Tensors and neural networks in Python with strong hardware acceleration. PyTorch is an open source project at the Linux Foundation. #PyTorchFoundation
Joined September 2016
- Tweets3.9K
- Following106
- Followers521K
- Likes1.2K
Fast kernels, without the tuning wait: Helion's new compile-time heuristics automatically choose high-performance configurations, with no autotuning required.
On H100 benchmarks, they beat handwritten Triton kernels for linear-attention forward passes by 18% and vLLM's handwritten CUDA reduction kernels by 27%.
Read the complete technical breakdown to learn more: bit.ly/3Vja0dq
Ron Kahn of @nvidia, Andrey Velichkevich of @Apple, Vara Bonthu of @awscloud, Ekin Karabulut of @nvidia, & Manabu McCloskey of @awscloud discuss powering the next era of cloud native AI with an open community at #KubeCon + #CloudNativeCon today, citing @kubeflow's @CloudNativeFdn graduation with shout outs to PyTorch Foundation projects - @PyTorch and @DeepSpeedAI. The importance of working across foundations is vital for a thriving ecosystem.
With help from AI coding agents, SGLang was integrated with @IBM Spyre in under a week using a 1,500-line plugin. Reusing existing PyTorch Spyre support and SGLang extension points delivered roughly 3.6x the throughput of the reference torch-spyre implementation.
Performance matched vLLM across Granite 3.3 8B and Gemma 4 26B A4B workloads, with RadixCache prefix sharing working out of the box without hardware-specific changes. This shows how open source AI standards and modular architectures allow developers to bring new silicon online quickly and expand hardware choice across the stack.
Read the full technical breakdown: bit.ly/4iMtA9B
@gal_bloch @RaghuGanti @mudhakar @alxnails
ALT Bar chart titled "Relative performance" showing comparison of two models: Granite 3.3B and Gemma 4 26B-A4B. Both models have similar performance values close to 1.0, with Gemma slightly higher at 1.008. Granite is marked as 0.993. The chart indicates that SGLang shows Granite is 0.7% faster and Gemma is 0.6% faster.
3
12
3,280
Thanks to PyTorch and TorchTitan, the Domyn Team successfully scaled up the training of a 1 trillion parameter MoE on Jupiter, a shared EuroHPC supercomputer, in the context of the European Frontier AI Grand Challenge
Per-GPU throughput on 384 GH200 GPUs per replica rose from 90 to ~1,300 tokens/s
Learn about all the details and the upstream contribution on PyTorch and TorchTitan: bit.ly/4eyDt9I
@domynai
ALT A digital graphic with blue diagonal stripes and text that reads, "Enabling 1T MoE training with PyTorch and TorchTitan on EU public infrastructure." The Domyn logo is at the bottom.
You don't measure an open source ecosystem in GitHub stars. You measure it by who still shows up on the day Beijing gets its first snow.
On Nov 7, Ascend for PyTorch Meetup brought maintainers from across the stack — PyTorch core and TorchNPU, @vllm_project and vLLM-Omni, MiniCPM / OpenBMB, LLaMA-Factory, PenguinHarness, FlagOS — together with more than a hundred developers in Beijing for a full day of talks, roundtables and hands-on work.
Morning: multi-backend integration, TorchNPU distributed training, vLLM-Omni inference, MiniCPM on-device models, FlagOS operators, Agent tooling.
Afternoon: an AI Infra roundtable, a contribution workshop walking through PR review to merge, an Agent build session — and 30+ community tasks opened on site.
At #PyTorchCon China this September, the Foundation shared what that energy is turning into: more than 250 organizations across China now contribute to Foundation projects including vLLM; Chinese developers now rank among the leading contributor groups on vLLM; and Chinese open-weight models went from 2% to 45% of OpenRouter token traffic between late 2024 and April 2026.
Enthusiasm is the starting point. Upstream commits are the proof.
That's why a vendor-neutral home matters: ANY MODEL · ANY CHIP · ANY CLOUD · ANY AGENT — so local energy feeds the global mainline instead of fragmenting into separate regional stacks.
