Lightning AI

Lightning AI sent me some swag! 🌩️⚡😍

Fresh from a week at #GTC2025, Luca Antiga and Thomas Viehmann breakdown what they learned at GTC, why Thunder is even more relevant with Blackwell, and a new high level API for getting started quickly. Check out the full episode here ➡️ https://lnkd.in/e-8_gH3N

We’re going deep on Robotics this April. If you’re in NYC, the Bay Area, or London and want to feature your work, shoot us a DM—meetups in each hub are happening at the end of the month. ⚡

🚀 𝐄𝐱𝐜𝐢𝐭𝐢𝐧𝐠 𝐔𝐩𝐝𝐚𝐭𝐞 : 𝐒𝐭𝐫𝐞𝐚𝐦 𝐇𝐮𝐠𝐠𝐢𝐧𝐠 𝐅𝐚𝐜𝐞 𝐃𝐚𝐭𝐚𝐬𝐞𝐭𝐬 𝐰𝐢𝐭𝐡 𝐋𝐢𝐭𝐃𝐚𝐭𝐚 Now you can 𝐬𝐭𝐫𝐞𝐚𝐦 𝐬𝐭𝐫𝐞𝐚𝐦 𝐦𝐚𝐬𝐬𝐢𝐯𝐞 𝐏𝐚𝐫𝐪𝐮𝐞𝐭 𝐝𝐚𝐭𝐚𝐬𝐞𝐭𝐬 directly from Hugging Face using LitData—all with just a few lines of code. 🎉 A new 𝐏𝐑 just landed this week—check it out for full details: 🔗 𝐏𝐑 #505: https://lnkd.in/g55TKgbQ Here’s what makes it awesome: 🔥 ✅ Handles Huge Datasets Seamlessly – Stream datasets like fineweb-edu (10BT or 100BT) effortlessly. ✅ Shuffled Mode for Flexibility – Easily randomize your data for better training. ✅ Efficient Downloads – Fetch only the chunks you need, saving time and resources. ✅ Low Memory Mode – No need to load entire Parquet files into memory—keep things light! 💡 Pro Tip: A faster internet connection helps when downloading larger chunks—stream smarter! ⚡ 📌 𝐆𝐞𝐭 𝐈𝐧𝐯𝐨𝐥𝐯𝐞𝐝 🔗 Explore the LitData repo: https://lnkd.in/gK9ZKnTw 🙌 𝐒𝐩𝐞𝐜𝐢𝐚𝐥 𝐓𝐡𝐚𝐧𝐤𝐬 A huge shoutout to Thomas Chaton for all the support and to Deependu Jha for laying the foundation for this feature. 🚀 I’d love to hear your thoughts! 💬 Got ideas to make this even better? Drop your suggestions in the comments—I’m all ears! 👂 If you find this useful, please ⭐ star the repo to support the project. It truly means a lot to me and the team! And if you have any questions, feel free to reach out—I’d be happy to chat. 😊 #DataScience #MachineLearning #HuggingFace #LitData #BigData #AI #Tech #OpenSource Hugging Face Lightning AI

Lightning’s Studios now offers reserve-ahead for A100s, H100s, and H200s for 50-80% off! ✅ Reserve by day up to a month ✅ Self-serve with pay-up-front billing ✅ Reservation prices up to 80% off Reserve yours ➡️ https://lnkd.in/d2gb397r

Deploy a background removal model. 🚀 Created by community member Sitam Meur, this studio uses BEN2, a deep learning model for background removal and foreground extraction, utilizing Confidence Guided Matting (CGM) for precise cut-out effects. You can duplicate the studio (link in comments) and try out this self-hosted and private API with LitServe.⚡

Lightning has partnered with NVIDIA for Startups - receive 80% off our enterprise Tier if you're part of the inception program! Lightning provides ALL tools to build AI products in one platform. Cloud GPUs, DevBoxes, train, deploy, and more with zero setup. https://lnkd.in/dPJuHYbR

GTC was a blast! 🚀 A huge thank you to everyone who attended our speaking sessions and stopped by our booth for live demos. Your support and curiosity made this event unforgettable. We had an action packed week of talks, meetings, and events, and we’re grateful to have engaged with so many amazing individuals. Looking forward to what’s next! ⚡

Another day, another breakthrough in computational biology! 🧬 Segger, a graph neural network that redefines transcript-to-cell assignment in spatial transcriptomics, solves common issues of cell segmentation. Faster, more accurate, & scalable—Segger integrates seamlessly into workflows. And it’s trained on PyTorch Lightning ⚡ Learn more about Segger here: https://lnkd.in/emsrD8Hf 🚀

Deploy an API template with Microsoft's Phi-4 Multimodal. 🚀 At just 5.6B params, this compact model for speech, vision and text is designed to push the boundaries of multimodal research. Self-hosted, lightning-fast & OpenAI-compatible— deploy on Lightning AI Hub! ⚡ Check it out here: https://lnkd.in/eE7-WhpB