Following the success of last year's blogpost track, we decided to host it again for ICLR 2023. More context about this initiative can be found on the main website of the Blog Post Track 2023.
Call for blogposts
We invite all researchers and practicioners to submit a blogpost discussing work previously published at ICLR, to the ICLR 2023 blogpost track.
The format and process for this blog post track is as follows:
- Write a post on a subject that has been published at ICLR relatively recently. The authors of the blog posts will have to declare their conflicts of interest (positive nor negative) with the paper (and their authors) they write about. Conflicts of interest include:
- Recent collaborators (less than 3 years)
- Current institution.
Blog Posts must not be used to highlight or advertise past publications of the authors or of their lab. Previously, we did not accept submissions with a conflict of interest, however this year we will only ask the authors to report if they have such a conflict. If so, reviewers will be asked to judge if the submission is sufficiently critical and objective of the papers addressed in the blog post.
The posts will be created and published under a unified template; see the submission instructions and the sample post hosted on the blog of this website.
- Blogs will be peer-reviewed (double-blind) for quality and novelty of the content: clarity and pedagogy of the exposition, new theoretical or practical insights, reproduction/extension of experiments, etc. We are slightly relaxing the double-blind constraints by assuming good faith from both submitters and reviewers (see the submission instructions for more details).
Submission deadline: February 2nd, 2023
OpenReview Registration deadline: February 2nd (please register your submission title by the orginal deadline)
GitHub Pull Request Deadline: February 10th
Notification of acceptance: March 31st, 2023
Camera-ready merge: TBD
See the submission instructions for more details.
For this edition of the Blogposts Track, we will forgo the requirement for total anonymity. The blog posts must be anonymized for the review process, but users will submit their anonymized blog posts via a pull request to a staging repository (in addition to a submission on OpenReview). The post will be merged into the staging repository, where it will be deployed to a separate Github Pages website. Reviewers will be able to access the posts directly through a public url on this staging website, and will submit their reviews on OpenReview. Reviewers should refrain from looking at the git history for the post, which may reveal information about the authors.
This still largely follows the Double-Blind reviewing principle; it is no less double-blind than when reviewers are asked to score papers that have previously been released to arXiv, an overwhelmingly common practice in the ML community. This approach was chosen to lower the burden on both the organizers and the authors; last year, many submissions had to be reworked once deployed due to a variety of reasons. By allowing the authors to render their websites to Github Pages prior to the review process, we hope to avoid this issue entirely. We also avoid the issue of having to host the submissions on a separate server during the reviewing process.
However, we understand the desire for total anonymity. Authors that wish to have a fully double-blind process might consider creating new GitHub accounts without identifying information which will only be used for this track. For an example of a submission in the past which used an anonymous account in this manner, you can check out the World Models blog post (Ha and Schmidhuber, 2018) and the accompanying repository.