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>