ChatGPT API Announcement & Code Walkthrough with LangChain
Ғылым және технология
Colab Link : rli.to/zG1DW
In the past day OpenAI have released their new ChatGPT API. I made a quick video to talk about the announcement and to show how to access it in Python with the OpenAI Package and with LangChai
My Links:
Twitter - / sam_witteveen
Linkedin - / samwitteveen
Github:
github.com/samwit/langchain-t...
github.com/samwit/llm-tutorials
00:00 OpenAI ChatGPT API Announcement & Blog
01:46 Chat Markup Language
04:21 GPT-4 Leaked Info
04:58 Code: Basic ChatGPT API use with OpenAI package
08:50 LangChain ChatGPT API usage
#LangChain #BuildingAppswithLLMs #chatgpttutorial #chatgpt #chatgptexplained
Пікірлер: 11
Loved it. Langchain will be the Gstreamer like framework where different developers can leverage their developmental skill to create products. Before this it seems like most of the things have been done by the Big 4, and we as developers have nothing left to be done. Langchain is changing that, and democratizing LLMs and designing cost optimized usage of LLMs. And also of course the Myriad of Sql, Api etc.
Thank you so much for your content, this is such a goldmine that you never had to release or create, yet you did it from the heart and for free. You clearly spent endless hours curating, recording, and uploading all this content. Keep doing you, thank you ❤
@samwitteveenai
Жыл бұрын
Thanks for the kind words.
loved it
can it be used with a vector store?
@samwitteveenai
Жыл бұрын
Yes I have some videos coming doing this kind of thing
@smtkumar007
Жыл бұрын
@@samwitteveenai does using LangChain optimize the use of token?
You've spelled interesting wrong in your prompt.
For a real commercial application with lots of data and inference, it seems LLMs are going to be cost prohibitive
@samwitteveenai
22 күн бұрын
That video is super old. Models are much cheaper and still going to get cheaper. Checkout my recent video about LLM Changes
The role: system is almost 101% useless. Tell it that it's a bad speller, or that it only responds like a 20's gangster, or -well - anything. Won't do it. As openai admits - it's not really adhered to.