gryf/coach
1
0
Fork 0
mirror of https://github.com/gryf/coach.git synced 2025-12-17 19:20:19 +01:00
Code Issues Projects Releases Wiki Activity

benchmarks and pip package updates

Browse Source
This commit is contained in:
Itai Caspi
2018-08-19 14:23:20 +03:00
parent 23d2945bf8
commit c5165cd7d6
14 changed files with 69 additions and 48 deletions
Download Patch File Download Diff File Expand all files Collapse all files

20
benchmarks/a3c/README.md
View File

@@ -6,11 +6,11 @@ The parameters used for Clipped PPO are the same parameters as described in the
### Inverted Pendulum A3C - 1/2/4/8/16 workers ### Inverted Pendulum A3C - 1/2/4/8/16 workers
```bash ```bash
python3 coach.py -p Mujoco_A3C -lvl inverted_pendulum -n 1 coach -p Mujoco_A3C -lvl inverted_pendulum -n 1
python3 coach.py -p Mujoco_A3C -lvl inverted_pendulum -n 2 coach -p Mujoco_A3C -lvl inverted_pendulum -n 2
python3 coach.py -p Mujoco_A3C -lvl inverted_pendulum -n 4 coach -p Mujoco_A3C -lvl inverted_pendulum -n 4
python3 coach.py -p Mujoco_A3C -lvl inverted_pendulum -n 8 coach -p Mujoco_A3C -lvl inverted_pendulum -n 8
python3 coach.py -p Mujoco_A3C -lvl inverted_pendulum -n 16 coach -p Mujoco_A3C -lvl inverted_pendulum -n 16
``` ```
<img src="inverted_pendulum_a3c.png" alt="Inverted Pendulum A3C" width="800"/> <img src="inverted_pendulum_a3c.png" alt="Inverted Pendulum A3C" width="800"/>
@@ -19,7 +19,7 @@ python3 coach.py -p Mujoco_A3C -lvl inverted_pendulum -n 16
### Hopper A3C - 16 workers ### Hopper A3C - 16 workers
```bash ```bash
python3 coach.py -p Mujoco_A3C -lvl hopper -n 16 coach -p Mujoco_A3C -lvl hopper -n 16
``` ```
<img src="hopper_a3c_16_workers.png" alt="Hopper A3C 16 workers" width="800"/> <img src="hopper_a3c_16_workers.png" alt="Hopper A3C 16 workers" width="800"/>
@@ -28,7 +28,7 @@ python3 coach.py -p Mujoco_A3C -lvl hopper -n 16
### Walker2D A3C - 16 workers ### Walker2D A3C - 16 workers
```bash ```bash
python3 coach.py -p Mujoco_A3C -lvl walker2d -n 16 coach -p Mujoco_A3C -lvl walker2d -n 16
``` ```
<img src="walker2d_a3c_16_workers.png" alt="Walker2D A3C 16 workers" width="800"/> <img src="walker2d_a3c_16_workers.png" alt="Walker2D A3C 16 workers" width="800"/>
@@ -37,7 +37,7 @@ python3 coach.py -p Mujoco_A3C -lvl walker2d -n 16
### Half Cheetah A3C - 16 workers ### Half Cheetah A3C - 16 workers
```bash ```bash
python3 coach.py -p Mujoco_A3C -lvl half_cheetah -n 16 coach -p Mujoco_A3C -lvl half_cheetah -n 16
``` ```
<img src="half_cheetah_a3c_16_workers.png" alt="Half Cheetah A3C 16 workers" width="800"/> <img src="half_cheetah_a3c_16_workers.png" alt="Half Cheetah A3C 16 workers" width="800"/>
@@ -46,7 +46,7 @@ python3 coach.py -p Mujoco_A3C -lvl half_cheetah -n 16
### Ant A3C - 16 workers ### Ant A3C - 16 workers
```bash ```bash
python3 coach.py -p Mujoco_A3C -lvl ant -n 16 coach -p Mujoco_A3C -lvl ant -n 16
``` ```
<img src="ant_a3c_16_workers.png" alt="Ant A3C 16 workers" width="800"/> <img src="ant_a3c_16_workers.png" alt="Ant A3C 16 workers" width="800"/>
@@ -56,7 +56,7 @@ python3 coach.py -p Mujoco_A3C -lvl ant -n 16
### Space Invaders A3C - 16 workers ### Space Invaders A3C - 16 workers
```bash ```bash
python3 coach.py -p Atari_A3C -lvl space_invaders -n 16 coach -p Atari_A3C -lvl space_invaders -n 16
``` ```
<img src="space_invaders_a3c_16_workers.png" alt="Space Invaders A3C 16 workers" width="800"/> <img src="space_invaders_a3c_16_workers.png" alt="Space Invaders A3C 16 workers" width="800"/>

