Introducing Pytorch for fast.ai

www.fast.ai

Fast.ai, one of the best online courses for learning Deep Learning, has decided to base their future courses on Pytorch rather than keras or Tensorflow. One of the main motivations for this change is the fact that dynamic methods such as attention and teacher forcing are hard to implement in the latter two, but are easy in Pytorch due to its dynamic computation graphs. DyNet shares similar benefits. For more information regarding Pytorch, have a look at the resources in this past newsletter.

Read more...
Linkedin

Want to receive more content like this in your inbox?