Video wird geladen...

Video konnte nicht geladen werden

Beim Laden dieses Videos ist ein Problem aufgetreten. Dies könnte an einem vorübergehenden Netzwerkproblem liegen oder das Video ist möglicherweise nicht verfügbar.

Long video generation is a systems problem. Introducing LongLive-2.0 from NVIDIA Research: an end-to-end NVFP4 training and inference system for long video generation. Low-precision deployment often relies on post-training quantization, creating a gap between how models are trained and how they run. LongLive-2.0 aligns NVFP4-aware training, distillation, and W4A4... show more

NVIDIA AI

343,167 subscribers

60,715 Aufrufe • vor 5 Monaten •via X (Twitter)

Bildung Wissenschaft & Technologie

Anya Rossi• Live Now

Private livecam show

24 Kommentare

Profilbild von NVIDIA AI

NVIDIA AIvor 5 Monaten

A closer look at LongLive-2.0: • Generates long video at 720p • Preserves subject and background consistency across multi-shot sequences • Able to switch prompts at chunk boundaries

Profilbild von NVIDIA AI

NVIDIA AIvor 5 Monaten

You can find the full project including paper, code, models, and demos here:

Profilbild von Ahmad

Ahmadvor 5 Monaten

Niceee, new toy for me to play with on the GPUs over the weekend :D

Profilbild von mr_r0b0t

mr_r0b0tvor 5 Monaten

@SamJWasserman don’t miss this 👀

Profilbild von micmann.eth

micmann.ethvor 5 Monaten

This is the part most people miss with video AI: the bottleneck is not just model quality, it is inference cost and memory. NVFP4-aware training matters if it makes longer generation practical in real production workflows.

Profilbild von Experiencer Lens

Experiencer Lensvor 5 Monaten

@grok explain this to me. Is this a new AI model?

Profilbild von Beginnersblog

Beginnersblogvor 5 Monaten

How long can I make a video?

Profilbild von Markets & Mayhem

Markets & Mayhemvor 5 Monaten

Very cool.

Profilbild von Paul Lu

Paul Luvor 5 Monaten

Useful framing long video is now a full-stack inference problem. The hard product layer is keeping control latency and review stable as context grows not just generating a nicer first clip.

Profilbild von James Clawn

James Clawnvor 5 Monaten

What happens to low-precision deployment once the workflow leaves the happy case? That is usually where small assumptions start compounding.

Profilbild von Neha Singh

Neha Singhvor 5 Monaten

Aligned NVFP4 Long Video

Profilbild von Aero - AI & Crypto

Aero - AI & Cryptovor 5 Monaten

Image Generation already is agentic. We need video to become agentic. That's a system.

Profilbild von Subhan Tariq

Subhan Tariqvor 5 Monaten

the AI industry really went from “can it generate 3 seconds” to optimizing end-to-end infrastructure for movie-length outputs in like 14 months

Profilbild von Shara

Sharavor 5 Monaten

Long video generation as a systems problem with LongLive-2.0 NVFP4 training is the infrastructure story

Profilbild von Naveen

Naveenvor 5 Monaten

Love how you're tackling long video generation as a systems problem, NVIDIA Research is crushing it with LongLive-2.0

Profilbild von Alberto Vélez

Alberto Vélezvor 5 Monaten

slop

Profilbild von Sumit Kumar

Sumit Kumarvor 5 Monaten

“Long video generation is a systems problem” is such an important framing. A lot of progress now is less about bigger models and more about optimizing the entire training → inference pipeline efficiently.

Profilbild von Subhan Tariq

Subhan Tariqvor 5 Monaten

“long video generation is a systems problem” is researcher code for “the GPUs were fighting for their lives”

Profilbild von Geraldo Pena

Geraldo Penavor 5 Monaten

👏🏻

Profilbild von Alice The Ai Expert

Alice The Ai Expertvor 5 Monaten

Smart systems-level improvement

Profilbild von Shivam Kumar

Shivam Kumarvor 5 Monaten

Benchmark of LTX 2.3 Video with native audio generation model in DGX Spark System

Profilbild von DrSadek 🪶

DrSadek 🪶vor 5 Monaten

did you guys vibe code this website?

Profilbild von Inflectiv AI ⧉

Inflectiv AI ⧉vor 5 Monaten

This advancement from NVIDIA Research solves key challenges in creating longer AI videos. It aligns training and inference in low-precision formats to achieve better speed and memory efficiency.

