PyTorch Day China 2026 is the premier gathering dedicated to open source AI and the impact of PyTorch Foundation projects. Scheduled for 12 December in China this community-driven event provides an unparalleled platform for PyTorch enthusiasts, machine learning engineers, AI researchers, and industry professionals.
Immerse yourself in a vibrant day of insightful technical talks and interactive discussions designed to foster knowledge exchange and collaboration. PyTorch Day China is your gateway to connecting with leading experts and peers in the open source AI community, offering you unique opportunities to explore cutting-edge advancements in training, inference, responsible AI, and more.
Please be aware that the LF Open Source LLC utilizes Sessionize for CFP submissions. Sessionize is a cloud-based event content management software designed to be intuitive and user-friendly. If you need guidance, please review how to submit your session for an event to see step-by-step instructions and helpful screenshots.
For questions about the CFP system, please get in touch with cfp@lfopensource.com.
- CFP Close: Tuesday, 10 November at 23:59 China Standard Time (UTC +8)
- CFP Notifications: Tuesday, 17 November
- Schedule Announced: Wednesday, 18 November
- Event Date: Saturday, 12 December
- Model Training & Post-Training: Model training, fine-tuning, and alignment using PyTorch and related open-source stacks. Topics include distributed training, supervised fine-tuning (SFT), reinforcement learning, preference optimization, and improvements in training efficiency, stability, and resource utilization.
- Efficient Inference & Deployment: Deploying models as reliable, continuously running services. Topics include LLM and multimodal inference, quantization, speculative decoding, KV cache optimization, cluster scheduling, and trade-offs among latency, throughput, and cost across cloud, on-premises, and local deployments.
- PyTorch Core & Hardware Acceleration: Advances in PyTorch compilation, execution, and hardware enablement. Topics include torch.compile, kernel and operator optimization, performance profiling, and adaptation, compatibility, and reproducible benchmarking across GPUs and AI accelerators.
- Physical AI: Applying PyTorch models to robotics and physical devices. Topics include Vision-Language-Action (VLA) models, perception and control, and simulation-based training.
- PyTorch Ecosystem & Applications: Tools, domain applications, and community contributions built on PyTorch and related open-source projects. Topics include domain-specific libraries, development and evaluation tools, and open-source collaboration.
- Session Presentation (20 minutes, Max of 3 Speakers): A focused talk covering a specific topic, concept, or case study, could also include brief Q&A.
- Note – LF Open Source LLC does not allow any talk with 3 or more speakers to only have men participating in an effort to increase equity & inclusion.
- Lightning Talk (10 minutes, Max of 2 Speakers): A fast-paced, high-impact presentation that delivers key insights or ideas in a concise format.
- Reviewers must be able to confidently assess a speaker’s subject-matter expertise from the proposal. Submissions that rely heavily on AI generated or templated content often lack the specificity needed to evaluate technical depth and may be harder to advance. Please ensure your proposal clearly reflects your own experience and understanding of the topic you intend to present.
- All speakers are required to adhere to our Code of Conduct. We also highly recommend that speakers take our online Inclusive Speaker Orientation Course.
- Focus on delivering high-quality, well-crafted sessions rather than submitting multiple proposals. Strong, engaging content stands out!
- Complimentary passes for the event will be provided for the accepted speaker(s) per submission.
- LF Open Source LLC does not allow any talk with 3 or more speakers to only have men participating in an effort to increase equity & inclusion.
- Avoid sales or marketing pitches and discussing unlicensed or potentially closed-source technologies when preparing your proposal; these talks are almost always rejected due to the fact that they take away from the integrity of our events, and are rarely well-received by conference attendees.
- All accepted speakers are required to submit their slides prior to the event.
Preparing to Submit Your Proposal
While it is not our intention to provide you with strict instructions on how to prepare your proposal, we hope you will take a moment to review the following guidelines that we have put together to help you prepare the best submission possible. To get started, here are three things that you should consider before submitting your proposal:
- What are you hoping to get from your presentation?
- How will your presentation benefit the PyTorch community
There are plenty of ways to give a presentation about projects and technologies without focusing on company-specific efforts. Remember the things to consider that we mentioned above when writing your proposal and think of ways to make it interesting for attendees while still letting you share your experiences, educate the community about an issue, or generate interest in a project.
How to Give a Great Talk
We want to ensure submitters receive resources to help put together a great submission and, if accepted, give the best presentation possible. To help do this, we recommend viewing seasoned speaker Dawn Foster’s in-depth session titled Overcoming Imposter Syndrome to Become a Conference Speaker!
Have More Questions? First Time Submitting? Don’t Feel Intimidated
LF Open Source events are an excellent way to get to know the community and share your ideas and the work you are doing, and we strongly encourage first-time speakers to submit talks for our events. If you aren’t sure about your abstract, reach out to us, and we will be more than happy to work with you on your proposal.
LF Open Source LLC is dedicated to providing a harassment-free experience for participants at all of our events. We encourage all submitters to review our complete Code of Conduct.
