Update app-backup.py

app-backup.py

@@ -10,18 +10,28 @@ HF_TOKEN = os.getenv("HF_TOKEN")
 target_models = {
     "openfree/flux-lora-korea-palace": "https://huggingface.co/openfree/flux-lora-korea-palace",
     "seawolf2357/hanbok": "https://huggingface.co/seawolf2357/hanbok",
+    "seawolf2357/ntower": "https://huggingface.co/seawolf2357/ntower",
+
     "LGAI-EXAONE/EXAONE-3.5-32B-Instruct": "https://huggingface.co/LGAI-EXAONE/EXAONE-3.5-32B-Instruct",
     "LGAI-EXAONE/EXAONE-3.5-2.4B-Instruct": "https://huggingface.co/LGAI-EXAONE/EXAONE-3.5-2.4B-Instruct",
     "LGAI-EXAONE/EXAONE-3.5-7.8B-Instruct": "https://huggingface.co/LGAI-EXAONE/EXAONE-3.5-7.8B-Instruct",
     "ginipick/flux-lora-eric-cat": "https://huggingface.co/ginipick/flux-lora-eric-cat",
     "seawolf2357/flux-lora-car-rolls-royce": "https://huggingface.co/seawolf2357/flux-lora-car-rolls-royce",
     "moreh/Llama-3-Motif-102B-Instruct": "https://huggingface.co/moreh/Llama-3-Motif-102B-Instruct",
-
-    
+    "OnomaAIResearch/Illustrious-xl-early-release-v0": "https://huggingface.co/OnomaAIResearch/Illustrious-xl-early-release-v0",
+    "upstage/solar-pro-preview-instruct": "https://huggingface.co/upstage/solar-pro-preview-instruct",   
     "NCSOFT/VARCO-VISION-14B": "https://huggingface.co/NCSOFT/VARCO-VISION-14B",
     "NCSOFT/Llama-VARCO-8B-Instruct": "https://huggingface.co/NCSOFT/Llama-VARCO-8B-Instruct",
     "NCSOFT/VARCO-VISION-14B-HF": "https://huggingface.co/NCSOFT/VARCO-VISION-14B-HF",
-    
+    "KAERI-MLP/llama-3.1-Korean-AtomicGPT-Bllossom-8B": "https://huggingface.co/KAERI-MLP/llama-3.1-Korean-AtomicGPT-Bllossom-8B",
+    "dnotitia/Llama-DNA-1.0-8B-Instruct": "https://huggingface.co/dnotitia/Llama-DNA-1.0-8B-Instruct",    
+    "Bllossom/llama-3.2-Korean-Bllossom-3B": "https://huggingface.co/Bllossom/llama-3.2-Korean-Bllossom-3B",
+
+    "unidocs/llama-3.1-8b-komedic-instruct": "https://huggingface.co/unidocs/llama-3.1-8b-komedic-instruct",
+    "unidocs/llama-3.2-3b-komedic-instruct": "https://huggingface.co/unidocs/llama-3.2-3b-komedic-instruct",    
+    "etri-lirs/eagle-3b-preview": "https://huggingface.co/etri-lirs/eagle-3b-preview",
+    "kakaobrain/kogpt": "https://huggingface.co/kakaobrain/kogpt", 
+
     "Saxo/Linkbricks-Horizon-AI-Korean-Gemma-2-sft-dpo-27B": "https://huggingface.co/Saxo/Linkbricks-Horizon-AI-Korean-Gemma-2-sft-dpo-27B",
     "AALF/gemma-2-27b-it-SimPO-37K": "https://huggingface.co/AALF/gemma-2-27b-it-SimPO-37K",
     "nbeerbower/mistral-nemo-wissenschaft-12B": "https://huggingface.co/nbeerbower/mistral-nemo-wissenschaft-12B",
@@ -319,28 +329,34 @@ def get_korea_models():
         print(f"Error fetching Korea models: {str(e)}")
         return []
 
-def get_all_models(limit=1000):
+def get_all_models(limit=3000):
     """모든 모델과 Korea 관련 모델 가져오기"""
     all_models = []
+    page_size = 1000  # API의 한 번 요청당 최대 크기
     
