2 ms·
Amazing work! It's something we knew would come at some point, but I didn't expect it to be that good so early! Since some of the authors are in this thread an
by littlestymaar 6y ago
Amazing work! It's something we knew would come at some point, but I didn't expect it to be that good so early!
Since some of the authors are in this thread answering questions, I have one: I wonder if such project requires Google-scale infrastructure or if it's something that can easily be replicated. For instance, how big is the training set? And how much compute was necessary to train the model?
- atorodius 6y agoHi! It depends on how far you want to go. For this project, we did a lot of exploration, because we had Google-scale infrastructure. Replicating all the exploration will need a lot of GPUs (in the 100s), replicating all the experiments that actually went into the paper maybe a few dozens. Training something similar as what's in the demo with the code we will release takes 1 V100 :) We can't release the internal training set, but expect a dataset of a few hundred thousand images (e.g., openimages) to be sufficient, maybe even less (AFAIK this has not been explored in a controlled setting).
- littlestymaar 6y agoThanks a lot!