diff --git a/benchmarks/README.md b/benchmarks/README.md new file mode 100644 index 0000000..031cb4c --- /dev/null +++ b/benchmarks/README.md @@ -0,0 +1,164 @@ +# Coach Benchmarks + +The following figures are training curves of some of the presets available through Coach. +The X axis in all the figures is the total steps (for multi-threaded runs, this is the accumulated number of steps over all the workers). +The Y axis in all the figures is the average episode reward with an averaging window of 11 episodes. +These are the result you can expect to get when running the pre-defined presets in Coach. + + +## A3C + +### Breakout_A3C with 16 workers + +```bash +python3 coach.py -p Breakout_A3C -n 16 -r +``` + +Breakout_A3C_16_workers + +### InvertedPendulum_A3C with 16 workers + +```bash +python3 coach.py -p InvertedPendulum_A3C -n 16 -r +``` + +Inverted_Pendulum_A3C_16_workers + +### Hopper_A3C with 16 workers + +```bash +python3 coach.py -p Hopper_A3C -n 16 -r +``` + +Hopper_A3C_16_workers + +### Ant_A3C with 16 workers + +```bash +python3 coach.py -p Ant_A3C -n 16 -r +``` + +Ant_A3C_16_workers + +## Clipped PPO + +### InvertedPendulum_ClippedPPO with 16 workers + +```bash +python3 coach.py -p InvertedPendulum_ClippedPPO -n 16 -r +``` + +InvertedPendulum_ClippedPPO_16_workers + +### Hopper_ClippedPPO with 16 workers + +```bash +python3 coach.py -p Hopper_ClippedPPO -n 16 -r +``` + +Hopper_Clipped_PPO_16_workers + +### Humanoid_ClippedPPO with 16 workers + +```bash +python3 coach.py -p Humanoid_ClippedPPO -n 16 -r +``` + +Humanoid_ClippedPPO_16_workers + +## DQN + +### Pong_DQN + +```bash +python3 coach.py -p Pong_DQN -r +``` + +Pong_DQN + +### Doom_Basic_DQN + +```bash +python3 coach.py -p Doom_Basic_DQN -r +``` + +Doom_Basic_DQN + +## Dueling DDQN + +### Doom_Basic_Dueling_DDQN + +```bash +python3 coach.py -p Doom_Basic_Dueling_DDQN -r +``` + +Doom_Basic_Dueling_DDQN + +## DFP + +### Doom_Health_DFP + +```bash +python3 coach.py -p Doom_Health_DFP -r +``` + +Doom_Health_DFP + +## MMC + +### Doom_Health_MMC + +```bash +python3 coach.py -p Doom_Health_MMC -r +``` + +Doom_Health_MMC + +## NEC + +## Doom_Basic_NEC + +```bash +python3 coach.py -p Doom_Basic_NEC -r +``` + +Doom_Basic_NEC + +## PG + +### CartPole_PG + +```bash +python3 coach.py -p CartPole_PG -r +``` + +CartPole_PG + +## DDPG + +### Pendulum_DDPG + +```bash +python3 coach.py -p Pendulum_DDPG -r +``` + +Pendulum_DDPG + + +## NAF + +### InvertedPendulum_NAF + +```bash +python3 coach.py -p InvertedPendulum_NAF -r +``` + +InvertedPendulum_NAF + +### Pendulum_NAF + +```bash +python3 coach.py -p Pendulum_NAF -r +``` + +Pendulum_NAF diff --git a/benchmarks/img/Ant_A3C_16_workers.png b/benchmarks/img/Ant_A3C_16_workers.png new file mode 100644 index 0000000..d677ab0 Binary files /dev/null and b/benchmarks/img/Ant_A3C_16_workers.png differ diff --git a/benchmarks/img/Breakout_A3C_16_workers.png b/benchmarks/img/Breakout_A3C_16_workers.png new file mode 100644 index 0000000..0f778e2 Binary files /dev/null and b/benchmarks/img/Breakout_A3C_16_workers.png differ diff --git a/benchmarks/img/CartPole_PG.png b/benchmarks/img/CartPole_PG.png