We also want to recognize @Huawei for years of work growing this ecosystem. We're looking forward to more days like this, and to reaching even more developers who care this much.
To every maintainer and developer who gave up a cold, sleety Saturday for open source: thank you.
What should we build together next? Tell us in the comments.
ALT A crowded lecture hall with attendees seated and focused on a presentation. Multiple screens display graphs and charts related to financial growth, labeled with "Friedberg Mercantile Group." The room is equipped with modern lighting and ceiling fixtures.
ALT A group of people is standing in front of a conference room holding a banner that reads, "PyTorch 社区贡献 Workshop." A screen in the background displays "Ascend for PyTorch Meetup." Bottles and nameplates are on the table in the foreground.
ALT A group of individuals is seated in a modern lecture hall, holding signs with various text. A long banner on the desk reads "Ascend for PyTorch Meetup." The setting is bright with overhead lighting, and the walls are a mix of gray and white.
ALT A group of people gathered indoors at a PyTorch Meetup event. They are holding various signs, including those with the PyTorch logo and others that read "LLM" and "Accelerate". The setting appears to be a casual dining or meeting space with tables and chairs.
🚀 Ready to explore what's next in open source AI? Join us at #PyTorchDayJapan 2026 on December 10 in Tokyo! Register today! bit.ly/4eYT0jb
5️⃣ reasons to attend:
- Explore AI innovations
- Learn from industry leaders
- Discover what's next in open source AI
- Get real-world insights
- Connect with the community
ALT Image promoting PyTorch Day Japan 2026. The text reads, "5 Reasons to Attend PyTorch Day Japan 2026. December 10 | Tokyo, Japan. Co-Hosted by Hugging Face, IBM, Mitsubishi Electric, PyTorch." Various people smiling and interacting at a conference are pictured.
What can you test on one GPU before a big PyTorch training run? PyTorch core maintainer @ezyang (@Meta) explains in this clip from #PyTorchCon North America 2026.
He describes how a fake process group lets developers stub out distributed communication and check for crashes and GPU out-of-memory errors. He also discusses using SPMD types to look for numerical bugs without running a full-scale distributed test.
In the full keynote, PyTorch core maintainer @albanDesmaison (@Meta) discusses hardware backend integration and testing. @ezyang also covers activation checkpointing with torch_remat and discusses FlexShard.
Watch the full keynote: youtu.be/EwZafo8QDDE?si=-Y2b…
At @AIatAMD @AMD AI DevDay in Austin @sparkycollier spoke to a room full of developers on how the PyTorch Foundation is powering training, inference, and agents as the open source center of gravity.
ExecuTorch WebGPU, PyTorch models on the GPU, in the browser.
Running a PyTorch model in the browser has usually meant converting it to another framework's format first. With ExecuTorch's new WebGPU backend, you export once with torch.export, and the same .pte file runs in WebGPU browsers such as Chrome and Safari. Everything runs locally in the tab, so prompts, photos and audio never leave the device.
The backend already runs Llama 3.2, Qwen3, Gemma 3, SmolVLM, SAM 2, Depth Anything V2, D-FINE, Real-ESRGAN and Whisper, and supports LoRA and DiReFT fine-tuning on-device.
By Julian Ng-Thow-Hing (@hing3074), Siddartha Pothapragada, Stephen Jia, Digant Desai (@digant23), and Bilgin Cagatay (@BilginCatay).
Read the post: pytorch.org/blog/executorch-…
Try the models in the ExecuTorch WebGPU Playground: executorch.ai/playground
With AI models scaling fast, seamless orchestration is key. Live from #KubeCon + #CloudNativeCon @robertnishihara announces AI & cloud-native ecosystem KubeRay joins @CloudNativeFdn as the Kubernetes-native lifecycle management, autoscaling, scheduling, reliability, and observability for Ray workloads. KubeRay is the standard way to run @raydistributed on Kubernetes & is the foundation of the managed Ray offerings from major cloud providers. Collaboration between CNCF and PyTorch Foundation is helping bridge the gap between large-scale distributed AI training/inference and robust Kubernetes infrastructure.