6
benchmarks/bootstrapped_dqn/README.md
View File

@@ -6,7 +6,7 @@ The parameters used for Bootstrapped DQN are the same parameters as described in
### Breakout Bootstrapped DQN - single worker ### Breakout Bootstrapped DQN - single worker
```bash ```bash
python3 coach.py -p Atari_Bootstrapped_DQN -lvl breakout coach -p Atari_Bootstrapped_DQN -lvl breakout
``` ```
<img src="breakout_bootstrapped_dqn.png" alt="Breakout Bootstrapped DQN" width="800"/> <img src="breakout_bootstrapped_dqn.png" alt="Breakout Bootstrapped DQN" width="800"/>
@@ -15,7 +15,7 @@ python3 coach.py -p Atari_Bootstrapped_DQN -lvl breakout
### Pong Bootstrapped DQN - single worker ### Pong Bootstrapped DQN - single worker
```bash ```bash
python3 coach.py -p Atari_Bootstrapped_DQN -lvl pong coach -p Atari_Bootstrapped_DQN -lvl pong
``` ```
<img src="pong_bootstrapped_dqn.png" alt="Pong Bootstrapped DQN" width="800"/> <img src="pong_bootstrapped_dqn.png" alt="Pong Bootstrapped DQN" width="800"/>
@@ -24,7 +24,7 @@ python3 coach.py -p Atari_Bootstrapped_DQN -lvl pong
### Space Invaders Bootstrapped DQN - single worker ### Space Invaders Bootstrapped DQN - single worker
```bash ```bash
python3 coach.py -p Atari_Bootstrapped_DQN -lvl space_invaders coach -p Atari_Bootstrapped_DQN -lvl space_invaders
``` ```
<img src="space_invaders_bootstrapped_dqn.png" alt="Space Invaders Bootstrapped DQN" width="800"/> <img src="space_invaders_bootstrapped_dqn.png" alt="Space Invaders Bootstrapped DQN" width="800"/>

