/coin

"Continual Optimistic Initialization for Value-Based Reinforcement Learning" accepted @ AAMAS-2024

Primary LanguagePythonMIT LicenseMIT

Continual Optimistic INitialization (COIN)

This codebase is based on the Spinning Up repository. Please follow the installation instructions provided here.

Running experiments

python -m spinup.run <algo> --env <env_name> --exp_name <log_folder> --epochs <num_epochs> --bonus <b> --bonus_freq <bonus_frequency> --seed <seed>

e.g.,

python -m spinup.run coin --env LunarLander-v2 --exp_name coin_lunarlander_b_0_2_freq_50000 --epochs 60 --bonus 0.2 --bonus_freq 50000 --seed 0