This repository is designed to allow projects seeking to join the PyTorch Foundation Ecosystem as a project to submit their applications. It is still a work in progress and we welcome PRs on this repo to automate assignments, improve the process, etc. If you have any questions, please post a message to the Slack Channel. For information on the process please check out the details on our application process below.
Please review the Ecosystem review process to ensure that you meet the minimum expectations before applying.
Applying is as easy as 1-2-3!
- Open a new Ecosystem Application using the linked issue form.
- Complete all sections.
- Submit the issue.
Once you have submitted your issue, your project's application is placed in the project backlog for triaging. You can view the status of your application and others at any time by checking out our Project Board.
As the community approaches the Ecosystem review meetings, PyTorch staff will move applications to the "in-progress" column on the Project Board (number dependent on how many can be reviewed in a given meeting). At this point, members of the PyTorch Ecosystem Working Group will ask questions, seek to learn more, and make a decision on the project.
Be sure to check your project issue periodically to see if the PyTorch Ecosystem Working Group has asked questions or for more information in the issue.
The PyTorch Ecosystem Working Group welcomes and appreciates community members support and exploration in surfacing items for technical consideration as part of application process because it allows the PyTorch Ecosystem Working Group members to have a more comprehensive perspective on each projects' potential and supports our understanding about these projects.
When the PyTorch Ecosystem Working Group members meet to review applications, your application will be pulled up and discussed. Your application may receive one of several status and/or labels, some of which are defined below.
Removing, updating, and adding labels as well as updating the project board are the responsibility of the PyTorch Ecosystem Working Group and PyTorch staff. Please refrain from adding additional labels.
The application is new and placed in the backlog.
The application has been approved and will be added. PyTorch staff will reach out to assist you in joining the Ecosystem.
The application has been declined. The reasons for decline will be provided in the comments of the issue and described in any additional labels applied to the issue.
If your project is assigned this label the PyTorch Ecosystem Working Group will comment on the issue with the specific additional information needed.
Please provide the additional information requested as a comment on the issue with links as appropriate.
If the issue has been affixed with this label it will be closed as the PyTorch Ecosystem Working Group has determined the project at the time of discussion is not ready for inclusion into PyTorch. There will be a comment on the issue that annotates the expectations in order for the project to be re-reviewed by the PyTorch Ecosystem Working Group. You may resubmit your project for review once you have completed the items noted in the issue.
If the issue has been affixed with this label the PyTorch Ecosystem Working Group has decided that the project is useful but too early in its lifecycle. They will request that you write a community blog post with the help of the PyTorch marketing team, who will reach out to assist you in creating the blog post.
The PyTorch Ecosystem Working Group reviews applications on a first-come first-served basis. It may take several months before an application may be reviewed, depending on community leadership availability.
The PyTorch Ecosystem Working Group attempts to work through approximately 7-10 applications per session. How many are actually reviewed in a session varies greatly between each session. Every project is unique and may warrant different areas of attention to be elevated for more in-depth discussion, depending on the nature and function of the project.
Each PyTorch Ecosystem Working Group member that is part of the review committee prepares for these sessions by performing an independent review of those projects scheduled for discussion at the next session prior to that session. They'll keep their notes on hand for when the project is up for discussion.
Applications are traditionally reviewed in a First In, First Out (FIFO) ordering as they appear on the project board by their corresponding issue number. The only exception is where projects are Returning for review. Projects that are Returning are prioritized for re-review as they had previously been subject to a review and have completed their outstanding asks. In cases where a project was Postponed and has substantial changes since last review, a new issue is opened, linked to the previous, and reviewed as a new application.
Applications are moved from New to in progress approximately two weeks prior to the next scheduled review session.
During the course of the session, the PyTorch Ecosystem Working Group will actively discuss the project based on the information provided in the issue as well as other observations the has about the project. Those observations vary greatly and may be subject to each member's area of domain expertise, personal experience, community feedback, or other factors. During the course of active discussion, the group members may identify areas the project needs to work on, complete, or allow to happen that could inhibit its acceptance into the PyTorch. These areas will be captured in the issue comments.
The PyTorch Ecosystem Working Group members will then vote on the project acceptance with a simple majority vote, one vote per member.
Once the voting is complete, the project is labeled with the status as accepted or declined based on the voting status. Projects will have information in the issue that describes feedback provided by the .
A project that has been declined may reapply for acceptance once the comments raised by the PyTorch Ecosystem Working Group as the reasons for the decision have been addressed.