gaspersic/baselines
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Update README

Browse Source
This commit is contained in:
Matthias Plappert
2018-03-26 16:50:22 +02:00
parent 3cc7df0608
commit 1f8a03f3a6
1 changed files with 7 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
baselines/her/README.md
View File

@@ -1,5 +1,5 @@
# Hindsight Experience Replay # Hindsight Experience Replay
For details on Hindsight Experience Replay (HER), please read the [paper](https://arxiv.org/pdf/1707.01495.pdf). For details on Hindsight Experience Replay (HER), please read the [paper](https://arxiv.org/abs/1707.01495).
## How to use Hindsight Experience Replay ## How to use Hindsight Experience Replay
@@ -22,14 +22,11 @@ You can try it right now with the results of the training step (the script print
This should visualize the current policy for 10 episodes and will also print statistics. This should visualize the current policy for 10 episodes and will also print statistics.
### Advanced usage ### Reproducing results
The train script comes with advanced features like MPI support, that allows to scale across all cores of a single machine. In order to reproduce the results from [Plappert et al. (2018)](https://arxiv.org/abs/1802.09464), run the following command:
To see all available options, simply run this command:
```bash ```bash
python -m baselines.her.experiment.train --help python -m baselines.her.experiment.train --num_cpu 19
``` ```
To run on, say, 20 CPU cores, you can use the following command: This will require a machine with sufficient amount of physical CPU cores. In our experiments,
```bash we used [Azure's D15v2 instances](https://docs.microsoft.com/en-us/azure/virtual-machines/linux/sizes),
python -m baselines.her.experiment.train --num_cpu 20 which have 20 physical cores. We only scheduled the experiment on 19 of those to leave some head-room on the system.
```
That's it, you are now running rollouts using 20 MPI workers and average gradients for network updates across all 20 core.