🚀A Simple Baseline for Streaming Video Understanding🚀 #SimpleStream reveals... a simple baseline of *recent N-frame sliding window* beats SOTA memory-based methods on standard streaming video benchmarks lmms-lab - Project: - Code:show more

Announcing Bespoke-MiniChart-7B, a new SOTA in chart understanding for... models of comparable size on seven benchmarks, on par with Gemini-1.5-Pro and Claude-3.5! 🚀 Beyond its real-world applications, chart understanding is a good challenging problem for VLMs, since it requires both mathematical as well as visual reasoning. 1/n🧵show more

Want to create a consistent AI influencer vlog? Stop... wasting time on random motions and awkward transitions. Use this simple workflow: ✨ AI Model → First Frame → Guide Image → Prompt → Video Create smooth, consistent videos in minutes. 🚀 Try it FREE:show more

(1/n) 🚀 With FastVideo, you can now generate a... 5-second video in 5 seconds on a single H200 GPU! Introducing FastWan series, a family of fast video generation models trained via a new recipe we term as “sparse distillation”, to speed up video denoising time by 70X! 🖥️ Live demo: (Thanks to @gmicloud for the support!) 🔗 Blog: 🔓 We fully open-source our models, code, and data with Apache-2.0 licensesshow more

Kling 2.6 Motion Capture is so good. It's a... huge leap forward. We've been able to do motion capture with AI for a while, but the quality bump finally made this approach useful. Here are the steps: - Get a reference video - record it or get a stock video - Grab the start frame of that video - Edit it using Nano Banana Pro. Ask to replace character and background. - Select the Kling 2.6 Motion Capture model in the video generator on Freepik (now Magnific) - Upload reference video + set the edited start frame - Video prompt can be simple e.g. "Gandalf dancing"show more

Do you use Runway Gen-2 to make AI videos?... This is for you! I built a simple tool to help "extend" Runway videos beyond the 4 second clip limit. Just choose a video, and the tool will give you the final frame -- so that you can drop this back into Runway, and generate another AI video based on it. Let me know if this is useful to you! Link is at bottom of the video and it's live to try right now.show more

🎥 Video generation is hitting the memory wall. As... videos get longer, the KV cache quietly explodes — and long-horizon consistency starts to break. We built Quant VideoGen: a training-free KV cache compression method for auto-regressive video diffusion. Instead of storing every KV in high precision, QVG exploits video’s spatiotemporal redundancy with semantic-aware smoothing + progressive residual quantization. 🚀 Up to 7× KV memory reduction ⚡ <4% overhead ✅ Strong long-video quality 🕹️ Deploy HYWorldPlay on your own RTX 5090 locally KV compression is becoming a core scaling primitive — not just for LLMs, but for video generation too. Paper: Code: (1/5)show more

🚀New paper out - We present Video-MSG (Multimodal Sketch... Guidance), a novel planning-based training-free guidance method for T2V models, improving control of spatial layout and object trajectories. 🔧 Key idea: • Generate a Video Sketch — a spatio-temporal plan with background, foreground, and motion in the pixel space. • Encode this structure directly into the latent space of the diffusion model during generation, which does not require fine-tuning or additional memory during inference. 🧵show more

Mesmerizing 45-degree infinity module of the GA3D project in... action. The GA3D Project is a “side-project” driven by a passion for the RepRap 3D printer theme, aiming to extend the RepRap/DIY (Do-It-Yourself) spirit to other processes and fields. The basic machine consists of a frame with a fixed table, and a 3-axis Cartesian robotic arm with CoreXY kinematics crossed in the XY plane. A very simple tool changer is located at both ends of the robotic arm. This makes it a modular machine that can be easily converted into any machine. Video Credit: Alain D. Geiser #3dprinting #3dprinter #additivetechnology #additivemanufacturing #engineering #technologyshow more

DimensionX: Create Any 3D and 4D Scenes from a... Single Image with Controllable Video Diffusion TL;DR: Create 3/4DGS from Video Diffusion Note: Some first inference code released (not all yet). Contributions (cited): • We present DimensionX, a novel framework for generating photorealistic 3D and 4D scenes from only a single image using controllable video diffusion. • We propose ST-Director, which decouples the spatial and temporal priors in video diffusion models by learning (spatial and temporal) dimension-aware modules with our curated datasets. We further enhance the hybriddimension control with a training-free composition approach according to the essence of video diffusion denoising process. • To bridge the gap between video diffusion and real-world scenes, we design a trajectory-aware mechanism for 3D generation and an identity-preserving denoising approach for 4D generation, enabling more realistic and controllable scene synthesis. • Extensive experiments manifest that our DimensionX delivers superior performance in video, 3D, and 4D generation compared with baseline methods.show more

