Re:Anthropic's original take home assignment open sourced
by 0xffff2 on Thursday January 22, 2026 @01:20PM
(#46723121)
> who is often easily beaten by LLMs
Is that really the case? My experience is fairly limited, but I've found that the LLM's willingness to fill in plausible sounding (but not necessarily at all accurate) numbers where it needs them to be a significant hindrance when asking it to think about performance.
Is that really the case? My experience is fairly limited, but I've found that the LLM's willingness to fill in plausible sounding (but not necessarily at all accurate) numbers where it needs them to be a significant hindrance when asking it to think about performance.