-    # 1. 일반 모델 리스트 가져오기
-    params = {
-        "limit": limit,
-        "full": "True",
-        "config": "True"
-    }
-    
-    response = requests.get(
-        "https://huggingface.co/api/models",
-        headers={'Accept': 'application/json'},
-        params=params
-    )
-    
-    if response.status_code == 200:
-        all_models.extend(response.json())
-        print(f"Fetched {len(all_models)} general models")
+    # 여러 페이지에 걸쳐 데이터 수집
+    for offset in range(0, limit, page_size):
+        params = {
+            'limit': min(page_size, limit - offset),
+            'full': 'True',
+            'config': 'True',
+            'offset': offset
+        }
+        
+        response = requests.get(
+            "https://huggingface.co/api/models",
+            headers={'Accept': 'application/json'},
+            params=params
+        )
+        
+        if response.status_code == 200:
+            all_models.extend(response.json())
+            print(f"Fetched models {offset+1} to {offset+len(response.json())}")
+        else:
+            print(f"Failed to fetch models at offset {offset}: {response.status_code}")
+            break
     
-    # 2. Korea 검색 결과 가져오기
+    # Korea 검색 결과도 동일하게 확장
     korea_params = {
         "search": "korea",
         "full": "True",
@@ -365,59 +381,70 @@ def get_all_models(limit=1000):
                 all_models.append(model)
                 existing_ids.add(model.get('id', ''))
     
-    # 3. Korean 검색 결과 가져오기
-    korean_params = {
-        "search": "korean",
-        "full": "True",
-        "config": "True",
-        "limit": limit
-    }
-    
-    korean_response = requests.get(
-        "https://huggingface.co/api/models",
-        headers={'Accept': 'application/json'},
-        params=korean_params
-    )
-    
-    if korean_response.status_code == 200:
-        korean_models = korean_response.json()
-        print(f"Fetched {len(korean_models)} Korean-related models")
-        
-        # 중복 제거하면서 Korean 모델 추가
-        for model in korean_models:
-            if model.get('id', '') not in existing_ids:
-                all_models.append(model)
-                existing_ids.add(model.get('id', ''))
-    
     print(f"Total unique models: {len(all_models)}")
     return all_models[:limit]
 
 def get_models_data(progress=gr.Progress()):
-    """모델 데이터 가져오기"""
+    def calculate_rank(model_id, all_global_models, korea_models):
+        # 글로벌 순위 확인
+        global_rank = next((idx for idx, m in enumerate(all_global_models, 1) 
+                          if m.get('id', '').strip() == model_id.strip()), None)
+        
+        # Korea 모델인 경우
+        is_korea = any(m.get('id', '').strip() == model_id.strip() for m in korea_models)
+        
+        if is_korea:
+            # Korea 모델 중에서의 순위 확인
+            korea_rank = next((idx for idx, m in enumerate(korea_models, 1)
+                             if m.get('id', '').strip() == model_id.strip()), None)
+            
+            if korea_rank:
+                return min(global_rank or 3001, korea_rank + 1000), True
+        
+        return global_rank if global_rank else 'Not in top 3000', is_korea
+
     try:
         progress(0, desc="Fetching models...")
         
-        # 모델 가져오기
-        all_global_models = get_all_models(limit=1000)
-        print(f"Actually fetched models count: {len(all_global_models)}")
-        
-        # API 응답 순서를 순위로 사용하여 순위 맵 생성
-        rank_map = {}
-        for rank, model in enumerate(all_global_models, 1):
-            model_id = model.get('id', '').strip()
-            rank_map[model_id] = {
-                'rank': rank,
-                'likes': model.get('likes', 0),
-                'downloads': model.get('downloads', 0),
-                'title': model.get('title', 'No Title')
-            }
-            print(f"Rank {rank}: {model_id}")
+        if not HF_TOKEN:
+            fig = create_error_plot()
+            error_html = """
+            <div style='padding: 20px; background: #fee; border-radius: 10px; margin: 10px 0;'>
+                <h3 style='color: #c00;'>⚠️ API 인증이 필요합니다</h3>
+                <p>HuggingFace API 토큰이 설정되지 않았습니다. 완전한 기능을 사용하기 위해서는 API 토큰이 필요합니다.</p>
+            </div>
+            """
+            empty_df = pd.DataFrame(columns=['Global Rank', 'Model ID', 'Title', 'Downloads', 'Likes', 'Korea Search', 'URL'])
+            return fig, error_html, empty_df
 