18
benchmarks/clipped_ppo/README.md
View File

@@ -6,7 +6,7 @@ The parameters used for Clipped PPO are the same parameters as described in the
### Inverted Pendulum Clipped PPO - single worker ### Inverted Pendulum Clipped PPO - single worker
```bash ```bash
python3 coach.py -p Mujoco_ClippedPPO -lvl inverted_pendulum coach -p Mujoco_ClippedPPO -lvl inverted_pendulum
``` ```
<img src="inverted_pendulum_clipped_ppo.png" alt="Inverted Pendulum Clipped PPO" width="800"/> <img src="inverted_pendulum_clipped_ppo.png" alt="Inverted Pendulum Clipped PPO" width="800"/>
@@ -15,7 +15,7 @@ python3 coach.py -p Mujoco_ClippedPPO -lvl inverted_pendulum
### Inverted Double Pendulum Clipped PPO - single worker ### Inverted Double Pendulum Clipped PPO - single worker
```bash ```bash
python3 coach.py -p Mujoco_ClippedPPO -lvl inverted_double_pendulum coach -p Mujoco_ClippedPPO -lvl inverted_double_pendulum
``` ```
<img src="inverted_double_pendulum_clipped_ppo.png" alt="Inverted Double Pendulum Clipped PPO" width="800"/> <img src="inverted_double_pendulum_clipped_ppo.png" alt="Inverted Double Pendulum Clipped PPO" width="800"/>
@@ -24,7 +24,7 @@ python3 coach.py -p Mujoco_ClippedPPO -lvl inverted_double_pendulum
### Reacher Clipped PPO - single worker ### Reacher Clipped PPO - single worker
```bash ```bash
python3 coach.py -p Mujoco_ClippedPPO -lvl reacher coach -p Mujoco_ClippedPPO -lvl reacher
``` ```
<img src="reacher_clipped_ppo.png" alt="Reacher Clipped PPO" width="800"/> <img src="reacher_clipped_ppo.png" alt="Reacher Clipped PPO" width="800"/>
@@ -33,7 +33,7 @@ python3 coach.py -p Mujoco_ClippedPPO -lvl reacher
### Hopper Clipped PPO - single worker ### Hopper Clipped PPO - single worker
```bash ```bash
python3 coach.py -p Mujoco_ClippedPPO -lvl hopper coach -p Mujoco_ClippedPPO -lvl hopper
``` ```
<img src="hopper_clipped_ppo.png" alt="Hopper Clipped PPO" width="800"/> <img src="hopper_clipped_ppo.png" alt="Hopper Clipped PPO" width="800"/>
@@ -42,7 +42,7 @@ python3 coach.py -p Mujoco_ClippedPPO -lvl hopper
### Half Cheetah Clipped PPO - single worker ### Half Cheetah Clipped PPO - single worker
```bash ```bash
python3 coach.py -p Mujoco_ClippedPPO -lvl half_cheetah coach -p Mujoco_ClippedPPO -lvl half_cheetah
``` ```
<img src="half_cheetah_clipped_ppo.png" alt="Half Cheetah Clipped PPO" width="800"/> <img src="half_cheetah_clipped_ppo.png" alt="Half Cheetah Clipped PPO" width="800"/>
@@ -51,7 +51,7 @@ python3 coach.py -p Mujoco_ClippedPPO -lvl half_cheetah
### Walker 2D Clipped PPO - single worker ### Walker 2D Clipped PPO - single worker
```bash ```bash
python3 coach.py -p Mujoco_ClippedPPO -lvl walker2d coach -p Mujoco_ClippedPPO -lvl walker2d
``` ```
<img src="walker2d_clipped_ppo.png" alt="Walker 2D Clipped PPO" width="800"/> <img src="walker2d_clipped_ppo.png" alt="Walker 2D Clipped PPO" width="800"/>
@@ -60,7 +60,7 @@ python3 coach.py -p Mujoco_ClippedPPO -lvl walker2d
### Ant Clipped PPO - single worker ### Ant Clipped PPO - single worker
```bash ```bash
python3 coach.py -p Mujoco_ClippedPPO -lvl ant coach -p Mujoco_ClippedPPO -lvl ant
``` ```
<img src="ant_clipped_ppo.png" alt="Ant Clipped PPO" width="800"/> <img src="ant_clipped_ppo.png" alt="Ant Clipped PPO" width="800"/>
@@ -69,7 +69,7 @@ python3 coach.py -p Mujoco_ClippedPPO -lvl ant
### Swimmer Clipped PPO - single worker ### Swimmer Clipped PPO - single worker
```bash ```bash
python3 coach.py -p Mujoco_ClippedPPO -lvl swimmer coach -p Mujoco_ClippedPPO -lvl swimmer
``` ```
<img src="swimmer_clipped_ppo.png" alt="Swimmer Clipped PPO" width="800"/> <img src="swimmer_clipped_ppo.png" alt="Swimmer Clipped PPO" width="800"/>
@@ -78,7 +78,7 @@ python3 coach.py -p Mujoco_ClippedPPO -lvl swimmer
### Humanoid Clipped PPO - single worker ### Humanoid Clipped PPO - single worker
```bash ```bash
python3 coach.py -p Mujoco_ClippedPPO -lvl humanoid coach -p Mujoco_ClippedPPO -lvl humanoid
``` ```
<img src="humanoid_clipped_ppo.png" alt="Humanoid Clipped PPO" width="800"/> <img src="humanoid_clipped_ppo.png" alt="Humanoid Clipped PPO" width="800"/>

