Connect Epicure Explorer to FlavourBench

README.md

@@ -8,7 +8,7 @@ sdk_version: 5.7.1
 app_file: app.py
 pinned: false
 license: cc-by-4.0
-short_description: Operators over the three Epicure ingredient embeddings
+short_description: Operators behind the FlavourBench culinary benchmark
 models:
   - Kaikaku/epicure-cooc
   - Kaikaku/epicure-core
@@ -19,6 +19,12 @@ datasets:
 
 # Epicure Explorer
 
+> **Epicure powers [FlavourBench](https://github.com/josefchen/flavourbench), the executable
+> culinary reasoning benchmark for frontier language models.** The benchmark turns versioned
+> Epicure operations into answer keys, evaluates 20 current endpoints on 32 tasks, and measures the
+> same model-task pairs with and without Epicure. [Read the paper](https://github.com/josefchen/flavourbench/blob/main/paper/build/flavourbench.pdf)
+> or [reproduce the leaderboard](https://github.com/josefchen/flavourbench#reproduce-the-result).
+
 Interactive chef-facing demo of the three Epicure sibling ingredient embeddings (Cooc, Core, Chem). Three operator tabs:
 
 1. **Pairings**: top-K cosine neighbours plus the closest emergent mode for any of 1,790 ingredients.

app.py

@@ -1742,6 +1742,51 @@ CUSTOM_CSS = f"""
 .gradio-container {{ max-width: 1280px !important; background: #ffffff !important; }}
 footer {{ visibility: hidden; }}
 
+/* FlavourBench research funnel */
+.benchmark-callout {{
+  align-items: center;
+  background: #f8fafc !important;
+  border: 1px solid #dbe4ec;
+  border-left: 4px solid {KAIKAKU_ACCENT};
+  display: grid;
+  gap: 20px;
+  grid-template-columns: minmax(0, 1fr) auto;
+  margin: 18px 0 14px;
+  padding: 18px 20px;
+}}
+.benchmark-callout-kicker {{
+  color: {KAIKAKU_ACCENT} !important;
+  font-size: 0.72rem;
+  font-weight: 800;
+  letter-spacing: 0.09em;
+  margin-bottom: 5px;
+  text-transform: uppercase;
+}}
+.benchmark-callout strong {{
+  color: #0f172a !important;
+  display: block;
+  font-size: 1.05rem;
+  line-height: 1.35;
+}}
+.benchmark-callout p {{
+  color: #475569 !important;
+  line-height: 1.45;
+  margin: 5px 0 0;
+  max-width: 820px;
+}}
+.benchmark-callout-links {{ display: flex; flex-wrap: wrap; gap: 8px; justify-content: flex-end; }}
+.benchmark-callout-links a {{
+  border: 1px solid {KAIKAKU_ACCENT};
+  color: {KAIKAKU_ACCENT} !important;
+  font-weight: 700;
+  padding: 9px 12px;
+  text-decoration: none !important;
+}}
+.benchmark-callout-links a:first-child {{
+  background: {KAIKAKU_ACCENT} !important;
+  color: #ffffff !important;
+}}
+
 /* Labels: plain dark text, no chip background */
 .gradio-container label, .gradio-container .label,
 .gradio-container [data-testid="block-label"], .gradio-container .block-label,
@@ -1808,6 +1853,8 @@ footer {{ visibility: hidden; }}
 
 /* Mobile responsive: stack the three sibling result tables on narrow screens */
 @media (max-width: 768px) {{
+  .benchmark-callout {{ grid-template-columns: 1fr; }}
+  .benchmark-callout-links {{ justify-content: flex-start; }}
   .gradio-container .compare-row {{ flex-direction: column !important; }}
   .gradio-container .compare-row > * {{ width: 100% !important; min-width: 0 !important; }}
   .spectrum-cell {{ padding: 6px 8px; }}
@@ -1844,6 +1891,21 @@ SPECTRUM_BAR = """
 </div>
 """
 
+BENCHMARK_CALLOUT = """
+<aside class="benchmark-callout">
+  <div>
+    <div class="benchmark-callout-kicker">Frontier culinary reasoning benchmark</div>
+    <strong>Epicure is the operator layer behind FlavourBench.</strong>
+    <p>See 20 current language-model endpoints on 32 executable culinary tasks, plus 640 matched
+    Model only / Model + Epicure pairs. No human or model judge scores the leaderboard.</p>
+  </div>
+  <div class="benchmark-callout-links">
+    <a href="https://github.com/josefchen/flavourbench" target="_blank" rel="noopener">Open FlavourBench</a>
+    <a href="https://github.com/josefchen/flavourbench/blob/main/paper/build/flavourbench.pdf" target="_blank" rel="noopener">Read paper</a>
+  </div>
+</aside>
+"""
+
 # ===== Pre-rendered killer demo on landing =====
 
 _DEFAULT_BASKET = ["chicken","lemon","garlic"]
@@ -1860,6 +1922,7 @@ Three sibling ingredient embeddings from [arXiv:2605.22391](https://arxiv.org/ab
 1,790 canonical ingredients across 7 languages; 300-D Metapath2Vec; controlled chemistry-vs-recipe-context spectrum.
 """
     )
+    gr.HTML(BENCHMARK_CALLOUT)
     gr.HTML(SPECTRUM_BAR)
 
     with gr.Tabs():
@@ -2275,7 +2338,7 @@ Three sibling ingredient embeddings from [arXiv:2605.22391](https://arxiv.org/ab
     gr.Markdown(
         """---
 **Cite:** Radzikowski and Chen, 2026, *Epicure: Navigating the Emergent Geometry of Food Ingredient Embeddings*, [arXiv:2605.22391](https://arxiv.org/abs/2605.22391).
-Models: [epicure-cooc](https://huggingface.co/Kaikaku/epicure-cooc) · [epicure-core](https://huggingface.co/Kaikaku/epicure-core) · [epicure-chem](https://huggingface.co/Kaikaku/epicure-chem) · [dataset](https://huggingface.co/datasets/Kaikaku/epicure-corpus-resources) · [API](/?view=api)
+Benchmark: [FlavourBench](https://github.com/josefchen/flavourbench) · Models: [epicure-cooc](https://huggingface.co/Kaikaku/epicure-cooc) · [epicure-core](https://huggingface.co/Kaikaku/epicure-core) · [epicure-chem](https://huggingface.co/Kaikaku/epicure-chem) · [dataset](https://huggingface.co/datasets/Kaikaku/epicure-corpus-resources) · [API](/?view=api)
 """
     )