Probabilistically Enriched Transformers Applied on Neural Spatio-temporal Point Processes

Papers

Erfanian, Negar, Santiago Segarra, and Maarten V. de Hoop. "Neural multi-event forecasting on spatio-temporal point processes using probabilistically enriched transformers." (2022).

1. Survey papers

Configure docker container on the server

First clone project in your desired path:

git clone [email protected]:Negar-Erfanian/Neural-spatio-temporal-probabilistic-transformers.git

In ssh command, first pull docker image

docker pull tensorflow/tensorflow:latest-gpu-py3-jupyter

Then launch the container

docker run --user $(id -u):$(id -g) --runtime=nvidia --rm -it -v ~/Neural-spatio-temporal-probabilistic-transformers:/tensorflow/Neural-spatio-temporal-probabilistic-transformers -w /tensorflow/Neural-spatio-temporal-probabilistic-transformers -p 8300:8888 tensorflow/tensorflow:latest-gpu-py3-jupyter

Access the jupyter interface of the container from browser, and launch a terminal from jupyter

pip3 install -r requirements.txt

Test if everything is ok from the jupyter terminal

python3 train.py

Install any missing packages in the error message

If test pass, commit the container from ssh command window

Check container id docker container ls, assuming it is c3f279d17e0a

Then commit the changes to the image of the same name

docker commit c3f279d17e0a tensorflow/tensorflow:latest-gpu-py3-jupyter

The newly installed python packages are now ready for next time you launch the container.