Great to see the open source community pushing inference performance forward on PyTorch:
A megakernel for Muse Spark Open on TPU v7 got 433 tok/s per user and up to 25.9K tok/s per chip on AgentX (1M context). All 62 layers of the 816B MoE run as one Pallas program on a single TPU v7 host, with SGLang on the TorchTPU PyTorch backend.
- 1.94x faster per token than our previous path
- DFlash2 speculative decoding: 2.0x per token
- fp8 KV, HiCache and session-aware eviction for long agent sessions
Read the full technical deep dive here: bit.ly/4zgrhBL
@radixark
Optimize kernels with a Python-based DSL and tile-centric primitives to automatically target Tensor Cores and other advanced GPU hardware features.
CUDA Tile can take direct PyTorch implementations that compute multiple terms separately, fuse them into one tiled kernel and keep that matrix in registers.
This blog shows how it enables you to write clear, readable, and highly performant kernels, including examples, docs, and sample code.
Read the full post:
bit.ly/3VigiKf
ALT A dynamic arrangement of green sticky notes displaying mathematical symbols, such as "f(x)" and geometric shapes. The notes are interspersed with rolled-up papers in varying shades of beige, green, and blue, set against a light background.
Welcome to our 2026 PyTorch Foundation Ambassadors! Today we feature:
Abraham Imani Bahati @imanibahati0 · Democratic Republic of Congo
Adesina Lekan Samuel @limitless__100 · Nigeria
Aditya Mehra · USA
Ali Gowani · USA
“I’m excited to contribute to the PyTorch community, learn alongside researchers and developers from around the world, and help make open-source AI more accessible to developers and learners across Africa.” - Abraham Imani Bahati
“I cannot wait to bridge the gap in AI access by democratizing PyTorch and its broader ecosystem for underrepresented students and researchers across all levels!” - Adesina Lekan Samuel
“I’m excited to help more engineers move PyTorch from experiments to reliable production systems, through hands-on tutorials, community events, and mentoring new contributors.” - Aditya Mehra
“PyTorch has fundamentally transformed how we teach, research, and deploy artificial intelligence. I’m excited to advocate for the ecosystem and drive hands-on adoption of open-source frameworks that keep AI innovation accessible, collaborative, and responsible.” - Ali Gowani
How does Shopify outperform frontier AI models at a fraction of the cost? During #PyTorchCon North America keynote, @cmazzaanthony of @ShopifyEng explained @Shopify's "Flywheel" a state-of-the-art continual learning loop built on top of @PyTorch and @vllm_project.
As enterprises scale generative AI, the bottleneck quickly shifts from building an initial model to maintaining and improving its performance in production. Shopify solves this by turning real-world failures into a self-improving engine.
Watch the full keynote video on YouTube to learn how this continuous learning loop operates in production: youtube.com/watch?v=2jOgIJRU…
This is your final opportunity to submit a session for #PyTorchDayChina taking place in Beijing on December 12. Bring your ideas and experiences to a community of PyTorch developers, machine learning engineers, AI researchers, and open source AI leaders from across China and beyond.
⏰ SUBMIT YOUR PROPOSAL by Tuesday, November 10, at 23:59 China Standard Time >> bit.ly/4imIAvV
"One thing that's special about the PyTorch Conference is that it really does focus on cutting-edge research and tools and trends that we see in AI."
Jana van Greunen, Director of Engineering, @Meta
At PyTorch Conference North America, our community gathered in person to accelerate open intelligence across the entire loop. From keynotes to technical deep dives, researchers, maintainers, and engineers came together to solve open bottlenecks and decide what comes next.
Read the full event recap: bit.ly/481GKe8
Catch the community highlights and hear directly from attendees: bit.ly/4zGi2vR
ALT The image shows an urban skyline at sunset with the PyTorch logo and text announcing "PyTorch Conference North America 2026." An American flag is visible on a building.