Create a chatbot linked to my data using Langchain

There are many limitations to creating chatbots using generative AI models. Language models are updateThere are many cases where I haven't been able to learn. AI models often respond inaccurately (halalization) to unlearned information. too textual I can't handle anything else, Internet searchThere are also models that can't do it. Response content save as fileI can't even do tasks like that.

Using LAN Chain, you can ask questions about documents, search the Internet, and save the results as a file. This class uses a chain that can link information unknown to the language model, A chatbot that searches within your own dataI will create

Wouldn't you like to create a chatbot where the chatbot asks questions based on a database of many of the data you have?

Python coding with AI

What is the most popular programming language these days PythonThis is it. There are so many things you can do with Python. AI service developmentis one of them. Also, generative AI can be used as a coding aid. Coding using AI can increase development speed and quality, and it is relatively easy to code.

Generative AI recommends code blocks. It's easy to code with the right context, comments, and prompts. AI coding aidIt also provides error correction and refactoring functions. The purpose of this class is to enable anyone who can code in Python to develop chatbots.

AI coding aid used in this lesson

When coding in Python, AI coding aidThey can help a lot. It is provided as an extension of the IDE or can be used in the form of an IDE, which is why Job workspacesYou can get suggestions for source code blocks by referring to all of the files in If you open all relevant files in an editor and get source code suggestions, you can get more accurate suggestions.

Most AI coding aids chatting It comes with features. You can also ask questions about things you don't know while coding, ask questions about the source code open in the editor, modify the source code, and generate it.

While coding The error is not well resolved Have you ever had trouble? From now on, AI coding aids and generative AI will help classmates code. Please freely implement the chatbot of your choice by freely coding in Python.

GitHub Copilot

It's a well-known AI coding aid, and I'm planning to install the VS Code extension and try it out. GitHub and lots of open source codeAfter learning, it suggests source code blocks and provides rich functionality.

Amazon Code Whisperer

Use Code Whisperer integrated into Amazon Q to receive code block suggestions. All features Totally freeIt can be used as, and can be used without troublesome procedures such as creating an account.

Cursor IDE

It was developed based on Visual Studio Code IDE equipped with AIThis is it. privacy modeSince it is provided, you can prevent the source code from being used to train the model. In the case of source code that must not be exposed to the outside world, security can be ensured.

I'm going to learn something like this

How to make good use of AI coding aids

AI coding aids can improve the accuracy and speed of coding, but always Accurate suggestionsIt's not about doing it. They also suggest source code that doesn't work, and they also generate code that is beyond the developer's intention. By handling AI coding aids a little better, Desired source codeIt covers how to get it.

To generative AI Coding questionsI'll also show you how to make it a little more effective.

Prompt engineering

In order to implement a chatbot, it is essential to use prompts. When querying an AI model, Properly designed promptsIf you send it, you can get slightly better results. The quality of AI model responses The quality of AI servicesSince it influences, here are some useful prompt engineering methodologies.

Creating a chatbot's UI with gradio

If you have implemented a chatbot, you need a user interface. To gradio Implement UI easily and simplyIt covers how to do it. You can implement the required interface very quickly. The UI implemented in gradio can be accessed via a browser web-based It is a UI, and public UI can also be created so that other users can access it.

Using generative AI APIs

Learn how to use generative AI APIs. A key is issued, and the issued key is externally without exposing How to use, model Set parametersI'll explain how to do it.

Creating chatbots using generative AI models

Let's call the OpenAI API and Claude API. Call the API by adjusting the parameters of the model according to the service you want to implement, and implement a psychological counseling chatbot and an English writing correction chatbot.

Implementing a chatbot is absolutely not difficult. The chatbot is completed by implementing only one source code file and one configuration file, and the source code is less than a few lines. Since AI models do most of the work, chatbots that seem to have many functions are actually very Simple implementationThis is possible. Once you've implemented a chatbot, you'll be amazed at how simple and easy it is to implement it!

Creating a chatbot using Langchain

There are many limitations to creating chatbots using only generative AI models. It's also a bit uncomfortable. By using LANchain, chatbots with complex functions can be implemented more easily and effectively. Prompt's templatesSpecify, or parse the model's response in an appropriate form, InternetYou can implement a chatbot that connects to and accesses files provided by users. Remember chat historyYou can implement a chatbot and configure multiple continuous tasks. Chatbot response results save as fileYou can do it. own OneDriveme Google DriveIt is also possible to access

Let's easily implement the functions that classmates want with Langchain~

Resources for students

We will provide materials to help classmates implement chatbots.

After watching a video once, you don't remember? Course presentation materialsWe will give you all in PDF.

I tried implementing a chatbot while watching a lecture, but I got an error? You can't solve the mission no matter what you do?

Implemented in a lecture Best practices for all source code and missionsWe will provide you with a file.

Attachments for the first lesson of each chapterPlease check it~ and Attachments to the mission explanation coursePlease check it out.

All source code implemented in the course and demonstration prompt files

To all source code implemented in the course Supplied by adding commentsI'll do it for you. Try a demonstration, or in the course materials prompt fileWe will also give you all of them. If it doesn't work after trying to code, open it and check it out.

Course materials PDF files

All presentation files used in the course will be converted to PDF and provided. Inside the data The URL you need to accessIna, Information to rememberIt is included, so please save it and try implementing the chatbot.