wip

app.py

@@ -11,6 +11,8 @@ document_embedding = None
 docs_df = None
 
 def text_embedding_batch():
+    global query_embedding
+    global docs_df
     model_name = "sentence-transformers/all-MiniLM-L6-v2"
     dataset = ir_datasets.load("beir/fiqa/dev")
     docs_df = pd.DataFrame(dataset.docs_iter()).set_index("doc_id").sample(frac=0.0001)
@@ -26,6 +28,7 @@ def text_embedding_batch():
 
 
 def text_embedding_single(query: str):
+    global document_embedding
     model_name = "sentence-transformers/all-MiniLM-L6-v2"
     predictor = MultiModalPredictor(
         pipeline="feature_extraction",
@@ -39,6 +42,11 @@ def text_embedding_single(query: str):
 
 
 def rank_document():
+    global query_embedding
+    global document_embedding
+    global docs_df
+    print('~~~~~here')
+    print('~~~~~~~~', query_embedding, document_embedding)
     q_norm = query_embedding / np.linalg.norm(query_embedding, axis=-1, keepdims=True)
     print(q_norm)
     d_norm = document_embedding / np.linalg.norm(document_embedding, axis=-1, keepdims=True)