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.

A global relay. A smarter model. A Mathletic victory. Team NemoSkills won a global Kaggle challenge with a reasoning model that thinks fast and generalizes faster. 34 Olympiad-level questions solved in 5 hours, across time zones, on 4 NVIDIA L4 GPUs. 🔗

NVIDIA Newsroom

58,952 subscribers

38,771 Aufrufe • vor 1 Jahr •via X (Twitter)

Wissenschaft & Technologie Bildung

Anya Rossi• Live Now

Private livecam show

6 Kommentare

Profilbild von RLopesRibeiro

RLopesRibeirovor 1 Jahr

@kaggle it would be nice if you guys reacted at least a bit to whats happening around, instead of posting shitty pointless videos...like, investors dont matter?

Profilbild von Reji Modiyil

Reji Modiyilvor 1 Jahr

@nvidia @kaggle @nvidia, what a remarkable achievement. teamwork truly paves the way to success.

Profilbild von OPEN

OPENvor 1 Jahr

Cinematic pedigree of the highest order meets innovative AAA gameplay in OP3N. Dive into the world of Ready Player One — Wishlist Now!

Profilbild von Marc Van der Weyde

Marc Van der Weydevor 1 Jahr

@kaggle Why is nvidia setting en factory in Texas when Trump does not allow you export chips to China ? It costs nvidia 5 billion dollars ! I think the shareholders do not agree what Trump is doing. Is this America first. You better plant a factory in China !

Profilbild von Electe

Electevor 1 Jahr

@nvidia @kaggle @nvidia, congratulations, Team NemoSkills. This victory highlights the power of collaboration and innovative thinking in complex challenges. #Teamwork

Profilbild von Jose María Torres Gamez

Jose María Torres Gamezvor 1 Jahr

@kaggle Yeaaa woow more AI!!!!!

Ähnliche Videos

never forget that time Nvidia hired Myhtbusters Team to demonstrate the difference between CPUs and GPUs (with a Mona Lisa painting)

never forget that time Nvidia hired Myhtbusters Team to demonstrate the difference between CPUs and GPUs (with a Mona Lisa painting)

Trung Phan

2,400,080 Aufrufe • vor 1 Jahr

never forget that time Nvidia hired Myhtbusters Team to demonstrate the difference between CPUs and GPUs (with a Mona Lisa painting)

never forget that time Nvidia hired Myhtbusters Team to demonstrate the difference between CPUs and GPUs (with a Mona Lisa painting)

Bearly AI

55,909 Aufrufe • vor 14 Tagen

A major moment - UNLIMITED Kling 2.5 exclusively inside Higgsfield. Live now. Day-0 access. A reasoning engine powers a video creation model, in one system. Higgsfield thinks, Kling creates. Retweet in 4 hours to receive 50 credits via DM.

A major moment - UNLIMITED Kling 2.5 exclusively inside Higgsfield. Live now. Day-0 access. A reasoning engine powers a video creation model, in one system. Higgsfield thinks, Kling creates. Retweet in 4 hours to receive 50 credits via DM.

Higgsfield AI 🧩

4,634,804 Aufrufe • vor 10 Monaten

Running a real-time experiment on Julius: Which model can find mag 7 financial data and create a dashboard from it faster and more effectively - Claude Fable 5 or GPT-5.5? Which do you think is Model A and which is Model B?

Running a real-time experiment on Julius: Which model can find mag 7 financial data and create a dashboard from it faster and more effectively - Claude Fable 5 or GPT-5.5? Which do you think is Model A and which is Model B?

Julius AI

159,113 Aufrufe • vor 2 Monaten

BOOM 💥 The Canadian women’s relay team cruises to a relay heat victory and punch their ticket into the Commonwealth Games final. This is a confident group of women who want to make global podiums a regular occurrence. They have another shot tonight in Glasgow.

