News for hackers. Stuff that matters.
Re:Learning to Reason with LLMs by seydor on Thursday September 12, 2024 @02:15PM (#41523830)
looks a bit like 'code', using keywords 'Hmm', 'Alternatively', 'Perfect'
Re:Learning to Reason with LLMs by legel on Thursday September 12, 2024 @04:01PM (#41524885)
As a technical engineer, I’ve learned the value of starting sentences with “basically”, even when I’m facing technical uncertainty. Basically, “basically” forces me to be simple.

Being trained to say words like “Alternatively”, “But…”, “Wait!”, “So,” … based on some metric of value in focusing / switching elsewhere / … is basically brilliant.

Re:Learning to Reason with LLMs by thomasahle on Thursday September 12, 2024 @02:18PM (#41523868)
Right, these are not mere "filler words", but initialize specific reasoning paths.
Re:Learning to Reason with LLMs by mewpmewp2 on Thursday September 12, 2024 @02:47PM (#41524178)
Hmm... you may be onto something here.
Re:Learning to Reason with LLMs by j_maffe on Thursday September 12, 2024 @03:29PM (#41524589)
Alternatively, these might not be "filler words", but instantiate paths of reasonsing.
Re:Learning to Reason with LLMs by squigz on Thursday September 12, 2024 @03:55PM (#41524823)
What a strange comment chain.
Re:Learning to Reason with LLMs by seydor on Thursday September 12, 2024 @03:59PM (#41524865)
Hmmm.
Re:Learning to Reason with LLMs by squigz on Thursday September 12, 2024 @04:06PM (#41524954)
Interesting.
Re:Learning to Reason with LLMs by wrs on Thursday September 12, 2024 @02:53PM (#41524234)
Interesting.