Update app.py

app.py

@@ -3,7 +3,8 @@ import numpy as np
 import h5py
 import faiss
 import json
-from sentence_transformers import SentenceTransformer
+from transformers import AutoTokenizer, AutoModel
+from sentence_transformers import SentenceTransformer, models
 from sklearn.feature_extraction.text import TfidfVectorizer
 from sklearn.metrics.pairwise import cosine_similarity
 import re
@@ -48,7 +49,11 @@ embeddings, patent_numbers, metadata, texts = load_data()
 
 # Load BERT model for encoding search queries
 try:
-    model = SentenceTransformer('anferico/bert-for-patents')
+    tokenizer = AutoTokenizer.from_pretrained('anferico/bert-for-patents')
+    bert_model = AutoModel.from_pretrained('anferico/bert-for-patents')
+    word_embedding_model = models.Transformer(model_name='anferico/bert-for-patents', tokenizer=tokenizer, model=bert_model)
+    pooling_model = models.Pooling(word_embedding_model.get_word_embedding_dimension())
+    model = SentenceTransformer(modules=[word_embedding_model, pooling_model])
 except Exception as e:
     print(f"Error loading anferico/bert-for-patents: {e}")
     print("Falling back to a general-purpose model.")