Spaces:
Runtime error
Runtime error
X commited on
Update app.py
Browse files
app.py
CHANGED
|
@@ -38,7 +38,7 @@ else:
|
|
| 38 |
query_logs = []
|
| 39 |
|
| 40 |
# ===================================================================
|
| 41 |
-
# 2. ЗАГРУЗКА ДАННЫХ (с кэшированием)
|
| 42 |
# ===================================================================
|
| 43 |
|
| 44 |
@st.cache_resource
|
|
@@ -65,11 +65,14 @@ def load_embedder():
|
|
| 65 |
return SentenceTransformer(MODEL_NAME)
|
| 66 |
|
| 67 |
@st.cache_resource
|
| 68 |
-
def create_embeddings(
|
| 69 |
-
"""
|
|
|
|
|
|
|
|
|
|
| 70 |
with st.spinner("🔢 Создаю эмбеддинги статей..."):
|
| 71 |
-
texts = [f"{a['title']}\n\n{a['text']}" for a in
|
| 72 |
-
return
|
| 73 |
|
| 74 |
# Загружаем всё
|
| 75 |
articles = load_wikipedia()
|
|
@@ -80,7 +83,7 @@ embeddings = create_embeddings(articles, embedder)
|
|
| 80 |
# 3. ФУНКЦИИ ПОИСКА И ЛОГИРОВАНИЯ
|
| 81 |
# ===================================================================
|
| 82 |
|
| 83 |
-
def search_wikipedia(query):
|
| 84 |
"""Ищет ответ на вопрос в Википедии"""
|
| 85 |
if not query:
|
| 86 |
return None
|
|
@@ -88,15 +91,15 @@ def search_wikipedia(query):
|
|
| 88 |
start_time = time.time()
|
| 89 |
|
| 90 |
# Поиск
|
| 91 |
-
query_vector =
|
| 92 |
-
scores =
|
| 93 |
top_indices = np.argsort(-scores)[:3]
|
| 94 |
|
| 95 |
results = []
|
| 96 |
for idx in top_indices:
|
| 97 |
score = float(scores[int(idx)])
|
| 98 |
if score > 0.3:
|
| 99 |
-
article =
|
| 100 |
results.append({
|
| 101 |
"title": article['title'],
|
| 102 |
"score": score,
|
|
@@ -250,7 +253,7 @@ if query and (search_clicked or query != st.session_state.get("last_query", ""))
|
|
| 250 |
st.session_state.last_query = query
|
| 251 |
|
| 252 |
with st.spinner("🔎 Ищу ответ..."):
|
| 253 |
-
results = search_wikipedia(query)
|
| 254 |
|
| 255 |
if results:
|
| 256 |
for i, result in enumerate(results, 1):
|
|
|
|
| 38 |
query_logs = []
|
| 39 |
|
| 40 |
# ===================================================================
|
| 41 |
+
# 2. ЗАГРУЗКА ДАННЫХ (с кэшированием) - ИСПРАВЛЕНО
|
| 42 |
# ===================================================================
|
| 43 |
|
| 44 |
@st.cache_resource
|
|
|
|
| 65 |
return SentenceTransformer(MODEL_NAME)
|
| 66 |
|
| 67 |
@st.cache_resource
|
| 68 |
+
def create_embeddings(_articles, _embedder):
|
| 69 |
+
"""
|
| 70 |
+
Создаёт эмбеддинги статей
|
| 71 |
+
_articles и _embedder - с подчёркиванием, чтобы Streamlit их не хешировал
|
| 72 |
+
"""
|
| 73 |
with st.spinner("🔢 Создаю эмбеддинги статей..."):
|
| 74 |
+
texts = [f"{a['title']}\n\n{a['text']}" for a in _articles]
|
| 75 |
+
return _embedder.encode(texts, normalize_embeddings=True)
|
| 76 |
|
| 77 |
# Загружаем всё
|
| 78 |
articles = load_wikipedia()
|
|
|
|
| 83 |
# 3. ФУНКЦИИ ПОИСКА И ЛОГИРОВАНИЯ
|
| 84 |
# ===================================================================
|
| 85 |
|
| 86 |
+
def search_wikipedia(query, _articles, _embeddings, _embedder):
|
| 87 |
"""Ищет ответ на вопрос в Википедии"""
|
| 88 |
if not query:
|
| 89 |
return None
|
|
|
|
| 91 |
start_time = time.time()
|
| 92 |
|
| 93 |
# Поиск
|
| 94 |
+
query_vector = _embedder.encode([query], normalize_embeddings=True)[0]
|
| 95 |
+
scores = _embeddings @ query_vector
|
| 96 |
top_indices = np.argsort(-scores)[:3]
|
| 97 |
|
| 98 |
results = []
|
| 99 |
for idx in top_indices:
|
| 100 |
score = float(scores[int(idx)])
|
| 101 |
if score > 0.3:
|
| 102 |
+
article = _articles[int(idx)]
|
| 103 |
results.append({
|
| 104 |
"title": article['title'],
|
| 105 |
"score": score,
|
|
|
|
| 253 |
st.session_state.last_query = query
|
| 254 |
|
| 255 |
with st.spinner("🔎 Ищу ответ..."):
|
| 256 |
+
results = search_wikipedia(query, articles, embeddings, embedder)
|
| 257 |
|
| 258 |
if results:
|
| 259 |
for i, result in enumerate(results, 1):
|