@SakanaAILabs These benchmarks compare harnessed results (Sakana's) to unharnessed results from other models. Apples to Oranges. When you compare Sakana's results to harnessed models they fall short even against Opus 4.8.
Hype not reality here unfortunately.
@fivewithflores I don't know exactly it works, but I assume it pulls the best year from the decade, so KD's 2008-09 (or 09-10?) season? Those aren't really top notch but ofc quibbles here, the other 4 are the best there is.
@MacroCRG@NYCMayor Pie has grown about 3.7X in 50 years. Median wages have grown almost nothing! It would be difficult to have a macro experiment thatβs failed worse than just grow the pie.
@toly I agree. No more 10s of billions in gov spending and credits for trillionaires. Time for real capitalism like they practice in sub-Saharan Africa! πͺπ»
@mert More like take more government subsidies, tax credits, and direct payments than anyone in the history of mankind, with considerably less profit generated (with a little fraud and grift thrown in).
More akin to communism, just aimed at wealthy people.