Update app.py

app.py

@@ -47,7 +47,7 @@ def load_doc(list_file_path, chunk_size, chunk_overlap):
 
 # Create vector database
 def create_db(splits, collection_name):
-    embedding = HuggingFaceEmbeddings()
+    embedding = HuggingFaceEmbeddings(model_name="sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2")
     new_client = chromadb.EphemeralClient()
     vectordb = Chroma.from_documents(
         documents=splits,
@@ -61,7 +61,7 @@ def create_db(splits, collection_name):
 
 # Load vector database
 def load_db():
-    embedding = HuggingFaceEmbeddings()
+    embedding = HuggingFaceEmbeddings(model_name="sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2")
     vectordb = Chroma(
         # persist_directory=default_persist_directory, 
         embedding_function=embedding)
@@ -132,8 +132,8 @@ def initialize_database(list_file_obj, chunk_size, chunk_overlap, progress=gr.Pr
     #file_path = file_obj.name
     list_file_path = [x.name for x in list_file_obj if x is not None]
     collection_name = Path(list_file_path[0]).stem
-    # print('list_file_path: ', list_file_path)
-    # print('Collection name: ', collection_name)
+    print('list_file_path: ', list_file_path)
+    print('Collection name: ', collection_name)
     progress(0.25, desc="Loading document...")
     # Load document and create splits
     doc_splits = load_doc(list_file_path, chunk_size, chunk_overlap)
@@ -174,8 +174,7 @@ def conversation(qa_chain, message, history):
     # Langchain sources are zero-based
     response_source1_page = response_sources[0].metadata["page"] + 1
     response_source2_page = response_sources[1].metadata["page"] + 1
-    # print ('chat response: ', response_answer)
-    # print('DB source', response_sources)
+    print ('Response: ', response)
     
     # Append user message and response to chat history
     new_history = history + [(message, response_answer)]