Adding intro text

app.py

@@ -76,9 +76,23 @@ def bot(history, hyde=False):
     history[-1][1] = response.split('### Response: [|AI|]')[-1]
     return history, prompt_html
 
+intro_md = """
+# Arabic RAG
+This is a project to demonstrate Retreiver Augmented Generation (RAG) in Arabic and English. It uses 
+[Arabic Wikipedia](https://ar.wikipedia.org/wiki) as a base to answer questions you have. 
+A retriever ([sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2](https://huggingface.co/sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2/discussions/8))
+ will find the articles relevant to your query and include them in a prompt so the reader ([core42/jais-13b-chat](https://huggingface.co/core42/jais-13b-chat)) 
+ can then answer your questions on it.
+ 
+You can see the prompt clearly displayed below the chatbot to understand what is going to the LLM. 
+
+Note I'm using Inference Endpoint's Scale to Zero to save money on GPUs. If the staus shows its not "Running" send a 
+chat to wake it up. You will get a `500 error` and it will take ~7 min to wake up.
+"""
 
 with gr.Blocks() as demo:
-    endpoint_status = gr.Textbox(check_endpoint_status, label="Endpoint Status (send chat to wake up)", every=1)
+    gr.Markdown(intro_md)
+    endpoint_status = gr.Textbox(check_endpoint_status, label="Inference Endpoint Status", every=1)
     with gr.Tab("Arabic-RAG"):
         chatbot = gr.Chatbot(
                 [],
@@ -152,7 +166,5 @@ with gr.Blocks() as demo:
         # Turn it back on
         hyde_txt_msg.then(lambda: gr.Textbox(interactive=True), None, [hyde_txt], queue=False)
 
-        # Examples
-
 demo.queue()
 demo.launch(debug=True)