Upload index.html with huggingface_hub

index.html

@@ -41,6 +41,7 @@
       margin-top: 1.8rem;
       margin-bottom: 0.8rem;
     }
+    p { margin-bottom: 0.8rem; }
     ul {
       padding-left: 1.2rem;
     }
@@ -85,18 +86,18 @@
 
 <hr />
 
-<h2>Disposition Distillation</h2>
+<p>We build on-device AI systems that reason, remember, and self-correct — small models designed to run autonomously at the edge with calibrated uncertainty and adversarial robustness.</p>
 
-<p>Tinman Lab develops <strong>Disposition Distillation (DD)</strong> — a multi-teacher distillation methodology that trains <em>how a model behaves</em> into weights, not system prompts. DD models plan before acting, acknowledge uncertainty, verify their own reasoning, and know what they don't know.</p>
+<h2>Research Areas</h2>
 
 <ul>
-  <li><strong>4-stage all-MIT pipeline</strong> — Kimi K2.5 → GLM-5 → MiniMax M2.7 → GLM-5</li>
-  <li><strong>7 behavioral dispositions</strong> — Eager, Deliberate, Adversarial, Curious, Self-Improving, Humble, Persistent</li>
-  <li><strong>On-device focus</strong> — 0.6B to 2B parameters, quantized for mobile and edge deployment</li>
-  <li><strong>100% open training data</strong> — MIT-licensed teachers only, zero proprietary model outputs</li>
+  <li><strong>Agent Memory</strong> — Encrypted semantic memory infrastructure for persistent agent context</li>
+  <li><strong>Adversarial Safety</strong> — Multi-agent stress-testing and trust verification for autonomous systems</li>
+  <li><strong>Perception Systems</strong> — On-device vision, voice, and multimodal understanding</li>
+  <li><strong>Disposition Distillation</strong> — Training behavioral tendencies (planning, uncertainty acknowledgment, self-verification) into sub-billion parameter model weights via all-MIT multi-teacher distillation</li>
 </ul>
 
-<h2>Models</h2>
+<h2>Open Models</h2>
 
 <table>
   <thead>