Hugging Face

Today, with over 100 industry partners, we introduced the NVIDIA Open Agent Safety Platform, bringing together OpenShell and Sentry. Artificial intelligence is extraordinary technology that will advance discovery, productivity, security, health, and prosperity for generations to come. But its full promise can only be realized when people have confidence that AI is being built to be safe and deployed with wisdom and responsibility.  NVIDIA Open Agent Safety Platform Reference Design combines NVIDIA OpenShell and NVIDIA Sentry. OpenShell is an open-source secure runtime that gives AI agents clear, enforceable boundaries. It traces their actions and enforces policy as they work. NVIDIA Sentry delivers added layer of security with hardware-based enforcement on NVIDIA BlueField, continuously monitoring agent activity through a trusted telemetry and detection pipeline and enabling millisecond-scale containment and quarantine. This is bigger than a single product. It's the beginning of an open ecosystem to build the trust layer for safe agent systems. Together, we are building the foundation of the AI economy. Trust and innovation are not in conflict. Safety is how trust is earned. We must build not only the most capable AI, but the most trusted AI, so that this extraordinary technology can realize its enormous promise for the world. https://nvda.ws/4hOkDx7

𝗫𝗶𝗮𝗼𝗺𝗶 𝗼𝗽𝗲𝗻-𝘀𝗼𝘂𝗿𝗰𝗲𝗱 𝟳𝟬𝟬𝟬+ 𝗥𝗟 𝗲𝗻𝘃𝗶𝗿𝗼𝗻𝗺𝗲𝗻𝘁𝘀. 𝗡𝗼𝘄 𝘆𝗼𝘂 𝗰𝗮𝗻 𝗲𝘅𝗽𝗹𝗼𝗿𝗲 𝘁𝗵𝗲𝗺 𝗮𝗻𝗱 𝗹𝗲𝗮𝗿𝗻 𝗵𝗼𝘄 𝗥𝗟 𝗲𝗻𝘃𝗶𝗿𝗼𝗻𝗺𝗲𝗻𝘁𝘀 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝘄𝗼𝗿𝗸.👀 The collection spans a surprisingly wide range of tasks, from: • Coding • Cybersecurity • Web development • General tasks • Music • And more But the interesting part is that you don't have to just read through the dataset. We built an 𝗥𝗟 𝗘𝗻𝘃𝗶𝗿𝗼𝗻𝗺𝗲𝗻𝘁 𝗘𝘅𝗽𝗹𝗼𝗿𝗲𝗿 where you can visualize the environments, understand how they're structured, pick a task, choose a model, and actually run a rollout to see what happens. Think of it as a more hands-on way to learn what an RL environment looks like under the hood. If you're new to RL, I'd especially recommend picking a few environments and playing around with them. You can learn a lot by seeing how the 𝘁𝗮𝘀𝗸, 𝘁𝗼𝗼𝗹𝘀, 𝗿𝗲𝘄𝗮𝗿𝗱𝘀, 𝗮𝗻𝗱 𝗲𝘃𝗮𝗹𝘂𝗮𝘁𝗶𝗼𝗻 fit together. #ReinforcementLearning #AI #MachineLearning #LLMs #OpenSourceAI #RL #GenerativeAI #HuggingFace

Voice agents still don’t understand who’s speaking to them. That’s a huge gap compared with humans, hidden by all the “phone-call” demos. But that changes today! NVIDIA is open-sourcing Nemotron 3 Diarization: a model that can reliably track speakers in live conversations, under a commercial-friendly license! In my tests, the quality is really good with one-second speech chunks. So we can use it for voice agents! I tested it with Reachy Mini and speech-to-speech running on a DGX Spark. It’s super fun to see the robot notice a new voice, ask for a name, and remember it. The model has day-zero integration with Transformers! Kudos to the NVIDIA team for shipping useful tools for the whole community!

Introducing FLUX 3 Action: An open weights 7B World Action Model that achieves first place on the RoboLab benchmark. It outperforms the previous best open model by 6.1 percentage points while using 56% fewer parameters and running up to 3.95x faster.⁠⁠ FLUX 3 Action removes the usual trade-off between world action model performance and VLA speed: it still predicts video and actions together, but plans more than twice as far ahead and runs faster per second of robot motion than the strongest open VLA. How it works: FLUX 3 Action takes recent camera frames, the system's current state, and a task description. It returns the next 32 actions and predicts how the scene will change. It observes, plans, acts, and adjusts, and it recovers from its own mistakes. The model builds on the same image, video, and audio pretraining as FLUX 3, in a smaller architecture designed for practical deployment. Our Self-Flow research made the smaller size possible, and in midtraining we taught it to predict actions and future frames together. Teams can fine-tune it on their own demonstrations to create policies for a specific robot and task. Together with NVIDIA, we integrated FLUX 3 Action natively into Hugging Face's LeRobot, with fine-tuning recipes included and edge deployment on NVIDIA Jetson. Beyond robotics, we’re also seeing promising results training task-specific policies for simulated environments like games, vehicle control, and computer use, and anywhere else a model needs to understand a visual environment and then decide what to do next. We're releasing the weights, code, fine-tuning recipe, benchmarks, and reproducible examples so researchers and developers can build on the model with their own robots, environments, and tasks.⁠⁠ Read the full blog: https://lnkd.in/eaqHUi9W Download the weights: https://lnkd.in/efwS4EqN Talk to our robotics team: https://bfl.ai/contact

We're proud to sponsor Open Together on Friday, October 16, where Hugging Face is kicking off Open Source AI Week at The Midway. The evening will be split into two parts: 🎉 6:00 PM – 9:00 PM: 36 live community demos, food, drinks, and time to connect with open-source builders. 🪩 9:00 PM – Midnight: Full dance floor with live DJ sets. Doors open at 6 PM, and the first 500 people through the door get collectible Hugging Face swag. 🤗 RSVP here: luma.com/opentogether