Update app.py

app.py

@@ -15,6 +15,8 @@ from langchain_core.runnables import RunnableParallel, RunnablePassthrough
 from langchain_core.documents import Document
 huggingface_token = os.environ.get("HUGGINGFACE_TOKEN")
 
+# Memory database to store question-answer pairs
+memory_database = {}
 def load_and_split_document_basic(file):
    """Loads and splits the document into pages."""
    loader = PyPDFLoader(file.name)
@@ -59,11 +61,11 @@ def get_model(temperature, top_p, repetition_penalty):
            "temperature": temperature,
            "top_p": top_p,
            "repetition_penalty": repetition_penalty,
-           "max_length": 1000
+           "max_length": 512
        },
        huggingfacehub_api_token=huggingface_token
    )
-def generate_chunked_response(model, prompt, max_tokens=1000, max_chunks=5):
+def generate_chunked_response(model, prompt, max_tokens=500, max_chunks=5):
    full_response = ""
    for i in range(max_chunks):
        chunk = model(prompt + full_response, max_new_tokens=max_tokens)
@@ -95,10 +97,17 @@ def update_vectors(files, use_recursive_splitter):
 def ask_question(question, temperature, top_p, repetition_penalty):
    if not question:
        return "Please enter a question."
+   # Check if the question exists in the memory database
+   if question in memory_database:
+       return memory_database[question]
    embed = get_embeddings()
    database = FAISS.load_local("faiss_database", embed, allow_dangerous_deserialization=True)
    model = get_model(temperature, top_p, repetition_penalty)
-   return response(database, model, question)
+   # Generate response from document database
+   answer = response(database, model, question)
+   # Store the question and answer in the memory database
+   memory_database[question] = answer
+   return answer
 def extract_db_to_excel():
    embed = get_embeddings()
    database = FAISS.load_local("faiss_database", embed, allow_dangerous_deserialization=True)