18
benchmarks/ddpg/README.md
View File

@@ -6,7 +6,7 @@ The parameters used for DDPG are the same parameters as described in the [origin
### Inverted Pendulum DDPG - single worker ### Inverted Pendulum DDPG - single worker
```bash ```bash
python3 coach.py -p Mujoco_DDPG -lvl inverted_pendulum coach -p Mujoco_DDPG -lvl inverted_pendulum
``` ```
<img src="inverted_pendulum_ddpg.png" alt="Inverted Pendulum DDPG" width="800"/> <img src="inverted_pendulum_ddpg.png" alt="Inverted Pendulum DDPG" width="800"/>
@@ -15,7 +15,7 @@ python3 coach.py -p Mujoco_DDPG -lvl inverted_pendulum
### Inverted Double Pendulum DDPG - single worker ### Inverted Double Pendulum DDPG - single worker
```bash ```bash
python3 coach.py -p Mujoco_DDPG -lvl inverted_double_pendulum coach -p Mujoco_DDPG -lvl inverted_double_pendulum
``` ```
<img src="inverted_double_pendulum_ddpg.png" alt="Inverted Double Pendulum DDPG" width="800"/> <img src="inverted_double_pendulum_ddpg.png" alt="Inverted Double Pendulum DDPG" width="800"/>
@@ -24,7 +24,7 @@ python3 coach.py -p Mujoco_DDPG -lvl inverted_double_pendulum
### Reacher DDPG - single worker ### Reacher DDPG - single worker
```bash ```bash
python3 coach.py -p Mujoco_DDPG -lvl reacher coach -p Mujoco_DDPG -lvl reacher
``` ```
<img src="reacher_ddpg.png" alt="Reacher DDPG" width="800"/> <img src="reacher_ddpg.png" alt="Reacher DDPG" width="800"/>
@@ -33,7 +33,7 @@ python3 coach.py -p Mujoco_DDPG -lvl reacher
### Hopper DDPG - single worker ### Hopper DDPG - single worker
```bash ```bash
python3 coach.py -p Mujoco_DDPG -lvl hopper coach -p Mujoco_DDPG -lvl hopper
``` ```
<img src="hopper_ddpg.png" alt="Hopper DDPG" width="800"/> <img src="hopper_ddpg.png" alt="Hopper DDPG" width="800"/>
@@ -42,7 +42,7 @@ python3 coach.py -p Mujoco_DDPG -lvl hopper
### Half Cheetah DDPG - single worker ### Half Cheetah DDPG - single worker
```bash ```bash
python3 coach.py -p Mujoco_DDPG -lvl half_cheetah coach -p Mujoco_DDPG -lvl half_cheetah
``` ```
<img src="half_cheetah_ddpg.png" alt="Half Cheetah DDPG" width="800"/> <img src="half_cheetah_ddpg.png" alt="Half Cheetah DDPG" width="800"/>
@@ -51,7 +51,7 @@ python3 coach.py -p Mujoco_DDPG -lvl half_cheetah
### Walker 2D DDPG - single worker ### Walker 2D DDPG - single worker
```bash ```bash
python3 coach.py -p Mujoco_DDPG -lvl walker2d coach -p Mujoco_DDPG -lvl walker2d
``` ```
<img src="walker2d_ddpg.png" alt="Walker 2D DDPG" width="800"/> <img src="walker2d_ddpg.png" alt="Walker 2D DDPG" width="800"/>
@@ -60,7 +60,7 @@ python3 coach.py -p Mujoco_DDPG -lvl walker2d
### Ant DDPG - single worker ### Ant DDPG - single worker
```bash ```bash
python3 coach.py -p Mujoco_DDPG -lvl ant coach -p Mujoco_DDPG -lvl ant
``` ```
<img src="ant_ddpg.png" alt="Ant DDPG" width="800"/> <img src="ant_ddpg.png" alt="Ant DDPG" width="800"/>
@@ -69,7 +69,7 @@ python3 coach.py -p Mujoco_DDPG -lvl ant
### Swimmer DDPG - single worker ### Swimmer DDPG - single worker
```bash ```bash
python3 coach.py -p Mujoco_DDPG -lvl swimmer coach -p Mujoco_DDPG -lvl swimmer
``` ```
<img src="swimmer_ddpg.png" alt="Swimmer DDPG" width="800"/> <img src="swimmer_ddpg.png" alt="Swimmer DDPG" width="800"/>
@@ -78,7 +78,7 @@ python3 coach.py -p Mujoco_DDPG -lvl swimmer
### Humanoid DDPG - single worker ### Humanoid DDPG - single worker
```bash ```bash
python3 coach.py -p Mujoco_DDPG -lvl humanoid coach -p Mujoco_DDPG -lvl humanoid
``` ```
<img src="humanoid_ddpg.png" alt="Humanoid DDPG" width="800"/> <img src="humanoid_ddpg.png" alt="Humanoid DDPG" width="800"/>

