Skip to content

Tags: HumanCompatibleAI/seals

Tags

v0.2.1

Toggle v0.2.1's commit message

Verified

This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
Fix typing issue in python 3.9 (#77)

v0.2

Toggle v0.2's commit message

Verified

This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
Migrate to gymnasium maintaining python 3.8 compatibility (#73)

* Initial commit

* py38 compatible type hints

* gymnasium compatible reset

* gymnasium compatibility changes

* gymnasium compatible reset and random

* Make type annotations python 3.8 compatible.

* Fix some grammar issues.

* Raise RuntimeErrors and ValueErrors in the proper places.

* Undoing unrelated formatting fixes in the readme.

* Remove unused ruff configuration.

* Add Adams wording suggestions.

* switch to alpha-version of the circle-ci image (reverst this before merge)

* Update Xdummy-entrypoint.py to python3

* Update Dockerfile to Ubuntu 20.04 and add ssh.

* Dont mention gym in inline comment but gymnasium.

* Absorb terminated AND truncated steps.

* Treat done == (terminated or truncated) and stop mentioning done in the documentation.

* Remove outdated make_env_no_wrappers

* Use registry keys instead of extracting env_id from the spec.

* Ensure to seed environments upon the first reset.

* Add missing shimmy dependency for atari.

* Detect atari envs by looking for shimmy entrypoint instead of gym entrypoint.

* Add missing observation space to TabularModelMDP

* Look for render modes in new location of the environment metadata.

* When testing the rollout schema, check for both termination and truncation.

* Remove outdated asserts on the result of env.reset().

* Adapt reset() of MaskScroeWrapper to new gymnasium API

* Switch to v4 versions of the MuJoCo environments.

* Simplify tests for render modes.

* Forward args and kwargs when constructing environments, so we can pass in the render mode.

* Add `Casino-Unmasked-v5` to the list of slow envs with randomness.

* Add some missing commas.

* Add pygame to setup.py

* Update ale-py version.

* Make test_sample_distribution deterministic by introducing a seed.

* Fixing isort issues.

* Add missing trailing commas.

* Minor formatting fixes.

* Fix trailing whitespace.

* Black fixes.

* Explicitly seed dummy environment.

* Remove unnecessary cast to int.

* Fix some typing issues.

* Fix more typing issues.

* Simplify ObsCastWrapper by inheriting from gym.ObservationWrapper instead of gym.Wrapper.

* Small typos in docstrings.

* Add reset info when generating rollouts.

* Remove unneeded default params to rand_gen.normal()

* Remove unneeded setter for the observation space property in a ResettableMDP.

* Ignore coverage for edge cases of where the observation space has no shape.

* Add a test case that ensures that options in the reset to a ResettablePOMDP are rejected.

* Remove rand_state property of ResettablePOMDP and use the canonical np_random of the superclass instead.

* Remove newline in base_envs.py

* Fix type annotations of FixedHorizonCartPole.reset()

* Remove leftover usages of rand_state.

* Fix quicks in dependencies that are no longer needed.

* Store unused info in _

* Make test_sample_distribution by seeding the used rng instead of setting the global seed.

* Add missing test dependency.

* Ensure we have the newest pip version to make the dependency resolution work.

* Make the dependencies cache also dependent on ci/build_venv.sh

---------

Co-authored-by: Juan Rocamonde <juancarlosrocamonde@gmail.com>
Co-authored-by: Maximilian Ernestus <maximilian@ernestus.de>

v0.1.5

Toggle v0.1.5's commit message

Verified

This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
Change type of `horizon` from `float` to `int`. (#66)

Or `None`, for infinite horizon.

v0.1.4

Toggle v0.1.4's commit message

Verified

This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
Migrate imitation envs to seals (#58)

v0.1.3

Toggle v0.1.3's commit message

Verified

This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
Automate PyPI uploads on release (#60)

* Add changes to automate PyPI uploads

* Update python version to 3.8

* Update docker image to base2

* Fix bug

* Bump cache version

* Remove .whl file

* Fix lint error

* Remove line deleting .whl

* Remove version.py

* Use base image now new image pushed

Co-authored-by: Adam Gleave <adam@gleave.me>

v0.1.2

Toggle v0.1.2's commit message

Verified

This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
Version 0.1.2 Release (#52)

* Bump version for new release

* Bump CircleCI

* Use free MuJoCo license key

v0.1.1

Toggle v0.1.1's commit message

Verified

This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
Test rendering of environments (#44)

* Add dummy X server

* Check rendering

* Amend Dockerfile to play nice with CircleCI

* Use larger resource class as X server is dieing on small machines, remove sharding which is not really justified given small test suite

* Fix Xorg config path

* Workaround MuJoCo bug

* Explicitly wrap in Xdummy-entrypoint.py (CircleCI ignores Dockerifle entrypoint)

* Fix typo

* Fix typo in Dockerfile

* Copy needed dependencies in Dockerfile for python-req stage

* Throttle number of CPUs to see if this resolves spurious failure

* Decrease Xorg size, bump CPUs back up to 4, warn of memory leak

* Fix type issue

* Improve docstring

* Fix typo

* Bump version