• 0 Posts
  • 263 Comments
Joined 2 years ago
cake
Cake day: July 9th, 2023

help-circle











  • Also, a key part of how GPT-based LLMs work today is they get the entire context window as their input all at once. Where as a human has to listen/read a word at a time and remember the start of the conversation on their own.

    I have a theory that this is one of the reasons LLMs don’t understand the progression of time.



  • the kind of stuff that people with no coding experience make

    The first complete program I ever wrote was in Basic. It took an input number and rounded it to the 10s or 100s digit. I had learned just enough to get it running. It was using strings and a bunch of if statements, so it didn’t work for more than 3 digit numbers. I didn’t learn about modulo operations until later.

    In all honesty, I’m still pretty proud of it, I was in 4th or 5th grade after all 😂. I’ve now been programming for 20+ years.


  • xthexder@l.sw0.comtoProgrammer Humor@programming.devAI in reality
    link
    fedilink
    arrow-up
    9
    arrow-down
    1
    ·
    edit-2
    19 days ago

    I think part of the problem is that LLMs stop learning at the end of the training phase, while a human never stops taking in new information.

    Part of why I think AGI is so far away is because to run the training in real-time like a human, it would take more compute than currently exists. They should be focusing on doing more with less compute to find new more efficient algorithms and architectures, not throwing more and more GPUs at the problem. Right now 10x the GPUs gets you like 5-10% better accuracy on whatever benchmarks, which is not a sustainable direction to go.