Seems like this will end up driving more research into very large ML models. Some of the performance benchmarks here are for trillion parameter models. I’ve obs

Seems like this will end up driving more research into very large ML models. Some of the performance benchmarks here are for trillion parameter models. I’ve observed that often Innovation directly follows silicon improvements.