Introduce EAGLE, a new method for fast LLM decoding... based on compression: - 3x🚀than vanilla - 2x🚀 than Lookahead (on its benchmark) - 1.6x🚀 than Medusa (on its benchmark) - provably maintains text distribution - trainable (in 1~2 days) and testable on RTX 3090s Playground: Blog: Code: ⚒️First Principle: Compression! Yi Ma We find that the sequence of second-top-layer features is compressible, making the prediction of subsequent feature vectors from previous ones easy by a small model. 🙏Acknowledge: This project is greatly inspired by the Medusa team (Tianle Cai ✈️ ICML🇰🇷 @yli3521 Zhengyang Geng ✈️ ICML Hongwu Peng Tri Dao), the Lookahead team (Hao Zhang LMSYS Org), and others. Joint work with Yuhui Li and Chao Zhangshow more

here's how the whole thing works. claude code doesn't... care what's behind the API. it just sends requests and expects responses. so i pointed it at my own machine instead of anthropic's servers. llama-server runs the model locally. LiteLLM sits in between and translates the API format. claude code thinks it's talking to claude. it's talking to qwen on localhost. the setup: 2x 3090s, 38 layers on GPU, 10 on CPU. 128K context window. generation is only 7 tok/s but the tradeoff is worth it. 128K means the agent can hold an entire project in memory without losing context midtask. claude code alone loads a 17.5K token system prompt on every request. tool definitions, safety rules, agent behavior. that's your baseline before you even say hello. pushed as far as i could tonight. what surprised me most wasn't the speed. it was the iteration quality. first prompt gave me a working particle sim. second prompt, the model read its own 564 lines, understood the architecture, and added trails, explosions, gravity wells, bloom effects. no handholding. 4bit quantized. 45GB on two consumer cards. running a full coding agent autonomously. detailed article coming. full benchmarks, hardware breakdowns, engine debugging, code quality. everything from setup to what broke and why.show more

This happened at a public school in Southern California,... students streaming to the altar to give their lives to Jesus! 😭 Watch it for yourself… Yesterday, I posted a full video sharing the vision to reach every school possible in 2026. It would be an honor to have you partner with us. How? Simple. We are believing for another 100 monthly financial supporters as we preach the gospel, disciple them and give students bibles on campus (yes it’s true) Would you partner with us on this initiative to save our generation?show more

You can't 3D reconstruct glass from images... ...WRONG! Thanks... for video diffusion, now just about anything is possible! Introducing...Diffusion Knows Transparency (DKT) Transparent and reflective objects usually break robot vision and photogrammetry pipelines because they don't follow the "solid object" rules standard cameras expect. DKT is a new AI model that repurposes the "internal physics engine" found in video generation models to solve this problem. Researchers took a massive video diffusion model (WAN) and fine-tuned it using a custom-built synthetic dataset to turn it into a high-precision depth sensor. To train the AI, they built the first massive synthetic video library of transparent objects, 1.32 million frames of perfectly labeled glass and metal objects in motion. Without ever seeing a "real" labeled video of glass during training, the model (DKT) outperformed all previous specialized systems on real-world benchmarks (ClearPose, DREDS). They created a "lightweight" 1.3B parameter version that runs fast enough (0.17s per frame) to be used on actual robot hardware. Two reasons I find this project important: 1. It further proves that synthetic data will be essential for training the next generation vision models. 2. In real-world robotic tests, using DKT's depth maps nearly doubled the success rate of robot arms trying to pick up objects on tricky reflective or translucent surfaces. At home robots will need to interact with these types of objects on a daily basis. Check out the project page here: Code is LIVE! #Computervision #Robotics #AIshow more

Excited that our paper StreamdiffusionV2 received the Best Research... Paper Award at #MLSys26! 🚀Video generation is quickly moving from demos to production-facing workloads. It is no longer a turn-based pipeline but should be a streaming pipeline to interact with users. 📖Our project page: and paper: 👂Come join the talk if you are interested in streaming video generation. Our talk will be at the Research Track Oral Presentation: Best Paper Session on Tue 8:45AM at #MLSys26 , I will talk about how we attacked the efficiency and quality challenges. Hope to see you there! ❤️Huge thanks to all authors! This work would not have been possible without the incredible effort from the entire team. Big shout out to Tianrui Feng, Zhi Li, Shuo Yang , Haocheng Xi, Muyang Li , lllyasviel , Xiuyu Li , Keting Yang, Kelly Peng, Song Han , Maneesh Agrawala, Kurt Keutzer , and あき先生(Aki)show more

World modeling and imitation learning have largely been considered... two disparate worlds. In our recent work, Unified World Models, just accepted to #RSS2025, Chuning Zhu provides a dead-simple unifying solution: just train a joint diffusion model over actions and future states, but with *decoupled* diffusion time steps across these modalities. Manipulating these decoupled time steps then allows for marginalization or conditioning on actions or states; a single model can serve as a policy, forward dynamics model, video prediction model, or inverse dynamics model by simply setting diffusion timesteps carefully. The resulting model can leverage video datasets along with robot training data much more effectively, and shows improved robustness, generalization, and flexibility. This is exciting because it is frustratingly simple, scalable, and shows strong improvement on real-world robotics problems. Please refer to Chuning Zhu 's excellent thread for more details! More details/code can be found on our website and in the paper -show more

