Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -77,17 +77,20 @@ def rut5_answer(question, context):
|
|
| 77 |
return tokenizer.decode(output_ids[0], skip_special_tokens=True)
|
| 78 |
|
| 79 |
def flatten_index(idx):
|
| 80 |
-
#
|
| 81 |
-
if isinstance(idx, (np.
|
| 82 |
return int(idx)
|
| 83 |
-
if isinstance(idx, (list, tuple
|
| 84 |
-
|
|
|
|
|
|
|
| 85 |
if hasattr(idx, "tolist"):
|
| 86 |
item = idx.tolist()
|
| 87 |
-
|
| 88 |
-
|
| 89 |
-
return int(
|
| 90 |
-
|
|
|
|
| 91 |
|
| 92 |
def ask_chatbot(question):
|
| 93 |
question = question.strip()
|
|
@@ -97,7 +100,7 @@ def ask_chatbot(question):
|
|
| 97 |
return "Ошибка: база знаний пуста. Проверьте .docx и перезапустите Space."
|
| 98 |
|
| 99 |
user_vec = vectorizer.transform([question.lower()])
|
| 100 |
-
sims = cosine_similarity(user_vec, matrix)
|
| 101 |
n_blocks = min(3, len(blocks))
|
| 102 |
if n_blocks == 0:
|
| 103 |
return "Ошибка: база знаний отсутствует или пуста."
|
|
@@ -105,11 +108,10 @@ def ask_chatbot(question):
|
|
| 105 |
context_blocks = []
|
| 106 |
for idx in sorted_idxs:
|
| 107 |
idx_int = flatten_index(idx)
|
| 108 |
-
if 0 <= idx_int < len(blocks):
|
| 109 |
context_blocks.append(blocks[idx_int])
|
| 110 |
context = " ".join(context_blocks)
|
| 111 |
# Ответ только из абзацев, не заголовков!
|
| 112 |
-
# Ищем наиболее релевантный "нормальный" блок
|
| 113 |
best_normal_block = ""
|
| 114 |
max_sim = -1
|
| 115 |
for nb in normal_blocks:
|
|
|
|
| 77 |
return tokenizer.decode(output_ids[0], skip_special_tokens=True)
|
| 78 |
|
| 79 |
def flatten_index(idx):
|
| 80 |
+
# Универсальный способ из всего достать int
|
| 81 |
+
if isinstance(idx, (int, float, np.integer, np.floating)):
|
| 82 |
return int(idx)
|
| 83 |
+
if isinstance(idx, (list, tuple, np.ndarray)):
|
| 84 |
+
if len(idx) == 0:
|
| 85 |
+
return 0
|
| 86 |
+
return flatten_index(idx)
|
| 87 |
if hasattr(idx, "tolist"):
|
| 88 |
item = idx.tolist()
|
| 89 |
+
return flatten_index(item)
|
| 90 |
+
try:
|
| 91 |
+
return int(idx)
|
| 92 |
+
except Exception:
|
| 93 |
+
return 0
|
| 94 |
|
| 95 |
def ask_chatbot(question):
|
| 96 |
question = question.strip()
|
|
|
|
| 100 |
return "Ошибка: база знаний пуста. Проверьте .docx и перезапустите Space."
|
| 101 |
|
| 102 |
user_vec = vectorizer.transform([question.lower()])
|
| 103 |
+
sims = cosine_similarity(user_vec, matrix)[0]
|
| 104 |
n_blocks = min(3, len(blocks))
|
| 105 |
if n_blocks == 0:
|
| 106 |
return "Ошибка: база знаний отсутствует или пуста."
|
|
|
|
| 108 |
context_blocks = []
|
| 109 |
for idx in sorted_idxs:
|
| 110 |
idx_int = flatten_index(idx)
|
| 111 |
+
if isinstance(idx_int, int) and 0 <= idx_int < len(blocks):
|
| 112 |
context_blocks.append(blocks[idx_int])
|
| 113 |
context = " ".join(context_blocks)
|
| 114 |
# Ответ только из абзацев, не заголовков!
|
|
|
|
| 115 |
best_normal_block = ""
|
| 116 |
max_sim = -1
|
| 117 |
for nb in normal_blocks:
|