ai-saas / saas-search

Commit 70a318c6391c338bc3a11330c0d931e6d081db5c

Authored by tangwang
1 parent 70dab99f

fix bug

Showing 1 changed file with 13 additions and 13 deletions   Show diff stats
Inline Side-by-side
embeddings/text_encoder.py
  Show/Hide comments   View file @70a318c
... ... @@ -113,7 +113,19 @@ class BgeEncoder:
113 113 uncached_indices: List[int] = []
114 114 uncached_texts: List[str] = []
115 115  
116   - # Prepare request data for uncached texts
  116 + # Process response
  117 + # Each element can be np.ndarray or None (表示该文本没有可用的向量）
  118 + embeddings: List[Optional[np.ndarray]] = [None] * len(sentences)
  119 +
  120 + for i, text in enumerate(sentences):
  121 + cached = self._get_cached_embedding(text, 'en') # Use 'en' as default language for title embedding
  122 + if cached is not None:
  123 + embeddings[i] = cached
  124 + else:
  125 + uncached_indices.append(i)
  126 + uncached_texts.append(text)
  127 +
  128 + # Prepare request data for uncached texts (after cache check)
117 129 request_data = []
118 130 for i, text in enumerate(uncached_texts):
119 131 request_item = {
... ... @@ -127,18 +139,6 @@ class BgeEncoder:
127 139 request_item["name_ru"] = None
128 140  
129 141 request_data.append(request_item)
130   -
131   - # Process response
132   - # Each element can be np.ndarray or None (表示该文本没有可用的向量）
133   - embeddings: List[Optional[np.ndarray]] = [None] * len(sentences)
134   -
135   - for i, text in enumerate(sentences):
136   - cached = self._get_cached_embedding(text, 'en') # Use 'en' as default language for title embedding
137   - if cached is not None:
138   - embeddings[i] = cached
139   - else:
140   - uncached_indices.append(i)
141   - uncached_texts.append(text)
142 142  
143 143 # If there are uncached texts, call service
144 144 if uncached_texts:
... ...