3 ms·
Then we'll create a distributed infrastructure for the creation of models. Run some program and donate spare GPU cycles to generate public AI tools that will be
by krunck 3y ago
Then we'll create a distributed infrastructure for the creation of models. Run some program and donate spare GPU cycles to generate public AI tools that will be made available to all.
- eurekin 3y agoI really really would like to believe this could work in practice. Given current data volume used during the training phase (tb/s), I highly doubt it's possible without two, magnitude changing, breakthroughs at once