dodox/fancy_rl
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Updated README

Browse Source
This commit is contained in:
Dominik Moritz Roth 2024-10-21 15:25:01 +02:00
parent 8eb9b384c7
commit df1ba6fe53
1 changed files with 20 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
README.md
View File

@ -52,17 +52,34 @@ To run the test suite:
pytest test/test_ppo.py pytest test/test_ppo.py
``` ```
## TODO ## Status
### Implemented Features
- Proximal Policy Optimization (PPO) algorithm
- Trust Region Policy Layers (TRPL) algorithm (WIP)
- Support for continuous and discrete action spaces
- Multiple projection methods (Rewritten for MIT License Compatability):
- KL Divergence projection
- Frobenius norm projection
- Wasserstein distance projection
- Identity projection (Eq to PPO)
- Configurable neural network architectures for actor and critic
- Logging support (Terminal and WandB, extendable)
### TODO
- [ ] All PPO Tests green
- [ ] Better / more logging - [ ] Better / more logging
- [ ] Test / Benchmark PPO - [ ] Test / Benchmark PPO
- [ ] Refactor Modules for TRPL - [ ] Refactor Modules for TRPL
- [ ] Get TRPL working - [ ] Get TRPL working
- [ ] Test / Benchmark TRPL - [ ] All TRPL Tests green
- [ ] Make contextual covariance optional - [ ] Make contextual covariance optional
- [ ] Allow full-cov via chol - [ ] Allow full-cov via chol
- [ ] Test / Benchmark TRPL
- [ ] Write docs / extend README - [ ] Write docs / extend README
- [ ] (Implement SAC?) - [ ] Test func of non-gym envs
- [ ] Implement SAC
- [ ] Implement VLEARN
## Contributing ## Contributing