😃Solved-St.chat_message，how to make google gemini chat.send_message() keep memory during multiple queries

Using Streamlit
,
1

I am working on a gemini-based chatbot . On the second query, the chat.send_message() can’t remember the context in the last query, gives totally new answer，can anybody tell me how to fix the problem？The code is following.

        # Display chat messages and bot response
    if st.session_state.messages[-1]["role"] != "assistant":
        with st.chat_message("assistant"):
            with st.spinner("Thinking..."):
                context = ""
                with open("context.txt", 'r', encoding='utf-8') as f:
                    text_docs = f.read()
                if len(text_docs) != 0:
                    context = text_docs
                    with open("context.txt", 'w', encoding='utf-8') as f:
                        f.truncate(0)
                response = chat.send_message(context + prompt)
                st.markdown(response.text)
                if response is not None:
                    message = {"role": "assistant", "content": response.text}
                    st.session_state.messages.append(message)
2

The context seems to be always the empty string, so there is nothing to remember.

3

The context get string from context.txt for the first time, then it is cleared by f.truncate(0).

4

So next time you read from context.txt you get the empty string.

5

This topic was automatically closed 180 days after the last reply. New replies are no longer allowed.