4 ms·
It's just a single linear layer and it's not clear to me that the technology is capable of anything more. If I'm reading it correctly it sounds like running the
by msgodel 1y ago
It's just a single linear layer and it's not clear to me that the technology is capable of anything more. If I'm reading it correctly it sounds like running the model forward couldn't even use the technology, they had to record the weights and do it the old fashion way.
- roflmaostc 1y agoWould you have discredited early AI work because they could only train and compute a couple of weights? This is about first prototypes and scaling is often easier than the basic principle.
- msgodel 1y agoIs this actually capable of propagating the gradient and training more complex layers though? A lot of these novel AI accelerators run into problems like that because they're not capable of general purpose computing. A good example of that are the boltzman machines on Dwave's stuff. Yeah it can do that but it can only do that because the machine is only capable of doing QUBO.
- roflmaostc 1y agoFor inference we do not care about training, right? But if we could make cheaper inference machines available, everyone would profit. Isn't it that LLMs use more energy in inference than training these days?