News for hackers. Stuff that matters.
Re:MemGPT – LLMs with self-editing memory for unbounded context by callalex on Monday October 16, 2023 @03:20PM (#37904870)
It’s not suspicious, the previous title calling it an operating system was extremely incorrect and intentionally misleading to generate buzz. Aka clickbait.
Re:MemGPT – LLMs with self-editing memory for unbounded context by xcv123 on Monday October 16, 2023 @04:22PM (#37905987)
The previous title used Operating System as an analogy to describe a novel concept. OS ideas were referenced in the paper to help the reader understand their new creation. If you're not that smart and not good at abstract thinking then of course it would only confuse you.
Re:MemGPT – LLMs with self-editing memory for unbounded context by dang on Monday October 16, 2023 @05:17PM (#37906769)
Please don't cross into personal attack, no matter how wrong someone is or you feel they are. It only makes things worse.

https://news.ycombinator.com/newsguidelines.html

Re:MemGPT – LLMs with self-editing memory for unbounded context by xcv123 on Wednesday October 18, 2023 @12:57AM (#37924982)
I had already edited my comment to make it non-personal. You saw it just before I fixed it.
Re:MemGPT – LLMs with self-editing memory for unbounded context by dang on Wednesday October 18, 2023 @02:07AM (#37925316)
I appreciate the edit, which was definitely an improvement. But "If you're not used to reading academic papers then of course it would confuse you" is still a personal swipe. At least that's how it lands with me (and probably at least some others as well).
Re:MemGPT – LLMs with self-editing memory for unbounded context by xcv123 on Wednesday October 18, 2023 @03:28PM (#37933553)
The comment I was replying to was accusing the authors of the paper of intentionally misleading people. In my mind that is someone who is confused and assuming bad faith of the authors without making a genuine effort to understand their paper beyond the headline.
Re:MemGPT – LLMs with self-editing memory for unbounded context by dang on Wednesday October 18, 2023 @08:15PM (#37936581)
I hear you, but that's an orthogonal point, in the sense that you should have expressed this without saying something that would land as a swipe with readers.
Re:MemGPT – LLMs with self-editing memory for unbounded context by xcv123 on Wednesday October 18, 2023 @11:08PM (#37937854)
Yes agreed