Jan Margeta

Introducing MONAI Physio at #MICCAI2026. 3D and 4D medical images ➡️ personalized cardiac and respiratory digital twins. This open-source Project MONAI toolkit lets researchers model how the heart beats and lungs move for simulation, visualization, and reproducible research. Explore the project 👉 nvda.ws/3VLmJFq

🤯 A localmaxxer hit ~381 tok/s on a SINGLE RTX 3090 with Qwen3.8-27B. This developer has turned a 24GB RTX 3090 into a monster Qwen inference box - w/some creativity. Four days ago 👉 ⚡ ~82 tok/s single-user Then 👉 ⚡ ~114 tok/s with optimized MTP ⚡ ~138 tok/s with DFlash2 + lookup drafting Now 👉 🔥 ~381 tok/s on ONE request How? The recipe combines ... 🧠 Qwen3.8-27B 🎮 1× RTX 3090 24GB @ 250W ⚙️ heavily optimized vLLM ⚡ DFlash2 speculative decoding 🔎 lookup-augmented drafting 📚 prefix caching 🧮 16-token verification blocks 💾 quantized KV / heads / activations DFlash2 normally proposes 7 tokens. The developer realized the verification block doesn't have to stop there. If Qwen is answering from a document already sitting in the prompt, the system can fill the remaining draft positions using tokens found directly in that context. 🎯 So the target model can verify 16 tokens at once. On a ~25K-token document reproduction task: Previous DFlash2 👉 ~260 tok/s Longer verification + context lookup 👉 🔥 ~382 tok/s Acceptance: 🤯 15 of 16 tokens per verification step That is where the crazy number comes from. ⚠️ On ordinary real-world chat prompts, the same setup is around ~133 tok/s Still extremely fast for a dense 27B model on an RTX 3090. The 381 tok/s mode shines when the answer largely comes from material already in context so these are best use cases 📚 RAG / document Q&A 💻 Coding assistants applying edits 📝 Quoting or rewriting documents 🔎 Extracting information from long prompts And another optimization 👉 With prefix caching, a second question against the same 25K-token document reportedly goes from: 🐌 22.4 sec TTFT → ⚡ 0.56 sec TTFT Because the model doesn't need to process the whole document again. 🎯 It's specifically a mode for RAG front ends and coding agents. Follow iamMess on Reddit or syv-ai on GitHub 🔗 Reddit: r/LocalLLaMA/comments/1vtup5s/ 🔗 GitHub: /syv-ai/qwen38-27b-rtx3090

Excited to announce MC-Calib v1.5.0! By upgrading to OpenCV 5, reprojection errors are almost halved across all tested scenarios. Shoutout to @LearnOpenCV for encouraging the upgrade. 🔗 github.org/rameau-fr/MC-Cali…

🤖🚀 𝗜𝗻𝘁𝗿𝗼𝗱𝘂𝗰𝗶𝗻𝗴 '𝗹𝗲𝗿𝗼𝗯𝗼𝘁-𝗿𝗼𝗹𝗹𝗼𝘂𝘁' — 𝗼𝗻𝗲 𝗖𝗟𝗜 𝘁𝗼 𝗱𝗲𝗽𝗹𝗼𝘆 𝗮𝗻𝘆 𝘁𝗿𝗮𝗶𝗻𝗲𝗱 𝗽𝗼𝗹𝗶𝗰𝘆 𝗼𝗻 𝗮𝗻𝘆 𝗿𝗲𝗮𝗹 𝗿𝗼𝗯𝗼𝘁. Until today, running a trained policy on a real robot meant repurposing 'lerobot-record' — the same script you used to collect data. Need a different inference engine? That was yet another script. 'lerobot-rollout' replaces all of that with one command: pick a rollout strategy, pick an inference engine, and go. 🔵 Base — Run your policy. See if it works. No recording, no setup friction. 🟢 Sentry — Leave your robot running overnight. Episodes auto-rotate, data streams to the 🤗 Hub in the background. Come back to a ready-made dataset. 🟠 Highlight — A ring buffer silently records the last N seconds. See something interesting? Press a key and that moment is saved retroactively. No more "I wish I was recording." 🟣 DAgger — When the policy fails, pause it, take over with the teleoperator, record the correction, and resume. Every intervention is tagged so you retrain on exactly the moments that matter. All four strategies work with both synchronous and Real-Time Chunking (RTC) inference, and you can even plug your own! Every combination of strategy × inference engine just works.

Codex is SOTA at CAD

Announcing Amazon S3 Files. The first and only cloud object store with fully-featured, high-performance file system access. Learn more here. go.aws/4tw17Zg

🚀 VIBOT 2026–2027 | Applications opening soon Read more: linkedin.com/posts/davidfofi… #AI #ComputerVision #Robotics #Master

We’re hiring research interns at @NVIDIA 🚀 You’ll work on core generative models, including autoregressive and diffusion, and help advance NVIDIA Nemotron. Interns are encouraged to publish in top-tier venues like ICLR, ICML, and NeurIPS. Apply or connect with us: forms.gle/8TbXoft7pNWd3Mav9 #GenAI #Nemotron #LLMs #Internships

it's crazy what a 1.5B model can do these days! "VibeThinker-1.5B is a 1.5-billion parameter dense language model. With a total training cost of only $7,800 USD, it achieves reasoning performance comparable to larger models like GPT OSS-20B Medium." runs perfectly on device!

supervision-0.27.0 lets you parse and visualize Qwen3-VL object detection results prompt: person between albert and marie. to answer this, Qwen3 needs prior visual knowledge of Marie Curie and Albert Einstein, and it needs to understand reference terms like between.

Ten years (!) after submitting the grant, NextBrain is published in @Nature . NextBrain is a brain atlas built from 3D histology that enables segmentation of 3D brain scans (MRI, Hip-CT, even photos!) into hundreds of regions. Article, data, code, videos, & more in this 🧵 (1/3)