Update app-backup.py

app-backup.py

@@ -15,10 +15,12 @@ target_models = {
     "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",
-    "moreh/Llama-3-Motif-102B": "https://huggingface.co/moreh/Llama-3-Motif-102B",
-    "Samsung/TinyClick": "https://huggingface.co/Samsung/TinyClick",
+
+    
+    "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",
     
     "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",
@@ -292,84 +294,166 @@ target_models = {
     "sel303/llama3-diverce-ver1.6": "https://huggingface.co/sel303/llama3-diverce-ver1.6"    
 }
 
-def get_models_data(progress=gr.Progress()):
-    """모델 데이터 가져오기"""
-    def normalize_model_id(model_id):
-        """모델 ID를 정규화"""
-        return model_id.strip().lower()
-        
+def get_korea_models():
+    """Korea 관련 모델 검색"""
+    params = {
+        "search": "korea",
+        "full": "True",
+        "config": "True",
+        "limit": 1000
+    }
+    
     try:
-        progress(0, desc="Fetching global rankings...")
-        
-        # 전체 모델 목록 가져오기 (좋아요 순으로 정렬)
-        global_params = {
-            'full': 'true',
-            'limit': 10000,  # 10000위까지 확장
-            'sort': 'likes',
-            'direction': -1
-        }
-        
-        global_response = requests.get(
+        response = requests.get(
             "https://huggingface.co/api/models",
             headers={'Accept': 'application/json'},
-            params=global_params
+            params=params
         )
         
-        if global_response.status_code != 200:
-            print(f"Failed to fetch global rankings: {global_response.status_code}")
-            return create_error_plot(), "<div>전체 순위 데이터를 가져오는데 실패했습니다.</div>", pd.DataFrame()
-            
-        all_global_models = global_response.json()
-        print(f"Fetched {len(all_global_models)} models")
-        
-        # 전체 순위 맵 생성
-        global_ranks = {}
-        for idx, model in enumerate(all_global_models, 1):
-            model_id = normalize_model_id(model.get('id', ''))
-            global_ranks[model_id] = {
-                'rank': idx,
-                'downloads': model.get('downloads', 0),
+        if response.status_code == 200:
+            return response.json()
+        else:
+            print(f"Failed to fetch Korea models: {response.status_code}")
+            return []
+    except Exception as e:
+        print(f"Error fetching Korea models: {str(e)}")
+        return []
+
+def get_all_models(limit=1000):
+    """모든 모델과 Korea 관련 모델 가져오기"""
+    all_models = []
+    
+    # 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")
+    
+    # 2. Korea 검색 결과 가져오기
+    korea_params = {
+        "search": "korea",
+        "full": "True",
+        "config": "True",
+        "limit": limit
+    }
+    
+    korea_response = requests.get(
+        "https://huggingface.co/api/models",
+        headers={'Accept': 'application/json'},
+        params=korea_params
+    )
+    
+    if korea_response.status_code == 200:
+        korea_models = korea_response.json()
+        print(f"Fetched {len(korea_models)} Korea-related models")
+        
+        # 중복 제거하면서 Korea 모델 추가
+        existing_ids = {model.get('id', '') for model in all_models}
+        for model in korea_models:
+            if model.get('id', '') not in existing_ids:
+                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"""
+    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')
             }
-        
-        # target_models의 상세 정보 가져오기
+            print(f"Rank {rank}: {model_id}")
+
+        # target_models의 순위 확인 및 정보 수집
         filtered_models = []
         for model_id in target_models.keys():
             try:
-                # 개별 모델 API 호출로 정확한 정보 가져오기
-                model_url_api = f"https://huggingface.co/api/models/{model_id}"
-                response = requests.get(model_url_api, headers={'Accept': 'application/json'})
+                # 개별 모델 API 호출
+                normalized_id = model_id.strip('/')
+                model_url_api = f"https://huggingface.co/api/models/{normalized_id}"
+                response = requests.get(
+                    model_url_api, 
+                    headers={'Accept': 'application/json'}
+                )
                 
                 if response.status_code == 200:
                     model_data = response.json()
-                    normalized_id = normalize_model_id(model_id)
+                    api_id = model_data.get('id', '').strip()
+                    
+                    # API 응답 순서에서 순위 찾기
+                    rank_info = rank_map.get(api_id)
                     
