正在加载视频...

视频加载失败

加载此视频时出现问题。这可能是由于临时网络问题，或视频可能不可用。

#VideoReason We are open-sourcing the entire VBVR stack to speed-up the arrival of video reasoning as the next fundamental paradigm of intelligence - 150+ synthetic generators - 1 million training clips - Cloud-scale data factory - Unified EvalKit - 100 rule-based evaluators - Strong baseline model Checkout at

Hokin Deng

1,202 subscribers

53,954 次观看 • 7 个月前 •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

30 条评论

Hokin Deng 的头像

Hokin Deng7 个月前

Full technical report at

Hokin Deng 的头像

Hokin Deng7 个月前

Other links: EvalKit: Cloud Infra: TrainingSet: EvalSet: Leaderboard: Model: Data Generators:

fuzzpuppy 的头像

fuzzpuppy7 个月前

Fantastic work! Here are some experiments I ran with the model - I can't believe it enabled the bear to sort the candies 🤯 thank you for the project!

Runming "Tony" Wang 的头像

Runming "Tony" Wang7 个月前

Exciting stuff!

Hokin Deng 的头像

Hokin Deng7 个月前

Thank you !

Pinyuan Feng 的头像

Pinyuan Feng7 个月前

That’s awesome, Hokin! Can’t imagine how much you and your team have done so far.

Hokin Deng 的头像

Hokin Deng7 个月前

Thank you so much!!! this is awesome ~~

Yichen LYU 的头像

Yichen LYU7 个月前

Worship!👍

Hokin Deng 的头像

Hokin Deng7 个月前

Thank you!

Paru Saha 的头像

Paru Saha7 个月前

opening the entire vbvr stack with 150 plus generators and a million clips accelerates video reasoning research effectively

Hokin Deng 的头像

Hokin Deng7 个月前

Thanh you !!!

Zeng Tao 的头像

Zeng Tao7 个月前

Recently read your paper on video reasoning and was thrilled to see that there's already new progress! Congrats!

Hokin Deng 的头像

Hokin Deng7 个月前

Thanks so much for your nice words !!!!!! Cheers !

Mark Wang 的头像

Mark Wang7 个月前

Great job

Hokin Deng 的头像

Hokin Deng7 个月前

@MarkWang971630g Thanks!

Qianli Ma 的头像

Qianli Ma7 个月前

Amazing!

Kelvin Mu 的头像

Kelvin Mu7 个月前

congrats Hokin!

Hokin Deng 的头像

Hokin Deng7 个月前

Thank you, Kelvin!

Chaoyue He 的头像

Chaoyue He7 个月前

huge contributions！！！

Hokin Deng 的头像

Hokin Deng7 个月前

Thank you!

Ran Ji 的头像

Ran Ji7 个月前

👍👍👍！

Hokin Deng 的头像

Hokin Deng7 个月前

Hhhh awesome!

Shin 的头像

Shin7 个月前

great work Hokin!!!

Gates C. Yin 的头像

Gates C. Yin7 个月前

human genome project for video reasoning. congrats!

Hokin Deng 的头像

Hokin Deng7 个月前

@SciOpt1mus Hhhhhh what an analogy

Marcus Xue 的头像

Marcus Xue7 个月前

Amazing projects

Hokin Deng 的头像

Hokin Deng7 个月前

@hao_xue_kidooo Thank you!

Tim Reha 的头像

Tim Reha7 个月前

Pefecto!

Arnas Uselis 的头像

Arnas Uselis7 个月前

Looks cool! Have you tried using this suite to assess (video) VLMs? (framing it as multiple choice questions over the next frame or similar)

Mark Wang 的头像

Mark Wang6 个月前

@caizhongang great work

相关视频

1/ We are speed running industrial robotics. It took us just 19 days from the first day of data collection to filming a 2.5-hour continuous run of our model autonomously breaking down industrial containers — zero human intervention. The data efficiency of our DVA model is fundamentally changing how fast we bring robots out of the lab and into the factory. Autonomous operation with 3 hours of data collection at a customer factory.

