Update app.py

app.py

@@ -9,52 +9,52 @@ from langchain_openai import ChatOpenAI
 from langchain.chains.combine_documents import create_stuff_documents_chain
 from langchain.embeddings import HuggingFaceEmbeddings
 
+# Embedding model
 embedding_model = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
 
+# Vector store setup
 persist_directory = 'vec_db'
+vectordb = Chroma(persist_directory=persist_directory, embedding_function=embedding_model)
+vectordb_retriever = vectordb.as_retriever(search_kwargs={'k': 5})
 
-vectordb = Chroma(persist_directory=persist_directory,
-                           embedding_function=embedding_model)
-
-vectordb_retriever = vectordb.as_retriever(search_kwargs={'k':5})
-
+# LLM
 llm = ChatOpenAI(model="gpt-4.1-nano", temperature=0.7)
 
+# Load instructions
 with open("instructions.txt", 'r') as file:
     instructions = file.read()
 
-
- # Custom prompt
+# Custom prompt
 custom_prompt = ChatPromptTemplate.from_messages([
     ("system", instructions),
     MessagesPlaceholder(variable_name="chat_history"),
     ("user", "Question: {input}\nContext: {context}")
 ])
 
- # Memory
+# Memory
 memory = ConversationBufferMemory(
     memory_key="chat_history",
     return_messages=True
 )
 
+# Chains
 question_answer_chain = create_stuff_documents_chain(llm, custom_prompt)
-
 chain = create_retrieval_chain(vectordb_retriever, question_answer_chain)
 
+# Main interaction function
 def conversate_assistant(query, history):
     greetings = {"hey", "hi", "hello"}
     normalized_query = query.strip().lower()
 
-    if len(memory.load_memory_variables({})["chat_history"]) >=6:
-        chat_history = memory.load_memory_variables({})["chat_history"][-6::]
-    else:
-        chat_history = memory.load_memory_variables({})["chat_history"]
+    # Load the last 6 messages from memory
+    chat_history = memory.load_memory_variables({})["chat_history"]
+    chat_history = chat_history[-6:] if len(chat_history) >= 6 else chat_history
 
-     # If greeting, skip retrieval and context
+    # If greeting, skip context retrieval
     if normalized_query in greetings:
         response = question_answer_chain.invoke({
             "input": query,
-            "context": [],  # empty context for greetings
+            "context": [],  # Empty context for greetings
             "chat_history": chat_history
         })
         answer = response
@@ -64,15 +64,17 @@ def conversate_assistant(query, history):
             "chat_history": chat_history
         })
         answer = response['answer']
-    
-    # Save to memory
+
+    # Save the interaction in memory
     memory.save_context({"input": query}, {"output": answer})
 
-     return answer
+    return answer
 
+# Gradio interface
 demo = gr.ChatInterface(
     conversate_assistant,
     type="messages"
 )
 
-demo.launch()
+# Launch the app
+demo.launch()