Skip to content

Latest commit

 

History

History
9 lines (6 loc) · 383 Bytes

README.md

File metadata and controls

9 lines (6 loc) · 383 Bytes

training ensemble nets

ensemble nets are a single pass way of representing M models in a neural net ensemble. see this blog post

note: blog post was based on a vmap version run on my local single GPU machine. the code for this first version is under tag v1.

the code as is now is a port of the model to run on a pod tpu slice using haiku