Remix.run Logo
▲ perching_aix 2 hours ago

... as a test / benchmark? Hard to make a comparison without a control...

▲suddenlybananas 2 hours ago | parent [-]

Yeah, but obviously doing a formalisation that has been done before is different than doing a new one since you have knowledge about the previous attempt. It also just seems less impressive than doing something new and the reason they are reporting this is marketing.

▲perching_aix 2 hours ago | parent [-]

do you assume or do you know that the previous attempt is in the training data?