- Pietro Carta, pietro.carta@epfl.ch
- Golinski Pawel, pawel.golinski@epfl.ch
- Jonathan Besomi, jonathan.besomi@epfl.ch
During the development phase we used different platforms to execute our code. We started testing out on our personal computers, then we moved to Google Cloud Platform and finally to Google Colab Notebook. Google Colab Notebook turned out to be the best since it provides a free GPU and can be share among multiple users.
There are thousands of papers submitted to the ICLR conference. Before deciding which paper to review, we spent some times looking for a convincing paper. Our approach has been first to download all papers locally, then order by number of comments and finally manually find an available paper that appeared interesting.
/find-papersfolder contains code to download and parse all papers submitted to the ICLR conference at openreview.net, order by number of comments
An executed version of the Jupyter Notebook is accessible on google colab platform at the following address: go.epfl.ch/neural-fingerprinting
The same version is also available locally at the following path: ./run.ipynb
Our research is coupled with a paper available here: report.pdf