-        # target_models의 순위 확인 및 정보 수집
+        # 일반 모델과 Korea 관련 모델 모두 가져오기 (3000위까지)
+        all_global_models = get_all_models(limit=3000)
+        korea_models = get_korea_models()
+        
+        print(f"Total global models fetched: {len(all_global_models)}")
+        print(f"Total Korea models fetched: {len(korea_models)}")
+        
+        # 모든 모델 통합 (중복 제거)
+        all_models = all_global_models.copy()
+        existing_ids = {model.get('id', '') for model in all_global_models}
+        
+        added_korea_models = 0
+        for korea_model in korea_models:
+            if korea_model.get('id', '') not in existing_ids:
+                all_models.append(korea_model)
+                existing_ids.add(korea_model.get('id', ''))
+                added_korea_models += 1
+        
+        print(f"Added {added_korea_models} unique Korea models")
+        print(f"Total combined models: {len(all_models)}")
+        
+        # 시각화를 위한 Figure 생성
+        fig = go.Figure()
+        
+        # 순위 정보 수집
         filtered_models = []
         for model_id in target_models.keys():
             try:
-                # 개별 모델 API 호출
                 normalized_id = model_id.strip('/')
                 model_url_api = f"https://huggingface.co/api/models/{normalized_id}"
                 response = requests.get(
@@ -427,100 +454,77 @@ def get_models_data(progress=gr.Progress()):
                 
                 if response.status_code == 200:
                     model_data = response.json()
-                    api_id = model_data.get('id', '').strip()
+                    rank, is_korea = calculate_rank(model_id, all_global_models, korea_models)
                     
-                    # API 응답 순서에서 순위 찾기
-                    rank_info = rank_map.get(api_id)
-                    
-                    model_info = {
+                    filtered_models.append({
                         'id': model_id,
-                        'global_rank': rank_info['rank'] if rank_info else 'Not in top 1000',
+                        'global_rank': rank,
                         'downloads': model_data.get('downloads', 0),
                         'likes': model_data.get('likes', 0),
-                        'title': model_data.get('title', 'No Title')
-                    }
-                    filtered_models.append(model_info)
-                    print(f"Model {model_id}: Rank={model_info['global_rank']}, Downloads={model_info['downloads']}, Likes={model_info['likes']}")
+                        'title': model_data.get('title', 'No Title'),
+                        'is_korea': is_korea
+                    })
+                    
+                    print(f"Model {model_id}: Rank={rank}, Is Korea={is_korea}")
                 else:
                     filtered_models.append({
                         'id': model_id,
-                        'global_rank': 'Not in top 1000',
+                        'global_rank': 'Not in top 3000',
                         'downloads': 0,
                         'likes': 0,
-                        'title': 'No Title'
+                        'title': 'No Title',
+                        'is_korea': False
                     })
             except Exception as e:
                 print(f"Error processing {model_id}: {str(e)}")
-                filtered_models.append({
-                    'id': model_id,
-                    'global_rank': 'Not in top 1000',
-                    'downloads': 0,
-                    'likes': 0,
-                    'title': 'No Title'
-                })
-        
+                continue
+
         # 순위로 정렬
         filtered_models.sort(key=lambda x: float('inf') if isinstance(x['global_rank'], str) else x['global_rank'])
         
-        progress(0.3, desc="Creating visualization...")
-        
-        # 시각화 생성
-        fig = go.Figure()
-        
-        # 순위권 내 모델만 필터링하여 시각화
+        # 시각화 데이터 준비
         valid_models = [m for m in filtered_models if isinstance(m['global_rank'], (int, float))]
         
         if valid_models:
             ids = [m['id'] for m in valid_models]
             ranks = [m['global_rank'] for m in valid_models]
-            likes = [m['likes'] for m in valid_models]
-            downloads = [m['downloads'] for m in valid_models]
-            
-            # Y축 값을 반전 (높은 순위가 위로 가도록)
-            y_values = [1001 - r for r in ranks]
             