BOOM 💥 The Canadian women’s relay team cruises to a relay heat victory and punch their ticket into the Commonwealth Games final. This is a confident group of women who want to make global podiums a regular occurrence. They have another shot tonight in Glasgow.

Devin Heroux

12,148 Aufrufe • vor 15 Tagen

leviatán 💙 a truly incredible run. in 5 maps, all the way from the lowers, against the games most consistent team, with 4 global rookies; the youngest team to ever lift a global valorant trophy.

leviatán 💙 a truly incredible run. in 5 maps, all the way from the lowers, against the games most consistent team, with 4 global rookies; the youngest team to ever lift a global valorant trophy.

mimi

19,705 Aufrufe • vor 1 Monat

Ex-NVIDIA engineer who built Unsloth explained RL, kernels, reasoning, quantization, and agents in 2 hours 42 minutes - better than $5000 fine-tuning bootcamps. pick the base model -> write triton kernels for 2x faster fine-tune -> quantize to 4-bit -> run GRPO/DPO -> ship a reasoning model on your single GPU. That loop is why Unsloth is the default way to fine-tune Llama, Qwen, Gemma, and Phi on hardware you already own. Unsloth + Triton kernels + 4-bit quantization + GRPO/DPO + single-GPU fine-tuning - that's the stack. Watch and save it, then fine-tune your first model tonight.

Ex-NVIDIA engineer who built Unsloth explained RL, kernels, reasoning, quantization, and agents in 2 hours 42 minutes - better than $5000 fine-tuning bootcamps. pick the base model -> write triton kernels for 2x faster fine-tune -> quantize to 4-bit -> run GRPO/DPO -> ship a reasoning model on your single GPU. That loop is why Unsloth is the default way to fine-tune Llama, Qwen, Gemma, and Phi on hardware you already own. Unsloth + Triton kernels + 4-bit quantization + GRPO/DPO + single-GPU fine-tuning - that's the stack. Watch and save it, then fine-tune your first model tonight.

h100envy

493,571 Aufrufe • vor 1 Monat

Introducing UI-TARS-1.5, a vision-language model that beats OpenAI Operator and Claude 3.7 on GUI Agent and Game Agent tasks. We've open-sourced a small-size version model for research purposes, more details can be found in our blog. TARS learns solely from a screen, but generalizes beyond a screen! Blog: Model: App:

Introducing UI-TARS-1.5, a vision-language model that beats OpenAI Operator and Claude 3.7 on GUI Agent and Game Agent tasks. We've open-sourced a small-size version model for research purposes, more details can be found in our blog. TARS learns solely from a screen, but generalizes beyond a screen! Blog: Model: App:

Yujia Qin

85,212 Aufrufe • vor 1 Jahr

Nvidia has just announced Alpamayo 2 Super, an open 34 billion parameter reasoning vision-language-action model designed to accelerate the development of autonomous vehicles. This new model combines the NVIDIA Cosmos 3 Super reasoning model with a 2 billion parameter diffusion-based action expert model, and is post trained with reinforcement learning. The model can return multiple outputs: future trajectory plans, reasoning traces, grounded answers to questions about the scenes, and auto label generation. The model weights are now available for anyone to download on Hugging Face, and the inference code has been posted to GitHub. Distilled models can be deployed commercially without any further permission from Nvidia, and model outputs carry no license conditions. Automakers can distill down a compact version of this model that can run on the Nvidia computer in the car. Major kudos to Nvidia and Jensen Huang for advancing the state of the industry by releasing this as an open model with permissive licensing. Jensen isn't just paying lip service to the idea of open models, Nvidia is actually contributing to the ecosystem — and it's great for their business, because it helps sell more Thor computers that go in the car. Anyone can go download the model and play with it. If you do, let me know what you think. Personally I think it's so cool that we have open weights models that are this advanced, for anyone to download.