Profilbild von Armin Catovic

Armin Catovicvor 5 Monaten

What’s a “long video”? 10 seconds? 30 seconds? Minutes?

Ähnliche Videos

🚀 Excited to release LongLive 2.0! 🎬 An end-to-end infrastructure for long video generation, with FP4 and parallelism at the core of both training and inference. ⚡45.7 FPS generation speed on 5B model⚡ ✨ LongLive 2.0 supports real-video training, few-step distillation, multi-shot training/inference, sequence-parallel acceleration, NVFP4 KV cache, and async VAE decoding deployment. 🧩 To our knowledge, this is the first open-source 4-bit long video generation infra that covers both training and inference. 🙌 Welcome to check it out, try it, and share feedback! 🔗 Code: 📰 Paper: 🎥 Demo: #LongVideoGeneration #VideoGeneration #Realtime #AIInfra #EfficientAI #FP4 #Parallel #NVIDIA

🚀 Excited to release LongLive 2.0! 🎬 An end-to-end infrastructure for long video generation, with FP4 and parallelism at the core of both training and inference. ⚡45.7 FPS generation speed on 5B model⚡ ✨ LongLive 2.0 supports real-video training, few-step distillation, multi-shot training/inference, sequence-parallel acceleration, NVFP4 KV cache, and async VAE decoding deployment. 🧩 To our knowledge, this is the first open-source 4-bit long video generation infra that covers both training and inference. 🙌 Welcome to check it out, try it, and share feedback! 🔗 Code: 📰 Paper: 🎥 Demo: #LongVideoGeneration #VideoGeneration #Realtime #AIInfra #EfficientAI #FP4 #Parallel #NVIDIA

Yukang Chen

60,351 Aufrufe • vor 5 Monaten

I spent my summer building TinyTPU : An open source ML inference and training chip. it can do end to end inference + training ENTIRELY on chip. here's how I did it👇:

I spent my summer building TinyTPU : An open source ML inference and training chip. it can do end to end inference + training ENTIRELY on chip. here's how I did it👇:

surya

356,700 Aufrufe • vor 1 Jahr

🚀 We open-sourced LongLive — interactive, real-time long-video generation. 👥Generates video in real time as users enter text prompts. ⚡️20.7 FPS on a single H100,⏱️up to 240s per clip. 🎬Fine-tunes SOTA short-video models (e.g., Wan) into long-video generators. 🌍One step closer to World Models. All code for training & inference, model weights, demo page, and videos released! Paper: Code: Model: Demo Page: Introduction Video:

🚀 We open-sourced LongLive — interactive, real-time long-video generation. 👥Generates video in real time as users enter text prompts. ⚡️20.7 FPS on a single H100,⏱️up to 240s per clip. 🎬Fine-tunes SOTA short-video models (e.g., Wan) into long-video generators. 🌍One step closer to World Models. All code for training & inference, model weights, demo page, and videos released! Paper: Code: Model: Demo Page: Introduction Video:

Yukang Chen

11,943 Aufrufe • vor 1 Jahr

Gemma and the power of post-training. 💪 Google DeepMind’s Léonard Hussenot details: 🔍 How Gemma facilitates post-training research 🚀 Best-of-N Distillation (BOND) for model efficiency 🏆 Improving reward models with weight averaging 👀 →

Gemma and the power of post-training. 💪 Google DeepMind’s Léonard Hussenot details: 🔍 How Gemma facilitates post-training research 🚀 Best-of-N Distillation (BOND) for model efficiency 🏆 Improving reward models with weight averaging 👀 →

Google for Developers

18,684 Aufrufe • vor 1 Jahr

Our research team just released Flex-Forcing: a video generation method that lets a single model switch between generation methods at inference time. Right now there are two main approaches to video generation. Bidirectional diffusion models attend to every frame at once, holding structure well at the cost of speed. Autoregressive models generate frame by frame, so they stream fast and scale to long clips, but accumulate error and drift over time. Flex-Forcing trains a single model to do both, letting you choose from the range at inference based on your compute budget.

Our research team just released Flex-Forcing: a video generation method that lets a single model switch between generation methods at inference time. Right now there are two main approaches to video generation. Bidirectional diffusion models attend to every frame at once, holding structure well at the cost of speed. Autoregressive models generate frame by frame, so they stream fast and scale to long clips, but accumulate error and drift over time. Flex-Forcing trains a single model to do both, letting you choose from the range at inference based on your compute budget.

