Official codebase for Improving Computational Efficiency in Visual Reinforcement Learning via Stored Embeddings. The Rainbow codebase was originally forked from Kaixhin's Rainbow and the CURL codebase was originally forked from CURL.
See instructions in Kaixhin's Rainbow. Additional hyperparameters are steps_until_freeze. Example scripts for using large replay buffers (run_lb_alien.sh) and small replay buffers (run_sb_alien.sh) can be found in the scripts folder.
See instructions in CURL. Additional hyperparameters are steps_until_freeze and num_copies. Example scripts for using large replay buffers (run_lb_cartpole.sh) and small replay buffers (run_sb_cartpole.sh) can be found in the scripts folder.