1/ We are speed running industrial robotics. It took us just 19 days from the first day of data collection to filming a 2.5-hour continuous run of our model autonomously breaking down industrial containers — zero human intervention. The data efficiency of our DVA model is fundamentally changing how fast we bring robots out of the lab and into the factory. Autonomous operation with 3 hours of data collection at a customer factory.

Rhoda AI

26,105 次观看 • 6 个月前

.Cohere is defining the future of sovereign AI. Autumn Moulder, VP of Engineering at Cohere, breaks down the requirements for a full AI sovereign stack at #NVIDIAGTC: ✅ Full-Stack Sovereignty: Deploying everything from the model to applications and reasoning traces within a single data center. ✅ The Power of Openness: Why open models like NVIDIA Nemotron are critical for data lineage and navigating evolving regulations.

.Cohere is defining the future of sovereign AI. Autumn Moulder, VP of Engineering at Cohere, breaks down the requirements for a full AI sovereign stack at #NVIDIAGTC: ✅ Full-Stack Sovereignty: Deploying everything from the model to applications and reasoning traces within a single data center. ✅ The Power of Openness: Why open models like NVIDIA Nemotron are critical for data lineage and navigating evolving regulations.

NVIDIA AI

10,965 次观看 • 6 个月前

$Introducing HRM-Text. An ultra-lean 1B-parameter reasoning language model designed to deliver strong general performance with a fraction of the data, compute, and infrastructure. Trained on just 40B structured tokens, HRM-Text achieves competitive performance while using ~1/1000 of the training data of comparable models. The kicker? The full model trains in roughly one day on a $1,000 budget. This opens the door to a new generation of AI that is powerful, accessible, and radically easier to adapt. Theories and research concepts once deemed too expensive to test are officially back in the game. Sapient Intelligence invites you to help us shape a new paradigm for general intelligence.$

Introducing HRM-Text. An ultra-lean 1B-parameter reasoning language model designed to deliver strong general performance with a fraction of the data, compute, and infrastructure. Trained on just 40B structured tokens, HRM-Text achieves competitive performance while using ~1/1000 of the training data of comparable models. The kicker? The full model trains in roughly one day on a $1,000 budget. This opens the door to a new generation of AI that is powerful, accessible, and radically easier to adapt. Theories and research concepts once deemed too expensive to test are officially back in the game. Sapient Intelligence invites you to help us shape a new paradigm for general intelligence.

Sapient Intelligence

518,390 次观看 • 4 个月前

Today we open-source LingBot-Video — the first MoE-based video foundation model built for embodied intelligence. 🔹30B params, only 3B active at inference. 🔹Augmented with 70K hours of embodied data on top of large-scale internet video pretraining. 🔹Already outperforming Wan2.6, Seedance 1.5 Pro, and Cosmos3 Super on RBench. 🧵👇

Today we open-source LingBot-Video — the first MoE-based video foundation model built for embodied intelligence. 🔹30B params, only 3B active at inference. 🔹Augmented with 70K hours of embodied data on top of large-scale internet video pretraining. 🔹Already outperforming Wan2.6, Seedance 1.5 Pro, and Cosmos3 Super on RBench. 🧵👇

Robbyant

1,183,330 次观看 • 2 个月前

1. Build II-Agent, possibly the best agent in the world 2. Open source it ... 4. Everyone profits Our mission at II is to increase humanity's intelligence II-Agent is the next part of the stack we are building to enable this, topping the agent benchmarks 🧵

1. Build II-Agent, possibly the best agent in the world 2. Open source it ... 4. Everyone profits Our mission at II is to increase humanity's intelligence II-Agent is the next part of the stack we are building to enable this, topping the agent benchmarks 🧵

Intelligent Internet

660,293 次观看 • 1 年前

ELON: AI WILL HARNESS THE POWER OF THE SUN AND GALAXY FOR COMPUTE "I think we'll see intelligence continue to scale all the way up to where most of the power of the sun is harnessed for compute and then ultimately most of the power of the galaxy, you know, sort of cloud Kardashev two, Kardashev three scale compute. So I guess once you think of artificial intelligence not as a destination that you reach, but really as part of the overall escalation of intelligence that we are aware of." Source: Jon Hernandez

