Update app.py

app.py

@@ -47,13 +47,19 @@ def invoke(openai_api_key, youtube_url, process_video, prompt):
     #print(result)
     return result["result"]
 
-description = """<strong>Overview:</strong> The app demonstrates how to use a <strong>Large Language Model</strong> (LLM) with <strong>Retrieval Augmented Generation 
-                 </strong> (RAG) on external data (YouTube videos in this case, but it could be PDFs, URLs, databases, or other structured/unstructured private/public 
-                 <a href='https://raw.githubusercontent.com/bstraehle/ai-ml-dl/c38b224c196fc984aab6b6cc6bdc666f8f4fbcff/langchain/document-loaders.png'>data sources</a>).\n\n
-                 <strong>Instructions:</strong> Enter an OpenAI API key, YouTube URL, and prompt to perform semantic search, sentiment analysis, summarization, 
-                 translation, etc. To ask multiple questions related to the same video, set "Process Video" to "True" on the first submission and then to "False". 
-                 The example is a 3:12 min. video about GPT-4 and takes less than 30 sec. to process. Experiment with different prompts, for example "what is gpt-4, 
-                 answer in german" or "write a haiku about gpt-4".\n\n
+description = """<strong>Overview:</strong> The app demonstrates how to use a <strong>Large Language Model</strong> (LLM) with <strong>Retrieval Augmented Generation
+                 </strong> (RAG) on external data (YouTube videos in this case, but it could be PDFs, URLs, databases, or other structured/unstructured and private/public 
+                 <a href='https://raw.githubusercontent.com/bstraehle/ai-ml-dl/c38b224c196fc984aab6b6cc6bdc666f8f4fbcff/langchain/document-loaders.png'>data sources</a>).
+                 \n\n<strong>Instructions:</strong> Enter an OpenAI API key, YouTube URL, and prompt to perform semantic search, sentiment analysis, summarization, 
+                 translation, etc. 
+                 <ol>
+                 <li>Submit prompt "what is gpt-4". The LLM without RAG does not know the answer.</li>
+                 <li>Select "Process Video" equals "True" and submit prompt "what is gpt-4". The LLM with RAG knows the answer.</li>
+                 <li>Experiment with different prompts, for example "what is gpt-4, answer in german" or "write a haiku about gpt-4".</li>
+                 </ol>
+                 In a production system processing external data is done in a batch process, while prompting is done in a user interaction.\n\nA sample system could load 
+                 all <a href='https://www.youtube.com/playlist?list=PL2yQDdvlhXf_hIzmfHCdbcXj2hS52oP9r'>AWS re:Invent 2022</a> YouTube videos and enable LLM use cases 
+                 related to them.\n\n
                  <strong>Technology:</strong> <a href='https://www.gradio.app/'>Gradio</a> UI using <a href='https://platform.openai.com/'>OpenAI</a> API 
                  via AI-first <a href='https://www.langchain.com/'>LangChain</a> toolkit with <a href='https://openai.com/research/whisper'>Whisper</a> (speech-to-text) 
                  and <a href='https://openai.com/research/gpt-4'>GPT-4</a> (LLM) foundation models as well as AI-native