Google released project Tensor2Tensor training tool kits Deep Learning. Modeling can do many things at the same time.

Google release a project Tensor2Tensor (T2T) tools help isolate the issue out of the deep learning model design makes it possible to experiment with new models, the original issues more. Allows researchers to design the model effectively with demo applications with deep learning model the MultiModel research multiple tasks simultaneously, include a caption, image classification, translate, convert audio to text, and the analysis of sentence structure.

MultiModel as model deep learning complex. Data encryption and decryption process took several results according to the work of various types. By also can work with various types. An accurate level good enough to use. And practice many problems along the way back to increase accuracy in various types.



Previously, Google had ever lived practice translate to create a model that can translate languages that do not have the information, training, but this time it's different types of internships, but can increase the accuracy of the model.

The project has created a T2T data received high popularity in the language and images. Google indicated that using this project makes Google can build language analysis on new data sets in a few days.