-                    # 전체 순위 정보와 개별 모델 정보 결합
-                    rank_info = global_ranks.get(normalized_id, {})
                     model_info = {
                         'id': model_id,
-                        'global_rank': rank_info.get('rank', 'Not in top 10000'),
+                        'global_rank': rank_info['rank'] if rank_info else 'Not in top 1000',
                         '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']}, Likes={model_info['likes']}, Downloads={model_info['downloads']}")
+                    print(f"Model {model_id}: Rank={model_info['global_rank']}, Downloads={model_info['downloads']}, Likes={model_info['likes']}")
                 else:
-                    print(f"Failed to fetch data for {model_id}: {response.status_code}")
                     filtered_models.append({
                         'id': model_id,
-                        'global_rank': 'Not in top 10000',
+                        'global_rank': 'Not in top 1000',
                         'downloads': 0,
                         'likes': 0,
                         'title': 'No Title'
                     })
             except Exception as e:
-                print(f"Error fetching data for {model_id}: {str(e)}")
+                print(f"Error processing {model_id}: {str(e)}")
                 filtered_models.append({
                     'id': model_id,
-                    'global_rank': 'Not in top 10000',
+                    'global_rank': 'Not in top 1000',
                     'downloads': 0,
                     'likes': 0,
                     'title': 'No Title'
@@ -378,38 +462,28 @@ def get_models_data(progress=gr.Progress()):
         # 순위로 정렬
         filtered_models.sort(key=lambda x: float('inf') if isinstance(x['global_rank'], str) else x['global_rank'])
         
-        if not filtered_models:
-            return create_error_plot(), "<div>모델 데이터를 찾을 수 없습니다.</div>", pd.DataFrame()
-        
         progress(0.3, desc="Creating visualization...")
         
         # 시각화 생성
         fig = go.Figure()
         
-        # 데이터 준비
-        ids = [model['id'] for model in filtered_models]
-        ranks = [model['global_rank'] for model in filtered_models]
-        likes = [model['likes'] for model in filtered_models]
-        downloads = [model['downloads'] for model in filtered_models]
-        
-        # 순위권 내 모델만 필터링
-        valid_indices = [i for i, rank in enumerate(ranks) if isinstance(rank, (int, float))]
-        
-        if valid_indices:  # 순위권 내 모델이 있는 경우만 그래프 생성
-            valid_ids = [ids[i] for i in valid_indices]
-            valid_ranks = [ranks[i] for i in valid_indices]
-            valid_likes = [likes[i] for i in valid_indices]
-            valid_downloads = [downloads[i] for i in valid_indices]
+        # 순위권 내 모델만 필터링하여 시각화
+        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
-            y_values = [10001 - r if isinstance(r, (int, float)) else 0 for r in valid_ranks]
+            # Y축 값을 반전 (높은 순위가 위로 가도록)
+            y_values = [1001 - r for r in ranks]
             
-            # 막대 그래프 생성
             fig.add_trace(go.Bar(
-                x=valid_ids,
+                x=ids,
                 y=y_values,
-                text=[f"Global Rank: {r}<br>Likes: {format(l, ',')}<br>Downloads: {format(d, ',')}" 
-                      for r, l, d in zip(valid_ranks, valid_likes, valid_downloads)],
+                text=[f"Global Rank: #{r}<br>Downloads: {format(d, ',')}<br>Likes: {format(l, ',')}" 
+                      for r, d, l in zip(ranks, downloads, likes)],
                 textposition='auto',
                 marker_color='rgb(158,202,225)',
                 opacity=0.8
@@ -417,7 +491,7 @@ def get_models_data(progress=gr.Progress()):
             
