This isn’t specific to GPT-3. But it makes me wonder how long it will be until mobile devices could run something like GPT-3? I’m guessing 20 to 30 years. Maybe less if there are some breakthroughs on new materials or photonic processors.
I just read the news on Twitter, it’s great how the OpenAI community quickly picks up on these things too.
Of course, it will be a breakthrough. I was already in awe of the power of TinyML, an extremely lightweight implementation for embedded devices.
I can’t wait to see that transformative experience for neural networks (beyond the recognition of the smartphone camera itself) that this new chip will bring us