1.24 MB

1 contributor

It seems that DQN performs the worst if trained for 1e6 timesteps. But it did train quicker, taking about 17 min, as opposed to 20-22.

3f14e63 about 3 years ago

DQN-1e6
It seems that DQN performs the worst if trained for 1e6 timesteps. But it did train quicker, taking about 17 min, as opposed to 20-22. about 3 years ago
a2c-1e6
A2C trained on LunarLander-v2 for 1e6 timesteps about 3 years ago
lander-MlpPolicy-1
The initial run of the lander, after training for 1M timestamps. about 3 years ago
ppo-1e6
Trained on my local machine. about 3 years ago
.gitattributes
1.48 kB

initial commit about 3 years ago
DQN-1e6.zip
110 kB
xet

It seems that DQN performs the worst if trained for 1e6 timesteps. But it did train quicker, taking about 17 min, as opposed to 20-22. about 3 years ago
README.md
784 Bytes

It seems that DQN performs the worst if trained for 1e6 timesteps. But it did train quicker, taking about 17 min, as opposed to 20-22. about 3 years ago
a2c-1e6.zip
102 kB
xet

A2C trained on LunarLander-v2 for 1e6 timesteps about 3 years ago
config.json
19.4 kB

It seems that DQN performs the worst if trained for 1e6 timesteps. But it did train quicker, taking about 17 min, as opposed to 20-22. about 3 years ago
lander-MlpPolicy-1.zip
147 kB
xet

The initial run of the lander, after training for 1M timestamps. about 3 years ago
ppo-1e6.zip
148 kB
xet

Trained on my local machine. about 3 years ago
replay.mp4
207 kB

It seems that DQN performs the worst if trained for 1e6 timesteps. But it did train quicker, taking about 17 min, as opposed to 20-22. about 3 years ago
results.json
165 Bytes

It seems that DQN performs the worst if trained for 1e6 timesteps. But it did train quicker, taking about 17 min, as opposed to 20-22. about 3 years ago