@giffmana @teortaxesTex Again, I have trained eyeballs with simple RNNs in simulators to point around and only render a few pixels to bring down compute costs
Ahh if only SpatialTransformerNet (not a transformer lol) actually worked at scale. They are the perfect match for this in theory, and beautiful.
In practice instead we have VLMs going "hmm thinking hmmm let me use my zoom and crop tool here hmmm"
Users in the replies dismissed discussion of RNNs training selective eyeballs in simulators to cut vision compute as repetitive rambling on familiar topics.
No Digg Deeper questions have been answered for this story yet.
@giffmana @teortaxesTex (to do a task such as flying a drone, this policy transfers to real life, the drone is a kind of eyeball!)
@giffmana @teortaxesTex Again, I have trained eyeballs with simple RNNs in simulators to point around and only render a few pixels to bring down compute costs
@yacineMTB @giffmana @teortaxesTex How's it going, @yacineMTB? @teortaxesTex, still wittering about the same two or three topics as per usual.