How to Make Your Own Custom ChatGPT With Internet Access + Memory Tutorial 

thank you!

😂😂😂 I will close the blinds my b

hmmm that's interesting. what are you passing in as tools? Are you adding new tools? what happens if you do something like tools=_my_create_csv_agent.tools in the from_agent_and_tools fn? i'm not sure that's ideal but i don't have a chance to test it rn lmk if it works haha

yes coming soon i apologize for the delay 🙏 🙏 thank u for patience

@@tylerwhatsgood sure! Also it will be helpful if you can show how we can embedded this as a chatbot on a website landing page.

I would recommend using chatgpt to generate the frontend. It’s pretty good at building this kind of code

.

hey! would love to help. i have many more videos coming, hopefully they are useful. if you have any specific questions just shoot me a DM on twitter @tylerwhatsgood_ Thank you!

@@tylerwhatsgood Hey Tyler, we should collaborate man! happy to help you with code!

yes Yuva you are absolutely right, this is one of the greatest parts! will make a video shortly about asking questions with Javascript to retrieve information from a CUSTOM document

@@tylerwhatsgood awesome !! Will be waiting for it.

Hey Jason! This should be possible, can you provide a little more insight into what your goal is? You want to use it for “story writing with my own material” meaning you would like it to write so that the output “sounds like you” (has some same writing style), do I have it right? Thanks for the comment, would love to help and I think we can probably figure this out!

@@tylerwhatsgood Say I want to load the pdf that I wrote into a chatgtp, I want to use it offline and have it remember what is in that pdf and be able to help me. maybe have a conversation with my characters or add more details. I want something that I can keep adding more information to with provided pdfs and such. like books or whatnot that I have written. Something also like a one-click install similar to nmkd sd.

@@tylerwhatsgood Something that would allow us to add in our own source material would be an awesome resource. Engineering books, history books, etc.

I believe it might be done with GPT4ALL and Langchain. I'm still trying to figure it out.

@@userx6679 hm ok i think there are a few options here. like Mohamed mentioned in comment below there are a few open source models that are not as good but maybe comparable to GPT4 capabilities that you can run offline. I like GPT4All and Vicuna, will have videos on those asap! I think however that your requirement of wanting to remain offline may make things a bit difficult: The "keep adding more information to with provided pdfs and such. like books or whatnot that I have written" will probably cause you to run into space contraints since your goal is to be totally offline in which case at some point you will have too many documents to save and load embeddings locally (Embeddings represent your documents). You can remedy this by saving embeddings in a vector store like Supabase or Pinecone then retrieving on-the-fly when you ask the question. Memory: You'd ultimately want a conversational AI that you can ask follow-up questions to, however the "memory" that your chatbot has will eventually make it slow as it grows, you're going to want to offload this somewhere and only retrieve relevant pieces of past conversation. Without doing this, your bot will only be able to query based on information in the last few chat messages since you'll run out of space both on computer and also in the input (# of tokens model accepts) for the prompt. Sorry for non-answer, I will try to get some building-blocks for this going but 100% offline is a bit of a mission but should be possible. I hope could help at least a little and i think what youre trying to accomplish is doable just might take a little searching for something similar and a little figuring out 😊 thank you for feedback it helps a ton!!