/TECHGoogle DeepMind's Dan Roy argues recent AI performance gains stem from better pre-training rather than new model capabilitiesDRSBThe claim sparked a technical debate with researcher Sebastien BubeckCombined views9152 posts, first seen 4h ago3 likesGoogle DeepMind's Dan Roy argues recent AI performance gains stem from better pre-training rather than new model capabilitiesDRSBThe claim sparked a technical debate with researcher Sebastien BubeckSource Posts on XCluster sourcesDan Roy@roydanroyTECH@SebastienBubeck @brianryhuang This story can be explained by better pre training alone. What makes you think it has to do with capabilities?Combined views9152 posts, first seen 4h ago3 likes comments
Dan Roy@roydanroyTECH@SebastienBubeck @brianryhuang This story can be explained by better pre training alone. What makes you think it has to do with capabilities?
Dan Roy@roydanroyTECH@SebastienBubeck @brianryhuang This story can be explained by better pre training alone. What makes you think it has to do with capabilities?
Sebastien Bubeck@SebastienBubeckTECH@roydanroy @brianryhuang I do not understand the question, can you rephrase it?
Sebastien Bubeck@SebastienBubeckTECH@roydanroy @brianryhuang I do not understand the question, can you rephrase it?