             fig.update_layout(
                 title={
-                    'text': 'Hugging Face Models Global Rankings (by Likes)',
+                    'text': 'Hugging Face Models Global Rankings',
                     'y':0.95,
                     'x':0.5,
                     'xanchor': 'center',
@@ -426,9 +500,9 @@ def get_models_data(progress=gr.Progress()):
                 xaxis_title='Model ID',
                 yaxis_title='Global Rank',
                 yaxis=dict(
-                    ticktext=[str(i) for i in range(1, 10001, 500)],
-                    tickvals=[10001 - i for i in range(1, 10001, 500)],
-                    range=[0, 10000]
+                    ticktext=[f"#{i}" for i in range(1, 1001, 50)],
+                    tickvals=[1001 - i for i in range(1, 1001, 50)],
+                    range=[0, 1000]
                 ),
                 height=800,
                 showlegend=False,
@@ -441,18 +515,12 @@ def get_models_data(progress=gr.Progress()):
         # HTML 카드 생성
         html_content = """
         <div style='padding: 20px; background: #f5f5f5;'>
-            <h2 style='color: #2c3e50;'>Models Global Rankings (by Likes)</h2>
+            <h2 style='color: #2c3e50;'>Models Global Rankings</h2>
             <div style='display: grid; grid-template-columns: repeat(auto-fill, minmax(300px, 1fr)); gap: 20px;'>
         """
         
         for model in filtered_models:
-            model_id = model['id']
-            global_rank = model['global_rank']
-            likes = model['likes']
-            downloads = model['downloads']
-            title = model.get('title', 'No Title')
-            
-            rank_display = f"Global Rank #{global_rank}" if isinstance(global_rank, (int, float)) else global_rank
+            rank_display = f"Global Rank #{model['global_rank']}" if isinstance(model['global_rank'], (int, float)) else "Not in top 1000"
             
             html_content += f"""
             <div style='
@@ -462,11 +530,11 @@ def get_models_data(progress=gr.Progress()):
                 box-shadow: 0 2px 4px rgba(0,0,0,0.1);
                 transition: transform 0.2s;
             '>
-                <h3 style='color: #34495e;'>{rank_display} - {model_id}</h3>
-                <p style='color: #2c3e50;'>{title}</p>
-                <p style='color: #7f8c8d;'>👍 Likes: {format(likes, ',')}</p>
-                <p style='color: #7f8c8d;'>⬇️ Downloads: {format(downloads, ',')}</p>
-                <a href='{target_models[model_id]}' 
+                <h3 style='color: #34495e;'>{rank_display}</h3>
+                <h4 style='color: #2c3e50;'>{model['id']}</h4>
+                <p style='color: #7f8c8d;'>⬇️ Downloads: {format(model['downloads'], ',')}</p>
+                <p style='color: #7f8c8d;'>👍 Likes: {format(model['likes'], ',')}</p>
+                <a href='{target_models[model['id']]}' 
                    target='_blank' 
                    style='
                     display: inline-block;
@@ -485,16 +553,14 @@ def get_models_data(progress=gr.Progress()):
         html_content += "</div></div>"
         
         # 데이터프레임 생성
-        df_data = [{
-            'Global Rank': model['global_rank'],
-            'Model ID': model['id'],
-            'Title': model.get('title', 'No Title'),
-            'Likes': format(model['likes'], ','),
-            'Downloads': format(model['downloads'], ','),
-            'URL': target_models[model['id']]
-        } for model in filtered_models]
-        
-        df = pd.DataFrame(df_data)
+        df = pd.DataFrame([{
+            'Global Rank': f"#{m['global_rank']}" if isinstance(m['global_rank'], (int, float)) else m['global_rank'],
+            'Model ID': m['id'],
+            'Title': m['title'],
+            'Downloads': format(m['downloads'], ','),
+            'Likes': format(m['likes'], ','),
+            'URL': target_models[m['id']]
+        } for m in filtered_models])
         
         progress(1.0, desc="Complete!")
         return fig, html_content, df
@@ -502,6 +568,7 @@ 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()
+
         
