PyTorch

👀 Take a first look at the future of AI. The #PyTorchConf schedule is now live: hubs.la/Q03yGz5z0 Join the PyTorch community on October 22–23 in San Francisco for two days of practical sessions, real-world use cases, and insights from the engineers and researchers shaping…

#ML minds unite. #PyTorchConf 2025 is coming to SF Oct 22-23 and the most exciting part of this summer? 🤔 The schedule is dropping soon! 👀 Don’t wait. ⏳ Register by Aug 6 to save $400 and lock in your spot for the most practical, #OpenSource #GenAI event of the year. 🎟️…

Opacus now supports FSDP2 with Fast Gradient Clipping and Ghost Clipping, enabling scalable, private training of large models like Llama-3 8B. This unlocks 2D/4D parallelism and overcomes DP-DDP memory limits. 🔗 hubs.la/Q03w5vnt0 #PyTorch #Opacus #MachineLearning

👀 ICYMI: The #PyTorchConf schedule is now live! hubs.la/Q03yNxFq0 We’re talking 2️⃣ days of cutting-edge talks on #LLM scaling, real-time inference, model optimization & more — straight from the #PyTorch community. 📍 Oct 22–23 | San Francisco 🎟️ Register:…

SmolLM3-3B-8da4w: With #TorchAO & optimum-executorch, quantizing and exporting for mobile is a breeze. Now ready for on-device deployment with #ExecuTorch, running at 15 tokens/sec on Galaxy S22. 🔗 Model card with recipes + checkpoints: hubs.la/Q03yGyTN0 #EdgeAI #PyTorch

🌉 San Francisco becomes the global hub for #OpenSource #AI this October. Join us for #OpenSourceAIWeek from Oct 18–26. Connect with top minds, cutting-edge tools, and real-world use cases in #GenAI & #ML. 📍 hubs.la/Q03yvLnR0

👀 The wait is over The #PyTorchConf schedule is LIVE 🗓️ hubs.la/Q03yB-jN0 Join us Oct 22–23 in San Francisco for 2 days of #AI innovation, real-world insights, and breakthrough tools from the #PyTorch community. 🎟️ Register: hubs.la/Q03yB-rf0

ICYMI: @CRN featured PyTorch in its list of “The 10 Hottest Data Science And Machine Learning Tools Of 2025 (So Far).” A reflection of PyTorch’s role in accelerating open source AI: 🔗 hubs.la/Q03xX-YY0 🖊️ Rick Whiting #PyTorch #OpenSourceAI #MachineLearning #DataScience

Explore wheel variant support in #PyTorch 2.8 and what it means for future installs. Hear from experts at @Meta, @NVIDIA, @astral_sh, and @quansightai at our upcoming live Q&A on PyTorch packaging. 🔗 Register: hubs.la/Q03xV-Mz0 #OpenSourceAI #Python

How can you maximize performance with torch.compile when working with Diffusers across different use cases? This blog shows how torch.compile can deliver significant speedups—even when using offloading and LoRAs. 🔗 Read here: hubs.la/Q03xKQTq0 From: @RisingSayak…

From @nvidia: new cuda.cccl library introduces Pythonic building blocks for CUDA kernel fusion, making it easier to implement or extend libraries like PyTorch and compose efficient GPU algorithms using PyTorch tensors. 🔗 hubs.la/Q03xxY_g0 #PyTorch #CUDA #Python…

Discover how #verl simplifies #ReinforcementLearning for advanced #LLM reasoning and tool use in our Aug 6 Expert Exchange with Haibin Lin (ByteDance). Supports PPO/GRPO/DAPO, async rollout, expert parallelism for MoE, and more. #PyTorch #OpenSourceAI 🔗 hubs.la/Q03xkQW-0

Less than 3 weeks left to submit your #PyTorchConf poster! 🪧 This is your moment to spotlight research, tools, or innovative uses of #PyTorch in a high-energy, interactive format. 🗓️ Event: Oct 22–23, San Francisco ✒️ #CFP closes Aug 1 🔗 hubs.ly/Q03wXdpH0

Excited to be at #ICML2025, presenting a poster with the @PyTorch TorchAO team at the @icmlconf CodeML workshop! 🧠 "TorchAO: PyTorch-Native Training-to-Serving Model Optimization" 📅 July 18 | 2:15–3:00 PM 📍 West Meeting Room 211–214 🔗 Session: icml.cc/virtual/2025/w… Catch…

At the #WeAreDevelopers World Congress, Executive Director @matthew_d_white spoke on the future of #OpenSourceAI, exploring the shift from open weights models like #Llama to permissive models like DeepSeek‑r2. He also joined a panel on how foundation models like Gemma and…

New in the #PyTorch Compiler Series: Making Flux Run Fast Core developer Joel Schlosser walks through how torch.compile, torch.export, and quantization via torchao brought #transformers-based inference to under 0.5s. 📺 hubs.la/Q03wv8zT0 #AIperformance

Open Source AI Week is coming to the San Francisco Bay Area, October 18–26, 2025. This week-long celebration brings together conferences, hackathons, showcases, and networking events focused on the intersection of AI, ML, and open source: with the PyTorch Conference as the…

verl: Flexible and Scalable Reinforcement Learning Library for LLM Reasoning and Tool-Calling verl is a flexible, efficient framework for building RL pipelines for LLMs, supporting PPO/GRPO/DAPO and scaling to MoE models like DeepSeek. It tackles new reasoning challenges with…

A convolutional neural network built with PyTorch is supporting marine conservation efforts by detecting ghost nets in sonar scans with 94% accuracy. Trained and deployed on Azure using NVIDIA A100 GPUs, the model powers GhostNetZero ai. 🔗 Read @NVIDIA's blog to learn more:…