4 ms·
Is this just go-faster-juice for Tensorflow code or does it have other implications? If you train on TPUs can you still run the model efficiently elsewhere?
by tveita 9y ago
Is this just go-faster-juice for Tensorflow code or does it have other implications? If you train on TPUs can you still run the model efficiently elsewhere?