zsss0316 commited on
Commit
ac02d4f
·
verified ·
1 Parent(s): 8c05cbf

Update index.html

Browse files
Files changed (1) hide show
  1. index.html +21 -18
index.html CHANGED
@@ -557,13 +557,15 @@
557
  </div>
558
  </header>
559
 
 
560
  <div class="tab-bar">
561
- <button class="tab active" data-idx="0">Compilation Success</button>
562
- <button class="tab" data-idx="1">Partial Replication</button>
563
  <button class="tab" data-idx="2">Coefficient Direction</button>
564
  </div>
565
 
566
- <div class="panel show" id="panel-compile">
 
567
  <div class="leaderboard-card">
568
  <div class="leaderboard-header">
569
  <h2>
@@ -577,11 +579,11 @@
577
  <span>Model</span>
578
  <span class="col-score">Score</span>
579
  </div>
580
- <div id="rows-compile"><div class="loading"><div class="loading-spinner"></div>Loading rankings…</div></div>
581
  </div>
582
  </div>
583
 
584
- <div class="panel" id="panel-rep">
585
  <div class="leaderboard-card">
586
  <div class="leaderboard-header">
587
  <h2>
@@ -595,7 +597,7 @@
595
  <span>Model</span>
596
  <span class="col-score">Score</span>
597
  </div>
598
- <div id="rows-rep"><div class="loading"><div class="loading-spinner"></div>Loading rankings…</div></div>
599
  </div>
600
  </div>
601
 
@@ -625,14 +627,15 @@
625
  Metric Definitions
626
  </h3>
627
  <ul class="metric-list">
628
- <li><strong>Compilation Success:</strong> Generated code runs without runtime errors.</li>
629
- <li><strong>Partial Replication:</strong> Target treatment coefficient reproduced within 10% relative error.</li>
630
- <li><strong>Correct Coefficient Direction:</strong> Correct sign of estimated coefficients.</li>
631
  </ul>
632
- <p class="metric-note">
633
- Native Hugging Face Community Evals leaderboard pending official allowlist.
 
634
  Dataset: <a href="https://huggingface.co/datasets/CamoAiLab/Infernet" target="_blank">CamoAiLab/Infernet</a>
635
- </p>
636
  </div>
637
  </div>
638
 
@@ -714,9 +717,8 @@ function buildRows(data, sortKey, containerId) {
714
  <div class="model-info">
715
  <div class="model-name" title="${modelId}">${modelId}</div>
716
  <div class="model-meta">
717
- <span class="status-badge ${isTop ? "verified" : "self-reported"}">
718
- ${isTop ? " verified" : "self-reported"}
719
- </span>
720
  </div>
721
  </div>
722
  </div>
@@ -760,15 +762,16 @@ async function loadData() {
760
  throw new Error("CSV parsed zero rows");
761
  }
762
 
763
- buildRows(data, "Compilation Success (%)", "rows-compile");
764
  buildRows(data, "Partial Replication (%)", "rows-rep");
 
765
  buildRows(data, "Correct Coefficient Direction (%)", "rows-dir");
766
  } catch (err) {
767
  console.error("load csv error", err);
768
  const banner = document.getElementById("error-banner");
769
  banner.textContent = `Failed to load ranking data: ${err.message}`;
770
  banner.classList.remove("hidden");
771
- ["rows-compile", "rows-rep", "rows-dir"].forEach(id => {
772
  document.getElementById(id).innerHTML = '<div class="loading">Unable to load data</div>';
773
  });
774
  }
@@ -777,4 +780,4 @@ async function loadData() {
777
  loadData();
778
  </script>
779
  </body>
780
- </html>
 
557
  </div>
558
  </header>
559
 
560
+ <!-- Tab顺序修改:0 Partial Replication(默认active),1 Compilation Success,2 Coefficient Direction -->
561
  <div class="tab-bar">
562
+ <button class="tab active" data-idx="0">Partial Replication</button>
563
+ <button class="tab" data-idx="1">Compilation Success</button>
564
  <button class="tab" data-idx="2">Coefficient Direction</button>
565
  </div>
566
 
567
+ <!-- panel顺序同步调换,第一个show是Partial Replication -->
568
+ <div class="panel show" id="panel-rep">
569
  <div class="leaderboard-card">
570
  <div class="leaderboard-header">
571
  <h2>
 
579
  <span>Model</span>
580
  <span class="col-score">Score</span>
581
  </div>
582
+ <div id="rows-rep"><div class="loading"><div class="loading-spinner"></div>Loading rankings…</div></div>
583
  </div>
584
  </div>
585
 
586
+ <div class="panel" id="panel-compile">
587
  <div class="leaderboard-card">
588
  <div class="leaderboard-header">
589
  <h2>
 
597
  <span>Model</span>
598
  <span class="col-score">Score</span>
599
  </div>
600
+ <div id="rows-compile"><div class="loading"><div class="loading-spinner"></div>Loading rankings…</div></div>
601
  </div>
602
  </div>
603
 
 
627
  Metric Definitions
628
  </h3>
629
  <ul class="metric-list">
630
+ <li><strong>Compilation Success:</strong> Generated econometric code executes completely without runtime or syntax errors.</li>
631
+ <li><strong>Partial Replication:</strong> The target treatment coefficient can be reproduced within a 10% relative error threshold.</li>
632
+ <li><strong>Correct Coefficient Direction:</strong> The sign (positive/negative) of the treatment‑effect coefficient matches ground‑truth results.</li>
633
  </ul>
634
+ <div class="metric-note">
635
+ <em>Note:</em> Codex serves as a strong code‑specialized upper‑bound baseline with high overall metrics across all three evaluation dimensions. However, it only supports one‑shot code generation without agent‑level interactive planning or multi‑round revision capabilities, and its public API is no longer available. MetricsAI outperforms vanilla‑LLM and general‑purpose‑agent baselines for interactive real‑world econometric‑research workflows.
636
+ <br><br>
637
  Dataset: <a href="https://huggingface.co/datasets/CamoAiLab/Infernet" target="_blank">CamoAiLab/Infernet</a>
638
+ </div>
639
  </div>
640
  </div>
641
 
 
717
  <div class="model-info">
718
  <div class="model-name" title="${modelId}">${modelId}</div>
719
  <div class="model-meta">
720
+ <!-- 全部硬编码 verified;后续csv增加verified字段后,这里改成读取row["verified"]做判断 -->
721
+ <span class="status-badge verified">✓ verified</span>
 
722
  </div>
723
  </div>
724
  </div>
 
762
  throw new Error("CSV parsed zero rows");
763
  }
764
 
765
+ // 对应panel‑idx 0:rep,1:compile,2:dir
766
  buildRows(data, "Partial Replication (%)", "rows-rep");
767
+ buildRows(data, "Compilation Success (%)", "rows-compile");
768
  buildRows(data, "Correct Coefficient Direction (%)", "rows-dir");
769
  } catch (err) {
770
  console.error("load csv error", err);
771
  const banner = document.getElementById("error-banner");
772
  banner.textContent = `Failed to load ranking data: ${err.message}`;
773
  banner.classList.remove("hidden");
774
+ ["rows-rep","rows-compile", "rows-dir"].forEach(id => {
775
  document.getElementById(id).innerHTML = '<div class="loading">Unable to load data</div>';
776
  });
777
  }
 
780
  loadData();
781
  </script>
782
  </body>
783
+ </html>