ELON: AI WILL HARNESS THE POWER OF THE SUN AND GALAXY FOR COMPUTE "I think we'll see intelligence continue to scale all the way up to where most of the power of the sun is harnessed for compute and then ultimately most of the power of the galaxy, you know, sort of cloud Kardashev two, Kardashev three scale compute. So I guess once you think of artificial intelligence not as a destination that you reach, but really as part of the overall escalation of intelligence that we are aware of." Source: Jon Hernandez

Mario Nawfal

1,668,111 次观看 • 10 个月前

SemiAnalysis's Dylan Patel says AI models will improve faster in the next 6 month to a year than we saw in the past year because there's a new axis of scale that has been unlocked in the form of synthetic data generation, that we are still very early in scaling up

SemiAnalysis's Dylan Patel says AI models will improve faster in the next 6 month to a year than we saw in the past year because there's a new axis of scale that has been unlocked in the form of synthetic data generation, that we are still very early in scaling up

Tsarathustra

174,883 次观看 • 1 年前

LP-MusicCaps: LLM-Based Pseudo Music Captioning paper page: Automatic music captioning, which generates natural language descriptions for given music tracks, holds significant potential for enhancing the understanding and organization of large volumes of musical data. Despite its importance, researchers face challenges due to the costly and time-consuming collection process of existing music-language datasets, which are limited in size. To address this data scarcity issue, we propose the use of large language models (LLMs) to artificially generate the description sentences from large-scale tag datasets. This results in approximately 2.2M captions paired with 0.5M audio clips. We term it Large Language Model based Pseudo music caption dataset, shortly, LP-MusicCaps. We conduct a systemic evaluation of the large-scale music captioning dataset with various quantitative evaluation metrics used in the field of natural language processing as well as human evaluation. In addition, we trained a transformer-based music captioning model with the dataset and evaluated it under zero-shot and transfer-learning settings. The results demonstrate that our proposed approach outperforms the supervised baseline model.

LP-MusicCaps: LLM-Based Pseudo Music Captioning paper page: Automatic music captioning, which generates natural language descriptions for given music tracks, holds significant potential for enhancing the understanding and organization of large volumes of musical data. Despite its importance, researchers face challenges due to the costly and time-consuming collection process of existing music-language datasets, which are limited in size. To address this data scarcity issue, we propose the use of large language models (LLMs) to artificially generate the description sentences from large-scale tag datasets. This results in approximately 2.2M captions paired with 0.5M audio clips. We term it Large Language Model based Pseudo music caption dataset, shortly, LP-MusicCaps. We conduct a systemic evaluation of the large-scale music captioning dataset with various quantitative evaluation metrics used in the field of natural language processing as well as human evaluation. In addition, we trained a transformer-based music captioning model with the dataset and evaluated it under zero-shot and transfer-learning settings. The results demonstrate that our proposed approach outperforms the supervised baseline model.

AK

78,794 次观看 • 3 年前

🚀 1/7 We are thrilled to launch LLM360 — pushing the frontier of open-source & transparent LLMs! Starting with Amber (7B) & CrystalCoder (7B), we are releasing brand new pre-trained LLMs with all training code, data, and up to 360 model checkpoints. 🔗

LLM360

329,538 次观看 • 2 年前

Elon Musk: Artificial intelligence will continue to scale until we've harnessed all the energy of our galaxy. “I think we'll see intelligence continue to scale all the way up to where most of the power of the sun is harnessed for compute, and then ultimately most of the power of the galaxy, sort of Kardashev 2, Kardashev 3 scale compute. One should think about artificial intelligence, not as a destination that you reach, but really as part of the overall escalation of intelligence that we are aware of. Human intelligence has also scaled as the population has increased and we've been able to store more and more information.” Interview at the All-In Summit, September 10, 2025

Elon Musk: Artificial intelligence will continue to scale until we've harnessed all the energy of our galaxy. “I think we'll see intelligence continue to scale all the way up to where most of the power of the sun is harnessed for compute, and then ultimately most of the power of the galaxy, sort of Kardashev 2, Kardashev 3 scale compute. One should think about artificial intelligence, not as a destination that you reach, but really as part of the overall escalation of intelligence that we are aware of. Human intelligence has also scaled as the population has increased and we've been able to store more and more information.” Interview at the All-In Summit, September 10, 2025

