mirror of
https://github.com/wassname/DeepRL.git
synced 2026-09-09 11:13:47 +08:00
75 lines
3.8 KiB
Markdown
75 lines
3.8 KiB
Markdown
# DeepRL
|
|
Highly modularized implementation of popular deep RL algorithms by PyTorch. My principal here is to
|
|
reuse as much components as I can through different algorithms, use as less tricks as I can and switch
|
|
easily between classical control tasks like CartPole and Atari games with raw pixel inputs.
|
|
|
|
Implemented algorithms:
|
|
* Deep Q-Learning (DQN)
|
|
* Double DQN
|
|
* Dueling DQN
|
|
* Async Advantage Actor Critic (A3C)
|
|
* Async One-Step Q-Learning
|
|
* Async One-Step Sarsa
|
|
* Async N-Step Q-Learning
|
|
* Continuous A3C
|
|
* Deep Deterministic Policy Gradient (DDPG)
|
|
|
|
# Curves
|
|
> Curves for CartPole are trivial so I didn't place it here.
|
|
## DQN, Double DQN, Dueling DQN
|
|
|
|
|
|
|
|
The network and parameters here are exactly same as the [DeepMind Nature paper](https://www.nature.com/nature/journal/v518/n7540/full/nature14236.html).
|
|
Training curve is smoothed by a window of size 100. All the models are trained in a server with
|
|
Xeon E5-2620 v3 and Titan X. For Breakout, test is triggered every 1000 episodes with 50 repetitions.
|
|
In total, 16M frames cost about 4 days and 10 hours. For Pong, test is triggered
|
|
every 10 episodes with no repetition. In total, 4M frames cost about 18 hours.
|
|
|
|
## Discrete A3C
|
|
|
|
|
|
|
|
|
|
The network I used here is a smaller network with only 42 * 42 input, alougth the network for DQN can also work here,
|
|
it's quite slow.
|
|
|
|
Training of A3C took about 2 hours (16 processes) in a server with two Xeon E5-2620 v3. While other async methods took about 1 day.
|
|
Those value based async methods do work but I don't know how to make them stable.
|
|
This is the test curve. Test is triggered in a separate deterministic test process every 50K frames.
|
|
|
|
## Continuous A3C
|
|
|
|
|
|
Sometimes _Bipedal Walker_ may run into _NAN_, I'm still not able to totally solve it. And continuous A3C is very sensible to hyper parameters.
|
|
|
|
# Dependency
|
|
* Open AI gym
|
|
* PyTorch
|
|
* PIL (pip install Pillow)
|
|
* Python 2.7 (I didn't test with Python 3)
|
|
* Tensorflow (We need tensorboard)
|
|
|
|
# Usage
|
|
Detailed usage and all training parameters can be found in ```main.py```
|
|
And you need to create following directories before running the program:
|
|
```
|
|
cd DeepRL
|
|
mkdir data log evaluation_log
|
|
```
|
|
|
|
# References
|
|
* [Human Level Control through Deep Reinforcement Learning](https://www.nature.com/nature/journal/v518/n7540/full/nature14236.html)
|
|
* [Asynchronous Methods for Deep Reinforcement Learning](https://arxiv.org/abs/1602.01783)
|
|
* [Deep Reinforcement Learning with Double Q-learning](https://arxiv.org/abs/1509.06461)
|
|
* [Dueling Network Architectures for Deep Reinforcement Learning](https://arxiv.org/abs/1511.06581)
|
|
* [Playing Atari with Deep Reinforcement Learning](https://arxiv.org/abs/1312.5602)
|
|
* [HOGWILD!: A Lock-Free Approach to Parallelizing Stochastic Gradient Descent](https://arxiv.org/abs/1106.5730)
|
|
* [Deterministic Policy Gradient Algorithms](http://proceedings.mlr.press/v32/silver14.pdf)
|
|
* [Continuous control with deep reinforcement learning](https://arxiv.org/abs/1509.02971)
|
|
* [High-Dimensional Continuous Control Using Generalized Advantage Estimation](https://arxiv.org/abs/1506.02438)
|
|
* [transedward/pytorch-dqn](https://github.com/transedward/pytorch-dqn)
|
|
* [ikostrikov/pytorch-a3c](https://github.com/ikostrikov/pytorch-a3c)
|
|
* [ghliu/pytorch-ddpg](https://github.com/ghliu/pytorch-ddpg)
|
|
* [MorvanZhou/Reinforcement-learning-with-tensorflow](https://github.com/MorvanZhou/Reinforcement-learning-with-tensorflow)
|