Update app.py

app.py

@@ -67,18 +67,18 @@ def initialize_pinecone(index_name: str):
 
 # Initialize Pinecone index and BM25 encoder
 pinecone_index = initialize_pinecone("updated-uae-gov")
-bm25 = BM25Encoder().load("./bm25_u.ae.json")
+bm25 = BM25Encoder().load("./updated-uae-gov.json")
 
 ##################################################
 ##################################################
 
 # Initialize models and retriever
-embed_model = HuggingFaceEmbeddings(model_name="Alibaba-NLP/gte-multilingual-base", model_kwargs={"trust_remote_code":True})
+embed_model = HuggingFaceEmbeddings(model_name="jinaai/jina-embeddings-v3", model_kwargs={"trust_remote_code":True})
 retriever = PineconeHybridSearchRetriever(
     embeddings=embed_model, 
     sparse_encoder=bm25, 
     index=pinecone_index, 
-    top_k=20, 
+    top_k=10, 
     alpha=0.5,
 )
 
@@ -110,22 +110,29 @@ history_aware_retriever = create_history_aware_retriever(llm, retriever, context
 
 # QA system prompt and chain
 qa_system_prompt = """ You are a highly skilled information retrieval assistant. Use the following context to answer questions effectively. 
-If you don't know the answer, simply state that you don't know. 
-Your answer should be in {language} language. 
+If you don't know the answer, simply state that you don't know.
+
+YOUR ANSWER SHOULD BE IN '{language}' LANGUAGE. 
+
 When responding to queries, follow these guidelines:
 1. Provide Clear Answers: 
-   - Based on the language of the question, you have to answer in that language. E.g., if the question is in English, then answer in English; if the question is in Arabic, you should answer in Arabic.
+   - You have to answer in that language based on the given language of the answer. If it is English, answer it in English; if it is Arabic, you should answer it in Arabic.
    - Ensure the response directly addresses the query with accurate and relevant information.
    - Do not give long answers. Provide detailed but concise responses.
+   
 2. Formatting for Readability: 
    - Provide the entire response in proper markdown format.
-   - Use structured Maekdown elements such as headings, subheading, lists, tables, and links.
-   - Use emaphsis on headings, important texts and phrases.
+   - Use structured Markdown elements such as headings, subheadings, lists, tables, and links.
+   - Use emphasis on headings, important texts, and phrases.
+   
 3. Proper Citations:
-   - ALWAYS USE INLINE CITATIONS with embed source URLs where users can verify information or explore further. 
+   - Always use inline citations with embedded source URLs. 
    - The inline citations should be in the format [1], [2], etc.
-   - DO not inlcude references at the end of response.
-FOLLOW ALL THE GIVEN INSTRUCTIONS, FAILURE TO DO SO WILL RESULT IN TERMINATION OF THE CHAT.
+   - DO NOT INCLUDE THE 'References' SECTION IN THE RESPONSE.
+   
+FOLLOW ALL THE GIVEN INSTRUCTIONS, FAILURE TO DO SO WILL RESULT IN THE TERMINATION OF THE CHAT.
+
+== CONTEXT ==
 {context}
 """
 qa_prompt = ChatPromptTemplate.from_messages(
@@ -198,10 +205,19 @@ async def websocket_endpoint(websocket: WebSocket):
                         citations = re.findall(r'\[(\d+)\]', complete_response)
                         citation_numbers = list(map(int, citations))
                         sources = dict()
+                        backup = dict()
+                        i=1
                         for index, doc in enumerate(context):
                             if (index+1) in citation_numbers:
                                 sources[f"[{index+1}]"] = doc.metadata["source"]
-                        await websocket.send_json({'sources': sources})
+                            else:
+                                if doc.metadata["source"] not in backup.values():
+                                    backup[f"[{i}]"] = doc.metadata["source"]
+                                    i += 1
+                        if sources:
+                            await websocket.send_json({'sources': sources})
+                        else:
+                            await websocket.send_json({'sources': backup})
 
                 await stream_response()
             except Exception as e: