en_version & kr_version

app.py

@@ -2,9 +2,10 @@ import warnings
 warnings.filterwarnings("ignore")
 
 import gradio as gr
-from tabs.leaderboard_v1 import create_leaderboard_v2_interface
+from tabs.leaderboard_v1_kr import create_leaderboard_v2_interface as leaderboard_kr
+from tabs.leaderboard_v1_en import create_leaderboard_v2_interface as leaderboard_en
 
-# 다크 모드에서 프로즈/마크다운 텍스트를 확실히 밝게 고정하는 CSS 보정
+# 다크 모드 텍스트 색상 보정
 FIX_DARK_TEXT_CSS = """
 html.dark .gr-prose, 
 html.dark .gr-prose p, 
@@ -17,14 +18,36 @@ html.dark .gr-markdown * {
 """
 
 def create_app():
-    # 권장: 명시적인 테마 객체 사용 (Default, Soft, Origin 등)
-    theme = gr.themes.Default()  # 필요 시 gr.themes.Origin() 등으로 변경
+    theme = gr.themes.Default()
 
     with gr.Blocks(theme=theme, css=FIX_DARK_TEXT_CSS) as app:
-        create_leaderboard_v2_interface()
+        # 🔹 왼쪽 상단에 언어 전환 버튼
+        with gr.Row():
+            lang_btn = gr.Button("🌍 English", scale=0, elem_id="lang-toggle-btn")
+
+        # 🔹 기본은 한국어 UI
+        with gr.Column(visible=True) as kr_view:
+            leaderboard_kr()
+
+        # 🔹 영어 UI는 숨김
+        with gr.Column(visible=False) as en_view:
+            leaderboard_en()
+
+        # 🔹 버튼 클릭 시 토글
+        def toggle_language(current_label):
+            if "English" in current_label:
+                return "🇰🇷 Korean", gr.update(visible=False), gr.update(visible=True)
+            else:
+                return "🌍 English", gr.update(visible=True), gr.update(visible=False)
+
+        lang_btn.click(
+            toggle_language,
+            inputs=[lang_btn],
+            outputs=[lang_btn, kr_view, en_view],
+        )
+
     return app
 
-demo = create_app()
 
-# Spaces/Gradio5에서 SSR이 꼬이면 일단 꺼서 확인
-demo.launch(ssr_mode=False)
+demo = create_app()
+demo.launch(ssr_mode=False)