new file mode 100644 index 0000000..46779dc Binary files /dev/null and b/benchmarks/img/CartPole_PG.png differ diff --git a/benchmarks/img/Doom_Basic_DQN.png b/benchmarks/img/Doom_Basic_DQN.png new file mode 100644 index 0000000..5f9382f Binary files /dev/null and b/benchmarks/img/Doom_Basic_DQN.png differ diff --git a/benchmarks/img/Doom_Basic_Dueling_DDQN.png b/benchmarks/img/Doom_Basic_Dueling_DDQN.png new file mode 100644 index 0000000..34478f7 Binary files /dev/null and b/benchmarks/img/Doom_Basic_Dueling_DDQN.png differ diff --git a/benchmarks/img/Doom_Basic_NEC.png b/benchmarks/img/Doom_Basic_NEC.png new file mode 100644 index 0000000..79b5c6f Binary files /dev/null and b/benchmarks/img/Doom_Basic_NEC.png differ diff --git a/benchmarks/img/Doom_Health_DFP.png b/benchmarks/img/Doom_Health_DFP.png new file mode 100644 index 0000000..3f8e16c Binary files /dev/null and b/benchmarks/img/Doom_Health_DFP.png differ diff --git a/benchmarks/img/Doom_Health_MMC.png b/benchmarks/img/Doom_Health_MMC.png new file mode 100644 index 0000000..d43f66b Binary files /dev/null and b/benchmarks/img/Doom_Health_MMC.png differ diff --git a/benchmarks/img/Hopper_A3C_16_workers.png b/benchmarks/img/Hopper_A3C_16_workers.png new file mode 100644 index 0000000..2c2efa7 Binary files /dev/null and b/benchmarks/img/Hopper_A3C_16_workers.png differ diff --git a/benchmarks/img/Hopper_ClippedPPO_16_workers.png b/benchmarks/img/Hopper_ClippedPPO_16_workers.png new file mode 100644 index 0000000..e9821d9 Binary files /dev/null and b/benchmarks/img/Hopper_ClippedPPO_16_workers.png differ diff --git a/benchmarks/img/Humanoid_ClippedPPO_16_workers.png b/benchmarks/img/Humanoid_ClippedPPO_16_workers.png new file mode 100644 index 0000000..0488c98 Binary files /dev/null and b/benchmarks/img/Humanoid_ClippedPPO_16_workers.png differ diff --git a/benchmarks/img/InvertedPendulum_ClippedPPO_16_workers.png b/benchmarks/img/InvertedPendulum_ClippedPPO_16_workers.png new file mode 100644 index 0000000..b563024 Binary files /dev/null and b/benchmarks/img/InvertedPendulum_ClippedPPO_16_workers.png differ diff --git a/benchmarks/img/InvertedPendulum_NAF.png b/benchmarks/img/InvertedPendulum_NAF.png new file mode 100644 index 0000000..9b8b6f6 Binary files /dev/null and b/benchmarks/img/InvertedPendulum_NAF.png differ diff --git a/benchmarks/img/Inverted_Pendulum_A3C_16_workers.png b/benchmarks/img/Inverted_Pendulum_A3C_16_workers.png new file mode 100644 index 0000000..d459990 Binary files /dev/null and b/benchmarks/img/Inverted_Pendulum_A3C_16_workers.png differ diff --git a/benchmarks/img/Pendulum_DDPG.png b/benchmarks/img/Pendulum_DDPG.png new file mode 100644 index 0000000..89abbac Binary files /dev/null and b/benchmarks/img/Pendulum_DDPG.png differ diff --git a/benchmarks/img/Pendulum_NAF.png b/benchmarks/img/Pendulum_NAF.png new file mode 100644 index 0000000..0faca93 Binary files /dev/null and b/benchmarks/img/Pendulum_NAF.png differ diff --git a/benchmarks/img/Pong_DQN.png b/benchmarks/img/Pong_DQN.png new file mode 100644 index 0000000..6122c78 Binary files /dev/null and b/benchmarks/img/Pong_DQN.png differ