Allow `alpha` (action precision) to be passed into `Agent()` constructor as optional argument

infer-actively / pymdp

A Python implementation of active inference for Markov Decision Processes

MIT License

420 stars 83 forks source link

Allow `alpha` (action precision) to be passed into `Agent()` constructor as optional argument #81

Closed conorheins closed 1 year ago

conorheins commented 2 years ago

Right now, action precision (alpha) in case of stochastic action sampling, is defaulted to alpha = 16.0 within the function of the control module sample_action(..., alpha = 16.0). Change this so that an instance of Agent.agent itself stores the action precision, and calls it in its method self.sample_action().

Originally posted by @helenegu in https://github.com/infer-actively/pymdp/discussions/4#discussioncomment-3018137

conorheins commented 1 year ago

This is now fixed in master; will close this after 0.0.6 is released so that this enhancement is applied to the registered version of the package.