Nvidia has just announced Alpamayo 2 Super, an open 34 billion parameter reasoning vision-language-action model designed to accelerate the development of autonomous vehicles. This new model combines the NVIDIA Cosmos 3 Super reasoning model with a 2 billion parameter diffusion-based action expert model, and is post trained with reinforcement learning. The model can return multiple outputs: future trajectory plans, reasoning traces, grounded answers to questions about the scenes, and auto label generation. The model weights are now available for anyone to download on Hugging Face, and the inference code has been posted to GitHub. Distilled models can be deployed commercially without any further permission from Nvidia, and model outputs carry no license conditions. Automakers can distill down a compact version of this model that can run on the Nvidia computer in the car. Major kudos to Nvidia and Jensen Huang for advancing the state of the industry by releasing this as an open model with permissive licensing. Jensen isn't just paying lip service to the idea of open models, Nvidia is actually contributing to the ecosystem — and it's great for their business, because it helps sell more Thor computers that go in the car. Anyone can go download the model and play with it. If you do, let me know what you think. Personally I think it's so cool that we have open weights models that are this advanced, for anyone to download.

Whole Mars Catalog

45,459 Aufrufe • vor 12 Tagen

Made a demo to do test-time-scaling with mlx-lm and a R1-based reasoning model. Same idea as S1: - To force a response, swap "Wait" for "</think>" - To think more, swap "</think>" for "Wait" Runs fast with 4-bit Qwen 32B on an M3 max:

Made a demo to do test-time-scaling with mlx-lm and a R1-based reasoning model. Same idea as S1: - To force a response, swap "Wait" for "</think>" - To think more, swap "</think>" for "Wait" Runs fast with 4-bit Qwen 32B on an M3 max:

Awni Hannun

59,790 Aufrufe • vor 1 Jahr

And Siddhartha Venkatayogi built a fine-tuned Qwen3-1.7B model that recommends 6th-8th grade level math problems, optimized for landing questions in a student’s zone of proximal development. Significantly beats Opus-4.8 on recall@16 (96% vs 17%) and is 8x faster (1.10s vs 8.57s)

And Siddhartha Venkatayogi built a fine-tuned Qwen3-1.7B model that recommends 6th-8th grade level math problems, optimized for landing questions in a student’s zone of proximal development. Significantly beats Opus-4.8 on recall@16 (96% vs 17%) and is 8x faster (1.10s vs 8.57s)

Austen Allred

17,680 Aufrufe • vor 27 Tagen

AFTER LIKE 4 OR 5 HOURS AND WITH TWO PERSONS EXPLAIN HOW TO CODE I FINALLY DO THIS, CUSTOM SFX AND A CPM MODEL AT THE SAME TIME

AFTER LIKE 4 OR 5 HOURS AND WITH TWO PERSONS EXPLAIN HOW TO CODE I FINALLY DO THIS, CUSTOM SFX AND A CPM MODEL AT THE SAME TIME

samurott lover #1

18,037 Aufrufe • vor 1 Jahr

NVIDIA just revealed MotionBricks, a new AI model trained on more than 350,000 motion clips. What’s really impressive is that the same model can animate a digital character and control a humanoid robot in the real world.

NVIDIA just revealed MotionBricks, a new AI model trained on more than 350,000 motion clips. What’s really impressive is that the same model can animate a digital character and control a humanoid robot in the real world.

Justin Ryan

84,730 Aufrufe • vor 27 Tagen

My biz partner Danny White and I acquired a Pro Volley Ball team (Nebraska). Pls help us name this team in the comments. Excited to be a part of the rise of women’s sports on a global level. Joey Burrow Stake Giriş #Global #2026 #realprovolleyball

My biz partner Danny White and I acquired a Pro Volley Ball team (Nebraska). Pls help us name this team in the comments. Excited to be a part of the rise of women’s sports on a global level. Joey Burrow Stake Giriş #Global #2026 #realprovolleyball

Jason Derulo

297,986 Aufrufe • vor 3 Jahren

Throwback to the very first #Swift. We had quite the coup then with the whole engg team that had worked on the global model. It was a first for Suzuki to have team India involved to that level. Look how far we have come since! SVP #SuzukiSwift #ThrowbackThursday #Swifting

