https://github.com/takuseno/mvc-drl-nnabla

NNabla implementation of mvc-drl
https://github.com/takuseno/mvc-drl-nnabla

deep-reinforcement-learning nnabla

Last synced: about 2 months ago
JSON representation

NNabla implementation of mvc-drl

• Host: GitHub
• URL: https://github.com/takuseno/mvc-drl-nnabla
• Owner: takuseno
• License: mit
• Created: 2019-04-25T11:28:33.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2019-06-06T10:36:42.000Z (over 5 years ago)
• Last Synced: 2024-10-23T04:06:28.197Z (3 months ago)
• Topics: deep-reinforcement-learning, nnabla
• Language: Python
• Size: 10.7 KB
• Stars: 2
• Watchers: 2
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# mvc-drl-nnabla

[NNabla](https://nnabla.org) implementation with [mvc-drl](https://github.com/takuseno/mvc-drl).

Algorithms implemented in this repository are faster than the original TensorFlow version.

## installation

### nvidia-docker

```

$ ./scripts/build.sh

```

### manual

```

$ pip install -r requirements.txt

$ pip install nnabla

# if you run example scripts

$ pip install pybullet roboschool

```

If you use GPU, see [here](https://nnabla.readthedocs.io/en/latest/python/pip_installation_cuda.html).

## algorithms

For academic usage, we provide baseline implementations that you might need to compare.

- [ ] Proximal Policy Optimization

- [x] Deep Deterministic Policy Gradients

- [ ] Soft Actor-Critic

## speed comparison

TODO.