To evaluate on Aftershock, we ask you to submit a Docker container image containing the inference algorithm you'd like to submit. The inclusion of training code is encouraged but not required.

You can submit an algorithm for evaluation on one or more tasks. For each task under the "Tasks" section on this website, you will find a link to a development dataset for the purposes of fine-tuning prior to submission, and information about the semantic contract your Docker image must obey in order to read the hidden testing dataset during evaluation, and submit predictions.

A sample submission with the necessary formatting will be released soon. To render your submission, please use this form.


We use Docker to orchestrate and evaluate submissions in a uniform fashion. To contribute your models to Aftershock, we ask that you submit a docker container image. If you are not familiar with Docker, we invite you to take a look at the excellent official tutorials to help you set up docker and build your container image.

We have published a sample submission to help you get started and we are open to requests for more specialised tutorials relating to specific environments (such as Matlab). The authors may also be able to personally assist you on a best-effort basis.