ELON CLIPS

41,848 次观看 • 1 年前

Introducing Praxis-1, an open-weight World Action Model that turns our video pre-training expertise into real world control for robots. Robot demonstration data is limited. But video data is infinite. Praxis-1 extends our bet that the best policy models will learn from video, and use that scale to bring embodied intelligence to every industry. Praxis-1 is built on our large-scale video pre-training and real-time infrastructure, and acts as a policy model for robotics developers and researchers that works across any embodiment or environment, no matter how complex. We are actively testing with early partners including Noble Machines, Standard Bots and Ultra, and will be releasing the weights publicly in the coming months. Learn more and request early access at

Introducing Praxis-1, an open-weight World Action Model that turns our video pre-training expertise into real world control for robots. Robot demonstration data is limited. But video data is infinite. Praxis-1 extends our bet that the best policy models will learn from video, and use that scale to bring embodied intelligence to every industry. Praxis-1 is built on our large-scale video pre-training and real-time infrastructure, and acts as a policy model for robotics developers and researchers that works across any embodiment or environment, no matter how complex. We are actively testing with early partners including Noble Machines, Standard Bots and Ultra, and will be releasing the weights publicly in the coming months. Learn more and request early access at

Runway

55,900 次观看 • 17 小时前

We are thrilled to announce the open-source release of Hallo2: a new avatar video generative model capable of generating stunning 4K resolution videos for up to 1 hour! 🎥✨ Explore the project for the open source code: Let's push the boundaries of avatar video generative model! #OpenSource #AI #FaceAnimation #GenerativeArt

Siyu ZHU

58,323 次观看 • 1 年前

Jason Ma reveals the first true scaling law in robotics: robots trained on 1 million hours of first-person human video predictably get better without ever seeing robot data "Yesterday we announced our new flagship robot foundation model called Dyna-2. It's very significant for the entire field for many different reasons. First of all, it's the first robot foundation model trained on at least one million hours of data. That in itself is a very challenging infrastructure challenge." "What we demonstrated is that even if the one million hours of data is entirely first-person video of humans doing manipulation tasks, we actually see scaling law for transferring to robot embodiments that the model has never seen before." "By training on more human data, we actually see predictable performance improvement on robots. That's a big deal because the biggest challenge facing robot foundation models is that we don't have enough data." "Unlike language models, there are just not enough robotics data out there, and it's very difficult to collect these robot data. Only by passively observing humans doing things can we scale this kind of data, and this is the first time we showed there is a lot of improvement that can transfer." Jason Ma Dyna Robotics

Jason Ma reveals the first true scaling law in robotics: robots trained on 1 million hours of first-person human video predictably get better without ever seeing robot data "Yesterday we announced our new flagship robot foundation model called Dyna-2. It's very significant for the entire field for many different reasons. First of all, it's the first robot foundation model trained on at least one million hours of data. That in itself is a very challenging infrastructure challenge." "What we demonstrated is that even if the one million hours of data is entirely first-person video of humans doing manipulation tasks, we actually see scaling law for transferring to robot embodiments that the model has never seen before." "By training on more human data, we actually see predictable performance improvement on robots. That's a big deal because the biggest challenge facing robot foundation models is that we don't have enough data." "Unlike language models, there are just not enough robotics data out there, and it's very difficult to collect these robot data. Only by passively observing humans doing things can we scale this kind of data, and this is the first time we showed there is a lot of improvement that can transfer." Jason Ma Dyna Robotics

MTS

17,740 次观看 • 1 个月前

We are teaming up with @Databricks to supercharge AI workflows. ⚡ #CUDA will soon be at the heart of Databricks' Data Intelligence Platform, starting with AI-accelerated Photon, delivering improved speed and efficiency for customers’ data warehousing and analytics workflows.

We are teaming up with @Databricks to supercharge AI workflows. ⚡ #CUDA will soon be at the heart of Databricks' Data Intelligence Platform, starting with AI-accelerated Photon, delivering improved speed and efficiency for customers’ data warehousing and analytics workflows.