NVIDIA AI

31,412 Aufrufe • vor 3 Monaten

Meta announces Movie Gen A Cast of Media Foundation Models We present Movie Gen, a cast of foundation models that generates high-quality, 1080p HD videos with different aspect ratios and synchronized audio. We also show additional capabilities such as precise instruction-based video editing and generation of personalized videos based on a user’s image. Our models set a new state-of-the-art on multiple tasks: text-to-video synthesis, video personalization, video editing, video-to-audio generation, and text-to-audio generation. Our largest video generation model is a 30B parameter transformer trained with a maximum context length of 73K video tokens, corresponding to a generated video of 16 seconds at 16 frames-per-second. We show multiple technical innovations and simplifications on the architecture, latent spaces, training objectives and recipes, data curation, evaluation protocols, parallelization techniques, and inference optimizations that allow us to reap the benefits of scaling pre-training data, model size, and training compute for training large scale media generation models. We hope this paper helps the research community to accelerate progress and innovation in media generation models

Meta announces Movie Gen A Cast of Media Foundation Models We present Movie Gen, a cast of foundation models that generates high-quality, 1080p HD videos with different aspect ratios and synchronized audio. We also show additional capabilities such as precise instruction-based video editing and generation of personalized videos based on a user’s image. Our models set a new state-of-the-art on multiple tasks: text-to-video synthesis, video personalization, video editing, video-to-audio generation, and text-to-audio generation. Our largest video generation model is a 30B parameter transformer trained with a maximum context length of 73K video tokens, corresponding to a generated video of 16 seconds at 16 frames-per-second. We show multiple technical innovations and simplifications on the architecture, latent spaces, training objectives and recipes, data curation, evaluation protocols, parallelization techniques, and inference optimizations that allow us to reap the benefits of scaling pre-training data, model size, and training compute for training large scale media generation models. We hope this paper helps the research community to accelerate progress and innovation in media generation models

AK

62,719 Aufrufe • vor 2 Jahren

This is why Nvidia bought Groq. If Groq deploys a ton of low-cost inference chips, demand for training will increase massively. The more inference you have, the more training you need, and vice versa. Nvidia can keep its high margins selling chips for training forever.

This is why Nvidia bought Groq. If Groq deploys a ton of low-cost inference chips, demand for training will increase massively. The more inference you have, the more training you need, and vice versa. Nvidia can keep its high margins selling chips for training forever.

Lin

92,571 Aufrufe • vor 10 Monaten

The Inference Engineering Masterclass: 10x faster models, quantization, speculative decoding, Rubin, & self-optimizing AI Baseten Philip Kiely and ali explain what actually happens after a model is trained, why turning weights into a fast and reliable product creates an entirely new optimization problem, how quantization errors can cancel out to unlock more throughput, why inference teams are still finding 20–200% performance gains, how video generation runs into a quadratic attention wall, and how GLM-5.2 helped rewrite and optimize the GPU kernels serving GLM-5.2 itself.

The Inference Engineering Masterclass: 10x faster models, quantization, speculative decoding, Rubin, & self-optimizing AI Baseten Philip Kiely and ali explain what actually happens after a model is trained, why turning weights into a fast and reliable product creates an entirely new optimization problem, how quantization errors can cancel out to unlock more throughput, why inference teams are still finding 20–200% performance gains, how video generation runs into a quadratic attention wall, and how GLM-5.2 helped rewrite and optimize the GPU kernels serving GLM-5.2 itself.

Latent.Space

149,249 Aufrufe • vor 3 Monaten

(1/5) 5 seconds of video. 1.8s seconds of generation. One NVIDIA GeForce RTX 5090 on FastVideo. 🤯🚀 - FastWan-QAD, a new family of video generation models - Trained with FastVideo's Quantization-Aware Distillation (QAD) recipe. - Powered by FastVideo, we push a single NVIDIA GeForce RTX 5090 to its absolute limit: generating a 5-second 480P video in 1.8s end-to-end! 📜 Blog: 💻 Code: 💽 Model:

(1/5) 5 seconds of video. 1.8s seconds of generation. One NVIDIA GeForce RTX 5090 on FastVideo. 🤯🚀 - FastWan-QAD, a new family of video generation models - Trained with FastVideo's Quantization-Aware Distillation (QAD) recipe. - Powered by FastVideo, we push a single NVIDIA GeForce RTX 5090 to its absolute limit: generating a 5-second 480P video in 1.8s end-to-end! 📜 Blog: 💻 Code: 💽 Model:

Hao AI Lab

13,468 Aufrufe • vor 4 Monaten

Scaling enterprise AI takes performance, flexibility and efficiency. Hear AT&T CTO Jeremy Legg share how AMD solutions are helping AT&T support post-training and inference workloads while prioritizing AI data sovereignty and cost efficiency.