🚀 Welcome to Luxxcoin - Where Meme Culture Becomes... Competition Luxxcoin ($LUX) is the first MemeFi project transforming memes into a real, rules-based on-chain competition. Every 60 days, the Hero Rewards engine ranks wallets through a transparent scoring model and automatically distributes $LUX to the top performers - all visible on a live leaderboard. Built on Solana. Audited. Multi-sig secured. Designed for long-term gameplay, not hype cycles. With a 7.5-year competitive structure, 45 reward rounds, and deterministic scoring, Luxxcoin introduces a new standard in meme culture: 🔸 Hold with strategy 🔸 Climb the leaderboard 🔸 Earn like a hero If you're here for the future of memes - competitive, transparent, and on-chain - welcome to the arena. 👑 The Meme Competition Has Begun. Join Luxx. Website: Community chat: #LuxxUp #MemeFi $LUXshow more

𝐀𝐍𝐎𝐍𝐆 𝐏𝐑𝐎𝐁𝐋𝐄𝐌𝐀 𝐌𝐎 𝐄𝐇 𝐌𝐀𝐇𝐀𝐋 𝐃𝐈𝐍 𝐊𝐈𝐓𝐀 As we... enter a 𝗻𝗲𝘄 𝗲𝗿𝗮 of 𝙔𝙤𝙪’𝙧𝙚 𝙈𝙮 𝙁𝙖𝙫𝙤𝙧𝙞𝙩𝙚 𝙎𝙤𝙣𝙜, we just want to sincerely thank everyone for the love and support you’ve shown. Thank you for streaming the 𝙏𝙚𝙠𝙖 𝙇𝙖𝙣𝙜 music video and helping it reach 𝟭.𝟳𝗠 𝘃𝗶𝗲𝘄𝘀 𝗮𝗻𝗱 𝗰𝗼𝘂𝗻𝘁𝗶𝗻𝗴. We truly appreciate every single one of you who took the time to watch, stream, and support the project. As a small way of giving back, here’s our little 𝙖𝙮𝙪𝙙𝙖 for all of you, a 𝗧𝗶𝗸𝗧𝗼𝗸 𝘃𝗶𝗱𝗲𝗼 𝗳𝗲𝗮𝘁𝘂𝗿𝗶𝗻𝗴 𝗔𝗻𝗴𝗲𝗹. We hope you enjoy it! 🥂 The question is, 𝙗𝙖𝙠𝙞𝙩 𝙠𝙖 𝙣𝙖𝙜𝙠𝙖𝙠𝙖𝙜𝙖𝙣𝙞𝙮𝙖𝙣? 𝙔𝙤𝙪’𝙧𝙚 𝙈𝙮 𝙁𝙖𝙫𝙤𝙧𝙞𝙩𝙚 𝙎𝙤𝙣𝙜, this 𝐉𝐮𝐧𝐞 on 𝐆𝐌𝐀.show more

🚀 Excited to share our #ICLR2025 work on planning... with neural dynamics models! While our lab has developed diverse neural dynamics models for manipulating rigid, deformable, and granular objects, having the model alone doesn’t solve the problem—planning with it remains a challenge. 💡 Enter BaB-ND, led by Keyi and Jiangwei! We propose a scalable, GPU-accelerated branch-and-bound algorithm, inspired by neural network verification, to enable effective planning for diverse objects modeled with neural dynamics. 🔗 Project page (open-source + detailed docs!): 🎥 Watch the video to see T being pushed around obstacles, and check out Keyi’s thread for more details!show more

Streaming iPhone data in real-time directly to Rerun 🚀... The collection process is one of the most frustrating parts of building imitation-learning datasets. I’ve got a little army of sensors—📱 iPhone, iPad, Quest 3—but getting them temporally aligned, spatially aligned, AND seeing real-time feedback while recording is tough. I stumbled on a great library from Cake Lab (WPI) called ARFlow. It’s a thin client built on Unity’s ARFoundation that connects over gRPC to a server running Rerun for live data logging. I forked it to: - Log the SLAM translation poses, and - Upgrade rerun to v0.23 for my use case. So far, it works well, but there are still a few hitches: 1. Right now, it’s solid on iPhone and iPad; my Quest 3 client is still slow and not super reliable. 2. I’m using an older ARFlow branch focused on real-time streaming only—no spatial or temporal sync yet. Unity builds for iOS keep failing. 🛠️ 3. Nothing is saved locally to the client, so packet loss is a risk on shaky networks. There’s huge potential in tapping the ubiquitous sensors we carry around every day, and ARFlow is a big step toward making that easyshow more