ys1087@partner.kit.edu 1caaa9d01f Add experiment infrastructure and production scripts

- Fix 6 critical bugs in original REPPO repository
- Add comprehensive README documentation
- Create production SLURM script for accelerated partition
- Add experiment submission script for batch jobs
- Algorithm now runs successfully with strong performance
- Ready for paper replication experiments on Brax suite

2025-07-22 18:47:43 +02:00

907 B

Raw Blame History

REPPO Experiment Plan

Proof of Concept Success

✅ Working Implementation: https://wandb.ai/dominik_roth/reppo_dev_test?nw=nwuserdominik_roth

Experiments To Run

1. Reproduce Paper Results

Brax Suite: 5 tasks (test first - already working)

ant, cheetah, humanoid, walker, hopper

DMC Suite (mujoco_playground): 23 tasks

AcrobotSwingup, CartpoleBalance, CheetahRun, FingerSpin, HumanoidRun, WalkerRun, etc.

ManiSkill Suite: 8 tasks (need wrapper first)

PickSingleYCB-v1, PegInsertionSide-v1, UnitreeG1TransportBox-v1, etc.

Settings: 50M steps, 1024 envs, 5 seeds each, paper hyperparameters

Scripts Needed

`submit_experiments.py`

Uses existing working SLURM script:

python submit_experiments.py --experiment brax --seeds 5
python submit_experiments.py --experiment mjx --seeds 5  
python submit_experiments.py --experiment maniskill --seeds 5

907 B Raw Blame History