tabs/{leaderboard_v1.py → leaderboard_v1_kr.py}

@@ -1103,7 +1103,7 @@ def create_leaderboard_v2_tab():
                 <path d="M15 7h3a5 5 0 0 1 5 5 5 5 0 0 1-5 5h-3m-6 0H6a5 5 0 0 1-5-5 5 5 0 0 1 5-5h3"/>
                 <line x1="8" y1="12" x2="16" y2="12"/>
             </svg>
-            <span>Blog</span>
+            <span>블로그</span>
         </a>
         <a href="https://github.com/Hugging-Face-KREW/Ko-AgentBench" target="_blank" rel="noopener noreferrer" class="hero-action-button">
             <svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round">
@@ -1118,7 +1118,7 @@ def create_leaderboard_v2_tab():
                 <polyline points="7 10 12 15 17 10"/>
                 <line x1="12" y1="15" x2="12" y2="3"/>
             </svg>
-            <span>Dataset</span>
+            <span>데이터셋</span>
         </a>
         <a href="https://github.com/Hugging-Face-KREW/Ko-AgentBench/blob/main/evaluate_model_run.py#L55" target="_blank" rel="noopener noreferrer" class="hero-action-button">
             <svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2" stroke-linecap="round" stroke-linejoin="round">
@@ -1127,7 +1127,7 @@ def create_leaderboard_v2_tab():
                 <path d="M12 17V7"/>
                 <path d="M17 17v-3"/>
             </svg>
-            <span>Metrics</span>
+            <span>평가 지표</span>
         </a>
     </div>
     """)
@@ -1141,7 +1141,7 @@ def create_leaderboard_v2_tab():
         <p class="section-lead" style="text-align: center; margin: 0 auto 24px auto; max-width: 720px; line-height: 1.7; word-break: keep-all;">단순 도구 호출부터 장기적 맥락 능력, 강건성 처리 능력까지 에이전트의 능력을 7단계로 입체적으로 분석하였습니다.</p>
         <div class="phase-grid">
             <div class="phase-card">
-                <h3>Single-Turn</h3>
+                <h3>단일 턴</h3>
                 <div class="phase-chart" style="--progress:80%;">
                     <span>80%</span>
                 </div>
@@ -1154,7 +1154,7 @@ def create_leaderboard_v2_tab():
                 </ul>
             </div>
             <div class="phase-card">
-                <h3>Multi-Turn</h3>
+                <h3>다중 턴</h3>
                 <div class="phase-chart" style="--progress:20%;">
                     <span>20%</span>
                 </div>
@@ -1944,8 +1944,8 @@ def create_leaderboard_v2_tab():
     with gr.Column(elem_classes=["domain-selector-container"], elem_id="task-level-selector"):
         gr.HTML("""
         <div class="domain-header">
-            <h2 class="domain-title" style="color: white;">🧠 Select Task Level</h2>
-            <p class="domain-subtitle" style="color: white;">Ko-AgentBench의 ALL · L1~L7 단계별 에이전트 성능을 손쉽게 비교하세요.</p>
+            <h2 class="domain-title" style="color: white;">🧠 태스크 레벨 선택</h2>
+            <p class="domain-subtitle" style="color: white;">Ko-AgentBench의 ALL · 모델별 에이전트 수행 능력을 단계별로 측정해보세요.</p>
         </div>
         """)
         domain_filter = gr.Radio(
@@ -1961,14 +1961,14 @@ def create_leaderboard_v2_tab():
     with gr.Column(elem_classes=["domain-selector-container", "filters-sorting-container"], elem_id="filters-sorting-container"):
         gr.HTML("""
         <div class="domain-header">
-            <h2 class="domain-title" style="color: white;">🔍 Filters & Sorting</h2>
+            <h2 class="domain-title" style="color: white;">🔍 필터 및 정렬</h2>
             <p class="domain-subtitle" style="color: white;">모델 접근 방식과 정렬 순서를 선택해 맞춤 뷰를 구성하세요.</p>
         </div>
         """)
         with gr.Row(elem_classes=["filters-sorting-row"]):
             with gr.Column(scale=1, elem_classes=["filter-group"]):
                 with gr.Row(elem_classes=["filter-group-row"]):
-                    gr.HTML("<span class='filter-group-label' style='color: white;'>Model Access</span>")
+                    gr.HTML("<span class='filter-group-label' style='color: white;'>모델 접근</span>")
                     model_type_filter = gr.Radio(
                         choices=["All", "OSS", "API"],
                         value="All",
@@ -1978,7 +1978,7 @@ def create_leaderboard_v2_tab():
                     )
             with gr.Column(scale=1, elem_classes=["filter-group"]):
                 with gr.Row(elem_classes=["filter-group-row"]):
-                    gr.HTML("<span class='filter-group-label' style='color: white;'>Sort Order</span>")
+                    gr.HTML("<span class='filter-group-label' style='color: white;'>정렬 순서</span>")
                     sort_order = gr.Radio(
                         choices=["Descending", "Ascending"],
                         value="Descending",
@@ -2000,16 +2000,16 @@ def create_leaderboard_v2_tab():
     gr.HTML("""
     <div class="domain-selector-container domain-performance-container">
         <div class="domain-header">
-            <h2 class="domain-title" style="color: white;">Core Capability Radar</h2>
-            <p class="domain-subtitle" style="color: white;">Track six essential pillars: Success, Execution, Reasoning, Robustness, Efficiency, and Call Validity.</p>
+            <h2 class="domain-title" style="color: white;">핵심 역량 레이더</h2>
+            <p class="domain-subtitle" style="color: white;">6가지 필수 핵심 요소(성공, 실행, 추론, 강건성, 효율성, 호출 유효성)를 추적합니다.</p>
         </div>
     """)
     