8
benchmarks/ddpg_her/README.md
View File

@@ -6,7 +6,7 @@ The parameters used for DDPG HER are the same parameters as described in the [fo
### Fetch Reach DDPG HER - single worker ### Fetch Reach DDPG HER - single worker
```bash ```bash
python3 coach.py -p Fetch_DDPG_HER_baselines -lvl reach coach -p Fetch_DDPG_HER_baselines -lvl reach
``` ```
<img src="fetch_ddpg_her_reach_1_worker.png" alt="Fetch DDPG HER Reach 1 Worker" width="800"/> <img src="fetch_ddpg_her_reach_1_worker.png" alt="Fetch DDPG HER Reach 1 Worker" width="800"/>
@@ -15,7 +15,7 @@ python3 coach.py -p Fetch_DDPG_HER_baselines -lvl reach
### Fetch Push DDPG HER - 8 workers ### Fetch Push DDPG HER - 8 workers
```bash ```bash
python3 coach.py -p Fetch_DDPG_HER_baselines -lvl push -n 8 coach -p Fetch_DDPG_HER_baselines -lvl push -n 8
``` ```
<img src="fetch_ddpg_her_push_8_workers.png" alt="Fetch DDPG HER Push 8 Worker" width="800"/> <img src="fetch_ddpg_her_push_8_workers.png" alt="Fetch DDPG HER Push 8 Worker" width="800"/>
@@ -24,7 +24,7 @@ python3 coach.py -p Fetch_DDPG_HER_baselines -lvl push -n 8
### Fetch Slide DDPG HER - 8 workers ### Fetch Slide DDPG HER - 8 workers
```bash ```bash
python3 coach.py -p Fetch_DDPG_HER_baselines -lvl slide -n 8 coach -p Fetch_DDPG_HER_baselines -lvl slide -n 8
``` ```
<img src="fetch_ddpg_her_slide_8_workers.png" alt="Fetch DDPG HER Slide 8 Worker" width="800"/> <img src="fetch_ddpg_her_slide_8_workers.png" alt="Fetch DDPG HER Slide 8 Worker" width="800"/>
@@ -33,7 +33,7 @@ python3 coach.py -p Fetch_DDPG_HER_baselines -lvl slide -n 8
### Fetch Pick And Place DDPG HER - 8 workers ### Fetch Pick And Place DDPG HER - 8 workers
```bash ```bash
python3 coach.py -p Fetch_DDPG_HER -lvl pick_and_place -n 8 coach -p Fetch_DDPG_HER -lvl pick_and_place -n 8
``` ```
<img src="fetch_ddpg_her_pick_and_place_8_workers.png" alt="Fetch DDPG HER Pick And Place 8 Workers" width="800"/> <img src="fetch_ddpg_her_pick_and_place_8_workers.png" alt="Fetch DDPG HER Pick And Place 8 Workers" width="800"/>

6
benchmarks/dfp/README.md
View File

@@ -6,7 +6,7 @@ The parameters used for DFP are the same parameters as described in the [origina
### Doom Basic DFP - 8 workers ### Doom Basic DFP - 8 workers
```bash ```bash
python3 coach.py -p Doom_Basic_DFP -n 8 coach -p Doom_Basic_DFP -n 8
``` ```
<img src="doom_basic_dfp_8_workers.png" alt="Doom Basic DFP 8 workers" width="800"/> <img src="doom_basic_dfp_8_workers.png" alt="Doom Basic DFP 8 workers" width="800"/>
@@ -15,7 +15,7 @@ python3 coach.py -p Doom_Basic_DFP -n 8
### Doom Health (D1: Basic) DFP - 8 workers ### Doom Health (D1: Basic) DFP - 8 workers
```bash ```bash
python3 coach.py -p Doom_Health_DFP -n 8 coach -p Doom_Health_DFP -n 8
``` ```
<img src="doom_health_dfp_8_workers.png" alt="Doom Health DFP 8 workers" width="800"/> <img src="doom_health_dfp_8_workers.png" alt="Doom Health DFP 8 workers" width="800"/>
@@ -25,7 +25,7 @@ python3 coach.py -p Doom_Health_DFP -n 8
### Doom Health Supreme (D2: Navigation) DFP - 8 workers ### Doom Health Supreme (D2: Navigation) DFP - 8 workers
```bash ```bash
python3 coach.py -p Doom_Health_Supreme_DFP -n 8 coach -p Doom_Health_Supreme_DFP -n 8
``` ```
<img src="doom_health_supreme_dfp_8_workers.png" alt="Doom Health Supreme DFP 8 workers" width="800"/> <img src="doom_health_supreme_dfp_8_workers.png" alt="Doom Health Supreme DFP 8 workers" width="800"/>

