pytorch-DDPG

KO | EN

gitlite — search

#javascript #python #hacktoberfest #react #ai #typescript #llm #go #golang #android #machine-learning #rust #deep-learning #linux

★ 9 Open GitHub ↗

Using pytorch to implement DDPG

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

Using pytorch to implement [Deep Deterministic Policy Gradient(DDPG)](https://arxiv.org/abs/1509.02971).

## Denpendency
- python 3.6
- pytorch 0.4+
- tensorboard
- gym

## Train
```
main.py --train --env MountainCarContinuous-v0 --cuda
```
Parameters:

(more parameters see the file)

You can use the tensorboard to see the training.
```
tensorboard --logdir=out/MountainCarContinuous-v0
```

## Test
You can test your model with `--test` like this:
```
main.py --test --env MountainCarContinuous-v0 --model_path out/MountainCarContinuous-v0-run0
```
It will render graphical interface.

## Result
It turns out that tuning parameters are very important, especially `eps_decay`.  I use the simple linear noise decay such as `epsilon -= eps_decay` every episode.

- Pendulum-v0

```
main.py --train --env Pendulum-v0 --cuda --eps_decay 0.01
```

- MountainCarContinuous-v0

```
main.py --train --env MountainCarContinuous-v0 --cuda --eps_decay 0.001
```

## Reference
- paper [Continuous control with deep reinforcement learning](https://arxiv.org/abs/1509.02971)

pytorch-DDPG

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README