How to Build a CHAT BOT with ChatGPT API (GPT-3.5-TURBO) having CONVERSATIONAL MEMORY in Python

In this tutorial, we will use the #langchain, #openai, and Streamlit libraries to create a #chatbot - which is very different from our traditional chatbot in the aspect of remembering the previous context or conversation.

📹 Related Videos to watch :

🤖 Related Playlist :

📝 Blog Post to refer :

🔴 Don't forget to subscribe to my channel if you like my videos!

🔥 By becoming a patron, you'll get access to exclusive content, codes, or videos beforehand, one-to-one web app development / relevant discussion, live-chat with me on specific videos and other perks. ( FYI : Basic Tier is 50% cheaper than ChatGPT/monthly with benefits which an AI can't help with 😉 ):

#langchain #openai #streamlit #nlp #openaigpt3 #pythonprogramming #python #pythonforbeginners #webdevelopment #datascience #openaimodels #machinelearning #webapp #openaimodels #avra #avrastreamlit #chatbotgpt #chatbotdevelopment

Video Segments
0:00 Intro and What we are building today
3:54 Packages Necessary and Why
4:48 Initializing Session States
5:57 Start to write the App
7:30 Creating @OpenAI Instance
8:45 Implement LangChain modules : Conversation Memory & Chain
12:35 Other aspects of the App Development
18:36 Wrap up

[MuhammadBakhsh-go]

bundle of thanks for making a video .I really appreciate your efforts love from pakistan

[oryxchannel]

uuuh. downloading this for my native drive.

[vmxxwicbz]

Is there a need to use Langchain to ensure state? Why not just rely on OpenAI's append messages functionality and pass the entire thread back each time?

[swatiganji]

Hi Avra, its helpful video can you please make video on Chatbot with our own database with memory i.e it should answer the question related to database and memorize the context

[mjmammoth]

Great content! But the slow zoom focus transition effect made it unwatchable for me :/

[dairyprods]

in providing the api with context, does every single previous message get sent to the ai as well? for example, if I send something and I get a response back, and the interaction uses 10 tokens, and then I send something else, will the new message use more than 10 tokens because the context of the previous interaction needs to be sent as well? does that mean the longer a conversation goes on for, the higher the price for each message becomes?

[chronodeluxe]

"Hello! A few weeks ago, I came across a video and I'm interested in integrating its memory feature with a personal data source. I've been actively searching for ways to establish a connection between the memory function and my personal data source. By doing so, I aim to enable the memory function to reference past conversations. If possible, could you please provide me with the link to the video you mentioned?"

[manish]

It works fine with gpt3.5 turbo .. How to make it work with FAISS DB using local data(pdf)..

[tipsstudio]

Hello, I seen this video on some days ago, How can I connect this memory feature with personal data source. I am looking for that and trying from many times related this so I can connect my personal data source with embedding with the memory feature so it can also refer the past conversations.

[Trashpanda_]

Haven’t dug into this, but let’s say the bot only talks to users who are logged in, would it be possible to keep the chat history going?

[genelin]

is it possible to modify the chatroom interface to resemble a typical chatroom, where messages are displayed in a section and arranged from the bottom to the top on streamlit?

[sparshagarwal]

Hi there, How can one further personalize this based on each user and their micro interactions on the overall platform?
For context - we are building a social network around teenager' careers and introducing a chatbot to answer all their career and anxiety related things, but we also want our chatbot to learn from the activity the user is performing on the overall platform

[nedu]

Lovely work! Meanwhile, I am getting this error "AttributeError: 'ConversationEntityMemory' object has no attribute 'store'". Did you forget to initialize it? " especially when I want to go to the next chat. I have checked but can't seem to resolve it. Please, can you help?

[chrisnickerson]

Hi, is it possible to stream the text using the gpt3.5 model, or will it only return the text when the response is complete?

[courses]

How to train it on my own knowledge base?

[nmnicholas]

Thanks for the tutorial. Is it possible to use your own prompt? so that you can control the behavior of the chatbot

[sso]

Thanks for the great video. I have a question about token limits. For GPT-3.5-TURBO, the token limit for one session is 4096 t tokens. Does it mean that if the tokens for the question and the previous context are over the limit, the bot cannot remember the previous context(or it will not work)?

[TheKnowledgeAlchemist]

so if I want to write a novel and I'm worried about writing a scene that conflicts with another chapter, would I be able to ask "Can I kill this character", and have it say, "No, xyz is in the next scene"? Second, can I say "Write the link between this scene and the next", have it write it out, then add it into the memory?

[Xyotel]

Ive copied the video and the chatbot is working, but it seems its running on gpt3 not gpt3.5 as when i ask it what model it is, it says 3. how can i fix this?

[radbirden]

Hi can we combine this entity module with our embedded personal data ?