     with gr.Column(elem_classes=["domain-selector-container", "model-selector-container"], elem_id="radar-model-selector"):
         gr.HTML("""
         <div class="domain-header">
-            <h2 class="domain-title" style="color: white;">🎯 Select Models for Comparison</h2>
-            <p class="domain-subtitle" style="color: white;">Choose up to 5 models to map on the capability radar.</p>
+            <h2 class="domain-title" style="color: white;">🎯 비교할 모델 선택</h2>
+            <p class="domain-subtitle" style="color: white;">역량 레이더에 표시할 모델을 최대 5개까지 선택하세요.</p>
         </div>
         """)
         model_selector = gr.Dropdown(
@@ -2040,16 +2040,16 @@ def create_leaderboard_v2_tab():
     gr.HTML("""
     <div class="domain-selector-container domain-performance-container level-metrics-wrapper">
         <div class="domain-header">
-            <h2 class="domain-title" style="color: white;">Level-Specific Metric Spotlight</h2>
-            <p class="domain-subtitle" style="color: white;">Dive deeper into each Ko-AgentBench stage and compare model scores across its unique evaluation metrics.</p>
+            <h2 class="domain-title" style="color: white;">레벨별 상세 지표</h2>
+            <p class="domain-subtitle" style="color: white;">각 Ko-AgentBench 단계별 고유 평가 지표를 통해 모델 점수를 비교하고 더 자세히 살펴보세요.</p>
         </div>
     """)
     
     with gr.Column(elem_classes=["domain-selector-container", "level-selector-container"], elem_id="level-selector-box"):
         gr.HTML("""
         <div class="domain-header">
-            <h2 class="domain-title" style="color: white;">🧭 Select Task Level and Models</h2>
-            <p class="domain-subtitle" style="color: white;">Choose a level and up to 5 models to explore their detailed SR-driven metrics.</p>
+            <h2 class="domain-title" style="color: white;">🧭 태스크 레벨 및 모델 선택</h2>
+            <p class="domain-subtitle" style="color: white;">레벨과 모델(최대 5개)을 선택하여 상세 SR(성공률) 기반 지표를 탐색하세요.</p>
         </div>
         """)
         level_metric_selector = gr.Dropdown(
@@ -2086,8 +2086,8 @@ def create_leaderboard_v2_tab():
     gr.HTML("""
     <div class="domain-selector-container domain-performance-container heatmap-wrapper">
         <div class="domain-header">
-            <h2 class="domain-title" style="color: white;">Comprehensive Performance Heatmap</h2>
-            <p class="domain-subtitle" style="color: white;">View Ko-AgentBench SR scores across L1~L7 for each model in a single glance.</p>
+            <h2 class="domain-title" style="color: white;">종합 성능 히트맵</h2>
+            <p class="domain-subtitle" style="color: white;">각 모델의 L1~L7 Ko-AgentBench SR(성공률) 점수를 한눈에 보세요.</p>
         </div>
         <div class="chart-container heatmap-chart-container">
     """)
@@ -2452,8 +2452,8 @@ def create_leaderboard_v2_tab():
     gr.HTML("""
     <div class="domain-selector-container performance-card-container">
         <div class="domain-header">
-            <h2 class="domain-title" style="color: white;">Model Performance Card</h2>
-            <p class="domain-subtitle" style="color: white;">Comprehensive performance card for any model - perfect for presentations and reports</p>
+            <h2 class="domain-title" style="color: white;">모델 성능 카드</h2>
+            <p class="domain-subtitle" style="color: white;">모든 모델의 종합 성능 카드로, 발표 및 보고서에 적합합니다.</p>
         </div>
         <div class="performance-card-content">
     """)
@@ -3182,7 +3182,17 @@ def create_domain_radar_chart(df, selected_models=None, max_models=5):
         height=800,
         width=900,
         margin=dict(t=30, b=50, l=10, r=10),
-        autosize=True
+        autosize=True,
+        annotations=[
+            dict(
+                text="Galileo Agent Leaderboard",
+                xref="paper", yref="paper",
+                x=0.98, y=0.02,
+                xanchor='right', yanchor='bottom',
+                font=dict(size=10, color='#64748B'),
+                showarrow=False
+            )
+        ]
     )
     
     return fig