ReinforcementLearning.jl icon indicating copy to clipboard operation
ReinforcementLearning.jl copied to clipboard

Published 20 hours ago • verified publisher icon JuliaReinforcementLearning

Cross language support

Open findmyway opened this issue 4 years ago • 0 comments

This is an ambitious goal that workers and learners can be relatively independent.

One possible scenario may be, running fast evaluation in Julia within workers and running learner with TF/PyTorch/JAX within learners.

To reach this goal,

  1. The actor model should support such kinds of scheduling and communication.
  2. Data sharing between different languages must be addressed. Maybe using Arrow format?

findmyway avatar Nov 15 '20 03:11 findmyway