Add SchemeArena logo (nav, hero, favicon) and paper Figure 5 to Results

.gitattributes

@@ -34,3 +34,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 assets/framework.png filter=lfs diff=lfs merge=lfs -text
+assets/figure5_scheming_rates.png filter=lfs diff=lfs merge=lfs -text

css/main.css

@@ -454,3 +454,8 @@ table.data.fixed col.lab { width: 30%; }
 
 .verdict { align-items: flex-start; font-weight: 500; line-height: 1.5; }
 .verdict svg { flex: none; margin-top: 2px; }
+
+.brand-mark { width: 28px; height: 28px; object-fit: contain; }
+.hero-logo { display: block; width: 84px; height: auto; margin: 0 auto 18px; filter: drop-shadow(0 6px 14px rgba(75, 86, 98, .18)); }
+.result-fig { display: block; width: 100%; height: auto; border-radius: 8px; margin-top: 6px; }
+@media (max-width: 760px) { .hero-logo { width: 64px; margin-bottom: 14px; } }

index.html

@@ -8,19 +8,14 @@
   <link rel="preconnect" href="https://fonts.googleapis.com" />
   <link rel="preconnect" href="https://fonts.gstatic.com" crossorigin />
   <link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700&family=JetBrains+Mono:wght@400;500;600;700&family=Source+Serif+4:opsz,wght@8..60,500;8..60,600&display=swap" rel="stylesheet" />
+  <link rel="icon" type="image/svg+xml" href="assets/logo.svg" />
   <link rel="stylesheet" href="css/main.css" />
 </head>
 <body>
 
 <nav id="topNav">
   <a class="nav-brand" href="#overview">
-    <svg class="brand-mark" viewBox="0 0 32 32" aria-hidden="true">
-      <rect x="1" y="3" width="30" height="27" rx="8" fill="#b9470f"/>
-      <rect x="5" y="11" width="22" height="8" rx="4" fill="#15171b"/>
-      <circle cx="11" cy="15" r="2" fill="#f5b38c"/><circle cx="21" cy="15" r="2" fill="#f5b38c"/>
-      <rect x="11" y="23" width="10" height="2.4" rx="1.2" fill="#fbeee5"/>
-      <rect x="15" y="0" width="2" height="4" rx="1" fill="#b9470f"/>
-    </svg>
+    <img class="brand-mark" src="assets/logo.svg" alt="" />
     SchemeArena
   </a>
   <div class="nav-tabs" role="tablist">
@@ -46,6 +41,7 @@
 <section id="tab-overview" class="tab-panel active">
 
   <header class="hero">
+    <img class="hero-logo" src="assets/logo.svg" alt="SchemeArena logo" />
     <span class="pill"><span class="dot"></span>Benchmark · AI Safety · LLM Agents</span>
     <h1><span class="name">SchemeArena</span>: Factorized Stress Testing of Scheming in LLM Agents</h1>
     <p class="subtitle">A 400-scenario benchmark that varies instrumental goals, pressure, strategic hints, and oversight to study when LLM agents covertly pursue misaligned goals.</p>
@@ -325,11 +321,19 @@
       </div>
     </section>
 
+    <section class="block" style="margin-top:20px">
+      <div class="card chart-card">
+        <div class="chart-head"><h3>Scheming rates across all seven setups</h3></div>
+        <img class="result-fig" src="assets/figure5_scheming_rates.png" alt="Line chart of scheming rates with error bars for o4-mini, o1, Claude-3.7-Sonnet, Qwen3-8B and Qwen3-32B across Setups 1 to 7, with Setup 6 split into action-only and full oversight" loading="lazy" />
+        <p class="footnote">Scheming rates for all setups that evaluate propensity (blue background) and capability (yellow background). Strategic hints (Setups 5 and 7) strongly enable scheming, especially for Qwen models. Action-only monitoring (Setup 6 w/ Action) slightly increases scheming among the closed models, while full-trajectory oversight yields only limited reductions. Error bars: 95% bootstrap confidence intervals. (Paper, Figure 5.)</p>
+      </div>
+    </section>
+
     <section class="block" style="margin-top:20px">
       <div class="card chart-card">
         <div class="chart-head"><h3>Scheming capability (Setups 5–7)</h3></div>
         <div class="table-wrap"><table class="data fixed" id="capTable"></table></div>
-        <p class="footnote">Values reported in the text of the paper (§5.3); changes are relative to Setup 5. "—": shown only graphically in Figure 5 of the paper, where the Qwen models stay near saturation across all oversight conditions. Setup 6 here is action-only oversight.</p>
+        <p class="footnote">Values reported in the text of the paper (§5.3); changes are relative to Setup 5. "—": not stated numerically in the text; see the figure above, where the Qwen models stay near saturation across all oversight conditions. Setup 6 here is action-only oversight.</p>
       </div>
     </section>