Patapelo for Distributed Deep Learning

Pade's name is parallel distributed deep earning. We have a very strong focus on prosessing large amounts of data and to train very big models. Erne titan, yes, sure. Patapelo is an important part in training such models. These models are very hard to train, so it must have a strong deep learning framework to help them. This model is already very large. It is one single model that has 260 billion perometers. Is largest already. But nowadays there are many different models. That is not one single model. They use quite a few different models and merge them. The primaters may grow byit even more. So it is, in my opinion, it