docs(weave): fix feedback example to support text feedback by improving streamlit …app structure. standardize around .env instead of .toml for secrets.

docs/docs/reference/gen_notebooks/feedback_prod.md

@@ -18,22 +18,24 @@ title: Log Feedback from Production
 
 It is often hard to automatically evaluate a generated LLM response so, depending on your risk tolerance, you can gather direct user feedback to find areas to improve.
 
-In this tutorial, we'll use a custom RAG chatbot as an example app with which the users can interact and which allows us to collect user feedback.
+In this tutorial, we'll use a custom chatbot as an example app from which to collect user feedback.
 We'll use Streamlit to build the interface and we'll capture the LLM interactions and feedback in Weave.
 
 ## Setup
 
 
 ```python
-!pip install weave openai streamlit
+!pip install weave openai streamlit wandb
+!pip install set-env-colab-kaggle-dotenv -q # for env var
 ```
 
-First, create a file called `secrets.toml` and add an OpenAI key so it works with [st.secrets](https://docs.streamlit.io/develop/api-reference/connections/st.secrets). You can [sign up](https://platform.openai.com/signup) on the OpenAI platform to get your own API key. 
-
 
 ```python
-# secrets.toml
-OPENAI_API_KEY = "your OpenAI key"
+# Add a .env file with your OpenAI and WandB API keys
+from set_env import set_env
+
+_ = set_env("OPENAI_API_KEY")
+_ = set_env("WANDB_API_KEY")
 ```
 
 Next, create a file called `chatbot.py` with the following contents:
@@ -43,17 +45,28 @@ Next, create a file called `chatbot.py` with the following contents:
 # chatbot.py
 
 import streamlit as st
+import wandb
 from openai import OpenAI
+from set_env import set_env
 
 import weave
 
-st.title("Add feedback")
+_ = set_env("OPENAI_API_KEY")
+_ = set_env("WANDB_API_KEY")
+
+# highlight-next-line
+wandb.login()
+
+# highlight-next-line
+weave_client = weave.init("feedback-example")
+
+oai_client = OpenAI()
 
 
 # highlight-next-line
 @weave.op
 def chat_response(prompt):
-    stream = client.chat.completions.create(
+    stream = oai_client.chat.completions.create(
         model="gpt-4o",
         messages=[
             {"role": "user", "content": prompt},
@@ -68,60 +81,103 @@ def chat_response(prompt):
     return {"response": response}
 
 
-client = OpenAI(api_key=st.secrets["OPENAI_API_KEY"])
-# highlight-next-line
-weave_client = weave.init("feedback-example")
-
-
 def display_chat_messages():
-    for message in st.session_state.messages:
+    for idx, message in enumerate(st.session_state.messages):
         with st.chat_message(message["role"]):
             st.markdown(message["content"])
 
+            # Only show feedback options for assistant messages
+            if message["role"] == "assistant":
+                # Get index of this call in the session state
+                call_idx = (
+                    sum(
+                        m["role"] == "assistant"
+                        for m in st.session_state.messages[: idx + 1]
+                    )
+                    - 1
+                )
 
-def get_and_process_prompt():
+                # Create a container for feedback options
+                feedback_container = st.container()
+                with feedback_container:
+                    col1, col2, col3 = st.columns([1, 1, 4])
+
+                    # Thumbs up button
+                    with col1:
+                        # highlight-next-line
+                        if st.button("👍", key=f"thumbs_up_{idx}"):
+                            if "calls" in st.session_state and call_idx < len(
+                                st.session_state.calls
+                            ):
+                                # highlight-next-line
+                                st.session_state.calls[call_idx].feedback.add_reaction(
+                                    "👍"
+                                )
+                                st.success("Thanks for the feedback!")
+
+                    # Thumbs down button
+                    with col2:
+                        # highlight-next-line
+                        if st.button("👎", key=f"thumbs_down_{idx}"):
+                            if "calls" in st.session_state and call_idx < len(
+                                st.session_state.calls
+                            ):
+                                # highlight-next-line
+                                st.session_state.calls[call_idx].feedback.add_reaction(
+                                    "👎"
+                                )
+                                st.success("Thanks for the feedback!")
+
+                    # Text feedback
+                    with col3:
+                        feedback_text = st.text_input(
+                            "Feedback", key=f"feedback_input_{idx}"
+                        )
+                        if st.button("Submit Feedback", key=f"submit_feedback_{idx}"):
+                            if feedback_text and call_idx < len(st.session_state.calls):
+                                # highlight-next-line
+                                st.session_state.calls[call_idx].feedback.add_note(
+                                    feedback_text
+                                )
+                                st.success("Feedback submitted!")
+
+
+def show_chat_prompt():
     if prompt := st.chat_input("What is up?"):
         st.session_state.messages.append({"role": "user", "content": prompt})
 
-        with st.chat_message("user"):
-            st.markdown(prompt)
-
         with st.chat_message("assistant"):
-            # highlight-next-line
             with weave.attributes(
                 {"session": st.session_state["session_id"], "env": "prod"}
             ):
-                # This could also be weave model.predict.call if you're using a weave.Model subclass
-                result, call = chat_response.call(
-                    prompt
-                )  # call the function with `.call`, this returns a tuple with a new Call object
-                # highlight-next-line
-                st.button(
-                    ":thumbsup:",
-                    on_click=lambda: call.feedback.add_reaction("👍"),
-                    key="up",
-                )
-                # highlight-next-line
-                st.button(
-                    ":thumbsdown:",
-                    on_click=lambda: call.feedback.add_reaction("👎"),
-                    key="down",
-                )
+                result, call = chat_response.call(prompt)
+                st.write(result["response"])
                 st.session_state.messages.append(
                     {"role": "assistant", "content": result["response"]}
                 )
+                # highlight-next-line
+                st.session_state.calls.append(call)
+
 
+def init_session_state():
+    if "session_id" not in st.session_state:
+        st.session_state["session_id"] = "123abc"
 
-def init_chat_history():
     if "messages" not in st.session_state:
-        st.session_state.messages = st.session_state.messages = []
+        st.session_state.messages = []
+
+    if "calls" not in st.session_state:
+        st.session_state.calls = []
 
 
 def main():
-    st.session_state["session_id"] = "123abc"
-    init_chat_history()
+    st.title("Add feedback")
+
+    init_session_state()
     display_chat_messages()
-    get_and_process_prompt()
+    show_chat_prompt()
+
+    st.rerun()
 
 
 if __name__ == "__main__":