Throwback to the very first #Swift. We had quite the coup then with the whole engg team that had worked on the global model. It was a first for Suzuki to have team India involved to that level. Look how far we have come since! SVP #SuzukiSwift #ThrowbackThursday #Swifting

PatanCAR

47,862 Aufrufe • vor 2 Jahren

Get Raid Ready in ARC Raiders featuring DLSS 4 with Multi Frame Generation, RTX global illumination, and NVIDIA Reflex. For a limited time, Get ARC Raiders with select GeForce RTX 50 Series. #RTXOn →

Get Raid Ready in ARC Raiders featuring DLSS 4 with Multi Frame Generation, RTX global illumination, and NVIDIA Reflex. For a limited time, Get ARC Raiders with select GeForce RTX 50 Series. #RTXOn →

NVIDIA GeForce

73,405 Aufrufe • vor 10 Monaten

Cerebras inference is very fast. So fast that it changes how we think about configuring our LLMs for voice agent use cases. Kimi K2.6 is a 1T parameter reasoning model that Cerebras serves at 650 - 1,000 tokens per second (end-to-end throughput), with time to first token metrics as low as 150ms (latency). These numbers are two to three times faster than other similarly capable models. The biggest lever we get from this kind of speed is that we can use the model in reasoning mode, and still have excellent "time to first non-thinking token." This solves a big pain point we have in 2026 for voice agent use cases. Almost all recent innovation in post-training has focused on making models good at reasoning ("test time compute"). This is great, but it makes the user-facing model latency much, much slower. Which is a problem for conversational voice agents. We can run Kimi K2.6 with reasoning turned on, and get responses faster than other models produce with reasoning disabled. On my 30-turn voice agent benchmark, Kimi K2.6 with reasoning enabled ties GPT 5.1 and Haiku 4.5 with reasoning disabled, and is still about 200ms seconds faster! On my primary task agent benchmark, Kimi K2.6 is now the #2 model. It ranks just behind Gemini 3.5 Flash in "high" reasoning mode, and tied with GLM 5, Sonnet 4.6, and GPT 5.4 with reasoning set to "low." But Kimi K2.6 completes each turn in the agent loop in under 500ms. The other four models are all at least 3x slower. (Models only qualify for this benchmark if they can complete task turns at a P50 <4s.) A couple of other things that this speed buys us, for production voice agents: - Tool calls happen fast enough that we don't have to work around tool call latency in our pipeline design. - We can prompt the model to output structured data at the beginning of a response, followed by plain text for voice generation. This opens up possibilities like asking the model to do complex classification/generation tasks that influence the rest of the pipeline. For example, the model could create a detailed style prompt for a steerable TTS model, for each individual conversation turn. And, of course, you can use Kimi K2.6 with reasoning turned off. Cerebras calls this "instant" mode. Here's a video of a Cerebras Kimi K2.6 voice agent with voice-to-voice response time, measured at the client, under 500ms. This is the true response latency as perceived by the user, including all network and audio codec overhead, transcription and turn detection, Kimi K2.6 token generation, and voice generation. 500ms is, effectively, instant. So the Cerebras naming for this mode is a propos. :-)

