Python implementation of the PR-SSM.
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

Probabilistic Recurrent State-Space Models

This is the companion code for the dynamics model learning method reported in the paper Probabilistic Recurrent State-Space Models by Andreas Doerr et al., ICML 2018. The paper can be found here The code allows the users to reproduce the PR-SSM results reported in the benchmark and large-scale experiments. Please cite the above paper when reporting, reproducing or extending the results.

Purpose of the project

This software is a research prototype, solely developed for and published as part of the publication cited above. It will neither be maintained nor monitored in any way.

Requirements, how to build, test, install, use, etc.

The PR-SSM code depends on Tensorflow.


In order to train a PR-SSM model for a new dataset, a new task has to be derived from the task base class. See for example

A valid path must be provided to store the experimental results and log files. An example is given in

Reproducing PR-SSM results

The experiments reported in the publication can be run by executing

python benchmarks/run_real_world_tasks/
python benchmarks/run_real_world_tasks/

The individual datasets have to be provided in the datasets folder.


Probabilistic recurrent state-space models is open-sourced under the MIT license. See the LICENSE file for details.

For a list of other open source components included in Benchmarks, see the file 3rd-party-licenses.txt.