Scaling enterprise AI takes performance, flexibility and efficiency. Hear AT&T CTO Jeremy Legg share how AMD solutions are helping AT&T support post-training and inference workloads while prioritizing AI data sovereignty and cost efficiency.

AMD

24,133 Aufrufe • vor 1 Monat

Frontier models that use reasoning to "think" during inference are generating 5X more AI tokens per year. ✨ "Inference is now a thinking process. And in order to teach AI how to think, reinforcement learning and very significant computation was introduced into post-training," said NVIDIA CEO Jensen Huang in a recent keynote at #CES2026. Reinforcement learning is increasing computation demands across all AI scaling laws: pre-training, post-training, and test-time scaling. Learn more about reinforcement learning and AI scaling ➡️

Frontier models that use reasoning to "think" during inference are generating 5X more AI tokens per year. ✨ "Inference is now a thinking process. And in order to teach AI how to think, reinforcement learning and very significant computation was introduced into post-training," said NVIDIA CEO Jensen Huang in a recent keynote at #CES2026. Reinforcement learning is increasing computation demands across all AI scaling laws: pre-training, post-training, and test-time scaling. Learn more about reinforcement learning and AI scaling ➡️

NVIDIA AI Infrastructure

13,329 Aufrufe • vor 9 Monaten

NVIDIA Cosmos 3 launches with a full model family — Cosmos Super for highest-accuracy robotics and AV post-training, Cosmos Nano for high-speed video and action reasoning, and Cosmos Edge for real-time edge inference. Read the release ➡️

NVIDIA Cosmos 3 launches with a full model family — Cosmos Super for highest-accuracy robotics and AV post-training, Cosmos Nano for high-speed video and action reasoning, and Cosmos Edge for real-time edge inference. Read the release ➡️

NVIDIA Newsroom

47,166 Aufrufe • vor 5 Monaten

We are innovating the full-stack AI infrastructure platform on a one-year rhythm — pushing the boundaries of inference performance across chips, systems, and software. NVIDIA Blackwell is the most advanced AI platform ever built — continuously optimized for record-breaking performance. Groundbreaking advancements include the NVIDIA-designed NVFP4 format, enabling high inference performance and accuracy. #GTCParis at #VivaTech

We are innovating the full-stack AI infrastructure platform on a one-year rhythm — pushing the boundaries of inference performance across chips, systems, and software. NVIDIA Blackwell is the most advanced AI platform ever built — continuously optimized for record-breaking performance. Groundbreaking advancements include the NVIDIA-designed NVFP4 format, enabling high inference performance and accuracy. #GTCParis at #VivaTech

NVIDIA

80,665 Aufrufe • vor 1 Jahr

World models are moving beyond offline generation towards interactive, real-time experiences. Introducing ⚡FlashDreams⚡: an open-source high-performance inference and serving library built for autoregressive world models: 🔥 Up to 3.10× faster LingBot-World inference 🔥 Up to 2.12× faster Self-Forcing inference 🔥 Up to 1.40× faster Wan2.1 inference 🔥 8 integrated models 🔥 Multi-GPU, streaming, low-latency serving 🔥 Agentic skills that teach you how to use it FlashDreams is designed for a new generation of AI systems that continuously evolve over time while responding to user interactions. It powers applications across robotics, autonomous vehicle simulation, gaming, and virtual worlds. Github: Docs: Research page: Join the #flashdreams Discord channel at FlashDreams is also the runtime backbone behind NVIDIA OmniDreams ( 1/n #AI #WorldModels #FastInference #PhysicalAI #OpenSource #NVIDIA