 # 관심 스페이스 URL 리스트와 정보
 target_spaces = {
@@ -554,7 +621,7 @@ def get_spaces_data(sort_type="trending", progress=gr.Progress()):
     url = "https://huggingface.co/api/spaces"
     params = {
         'full': 'true',
-        'limit': 300
+        'limit': 400
     }
     
     if sort_type == "modes":
@@ -588,13 +655,11 @@ def get_spaces_data(sort_type="trending", progress=gr.Progress()):
         likes = [space.get('likes', 0) for space in spaces]
         titles = [space.get('cardData', {}).get('title') or space.get('title', 'No Title') for space in spaces]
         
-        # Y축 값을 반전
-        y_values = [301 - r for r in ranks]
-        
-        # 막대 그래프 생성
+        # 막대 그래프 생성 (각 순위에서 400까지의 길이로 막대 생성)
         fig.add_trace(go.Bar(
             x=ids,
-            y=y_values,
+            y=ranks,  # 실제 순위 사용
+            base=400,  # 막대의 기준점을 400으로 설정
             text=[f"Rank: {r}<br>Title: {t}<br>Likes: {l}" 
                   for r, t, l in zip(ranks, titles, likes)],
             textposition='auto',
@@ -604,7 +669,7 @@ def get_spaces_data(sort_type="trending", progress=gr.Progress()):
         
         fig.update_layout(
             title={
-                'text': f'Hugging Face Spaces {sort_type.title()} Rankings (Top 300)',
+                'text': f'Hugging Face Spaces {sort_type.title()} Rankings (Top 400)',
                 'y':0.95,
                 'x':0.5,
                 'xanchor': 'center',
@@ -613,9 +678,11 @@ def get_spaces_data(sort_type="trending", progress=gr.Progress()):
             xaxis_title='Space ID',
             yaxis_title='Rank',
             yaxis=dict(
-                ticktext=[str(i) for i in range(1, 301, 20)],
-                tickvals=[301 - i for i in range(1, 301, 20)],
-                range=[0, 300]
+                autorange='reversed',  # Y축을 반전
+                tickmode='linear',
+                tick0=0,
+                dtick=20,
+                range=[0, 400],  # Y축 범위
             ),
             height=800,
             showlegend=False,
@@ -638,7 +705,6 @@ def get_spaces_data(sort_type="trending", progress=gr.Progress()):
             title = space.get('cardData', {}).get('title') or space.get('title', 'No Title')
             likes = space.get('likes', 0)
 
-# 스페이스 함수의 HTML 카드 생성 부분 수정
             html_content += f"""
             <div style='
                 background: white;
@@ -675,8 +741,6 @@ def get_spaces_data(sort_type="trending", progress=gr.Progress()):
                 </a>
             </div>
             """
-
-       
         
         html_content += "</div></div>"
         
@@ -766,7 +830,7 @@ def get_trending_spaces_without_token():
         params = {
             'sort': 'likes',
             'direction': -1,
-            'limit': 1000,
+            'limit': 400,
             'full': 'true'
         }
         
@@ -908,7 +972,7 @@ def refresh_data():
 with gr.Blocks(theme=gr.themes.Soft()) as demo:
     gr.Markdown("""
     # 🤗 허깅페이스 '한국(언어) 리더보드'
-    실시간으로 Hugging Face의 Spaces와 Models 인기 순위를 분석합니다. 한국인(기업)이 공개 및 한국어 'LLM 리더보드'에 공개된 리스트 등을 등록. 신규 등록 요청: [email protected]
+    HuggingFace가 제공하는 Spaces와 Models 실시간 인기 순위를 반영: 한국인(기업)이 공개, 한국 'LLM 리더보드' 및 TAG 등을 참고해 리스트 갱신. 신규 등록 요청: [email protected]
     """)
     
     # 새로 고침 버튼을 상단으로 이동하고 한글로 변경