3 ms·For inference... Still accessible, but expensive to run at scale. And training even worse.by qeternity 5y agoFor inference... Still accessible, but expensive to run at scale. And training even worse.