NVIDIA AI Developer

22,937 次观看 • 2 年前

We’re excited to share that 🥇Llama Nemotron Super 49B v1.5 -- our latest open reasoning model -- is now #1 on the Artificial Analysis Intelligence Index - a leaderboard that spans advanced math, science, and agentic tasks, in the 70B open model category. Llama Nemotron Super 49B v1.5 is trained with high-quality reasoning synthetic data generated from models like Qwen3-235B and DeepSeek R1. It delivers state-of-the-art accuracy and throughput, running on a single H100. Key features: 🎯 Leading accuracy on multi-step reasoning, math, coding, and function-calling 🏗️ Post-trained using RPO, DPO, and RLVR across 26M+ synthetic examples 📊 Fully transparent training data and techniques If you're building AI agents and want a high accuracy, fully-open, and transparent reasoning model that you can deploy anywhere, try Super v1.5 on or download from Hugging Face 🤗 ➡️ Leaderboard:

We’re excited to share that 🥇Llama Nemotron Super 49B v1.5 -- our latest open reasoning model -- is now #1 on the Artificial Analysis Intelligence Index - a leaderboard that spans advanced math, science, and agentic tasks, in the 70B open model category. Llama Nemotron Super 49B v1.5 is trained with high-quality reasoning synthetic data generated from models like Qwen3-235B and DeepSeek R1. It delivers state-of-the-art accuracy and throughput, running on a single H100. Key features: 🎯 Leading accuracy on multi-step reasoning, math, coding, and function-calling 🏗️ Post-trained using RPO, DPO, and RLVR across 26M+ synthetic examples 📊 Fully transparent training data and techniques If you're building AI agents and want a high accuracy, fully-open, and transparent reasoning model that you can deploy anywhere, try Super v1.5 on or download from Hugging Face 🤗 ➡️ Leaderboard:

NVIDIA AI Developer

100,506 次观看 • 1 年前

Today we're launching Hebbian Robotics (YC S26) Hebbian Robotics . We build APIs for evaluating data quality, without training a robotics model. Data vendors use our API to search and analyze Physical AI data at scale, getting on-demand quality signals and metrics without managing infrastructure. High-quality data is the bottleneck to the next frontier of intelligence in robotics. Buyers want data they can trust. Data providers are focused on operational excellence. They don't have the bandwidth to understand the quality of their data at scale. We're fixing this. Vendors use our APIs to understand the impact of their decisions on their data quality: - Buyers demanding per-collector hour limits? Prove compliance while we stay ahead of the workarounds. - Changed your collection SOPs? Monitor for quality drift and regression. - Suspect duplicate samples? Verify with a single API call. Before Hebbian, I trained AI models for industrial cleaning robots, and Kingston Kuan spent years building large-scale infrastructure at Jane Street.

Today we're launching Hebbian Robotics (YC S26) Hebbian Robotics . We build APIs for evaluating data quality, without training a robotics model. Data vendors use our API to search and analyze Physical AI data at scale, getting on-demand quality signals and metrics without managing infrastructure. High-quality data is the bottleneck to the next frontier of intelligence in robotics. Buyers want data they can trust. Data providers are focused on operational excellence. They don't have the bandwidth to understand the quality of their data at scale. We're fixing this. Vendors use our APIs to understand the impact of their decisions on their data quality: - Buyers demanding per-collector hour limits? Prove compliance while we stay ahead of the workarounds. - Changed your collection SOPs? Monitor for quality drift and regression. - Suspect duplicate samples? Verify with a single API call. Before Hebbian, I trained AI models for industrial cleaning robots, and Kingston Kuan spent years building large-scale infrastructure at Jane Street.

Brandon Ong

166,693 次观看 • 1 个月前

Very happy to announce that we are open-sourcing the π₀ model, weights, and some fine-tuned checkpoints! Hoping this leads to lots of great follow-up research: Here is a fun test from our friends at UPenn.

Very happy to announce that we are open-sourcing the π₀ model, weights, and some fine-tuned checkpoints! Hoping this leads to lots of great follow-up research: Here is a fun test from our friends at UPenn.

Sergey Levine

72,579 次观看 • 1 年前