feat: 01_07e

README.md

@@ -2,37 +2,36 @@
 This is the repository for the LinkedIn Learning course `Hands-On AI: Building and Deploying LLM-Powered Apps`. The full course is available from [LinkedIn Learning][lil-course-url].
 
 _See the readme file in the main branch for updated instructions and information._
-## Lab1: Introduction to Chainlit
-We will be using [Chainlit](https://docs.chainlit.io/get-started/overview) as the frontend framework to develop our LLM Powered applications. Chainlit is an open-source Python package that makes it incredibly fast to build Chat GPT like applications with your own business logic and data.
+## Lab2: Adding LLM to Chainlit App
+Now we have a web interface working, we will now add an LLM to our Chainlit app to have our simplified version of ChatGPT. We will be using [Langchain](https://python.langchain.com/docs/get_started/introduction) as the framework for this course. It provides easy abstractions and a wide varieties of data connectors and interfaces for everything LLM app development.
 
-In this lab, we will put up a very simple Chainlit application that echos a user's query.
+In this lab, we will be adding an Chat LLM to our Chainlit app. 
 
-For example, if user says
-
-```
-hello
-```
+## Exercises
 
-Our Chainlit app will respond with 
+We will build on top of our existing chainlit app code in `app/app.py` in the `app` folder. As in our previous app, we added some template code and instructions in `app/app.py`
 
-```
-Received: hello
-```
+1. Please go through the exercises in `app/app.py`. 
 
-The learning objective is to familiarize with Chainlit's framework and to launch the application.
+2. Please lanuch the application by running the following command on the Terminal:
 
-## Exercises
+```bash
+chainlit run app/app.py -w
+```
 
-We have created some template code in `app/app.py` in the `app folder`.
+## Solution
 
-1. Please go through [Chainlit's documentation](https://docs.chainlit.io/get-started/pure-python) and answer the questions in `app/app.py`
+Please see `app/app.py`.
 
-2. Please lanuch the application by running the following command on the Terminal:
+Alternatively, to launch the application, please run the following command on the Terminal:
 
 ```bash
 chainlit run app/app.py -w
 ```
 
-## Solution
 
-Please see app/app.py.
+## References
+
+- [Langchain's Prompt Template](https://python.langchain.com/docs/modules/model_io/prompts/prompt_templates/#chatprompttemplate)
+- [Langchain documentation](https://python.langchain.com/docs/modules/chains/foundational/llm_chain#legacy-llmchain)
+- [Chainlit's documentation](https://docs.chainlit.io/get-started/pure-python)

app/app.py

@@ -1,19 +1,73 @@
 import chainlit as cl
+from langchain.chat_models import ChatOpenAI
+from langchain.prompts import ChatPromptTemplate
+from langchain.schema import StrOutputParser
+from langchain.chains import LLMChain
+
+@cl.on_chat_start
+async def on_chat_start():
+    ##########################################################################
+    # Exercise 1a:
+    # Our Chainlit app should initialize the LLM chat via Langchain at the
+    # start of a chat session.
+    # 
+    # First, we need to choose an LLM from OpenAI's list of models. Remember
+    # to set streaming=True for streaming tokens
+    ##########################################################################
+    model = ChatOpenAI(
+        model="gpt-4-1106-preview",
+        streaming=True
+    )
+
+    ##########################################################################
+    # Exercise 1b:
+    # Next, we will need to set the prompt template for chat. Prompt templates
+    # is how we set prompts and then inject informations into the prompt.
+    # 
+    # Please create the prompt template using ChatPromptTemplate. Use variable
+    # name "question" as the variable in the template.
+    # Refer to the documentation listed in the README.md file for reference.
+    ##########################################################################
+    prompt = ChatPromptTemplate.from_messages(
+        [
+            (
+                "system",
+                "You are Chainlit GPT, a helpful assistant.",
+            ),
+            (
+                "human",
+                "{question}"
+            ),
+        ]
+    )
+    ##########################################################################
+    # Exercise 1c:
+    # Now we have model and prompt, let's build our Chain. A Chain is one or a
+    # series of LLM calls.We will use the default StrOutputParser to parse the
+    # LLM outputs.
+    ##########################################################################
+    chain = LLMChain(llm=model, prompt=prompt, output_parser=StrOutputParser())
+
+    # We are saving the chain in user_session, so we do not have to rebuild
+    # it every single time.
+    cl.user_session.set("chain", chain)
 
 
-##############################################################################
-# Exercise 1a: 
-# Please add the proper decorator to this main function so Chainlit will call
-# this function when it receives a message
-##############################################################################
 @cl.on_message
 async def main(message: cl.Message):
 
-##############################################################################
-# Exercise 1b:
-# Please get the content of the chainlit Message and send it back as a
-# response
-##############################################################################
-    response = message.content
+    # Let's load the chain from user_session
+    chain = cl.user_session.get("chain")  # type: LLMChain
+
+    ##########################################################################
+    # Exercise 1d:
+    # Everytime we receive a new user message, we will get the chain from 
+    # user_session. We will run the chain with user's question and return LLM
+    # response to the user.
+    ##########################################################################
+    response = await chain.arun(
+        question=message.content, callbacks=[cl.LangchainCallbackHandler()]
+    )
+
+    await cl.Message(content=response).send()
 
-    await cl.Message(content=f"Received: {response}").send()