             fig.add_trace(go.Bar(
                 x=ids,
-                y=y_values,
-                text=[f"Global Rank: #{r}<br>Downloads: {format(d, ',')}<br>Likes: {format(l, ',')}" 
-                      for r, d, l in zip(ranks, downloads, likes)],
+                y=[3001 - r for r in ranks],  # Y축 범위 3000까지 확장
+                text=[f"Rank: #{r}<br>Downloads: {format(m['downloads'], ',')}<br>Likes: {format(m['likes'], ',')}"
+                      for r, m in zip(ranks, valid_models)],
                 textposition='auto',
-                marker_color='rgb(158,202,225)',
+                marker_color=['rgba(255,0,0,0.6)' if m['is_korea'] else 'rgba(0,0,255,0.6)' 
+                            for m in valid_models],
                 opacity=0.8
             ))
-            
+
             fig.update_layout(
-                title={
-                    'text': 'Hugging Face Models Global Rankings',
-                    'y':0.95,
-                    'x':0.5,
-                    'xanchor': 'center',
-                    'yanchor': 'top'
-                },
-                xaxis_title='Model ID',
-                yaxis_title='Global Rank',
+                title="HuggingFace Models Global Rankings (Up to #3000)",
+                xaxis_title="Model ID",
+                yaxis_title="Global Rank",
                 yaxis=dict(
-                    ticktext=[f"#{i}" for i in range(1, 1001, 50)],
-                    tickvals=[1001 - i for i in range(1, 1001, 50)],
-                    range=[0, 1000]
+                    ticktext=[f"#{i}" for i in range(1, 3001, 100)],
+                    tickvals=[3001 - i for i in range(1, 3001, 100)],
+                    range=[0, 3000]
                 ),
                 height=800,
                 showlegend=False,
                 template='plotly_white',
                 xaxis_tickangle=-45
             )
-        
-        progress(0.6, desc="Creating model cards...")
-        
+
         # HTML 카드 생성
         html_content = """
         <div style='padding: 20px; background: #f5f5f5;'>
-            <h2 style='color: #2c3e50;'>Models Global Rankings</h2>
+            <h2 style='color: #2c3e50;'>Models Rankings (Up to #3000)</h2>
             <div style='display: grid; grid-template-columns: repeat(auto-fill, minmax(300px, 1fr)); gap: 20px;'>
         """
         
         for model in filtered_models:
-            rank_display = f"Global Rank #{model['global_rank']}" if isinstance(model['global_rank'], (int, float)) else "Not in top 1000"
+            rank_display = f"Global Rank #{model['global_rank']}" if isinstance(model['global_rank'], (int, float)) else "Not in top 3000"
+            korea_badge = "🇰🇷 Korea Search Result" if model['is_korea'] else ""
             
             html_content += f"""
             <div style='
@@ -529,9 +533,11 @@ def get_models_data(progress=gr.Progress()):
                 border-radius: 10px;
                 box-shadow: 0 2px 4px rgba(0,0,0,0.1);
                 transition: transform 0.2s;
+                {f"border: 2px solid #e74c3c;" if model['is_korea'] else ""}
             '>
                 <h3 style='color: #34495e;'>{rank_display}</h3>
                 <h4 style='color: #2c3e50;'>{model['id']}</h4>
+                <p style='color: #e74c3c; font-weight: bold;'>{korea_badge}</p>
                 <p style='color: #7f8c8d;'>⬇️ Downloads: {format(model['downloads'], ',')}</p>
                 <p style='color: #7f8c8d;'>👍 Likes: {format(model['likes'], ',')}</p>
                 <a href='{target_models[model['id']]}' 
@@ -559,6 +565,7 @@ def get_models_data(progress=gr.Progress()):
             'Title': m['title'],
             'Downloads': format(m['downloads'], ','),
             'Likes': format(m['likes'], ','),
+            'Korea Search': '🇰🇷' if m['is_korea'] else '',
             'URL': target_models[m['id']]
         } for m in filtered_models])
         
@@ -567,12 +574,20 @@ def get_models_data(progress=gr.Progress()):
         
     except Exception as e:
         print(f"Error in get_models_data: {str(e)}")
-        return create_error_plot(), f"<div>에러 발생: {str(e)}</div>", pd.DataFrame()
+        error_fig = create_error_plot()
+        error_html = f"""
+        <div style='padding: 20px; background: #fee; border-radius: 10px; margin: 10px 0;'>
+            <h3 style='color: #c00;'>⚠️ 오류가 발생했습니다</h3>
+            <p>{str(e)}</p>
+        </div>
+        """
+        empty_df = pd.DataFrame(columns=['Global Rank', 'Model ID', 'Title', 'Downloads', 'Likes', 'Korea Search', 'URL'])
+        return error_fig, error_html, empty_df
 
