File size: 7,189 Bytes
499c4f2 99bc722 499c4f2 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 | <!doctype html>
<html lang="en">
<head>
<meta charset="utf-8" />
<meta name="viewport" content="width=device-width, initial-scale=1" />
<meta
name="description"
content="Traceable historic 100-ID Gaussian-mixture evidence for Diffusion differentiable resampling."
/>
<link rel="icon" href="data:," />
<title>DiffRes: traceable 100-ID Gaussian-mixture study</title>
<style>
:root {
color-scheme: light dark;
--paper: #f8fafc;
--ink: #172033;
--muted: #526076;
--panel: #ffffff;
--line: #d8e0ec;
--accent: #215bd8;
--warning: #7a4d00;
--warning-bg: #fff6e3;
}
@media (prefers-color-scheme: dark) {
:root {
--paper: #10151f;
--ink: #edf2ff;
--muted: #afbbce;
--panel: #182131;
--line: #2b384d;
--accent: #8bb2ff;
--warning: #ffce72;
--warning-bg: #2b2417;
}
}
* { box-sizing: border-box; }
body {
margin: 0;
background: var(--paper);
color: var(--ink);
font: 16px/1.55 system-ui, -apple-system, BlinkMacSystemFont, "Segoe UI", sans-serif;
}
main { max-width: 900px; margin: 0 auto; padding: 56px 24px 72px; }
h1, h2 { line-height: 1.2; }
h1 { font-size: clamp(2rem, 5vw, 3.25rem); margin: 0.45rem 0 1rem; }
h2 { margin: 2.4rem 0 0.7rem; font-size: 1.3rem; }
p, li { max-width: 76ch; }
a { color: var(--accent); }
.eyebrow { color: var(--muted); font-weight: 700; letter-spacing: 0.08em; text-transform: uppercase; }
.status {
display: inline-block;
border: 1px solid var(--line);
border-radius: 999px;
padding: 0.2rem 0.65rem;
color: var(--muted);
font-size: 0.85rem;
font-weight: 650;
}
.card, .warning {
border: 1px solid var(--line);
border-radius: 12px;
padding: 1.1rem 1.2rem;
background: var(--panel);
}
.warning { background: var(--warning-bg); color: var(--warning); }
table { border-collapse: collapse; width: 100%; margin: 1rem 0; }
th, td { border-bottom: 1px solid var(--line); padding: 0.7rem 0.55rem; text-align: left; vertical-align: top; }
th:first-child, td:first-child { padding-left: 0; }
th:last-child, td:last-child { padding-right: 0; }
.number { font-variant-numeric: tabular-nums; text-align: right; white-space: nowrap; }
footer { border-top: 1px solid var(--line); color: var(--muted); font-size: 0.9rem; margin-top: 3rem; padding-top: 1rem; }
</style>
</head>
<body>
<main>
<div class="eyebrow">ICML 2026 evidence package</div>
<h1>DiffRes: traceable 100-ID Gaussian-mixture study</h1>
<p class="status">Historic stored output · static package · no numerical execution</p>
<p>
A local study wrapper using released implementation, stored numerical random keys and pinned source
config recorded a traceable 100-ID Gaussian-mixture result; a separate validator recomputed
raw-record aggregates, paired intervals, and source hashes.
</p>
<section class="warning" aria-labelledby="boundary-title">
<h2 id="boundary-title">Interpretation boundary</h2>
<p>
This is not an exact reproduction, independent replication, complete Table 1 or Table 2
reproduction, publication score, or broad outperformance claim. It is a stored,
traceable comparison against the multinomial baseline only. This static package
does not verify a source checkout or numerical execution, and it does not establish
source-hash identity, execution provenance, paper-level results, or significance.
</p>
</section>
<section aria-labelledby="record-title">
<h2 id="record-title">Recorded configuration</h2>
<div class="card">
<ul>
<li>Released <code>zgbkdlm/diffres</code> source at <code>767effe3e755067eb8a04422597fbf37eb8ab754</code>.</li>
<li>Stored Monte Carlo IDs 0–99; 10,000 particles; 1,000 sliced-Wasserstein projections.</li>
<li>Probability-flow ODE, <code>T=3</code>, 128 steps, Jentzen--Kloeden integrator.</li>
<li>Diffusion resampling and multinomial baseline; no other Table 1 baselines.</li>
</ul>
</div>
</section>
<section aria-labelledby="results-title">
<h2 id="results-title">Historical stored aggregates</h2>
<table>
<thead>
<tr>
<th scope="col">Metric (lower is better)</th>
<th scope="col" class="number">Diffusion resampling</th>
<th scope="col" class="number">Multinomial baseline</th>
</tr>
</thead>
<tbody>
<tr>
<td>Sliced Wasserstein L1, mean ± population SD</td>
<td class="number">0.0807686 ± 0.0212899</td>
<td class="number">0.0824334 ± 0.0250323</td>
</tr>
<tr>
<td>Squared posterior-mean residual L2, mean ± population SD</td>
<td class="number">0.0373908 ± 0.0299358</td>
<td class="number">0.0378330 ± 0.0442877</td>
</tr>
</tbody>
</table>
<p>
The recorded paired 95% t intervals cross zero for both metrics. They are not evidence of
a statistically significant improvement.
</p>
</section>
<section aria-labelledby="evidence-title">
<h2 id="evidence-title">Inspect the evidence</h2>
<ul>
<li><a href="evidence/gms-selected-ids-0-99.json">Raw 100-ID result</a></li>
<li><a href="evidence/gms-selected-ids-0-99-validation.json">Historic validation record</a></li>
<li><a href="evidence/validate_gms_selected_result.py">Validator source</a></li>
<li><a href="evidence/verify_stored_gms_evidence.py">Self-contained stored-evidence verifier</a></li>
<li><a href="evidence/README.md">Evidence guide and validator limits</a></li>
<li><a href="MANIFEST.md">Source and artefact manifest</a></li>
<li><a href="README.md">Full README, safety note, and licensing</a></li>
</ul>
</section>
<section aria-labelledby="safety-title">
<h2 id="safety-title">Selection and safety</h2>
<p>
The package contains only static presentation files, two historical JSON records, a
non-networked audit script, and licensing/provenance notes. It deliberately excludes the
upstream source tree, virtual environments, caches, runners, model artefacts, partial
studies, and LGSSM material.
</p>
</section>
<footer>
<a href="https://arxiv.org/abs/2512.10401">Paper</a> ·
<a href="https://openreview.net/forum?id=M0e5XORjAW">OpenReview</a> ·
<a href="https://github.com/zgbkdlm/diffres">released source</a> ·
<a href="NOTICE">attribution and licence notice</a>
</footer>
</main>
</body>
</html>
|