ToolUse.1: leaderboard

app.js

@@ -303,13 +303,17 @@ async function loadResults() {
 }
 
 function applyMetricCards(data) {
-  const best = [...(data.rows || [])].sort((a, b) => (b.score ?? 0) - (a.score ?? 0))[0];
+  const best = [...(data.rows || [])]
+    .filter((r) => !r.self_check && typeof r.score === "number")
+    .sort((a, b) => (b.score ?? 0) - (a.score ?? 0))[0];
   if (!best) return;
   const map = { overall: best.score, call: best.correct_call, abstain: best.abstention };
   document.querySelectorAll("[data-metric]").forEach((el) => {
     const v = map[el.dataset.metric];
     if (typeof v === "number") el.textContent = pct(v);
   });
+  const leader = document.querySelector("[data-leader]");
+  if (leader) leader.textContent = best.model;
 }
 
 document.addEventListener("DOMContentLoaded", async () => {

index.html

@@ -46,20 +46,21 @@
 
   <section class="metrics reveal">
     <div class="metric">
-      <div class="label">Invocation</div>
-      <div class="value" data-metric="overall">—</div>
-      <div class="note">Right tool, semantically correct arguments. Expressions are evaluated — <code>45+12</code> equals <code>45 + 12</code>.</div>
+      <div class="label">Overall score</div>
+      <div class="value" data-metric="overall">64.0%</div>
+      <div class="note">Passed cases over all cases, three protocols combined. Each entry reports a Wilson 95% interval.</div>
     </div>
     <div class="metric">
-      <div class="label">Abstention</div>
-      <div class="value" data-metric="abstain">—</div>
-      <div class="note">No call when no tool applies: knowledge questions, absent capabilities, explicit no-tool instructions.</div>
+      <div class="label">Correct call</div>
+      <div class="value" data-metric="call">58.6%</div>
+      <div class="note">Right tool with semantically correct arguments when a tool applies. Expressions are evaluated — <code>45+12</code> equals <code>45 + 12</code>.</div>
     </div>
     <div class="metric">
-      <div class="label">Clarification</div>
-      <div class="value" data-metric="call">—</div>
-      <div class="note">A required argument is missing. Asking the user passes; guessing an argument fails.</div>
+      <div class="label">Abstention</div>
+      <div class="value" data-metric="abstain">75.9%</div>
+      <div class="note">No call emitted when no tool applies: knowledge questions, absent capabilities, explicit no-tool instructions.</div>
     </div>
+    <p class="metric-leader">Current leader: <b data-leader>HardcoreAI/Nova-1.1-0.8B</b> — the figures above are its run.</p>
   </section>
 
   <section class="grid reveal">
@@ -85,7 +86,7 @@
     <div class="board" data-board>
       <div class="board-head">
         <h2>ToolUse.1 — Leaderboard</h2>
-        <p class="stamp" data-board-stamp>loading results…</p>
+        <p class="stamp" data-board-stamp>3 entries · v1.0.0 · updated 2026-10-17</p>
       </div>
       <table>
         <thead>

leaderboard.html

@@ -41,7 +41,7 @@
       <div class="board-head">
         <h2>All submissions</h2>
         <div class="board-tools">
-          <p class="stamp" data-board-stamp>loading results…</p>
+          <p class="stamp" data-board-stamp>3 entries · v1.0.0 · updated 2026-10-17</p>
         </div>
       </div>
       <table>

styles.css

@@ -267,6 +267,8 @@ pre code { background: none; border: none; padding: 0; }
 .metric .value { font-family: var(--mono); font-size: 1.7rem; font-variant-numeric: tabular-nums; font-weight: 600; }
 .metric .note { margin-top: 0.35rem; font-size: 0.78rem; color: var(--dim); }
 .metric .note code { font-size: 0.78em; }
+.metric-leader { grid-column: 1 / -1; margin: 0.2rem 0 0; color: var(--dim); font-size: 0.8rem; }
+.metric-leader b { color: var(--text); font-weight: 500; }
 
 /* ------------------------------------------------------------------ footer */
 footer { border-top: 1px solid var(--line); padding: 2rem 0 3rem; color: var(--dim); font-size: 0.78rem; }