embedding

src/generate_queries_alternative.py

@@ -201,7 +201,7 @@ def similarity_question(question, questions_queries_dictionary, collection, n_re
     # Store each unique document in the vector embedding database
     for i, d in enumerate(masked_documents):
         response = get_embeddings(d)
-        embedding = response["embeddings"][0]  # Extract the first (and only) embedding from the nested list
+        embedding = response[0]  # Extract the first (and only) embedding from the nested list
         
         # Check if embedding is unique
         is_duplicate = any(np.allclose(embedding, np.array(e), atol=1e-6) for e in unique_embeddings.values())
@@ -217,7 +217,7 @@ def similarity_question(question, questions_queries_dictionary, collection, n_re
     # Compute the embedding for the input question
     masked_question = mask_entities(question, nlp)
     response = get_embeddings(d)
-    query_embedding = response["embeddings"][0]  # Extract embedding
+    query_embedding = response[0]  # Extract embedding
 
     results = collection.query(
         query_embeddings=[query_embedding],  # Ensure correct format
@@ -260,7 +260,7 @@ def similarity_question_no_masking(question, questions_queries_dictionary, colle
     # Store each unique document in the vector embedding database
     for i, d in enumerate(original_documents):
         response = get_embeddings(d)
-        embedding = response["embeddings"][0]  # Extract the first (and only) embedding from the nested list
+        embedding = response[0]  # Extract the first (and only) embedding from the nested list
         
         # Check if embedding is unique
         is_duplicate = any(np.allclose(embedding, np.array(e), atol=1e-6) for e in unique_embeddings.values())
@@ -276,7 +276,7 @@ def similarity_question_no_masking(question, questions_queries_dictionary, colle
     # Compute the embedding for the input question
     
     response = get_embeddings(question)
-    query_embedding = response["embeddings"][0]  # Extract embedding
+    query_embedding = response[0]  # Extract embedding
 
     results = collection.query(
         query_embeddings=[query_embedding],  # Ensure correct format