World models are moving beyond offline generation towards interactive, real-time experiences. Introducing ⚡FlashDreams⚡: an open-source high-performance inference and serving library built for autoregressive world models: 🔥 Up to 3.10× faster LingBot-World inference 🔥 Up to 2.12× faster Self-Forcing inference 🔥 Up to 1.40× faster Wan2.1 inference 🔥 8 integrated models 🔥 Multi-GPU, streaming, low-latency serving 🔥 Agentic skills that teach you how to use it FlashDreams is designed for a new generation of AI systems that continuously evolve over time while responding to user interactions. It powers applications across robotics, autonomous vehicle simulation, gaming, and virtual worlds. Github: Docs: Research page: Join the #flashdreams Discord channel at FlashDreams is also the runtime backbone behind NVIDIA OmniDreams ( 1/n #AI #WorldModels #FastInference #PhysicalAI #OpenSource #NVIDIA

Ruilong Li

100,490 Aufrufe • vor 5 Monaten

Introducing Modality Forcing, a recipe for post-training T2I models for SOTA RGB-Depth generation! Text-to-image (T2I) models learn rich representations of the spatial world. How do we build on this prior for high-quality depth generation? 🧵 [1/6]

Introducing Modality Forcing, a recipe for post-training T2I models for SOTA RGB-Depth generation! Text-to-image (T2I) models learn rich representations of the spatial world. How do we build on this prior for high-quality depth generation? 🧵 [1/6]

Bardienus Duisterhof

67,032 Aufrufe • vor 4 Monaten

Generation 𝗖𝗔𝗡 improve understanding! With the right recipe, we show that training on generation data can help understanding tasks. Introducing OmniTaskonomy: our year-long study of when and how generation helps understanding.

Generation 𝗖𝗔𝗡 improve understanding! With the right recipe, we show that training on generation data can help understanding tasks. Introducing OmniTaskonomy: our year-long study of when and how generation helps understanding.

Jiaxin Ge

61,374 Aufrufe • vor 29 Tagen

At Apsara2026, Alibaba’s chip design unit, T-Head, introduced the Zhenwu V900—its latest AI training and inference processor capable of handling both high-precision model training and ultra-low-precision inference, delivering three times the performance of its predecessor. Zhenwu AI chips are already powering 650+ customers across industries, from automobiles to LLMs and embodied intelligence. The roadmap for next-generation Yitian CPUs was also introduced, showcasing processors tailored for agentic AI tasks and scheduled for launch in 2027.🚀

At Apsara2026, Alibaba’s chip design unit, T-Head, introduced the Zhenwu V900—its latest AI training and inference processor capable of handling both high-precision model training and ultra-low-precision inference, delivering three times the performance of its predecessor. Zhenwu AI chips are already powering 650+ customers across industries, from automobiles to LLMs and embodied intelligence. The roadmap for next-generation Yitian CPUs was also introduced, showcasing processors tailored for agentic AI tasks and scheduled for launch in 2027.🚀

Alibaba Group

1,397,469 Aufrufe • vor 1 Monat

Groq founder Jonathan Ross explains why their architecture is better for inference than GPUs. Inference doesn't require as much compute power as training. What matters in inference is speed, cost, and energy consumption. $NVDA was going to lose market share as AI workloads shift to inference. They saw this and made a highly strategic move by striking a deal with Groq. $NVDA is already dominating training, and now it's also well-positioned to dominate inference.

Groq founder Jonathan Ross explains why their architecture is better for inference than GPUs. Inference doesn't require as much compute power as training. What matters in inference is speed, cost, and energy consumption. $NVDA was going to lose market share as AI workloads shift to inference. They saw this and made a highly strategic move by striking a deal with Groq. $NVDA is already dominating training, and now it's also well-positioned to dominate inference.

Oguz Erkan

97,186 Aufrufe • vor 10 Monaten

We're releasing Paris 2.0, which, to our knowledge, is the world's first decentralized trained video generation model. We benchmarked it against a monolithic model trained on the same data and compute budget, and Paris 2.0 outperformed the monolithic by ~2x on FVD benchmark.

We're releasing Paris 2.0, which, to our knowledge, is the world's first decentralized trained video generation model. We benchmarked it against a monolithic model trained on the same data and compute budget, and Paris 2.0 outperformed the monolithic by ~2x on FVD benchmark.

Bidhan Roy

457,856 Aufrufe • vor 5 Monaten

"the more effort you put into training the less effort you need in inference." Elon Musk on AI training and pursuing a dual path of NVIDIA and DOJO

"the more effort you put into training the less effort you need in inference." Elon Musk on AI training and pursuing a dual path of NVIDIA and DOJO

Adrian Dittmann

55,171 Aufrufe • vor 2 Jahren