On Mon, Apr 8, 2013 at 10:15 PM, Ben Goertzel <[email protected]> wrote: >> Then you need about a decade's worth of video to train the system. You >> probably don't want to wait a decade for each experiment, so you >> really want something faster than 200 TFLOPS. > > We can save and load DeSTIN states, and we have also worked out the > mathematics to merge DeSTIN states from different instances of DeSTIN > > So we could train multiple instances in parallel and merge their knowledge > into each other... > > But this is a problem for a little later...
How does that help? With a single 4 TFLOP GPU, that's 500 years for one experiment. Or are you planning some scaled down experiments like reading text or recognizing faces? Or something else? -- -- Matt Mahoney, [email protected] ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