6
benchmarks/dqn/README.md
View File

@@ -6,7 +6,7 @@ The parameters used for DQN are the same parameters as described in the [origina
### Breakout DQN - single worker ### Breakout DQN - single worker
```bash ```bash
python3 coach.py -p Atari_DQN -lvl breakout coach -p Atari_DQN -lvl breakout
``` ```
<img src="breakout_dqn.png" alt="Breakout DQN" width="800"/> <img src="breakout_dqn.png" alt="Breakout DQN" width="800"/>
@@ -14,7 +14,7 @@ python3 coach.py -p Atari_DQN -lvl breakout
### Pong DQN - single worker ### Pong DQN - single worker
```bash ```bash
python3 coach.py -p Atari_DQN -lvl pong coach -p Atari_DQN -lvl pong
``` ```
<img src="pong_dqn.png" alt="Pong DQN" width="800"/> <img src="pong_dqn.png" alt="Pong DQN" width="800"/>
@@ -22,7 +22,7 @@ python3 coach.py -p Atari_DQN -lvl pong
### Space Invaders DQN - single worker ### Space Invaders DQN - single worker
```bash ```bash
python3 coach.py -p Atari_DQN -lvl space_invaders coach -p Atari_DQN -lvl space_invaders
``` ```
<img src="space_invaders_dqn.png" alt="Space Invaders DQN" width="800"/> <img src="space_invaders_dqn.png" alt="Space Invaders DQN" width="800"/>

23
benchmarks/dueling_ddqn/README.md
View File

@@ -3,12 +3,33 @@
Each experiment uses 3 seeds and is trained for 10k environment steps. 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](https://arxiv.org/abs/1706.01502). The parameters used for Dueling DDQN are the same parameters as described in the [original paper](https://arxiv.org/abs/1706.01502).
### Pong Dueling DDQN - single worker
```bash
coach -p Atari_Dueling_DDQN -lvl pong
```
<img src="pong_dueling_ddqn.png" alt="Pong Dueling DDQN" width="800"/>
### Breakout Dueling DDQN - single worker ### Breakout Dueling DDQN - single worker
```bash ```bash
python3 coach.py -p Atari_Dueling_DDQN -lvl breakout coach -p Atari_Dueling_DDQN -lvl breakout
``` ```
<img src="breakout_dueling_ddqn.png" alt="Breakout Dueling DDQN" width="800"/> <img src="breakout_dueling_ddqn.png" alt="Breakout Dueling DDQN" width="800"/>
### Space Invaders Dueling DDQN - single worker
```bash
coach -p Atari_Dueling_DDQN -lvl space_invaders
```
<img src="space_invaders_dueling_ddqn.png" alt="Space Invaders Dueling DDQN" width="800"/>

BIN
benchmarks/dueling_ddqn/breakout_dueling_ddqn.png
View File

Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 131 KiB

After

Width:  |  Height:  |  Size: 84 KiB

BIN
benchmarks/dueling_ddqn/pong_dueling_ddqn.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 70 KiB

BIN
benchmarks/dueling_ddqn/space_invaders_dueling_ddqn.png Normal file
View File

