Update index.html

index.html

@@ -771,7 +771,7 @@
     ];
 
     const BENCHMARKS = [
-      { key: 'avg_4', label: 'Avg' },
+      { key: 'avg', label: 'Avg' },
       { key: 'arc_easy', label: 'ARC-Easy' },
       { key: 'arc_challenge', label: 'ARC-Challenge' },
       { key: 'hellaswag', label: 'HellaSwag' },
@@ -779,7 +779,7 @@
     ];
 
     const BENCHMARK_NAMES = {
-      avg_4: 'Average',
+      avg: 'Average',
       arc_easy: 'ARC-Easy',
       arc_challenge: 'ARC-Challenge',
       hellaswag: 'HellaSwag',
@@ -787,7 +787,7 @@
     };
 
     const MIN_PLOT_PARAMS = 500000;
-    let activeBenchmark = 'avg_4';
+    let activeBenchmark = 'avg';
     let activeSearch = '';
     let activeOrg = 'all';
     let activeOrgSearch = '';
@@ -796,6 +796,8 @@
     let lineEmphasis = false;
     let activeTheme = 'light';
 
+    const META_KEYS = new Set(['name', 'org', 'params', 'url']);
+
     const fmtParams = (n) => {
       if (n >= 1e9) return `${(n / 1e9).toFixed(2)}B`;
       if (n >= 1e6) return `${(n / 1e6).toFixed(2)}M`;
@@ -805,6 +807,17 @@
 
     const toPct = (v) => v * 100;
 
+    function getMetricValue(model, key) {
+      if (key === 'avg') {
+        const values = Object.entries(model)
+          .filter(([k, v]) => typeof v === 'number' && Number.isFinite(v) && !META_KEYS.has(k) && !k.startsWith('avg'))
+          .map(([, v]) => v);
+        if (!values.length) return null;
+        return values.reduce((a, b) => a + b, 0) / values.length;
+      }
+      return model[key];
+    }
+
     function linearRegression(points) {
       const n = points.length;
       let sumX = 0, sumY = 0, sumXY = 0, sumXX = 0;
@@ -846,13 +859,14 @@
     }
 
     function getEligibleModels() {
-      return MODELS.filter(m =>
-        Number.isFinite(m.params) &&
-        m.params >= MIN_PLOT_PARAMS &&
-        m[activeBenchmark] !== null &&
-        m[activeBenchmark] !== undefined &&
-        Number.isFinite(m[activeBenchmark])
-      );
+      return MODELS.filter(m => {
+        const score = getMetricValue(m, activeBenchmark);
+        return Number.isFinite(m.params) &&
+          m.params >= MIN_PLOT_PARAMS &&
+          score !== null &&
+          score !== undefined &&
+          Number.isFinite(score);
+      });
     }
 
     function getVisibleModels() {
@@ -954,28 +968,34 @@
       const fitModels = getEligibleModels();
       const visibleModels = getVisibleModels();
 
-      const fitData = fitModels.map(m => ({
-        name: m.name,
-        org: m.org,
-        params: m.params,
-        score: toPct(m[activeBenchmark]),
-        url: m.url,
-        x: Math.log10(m.params),
-        y: toPct(m[activeBenchmark])
-      }));
-
-      const data = visibleModels.map(m => ({
-        name: m.name,
-        org: m.org,
-        params: m.params,
-        score: toPct(m[activeBenchmark]),
-        url: m.url,
-        x: Math.log10(m.params),
-        y: toPct(m[activeBenchmark])
-      }));
+      const fitData = fitModels.map(m => {
+        const score = getMetricValue(m, activeBenchmark);
+        return {
+          name: m.name,
+          org: m.org,
+          params: m.params,
+          score: toPct(score),
+          url: m.url,
+          x: Math.log10(m.params),
+          y: toPct(score)
+        };
+      });
+
+      const data = visibleModels.map(m => {
+        const score = getMetricValue(m, activeBenchmark);
+        return {
+          name: m.name,
+          org: m.org,
+          params: m.params,
+          score: toPct(score),
+          url: m.url,
+          x: Math.log10(m.params),
+          y: toPct(score)
+        };
+      });
 
       const chartTitleMap = {
-        avg_4: 'Average score vs log parameters',
+        avg: 'Average score vs log parameters',
         arc_easy: 'ARC-Easy vs log parameters',
         arc_challenge: 'ARC-Challenge vs log parameters',
         hellaswag: 'HellaSwag vs log parameters',
@@ -1220,4 +1240,4 @@
     render();
   </script>
 </body>
-</html>
+</html>