Update build_quora_index.py

build_quora_index.py

@@ -6,6 +6,9 @@ from sentence_transformers import SentenceTransformer
 from qdrant_client import QdrantClient, models
 
 
+MAX_QUESTIONS = 1000
+
+
 def compute_embedding(sentences, emb_model):
     return emb_model.encode(sentences=sentences)
 
@@ -27,8 +30,6 @@ def build_index():
         api_key=os.environ['QDRANT_API_KEY'],
     )
     encoder = SentenceTransformer(model_name_or_path='BAAI/bge-small-en-v1.5')
-    MAX_QUESTIONS = 1000
-    BATCH_SIZE = 100
 
     quora_ds = load_dataset(path='quora', split='train', streaming=True)
     quora_questions = get_questions(ds=quora_ds)
@@ -41,6 +42,7 @@ def build_index():
       )
     )
 
+    BATCH_SIZE = 100
     question_batch = []
     for idx, entry in enumerate(tqdm(quora_questions, desc='Uploading vector embeddings in batch size of {}'.format(BATCH_SIZE))):
         if len(question_batch) < BATCH_SIZE: