NCSoft's 'Llama Varco 8B Instruct' recorded the lowest score at 2 points. When they made domestic AIs solve the CSAT (Korean SAT)... 'NCSoft Model 2 points, most others in the 20s.' They ended up creating Artificial Stupidity.
"The community is torn between roasting NCSoft for creating 'Artificial Stupidity' and technically defending them, pointing out that using an ancient 8B Llama model against brand-new 72B models is just an unfair comparison. Still, LOL at the 2-point score."
#FunContinue Browsing