         
 # 관심 스페이스 URL 리스트와 정보
 target_spaces = {
-
+    "fantos/x-mas": "https://huggingface.co/spaces/fantos/x-mas",
     "openfree/Korean-Leaderboard": "https://huggingface.co/spaces/openfree/Korean-Leaderboard",
     "ginipick/FLUXllama": "https://huggingface.co/spaces/ginipick/FLUXllama",
     "ginipick/SORA-3D": "https://huggingface.co/spaces/ginipick/SORA-3D",
@@ -605,6 +620,12 @@ target_spaces = {
     "upstage/open-ko-llm-leaderboard": "https://huggingface.co/spaces/upstage/open-ko-llm-leaderboard",
     "LGAI-EXAONE/EXAONE-3.5-Instruct-Demo": "https://huggingface.co/spaces/LGAI-EXAONE/EXAONE-3.5-Instruct-Demo",
 
+    "r3gm/DiffuseCraft": "https://huggingface.co/spaces/r3gm/DiffuseCraft",
+
+    "LeeSangHoon/HierSpeech_TTS": "https://huggingface.co/spaces/LeeSangHoon/HierSpeech_TTS",
+    "etri-vilab/KOALA": "https://huggingface.co/spaces/etri-vilab/KOALA",
+    "etri-vilab/Ko-LLaVA": "https://huggingface.co/spaces/etri-vilab/Ko-LLaVA",
+    
     "cutechicken/TankWar3D": "https://huggingface.co/spaces/cutechicken/TankWar3D",
     "kolaslab/RC4-EnDecoder": "https://huggingface.co/spaces/kolaslab/RC4-EnDecoder",
     "kolaslab/simulator": "https://huggingface.co/spaces/kolaslab/simulator",
@@ -621,7 +642,7 @@ def get_spaces_data(sort_type="trending", progress=gr.Progress()):
     url = "https://huggingface.co/api/spaces"
     params = {
         'full': 'true',
-        'limit': 400
+        'limit': 500
     }
     
     if sort_type == "modes":
@@ -668,7 +689,7 @@ def get_spaces_data(sort_type="trending", progress=gr.Progress()):
         
         fig.update_layout(
             title={
-                'text': f'Hugging Face Spaces {sort_type.title()} Rankings (Top 400)',
+                'text': f'Hugging Face Spaces {sort_type.title()} Rankings (Top 500)',
                 'y':0.95,
                 'x':0.5,
                 'xanchor': 'center',
@@ -679,9 +700,9 @@ def get_spaces_data(sort_type="trending", progress=gr.Progress()):
             yaxis=dict(
                 autorange='reversed',  # Y축을 반전
                 tickmode='array',
-                ticktext=[str(i) for i in range(1, 401, 20)],  # 1부터 400까지 20 간격으로 표시
-                tickvals=[i for i in range(1, 401, 20)],
-                range=[1, 400]  # Y축 범위를 1부터 400까지로 설정
+                ticktext=[str(i) for i in range(1, 501, 20)],  # 1부터 400까지 20 간격으로 표시
+                tickvals=[i for i in range(1, 501, 20)],
+                range=[1, 500]  # Y축 범위를 1부터 400까지로 설정
             ),
             height=800,
             showlegend=False,
@@ -829,7 +850,7 @@ def get_trending_spaces_without_token():
         params = {
             'sort': 'likes',
             'direction': -1,
-            'limit': 400,
+            'limit': 500,
             'full': 'true'
         }
         
@@ -971,7 +992,7 @@ def refresh_data():
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
     gr.Markdown("""
     # 🤗 허깅페이스 '한국(언어) 리더보드'
-    HuggingFace가 제공하는 Spaces와 Models 실시간 인기 순위를 반영: 한국인(기업)이 공개, 한국 'LLM 리더보드' 및 TAG 등을 참고해 리스트 갱신. 신규 등록 요청: arxivgpt@gmail.com
+    HuggingFace가 제공하는 Spaces와 Models 실시간 인기 순위 반영하여 '한국안(기업/언어)'의 리스트(공개,검색,리더보드 등)만 분석. (c)'한국인공지능진흥협회' / 요청: arxivgpt@gmail.com
     """)
     
     # 새로 고침 버튼을 상단으로 이동하고 한글로 변경