JohannesHeidecke / irl-benchmark

JohannesHeidecke/irl-benchmark Issues

Env not unwrappable to BaseWorldModelWrapper
Updated a year ago
Get >90% test coverage
Updated 6 years ago
Make documentation work perfectly
Updated 6 years ago
Host our docs on readthedocs.org
Closed 6 years ago
Update README.md
Closed 6 years ago1
RelEnt uses trajectories of varying length
Updated 6 years ago1
ILE calculated using initial/estimated rather than true reward
Updated 6 years ago1
Inconsistent use of discount factors
Updated 6 years ago
MaxEnt estimates the initial state distribution rather than getting it from the env
Updated 6 years ago
Value Iteration uses leaked true reward
Closed 6 years ago
Test `tests/ppo_test.py` fails
Closed 6 years ago1