Комментарии:
I think this is missing the most important and definitely most difficult part of this model design. How do you scale the ChatGPT model for inference?
ОтветитьWondering if the interviewee came prepared for this question or its spontaneous :) just kidding. great content and great share. Kudos
ОтветитьIf we design the AI model as what will the researchers at Stanford and Oxford do 🤣🤣
ОтветитьWhy does thumbsupthumbsdown use the message ID that is sent by a user? Shouldn’t it apply to the chatGPT’s response? Am I missing where the “answer” is? At some point we should GET a message too? 🤔
ОтветитьVery good discussion.
ОтветитьWhat was the tool used to draw in the video ?
Ответитьawesome video, I have never worked with ML, no idea of any piece in the last section where he explained the ChatGPT models setup. Just wondering how to tackle such blank areas in an interview?
ОтветитьIts interesting that the first thing he thinks about is how to do sensorship. I was about to go off, but they won me over by talking about the chatgpt model int the last 12min 😂
ОтветитьHe went deep into the ML part as well. Is that also required for Sys Design interviews? What if I am not familiar with the ML concepts like reward model, supervised fine tuning etc?
ОтветитьA question: when you send a message, you return a message ID, but you don't talk about what happens to ChatGPT's response to the user's question. Does it get shown to the user at some point?
ОтветитьWow! awesome topic and realy deep dive explanation
ОтветитьLove it!
ОтветитьWhat is the tool he is using?
ОтветитьLove this topic!!!
ОтветитьWhat is the tool he is using?
ОтветитьVery helpful video! I was wondering which platform did you use to create the mindmap?
Ответить