
.card {
display: flex;
flex-direction: column;
gap: 0.9rem;
padding: 1.35rem;
border: 1px solid var(--border);
border-radius: 18px;
background: linear-gradient(180deg, var(--panel), var(--panel-2));
}
.card h2 {
margin: 0;
font-size: 1.45rem;
color: var(--text);
}
.card p {
margin: 0;
color: var(--muted);
line-height: 1.6;
}
.meta {
color: var(--muted);
font-size: 0.9rem;
}
.cta {
display: inline-flex;
align-items: center;
width: fit-content;
padding: 0.7rem 1rem;
border-radius: 999px;
text-decoration: none;
color: white;
background: linear-gradient(135deg, var(--accent), var(--accent-2));
font-weight: 600;
}
.footer {
margin-top: 2rem;
color: var(--muted);
font-size: 0.95rem;
}
a {
color: var(--accent);
}
</style>
</head>
<body>
<main>
<p class="eyebrow">latent-inspector</p>
<h1>Representation analysis, presentation, and report</h1>
<p class="lede">This GitHub Pages deployment has two public entry points: the narrative presentation of the project and a self-contained sample compare report with the raw metrics, PCA projections, and heatmaps behind the claims.</p>
<div class="grid">
<section class="card">
<div class="meta">Presentation</div>
<h2>How AI Models See the World</h2>
<p>Reveal.js slide deck for latent-inspector. It walks through the project, the model comparison setup, the corrected EUPE and V-JEPA 2 interpretations, and the broader thesis about how different training objectives reshape representation geometry.</p>
<a class="cta" href="slides.html">Open presentation</a>
</section>
<section class="card">
<div class="meta">Sample Report</div>
<h2>Four-Model Elephant Compare Report</h2>
<p>Self-contained HTML report generated by <code>latent-inspector compare</code> for DINOv2, I-JEPA, V-JEPA 2, and EUPE on the canonical elephant image. Includes per-model metrics, pairwise CKA and k-NN overlap, PCA images, and exported artifact metadata.</p>
<a class="cta" href="reports/20260408-123006/report.html">Open sample report</a>
</section>
</div>
<p class="footer">Repository: <a href="https://github.com/AbdelStark/latent-inspector">github.com/AbdelStark/latent-inspector</a></p>
</main>
</body>
</html>