PyTorch Content Intake Process + Guidelines

The PyTorch blog serves as a central hub for sharing insights, updates, and technical content with our community.

Features content like technical deep dives, release announcements, roadmap updates, tutorials, case studies, project highlights, research applications, and real-world use cases that demonstrate how PyTorch Foundation projects are being applied across domains.

• Please allow a minimum of 2–3 weeks’ lead time so the technical review team has adequate time to complete a thorough review.
• Content will be vetted for story value, relevance, and timeliness. Blogs should follow a narrative format.
• PyTorch Foundation team will initiate the review process for each blog that includes a technical review and placement in the editorial calendar.

• Topic Area: Stick to topics directly relevant to the PyTorch Foundation and foundation-hosted projects. Posts must focus on technology that is open source and accessible to readers. Posts should be well scoped and focus on a single primary topic or idea.
• Tone: Friendly, yet authoritative with a preference for first-person voice. Tone should remain professional and low on marketing language. Claims should be supported with appropriate context or evidence where possible.
• Word Count: For summaries and updates – 500+ words. For technical deep-dives – 1,500+ words
• Style: Focus on readability. Write for the non-expert. Spell out acronyms upon first use. Break content into easily digestible parts with headings. Ensure a clear structure and sequencing, introducing concepts in the order they are needed.
• Technical Accuracy and Consistency: Content should be internally consistent and, where applicable, make a reasonable effort toward reproducibility. Use consistent capitalization, terminology, and nomenclature throughout.
• Attribution: Identify author(s) and affiliations. When possible, try to have authors from multiple organizations to demonstrate the breadth of support and collaboration.
• Intent: No sales pitches or product marketing pieces please. While it is ok to highlight the work of an individual company when it ties into the overall theme of the blog, any company or product mentions must be directly related to the PyTorch Foundation, remain balanced, and not be at the expense of others. Blogs exclusively or primarily about a for-profit-company’s products or services will not be accepted.
• Images: Relevant graphics like charts, graphs, and photos are encouraged.

• We offer this blog template if you’d like a starting place to ensure your blog post is informative, technically accurate, and valuable to the PyTorch community. 
• A strong blog post follows a clear narrative. Start with the motivation, guide the reader through your technical contribution, and end with results, practical guidance, and future opportunities. The narrative should remain focused on a single core contribution or idea. Your goal is to explain what you’ve built or discovered, why it matters, and how others can use or build on it.
• Please close out any comments

• Content submitted: Once submitted through the intake form, the content goes into a tracking system through the PyTorch Foundation (PTF).
  • Desired social media copy, including preferred hashtags, must be included in the indicated section of the content form.
• Content triaged: Content is vetted and scheduled into the existing editorial calendar. The PTF team to follow up with the submitter with any questions, concerns, or updates.
• Content reviewed: Content will be reviewed by PTF marketing and communications and the technical review panel. If necessary, content will be reviewed by Linux Foundation legal. 
• Content sent back: Upon approval, the content will be sent back to the submitter for a final look at the content before posting. 
• Content scheduled: PTF team will schedule the post once final, and share final links with the submitter to synchronize amplification whenever possible. 
• Content amplified: PTF team will amplify content on relevant PyTorch channels. We recommend that the submitting company amplifies on their channels as well.

The PyTorch Foundation team owns all PyTorch social media channels (X, LinkedIn, Facebook, BlueSky).

• The PyTorch Foundation team will work with member companies and community members to amplify content based on social media submissions included in the content form.
• Content will be vetted for story value, relevance, and timeliness.
• The PyTorch Foundation team will own all communications with other companies when the content is solely about said company and PyTorch as an open source project.

Typically, these post accompany a blog post, case study or similar content. You do not need to submit a separate form for social media if it is connected to a blog post.

 Please provide the following details in your request for amplification:

• About: Provide a brief description of what you are amplifying.
• Platform: Do you want it to post on Twitter, LinkedIn and/or Facebook?
• Date/time to post: Provide with the month, day and time, specifying time zone.
• Suggested copy: Let us know what you would like the post to say, and include the CTA and other links.
• Images: Let us know if you have an image to attach.
• Tags: Is there an account you want us to tag?
• Hashtags: Are there any hashtags that would be useful to include?
• Other notes: Is there anything helpful for us to know?

All content must be reviewed and final by the time you submit it to us.
If you have any questions or concerns about your content, please contact the foundation team.

• The language used should be as precise and clear as possible.
• We do not amplify or share content that could unduly promote or benefit one company in the PyTorch community over the other.

We are glad to repost and like content related to PyTorch on X, LinkedIn or Facebook. We are able to amplify content about open source projects for members, but no product announcements. The easiest and quickest way to do this is to ping the foundation team on Slack. We will aim to repost as soon as possible.

It is important to note that just because you request amplification, it does not mean that we will automatically post. We will do our best to follow up with you if we have any questions or concerns, and if we can post your content within the timeframe requested. Sometimes, the editorial calendar will fill up, and other announcements may conflict with yours. The team will use their best discretion on amplification.