-
Notifications
You must be signed in to change notification settings - Fork 97
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
feat: general wrapper framework #948
Conversation
Global state wrapper not yet working properly
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I can't approve 😢 but I am happy with this, just one question about the shared reward and a minor docstring thing
Just for bookkeeping #951 was discovered in this PR, but we're deciding to fix it in a separate PR so we can get this one in ASAP. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Can we add a feature called use_team_reward
as we need it for a fair comparison between our work and existing works that rely on giving the agents the same reward measured by mean(timestep.reward)
I made all the necessary changes in this PR #952, just to make this PR smaller.
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thank you for all the work! Super nice changes! Just some minor comments and suggestions here and there.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thank you @WiemKhlifi and @sash-a 🔥
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks @WiemKhlifi and @sash-a 🔥
feat: general wrapper framework
- feat(shaed.py): move the shared wrappers in a seperate file - docs: small edits based on review - fix: small edits - revert: revert some changes + doc changing based on review - Merge new changes from develop into feat/general-wrapper-framework - revert: revert the changes made on done variable - feat: edit rec systems - feat: edit types and wrapper for other systems - Merge branch develop into feat/general-wrapper-framework - feat: repeat reward and discount in wrapper - fix: centralized critic eval bug with new wrapper - feat: general wrapper mostly complete Co-authored-by: Sasha Abramowitz <reallysasha@gmail.com> Co-authored-by: RuanJohn <33461981+RuanJohn@users.noreply.github.com>
- feat(shaed.py): move the shared wrappers in a seperate file - docs: small edits based on review - fix: small edits - revert: revert some changes + doc changing based on review - Merge new changes from develop into feat/general-wrapper-framework - revert: revert the changes made on done variable - feat: edit rec systems - feat: edit types and wrapper for other systems - Merge branch develop into feat/general-wrapper-framework - feat: repeat reward and discount in wrapper - fix: centralized critic eval bug with new wrapper - feat: general wrapper mostly complete Co-authored-by: Sasha Abramowitz <reallysasha@gmail.com>
What?
General observation and example rware wrapper which makes it easy to support more jumanji envs in the future 👀
Why?
To support more jumanji envs
How?
Create an observation type with known fields. Wrap all new envs so that they return this observation type instead of their own custom observation.