Cerebras inference is very fast. So fast that it changes how we think about configuring our LLMs for voice agent use cases. Kimi K2.6 is a 1T parameter reasoning model that Cerebras serves at 650 - 1,000 tokens per second (end-to-end throughput), with time to first token metrics as low as 150ms (latency). These numbers are two to three times faster than other similarly capable models. The biggest lever we get from this kind of speed is that we can use the model in reasoning mode, and still have excellent "time to first non-thinking token." This solves a big pain point we have in 2026 for voice agent use cases. Almost all recent innovation in post-training has focused on making models good at reasoning ("test time compute"). This is great, but it makes the user-facing model latency much, much slower. Which is a problem for conversational voice agents. We can run Kimi K2.6 with reasoning turned on, and get responses faster than other models produce with reasoning disabled. On my 30-turn voice agent benchmark, Kimi K2.6 with reasoning enabled ties GPT 5.1 and Haiku 4.5 with reasoning disabled, and is still about 200ms seconds faster! On my primary task agent benchmark, Kimi K2.6 is now the #2 model. It ranks just behind Gemini 3.5 Flash in "high" reasoning mode, and tied with GLM 5, Sonnet 4.6, and GPT 5.4 with reasoning set to "low." But Kimi K2.6 completes each turn in the agent loop in under 500ms. The other four models are all at least 3x slower. (Models only qualify for this benchmark if they can complete task turns at a P50 <4s.) A couple of other things that this speed buys us, for production voice agents: - Tool calls happen fast enough that we don't have to work around tool call latency in our pipeline design. - We can prompt the model to output structured data at the beginning of a response, followed by plain text for voice generation. This opens up possibilities like asking the model to do complex classification/generation tasks that influence the rest of the pipeline. For example, the model could create a detailed style prompt for a steerable TTS model, for each individual conversation turn. And, of course, you can use Kimi K2.6 with reasoning turned off. Cerebras calls this "instant" mode. Here's a video of a Cerebras Kimi K2.6 voice agent with voice-to-voice response time, measured at the client, under 500ms. This is the true response latency as perceived by the user, including all network and audio codec overhead, transcription and turn detection, Kimi K2.6 token generation, and voice generation. 500ms is, effectively, instant. So the Cerebras naming for this mode is a propos. :-)

kwindla

40,593 Aufrufe • vor 2 Monaten

🧠 Chat with Reasoning A few days ago the DeepSeek team released a LLM model with reasoning in various sizes. This we show is an example of 1bl that can run on machines with low GPU power like a mobile, but have enough power to answer complex questions. With these advanced models it is possible to link it with #IoT equipment to control information and use it in advanced control environments. All this under Open Source models and decentralized networks such as #Neurai #XNA $XNA #DeepSeek #Reasoning #AIchat

🧠 Chat with Reasoning A few days ago the DeepSeek team released a LLM model with reasoning in various sizes. This we show is an example of 1bl that can run on machines with low GPU power like a mobile, but have enough power to answer complex questions. With these advanced models it is possible to link it with #IoT equipment to control information and use it in advanced control environments. All this under Open Source models and decentralized networks such as #Neurai #XNA $XNA #DeepSeek #Reasoning #AIchat

NeurAI Project / XNA

17,691 Aufrufe • vor 1 Jahr

French economist Thomas Piketty to Zack Polanski, "If we don't manage to articulate together a model of development, a model of energy production, a model of dematerialization of the economy, reduction in working time, which makes all of this aspiration compatible with planetary habitability, we are all going to suffer a lot." "We are all going to have three, four degrees or more in global warming." "And I can tell you the kind of heat wave we had, two weeks ago. It's going to look almost nothing at all as compared to what we'll have in a few decades." "So the challenge of the future is to try to have the kind of social and democratic and fiscal transformations that we had in the 20th century at the national level, to have it at a broader level." "Because the problem we have to solve, are global."

French economist Thomas Piketty to Zack Polanski, "If we don't manage to articulate together a model of development, a model of energy production, a model of dematerialization of the economy, reduction in working time, which makes all of this aspiration compatible with planetary habitability, we are all going to suffer a lot." "We are all going to have three, four degrees or more in global warming." "And I can tell you the kind of heat wave we had, two weeks ago. It's going to look almost nothing at all as compared to what we'll have in a few decades." "So the challenge of the future is to try to have the kind of social and democratic and fiscal transformations that we had in the 20th century at the national level, to have it at a broader level." "Because the problem we have to solve, are global."

Farrukh

21,107 Aufrufe • vor 7 Tagen