coach/benchmarks/dueling_ddqn/README.md

Dueling DDQN

Each experiment uses 3 seeds and is trained for 10k environment steps. The parameters used for Dueling DDQN are the same parameters as described in the original paper.

Breakout Dueling DDQN - single worker

python3 coach.py -p Atari_Dueling_DDQN -lvl breakout