Spaces:
Configuration error
Configuration error
Add org card HTML content
Browse files- index.html +68 -19
index.html
CHANGED
|
@@ -1,19 +1,68 @@
|
|
| 1 |
-
<!
|
| 2 |
-
<html>
|
| 3 |
-
|
| 4 |
-
|
| 5 |
-
|
| 6 |
-
|
| 7 |
-
|
| 8 |
-
|
| 9 |
-
|
| 10 |
-
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
<!DOCTYPE html>
|
| 2 |
+
<html lang="en">
|
| 3 |
+
<head>
|
| 4 |
+
<meta charset="UTF-8">
|
| 5 |
+
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
| 6 |
+
<title>Divinci AI</title>
|
| 7 |
+
<style>
|
| 8 |
+
body { font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', sans-serif; max-width: 860px; margin: 0 auto; padding: 2rem 1.5rem; color: #1a1a1a; line-height: 1.6; }
|
| 9 |
+
h1 { font-size: 1.8rem; font-weight: 700; margin-bottom: 0.25rem; }
|
| 10 |
+
h2 { font-size: 1.1rem; font-weight: 600; margin-top: 2rem; margin-bottom: 0.5rem; border-bottom: 1px solid #e5e7eb; padding-bottom: 0.25rem; }
|
| 11 |
+
p { margin: 0.5rem 0 1rem; }
|
| 12 |
+
table { border-collapse: collapse; width: 100%; margin: 1rem 0; font-size: 0.9rem; }
|
| 13 |
+
th, td { border: 1px solid #e5e7eb; padding: 0.5rem 0.75rem; text-align: left; }
|
| 14 |
+
th { background: #f9fafb; font-weight: 600; }
|
| 15 |
+
a { color: #2563eb; text-decoration: none; }
|
| 16 |
+
a:hover { text-decoration: underline; }
|
| 17 |
+
.tagline { color: #6b7280; font-size: 1rem; margin-bottom: 1.5rem; }
|
| 18 |
+
.footer { margin-top: 2.5rem; padding-top: 1rem; border-top: 1px solid #e5e7eb; font-size: 0.85rem; color: #6b7280; }
|
| 19 |
+
hr { border: none; border-top: 1px solid #e5e7eb; margin: 1.5rem 0; }
|
| 20 |
+
</style>
|
| 21 |
+
</head>
|
| 22 |
+
<body>
|
| 23 |
+
|
| 24 |
+
<h1>Divinci AI</h1>
|
| 25 |
+
<p class="tagline">Feature-level interpretability artifacts for open transformers — built openly, validated empirically.</p>
|
| 26 |
+
|
| 27 |
+
<p>A <strong>vindex</strong> is a transformer's weights decompiled into a queryable feature database. It exposes the entity associations, circuit structure, and knowledge-editing surfaces that live inside a model's FFN layers — without requiring GPU inference for most operations.</p>
|
| 28 |
+
<p>Think of it as the model's index: the thing you search before you run it.</p>
|
| 29 |
+
|
| 30 |
+
<hr>
|
| 31 |
+
|
| 32 |
+
<h2>Published Vindexes</h2>
|
| 33 |
+
<table>
|
| 34 |
+
<tr><th>Model</th><th>Architecture</th><th>Params</th><th>Vindex</th></tr>
|
| 35 |
+
<tr><td>Gemma 4 E2B-it</td><td>Dense (Gemma 4)</td><td>2B</td><td><a href="https://huggingface.co/Divinci-AI/gemma-4-e2b-vindex">gemma-4-e2b-vindex</a></td></tr>
|
| 36 |
+
<tr><td>Qwen3.6-35B-A3B</td><td>MoE (Qwen3.6)</td><td>35B / 3B active</td><td><a href="https://huggingface.co/Divinci-AI/qwen3.6-35b-a3b-vindex">qwen3.6-35b-a3b-vindex</a></td></tr>
|
| 37 |
+
<tr><td>GPT-OSS 120B</td><td>MoE (OpenAI)</td><td>120B / ~13B active</td><td><em>building</em></td></tr>
|
| 38 |
+
</table>
|
| 39 |
+
<p>Three organizations, three architectures: Gemma dense, Qwen MoE, OpenAI MoE.</p>
|
| 40 |
+
|
| 41 |
+
<hr>
|
| 42 |
+
|
| 43 |
+
<h2>What's a Vindex?</h2>
|
| 44 |
+
<p>Standard model weights tell you <em>what</em> a model computes. A vindex tells you <em>where</em> it stores specific knowledge and <em>which features</em> need to change for a targeted edit.</p>
|
| 45 |
+
<p>Concretely: given a query like <code>Paris → capital</code>, a vindex walk returns the layers, feature directions, and token associations that encode that fact. A patch operation writes a rank-1 ΔW that suppresses or overwrites that association — compiled back to standard HuggingFace safetensors for inference.</p>
|
| 46 |
+
<p>LarQL (the toolchain that builds vindexes) is open-source: <a href="https://github.com/chrishayuk/larql">chrishayuk/larql</a> · <a href="https://github.com/Divinci-AI/larql">Divinci-AI/larql</a>.</p>
|
| 47 |
+
|
| 48 |
+
<hr>
|
| 49 |
+
|
| 50 |
+
<h2>Research</h2>
|
| 51 |
+
<p><strong>Paper 1 — Architectural Invariants of Transformer Computation</strong> <em>(arXiv forthcoming)</em><br>
|
| 52 |
+
Five properties measured across every model in this collection. Three hold within ±15% coefficient of variation across architectures, organizations, and scales. One collapses under 1-bit quantization. One scales monotonically with model size.</p>
|
| 53 |
+
<p><strong>Paper 2 — Constellation Edits</strong> <em>(draft)</em><br>
|
| 54 |
+
Mechanistic knowledge editing in transformer feature space. Includes a negative result: why activation-space edits fail in 1-bit models, and what weight-space geometry reveals about why.</p>
|
| 55 |
+
<p>Working notebooks: <a href="https://github.com/Divinci-AI/server/tree/preview/notebooks">github.com/Divinci-AI/server/tree/preview/notebooks</a></p>
|
| 56 |
+
|
| 57 |
+
<hr>
|
| 58 |
+
|
| 59 |
+
<h2>Working in Public</h2>
|
| 60 |
+
<p>Every measurement in our papers traces back to a notebook and a commit. Negative results ship alongside positive ones — the compensation mechanism that defeats knowledge editing in 1-bit models is in the notebooks, not buried in a supplement.</p>
|
| 61 |
+
<p>If you replicate a result and find a discrepancy, open an issue on the LarQL repo.</p>
|
| 62 |
+
|
| 63 |
+
<div class="footer">
|
| 64 |
+
Vindexes on this org are free for academic and research use (CC-BY-NC 4.0). Commercial licensing: <a href="mailto:mike@divinci.ai">mike@divinci.ai</a>
|
| 65 |
+
</div>
|
| 66 |
+
|
| 67 |
+
</body>
|
| 68 |
+
</html>
|