Eclectic-Sheep / sheeprl

Distributed Reinforcement Learning accelerated by Lightning Fabric
https://eclecticsheep.ai
Apache License 2.0
274 stars 26 forks source link

Feature/dv3 parallel stochastic #225

Closed belerico closed 4 months ago

belerico commented 4 months ago

Summary

This PR introduces the possiblity to have the Dreamer-V3 representation model (the one responsible to output the posterior) to be independent from the recurrent model.

Type of Change

Please select the one relevant option below:

Checklist

Please confirm that the following tasks have been completed:

Screenshots or Visuals (Optional)

image

where the pink one is DV3 with the parallel stochastic while the blue is the standard one.

Additional Information (Optional)

Please provide any additional information that may be useful for the reviewer, such as:

Thank you for your contribution! Once you have filled out this template, please ensure that you have assigned the appropriate reviewers and that all tests have passed.