21–26 of 26 posts
Interesting that they are doubling down on censorship ("safety and robustness"), given that a major advantage of DeepSeek is its lack of refusals in deployed variants and open weights (can't patch-in more censorship in weights after the fact).
Earlier quoted context omitted.
Kind of like an ai “thinking fast and thinking slow”.
Sort of? I don't see why thinking slow should inhibit the ability to follow instructions.
"The book's main thesis is a differentiation between two modes of thought: "System 1" is fast, instinctive and emotional; "System 2" is slower, more deliberative, and more logical. "
More discussion on official post: https://news.ycombinator.com/item?id=42890627
Earlier quoted context omitted.
Sort of? I don't see why thinking slow should inhibit the ability to follow instructions.
I think they're referencing "Thinking, Fast and Slow" - https://en.wikipedia.org/wiki/Thinking,_Fast_and_Slow "The book's main thesis is a differentiation between two modes of thought: "System 1" is fast, instinctive and emotional; "System 2" is slower, more deliberative, and more logical. "
In this case "fast, instinctive, and emotional" models are better at instruction following than "slower, more deliberative, and more logical" models.