Update app.py

app.py

@@ -16,16 +16,16 @@ _ = load_dotenv(find_dotenv())
 
 #openai.api_key = os.environ["OPENAI_API_KEY"]
 
-template     = """Answer the question at the end. If you don't know the answer, just say that you don't know, don't try to make up an answer. Keep the answer as concise as 
-                  possible. Always say "🚀 Thanks for using the app - Bernd Straehle." at the end of the answer. 
-                  Question: {question} Helpful Answer: """
+template = """If you don't know the answer, just say that you don't know, don't try to make up an answer. Keep the answer as concise as possible. Always say 
+              "🔥 Thanks for using the app - Bernd Straehle." at the end of the answer. """
 
-rag_template = """Use the following pieces of context to answer the question at the end. If you don't know the answer, just say that you don't know, don't try to make up 
-                  an answer. Keep the answer as concise as possible. Always say "🔥 Thanks for using the app - Bernd Straehle." at the end of the answer. 
-                  {context} Question: {question} Helpful Answer: """
+llm_template = "Answer the question at the end. " + template + "Question: {question} Helpful Answer: "
+rag_template = "Use the following pieces of context to answer the question at the end. " + template + "{context} Question: {question} Helpful Answer: "
 
-CHAIN_PROMPT     = PromptTemplate(input_variables = ["question"], template = template)
-RAG_CHAIN_PROMPT = PromptTemplate(input_variables = ["context", "question"], template = rag_template)
+LLM_CHAIN_PROMPT = PromptTemplate(input_variables = ["question"], 
+                                  template = llm_template)
+RAG_CHAIN_PROMPT = PromptTemplate(input_variables = ["context", "question"], 
+                                  template = rag_template)
 
 CHROMA_DIR  = "docs/chroma"
 YOUTUBE_DIR = "docs/youtube"
@@ -35,21 +35,31 @@ YOUTUBE_URL = "https://www.youtube.com/watch?v=--khbXchTeE"
 MODEL_NAME  = "gpt-4"
 
 def invoke(openai_api_key, use_rag, prompt):
-    llm = ChatOpenAI(model_name = MODEL_NAME, openai_api_key = openai_api_key, temperature = 0)
+    llm = ChatOpenAI(model_name = MODEL_NAME, 
+                     openai_api_key = openai_api_key, 
+                     temperature = 0)
     if (use_rag):
         if (os.path.isdir(CHROMA_DIR)):
-            vector_db = Chroma(persist_directory = CHROMA_DIR, embedding_function = OpenAIEmbeddings())
+            vector_db = Chroma(embedding_function = OpenAIEmbeddings(),
+                               persist_directory = CHROMA_DIR)
         else:
-            loader = GenericLoader(YoutubeAudioLoader([YOUTUBE_URL], YOUTUBE_DIR), OpenAIWhisperParser())
+            loader = GenericLoader(YoutubeAudioLoader([YOUTUBE_URL], YOUTUBE_DIR), 
+                                   OpenAIWhisperParser())
             docs = loader.load()
-            text_splitter = RecursiveCharacterTextSplitter(chunk_size = 1500, chunk_overlap = 150)
+            text_splitter = RecursiveCharacterTextSplitter(chunk_overlap = 150,
+                                                           chunk_size = 1500)
             splits = text_splitter.split_documents(docs)
-            vector_db = Chroma.from_documents(documents = splits, embedding = OpenAIEmbeddings(), persist_directory = CHROMA_DIR)
-        rag_chain = RetrievalQA.from_chain_type(llm, retriever = vector_db.as_retriever(search_kwargs = {"k": 3}), return_source_documents = True, chain_type_kwargs = {"prompt": RAG_CHAIN_PROMPT})
+            vector_db = Chroma.from_documents(documents = splits, 
+                                              embedding = OpenAIEmbeddings(), 
+                                              persist_directory = CHROMA_DIR)
+        rag_chain = RetrievalQA.from_chain_type(llm, 
+                                                chain_type_kwargs = {"prompt": RAG_CHAIN_PROMPT}, 
+                                                retriever = vector_db.as_retriever(search_kwargs = {"k": 3}), 
+                                                return_source_documents = True)
         result = rag_chain({"query": prompt})
         result = result["result"]
     else:
-        chain = LLMChain(llm = llm, prompt = CHAIN_PROMPT)
+        chain = LLMChain(llm = llm, prompt = LLM_CHAIN_PROMPT)
         result = chain.run({"question": prompt})
     return result
 
@@ -59,8 +69,8 @@ description = """<strong>Overview:</strong> The app demonstrates how to use a La
                  <strong>Instructions:</strong> Enter an OpenAI API key and perform LLM use cases on a <a href='https://www.youtube.com/watch?v=--khbXchTeE'>short video about GPT-4</a> 
                  (semantic search, sentiment analysis, summarization, translation, etc.)
                  <ul style="list-style-type:square;">
-                 <li>Set "Use RAG" to "False" and submit prompt "what is gpt-4". The LLM <strong>without</strong> RAG does not know the answer.</li>
-                 <li>Set "Use RAG" to "True" and submit prompt "what is gpt-4". The LLM <strong>with</strong> RAG knows the answer.</li>
+                 <li>Set "Retrieval Augmented Generation" to "False" and submit prompt "what is gpt-4". The LLM <strong>without</strong> RAG does not know the answer.</li>
+                 <li>Set "Retrieval Augmented Generation" to "True" and submit prompt "what is gpt-4". The LLM <strong>with</strong> RAG knows the answer.</li>
                  <li>Experiment with different prompts, for example "what is gpt-4, answer in german" or "write a poem about gpt-4".</li>
                  </ul>
                  In a production system, processing external data would be done in a batch process. An idea for a production system would be to perform LLM use cases on the 
@@ -72,8 +82,8 @@ description = """<strong>Overview:</strong> The app demonstrates how to use a La
 
 gr.close_all()
 demo = gr.Interface(fn=invoke, 
-                    inputs = [gr.Textbox(label = "OpenAI API Key", value = "sk-", lines = 1), gr.Radio([True, False], label="Use RAG", value = False), gr.Textbox(label = "Prompt", value = "what is gpt-4", lines = 1)],
+                    inputs = [gr.Textbox(label = "OpenAI API Key", value = "sk-", lines = 1), gr.Radio([True, False], label="Retrieval Augmented Generation", value = False), gr.Textbox(label = "Prompt", value = "what is gpt-4", lines = 1)],
                     outputs = [gr.Textbox(label = "Completion", lines = 1)],
                     title = "Generative AI - LLM & RAG",
                     description = description)
-demo.queue().launch()
+demo.launch()