Binary file not shown.
Side by Side

After

Width:  |  Height:  |  Size: 79 KiB

6
benchmarks/dueling_ddqn_with_per/README.md
View File

@@ -6,7 +6,7 @@ The parameters used for Dueling DDQN with PER are the same parameters as describ
### Breakout Dueling DDQN with PER - single worker ### Breakout Dueling DDQN with PER - single worker
```bash ```bash
python3 coach.py -p Atari_Dueling_DDQN_with_PER_OpenAI -lvl breakout coach -p Atari_Dueling_DDQN_with_PER_OpenAI -lvl breakout
``` ```
<img src="breakout_dueling_ddqn_with_per.png" alt="Breakout Dueling DDQN with PER" width="800"/> <img src="breakout_dueling_ddqn_with_per.png" alt="Breakout Dueling DDQN with PER" width="800"/>
@@ -15,7 +15,7 @@ python3 coach.py -p Atari_Dueling_DDQN_with_PER_OpenAI -lvl breakout
### Pong Dueling DDQN with PER - single worker ### Pong Dueling DDQN with PER - single worker
```bash ```bash
python3 coach.py -p Atari_Dueling_DDQN_with_PER_OpenAI -lvl pong coach -p Atari_Dueling_DDQN_with_PER_OpenAI -lvl pong
``` ```
<img src="pong_dueling_ddqn_with_per.png" alt="Pong Dueling DDQN with PER" width="800"/> <img src="pong_dueling_ddqn_with_per.png" alt="Pong Dueling DDQN with PER" width="800"/>
@@ -24,7 +24,7 @@ python3 coach.py -p Atari_Dueling_DDQN_with_PER_OpenAI -lvl pong
### Space Invaders Dueling DDQN with PER - single worker ### Space Invaders Dueling DDQN with PER - single worker
```bash ```bash
python3 coach.py -p Atari_Dueling_DDQN_with_PER_OpenAI -lvl space_invaders coach -p Atari_Dueling_DDQN_with_PER_OpenAI -lvl space_invaders
``` ```
<img src="space_invaders_dueling_ddqn_with_per.png" alt="Space Invaders Dueling DDQN with PER" width="800"/> <img src="space_invaders_dueling_ddqn_with_per.png" alt="Space Invaders Dueling DDQN with PER" width="800"/>

4
benchmarks/qr_dqn/README.md
View File

@@ -6,7 +6,7 @@ The parameters used for QR-DQN are the same parameters as described in the [orig
### Breakout QR-DQN - single worker ### Breakout QR-DQN - single worker
```bash ```bash
python3 coach.py -p Atari_QR_DQN -lvl breakout coach -p Atari_QR_DQN -lvl breakout
``` ```
<img src="breakout_qr_dqn.png" alt="Breakout QR-DQN" width="800"/> <img src="breakout_qr_dqn.png" alt="Breakout QR-DQN" width="800"/>
@@ -15,7 +15,7 @@ python3 coach.py -p Atari_QR_DQN -lvl breakout
### Pong QR-DQN - single worker ### Pong QR-DQN - single worker
```bash ```bash
python3 coach.py -p Atari_QR_DQN -lvl pong coach -p Atari_QR_DQN -lvl pong
``` ```
<img src="pong_qr_dqn.png" alt="Pong QR-DQN" width="800"/> <img src="pong_qr_dqn.png" alt="Pong QR-DQN" width="800"/>

2
setup.py
View File

@@ -50,7 +50,7 @@ with open(path.join(here, 'README.md'), encoding='utf-8') as f:
install_requires=[ install_requires=[
'annoy==1.8.3', 'Pillow==4.3.0', 'matplotlib==2.0.2', 'numpy==1.14.5', 'pandas==0.22.0', 'annoy==1.8.3', 'Pillow==4.3.0', 'matplotlib==2.0.2', 'numpy==1.14.5', 'pandas==0.22.0',
'pygame==1.9.3', 'PyOpenGL==3.1.0', 'scipy==0.19.0', 'scikit-image==0.13.0', 'pygame==1.9.3', 'PyOpenGL==3.1.0', 'scipy==0.19.0', 'scikit-image==0.13.0',
'box2d==2.3.2', 'gym==0.10.5', 'gym[atari]==0.10.5', 'bokeh==0.13.0', 'futures==3.1.1', 'wxPython==4.0.1'] 'box2d==2.3.2', 'gym==0.10.5', 'bokeh==0.13.0', 'futures==3.1.1', 'wxPython==4.0.1']
# check if system has CUDA enabled GPU # check if system has CUDA enabled GPU
p = subprocess.Popen(['command -v nvidia-smi'], stdout=subprocess.PIPE, shell=True) p = subprocess.Popen(['command -v nvidia-smi'], stdout=subprocess.PIPE, shell=True)