Update auditqa/doc_process.py

auditqa/doc_process.py

@@ -41,13 +41,16 @@ def process_pdf():
         doc_processed = text_splitter.split_documents(value)
         for doc in doc_processed:
             doc.metadata["source"] = file
+            doc.metadata["year"] = file[-4:]
         all_documents[file] = doc_processed
         
     print(all_documents.keys())
   
     
     embeddings = HuggingFaceEmbeddings(
-      model_name="BAAI/bge-small-en-v1.5"
+        model_kwargs = {'device': 'cpu'},
+        encode_kwargs = {'normalize_embeddings': True},
+        model_name="BAAI/bge-small-en-v1.5"
     )
     
     qdrant_collections = {}