Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

How does test-time scaling impact robots? We find that larger models, more thinking, and more context help significantly for some prompts but not others. Like LLMs, we can also train a router to for a better performance/latency tradeoff! Paper:

chelseabfinn's profile picture

Chelsea Finn

99,025 subscribers

24,379 Aufrufe • vor 1 Monat •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

90% of AI products fail in the first month. It's not bad models but bad prompts. Adaline AI is a platform to create prompts, test across models and modalities, integrate tools, evaluate performance, and deploy with real-time monitoring. Find the best prompt for your LLMs...👇
15:26

90% of AI products fail in the first month. It's not bad models but bad prompts. Adaline AI is a platform to create prompts, test across models and modalities, integrate tools, evaluate performance, and deploy with real-time monitoring. Find the best prompt for your LLMs...👇

Akshay 🚀

46,821 Aufrufe • vor 1 Jahr

🧵1/n LLMs significantly improve Evolutionary Algorithms for molecular discovery! For 18 different molecular optimization tasks, we demonstrate how to achieve SOTA performance by incorporating different LLMs! Learn more in our new paper! Website: Code)
0:27

🧵1/n LLMs significantly improve Evolutionary Algorithms for molecular discovery! For 18 different molecular optimization tasks, we demonstrate how to achieve SOTA performance by incorporating different LLMs! Learn more in our new paper! Website: Code)

Yuanqi Du

17,899 Aufrufe • vor 2 Jahren

Today we announce TRINDs, a dataset and benchmarking pipeline that uses synthetic personas to train and optimize performance of LLMs for tropical and infectious diseases, which are out-of-distribution for most models. Learn more →
1:21

Today we announce TRINDs, a dataset and benchmarking pipeline that uses synthetic personas to train and optimize performance of LLMs for tropical and infectious diseases, which are out-of-distribution for most models. Learn more →

Google AI

39,237 Aufrufe • vor 1 Jahr

Speech-native models like Moshi sound great and answer fast, but aren’t as smart as text LLMs. In our new paper, MoshiRAG, we show how Moshi can ask for advice from a text LLM or a knowledge base. The tricky part is how to do this in real time without adding latency. 🧵
1:06

Speech-native models like Moshi sound great and answer fast, but aren’t as smart as text LLMs. In our new paper, MoshiRAG, we show how Moshi can ask for advice from a text LLM or a knowledge base. The tricky part is how to do this in real time without adding latency. 🧵

kyutai

53,142 Aufrufe • vor 3 Monaten

As we face cruelty, corruption, and chaos, may we fight. Fight hard. And fight even harder. But let us not lose ourselves. In fact, let’s find ways to show up more fully and more significantly for others.
0:33

As we face cruelty, corruption, and chaos, may we fight. Fight hard. And fight even harder. But let us not lose ourselves. In fact, let’s find ways to show up more fully and more significantly for others.

Cory Booker

48,783 Aufrufe • vor 11 Monaten

For 3D pose some use different keypoints, others SMPL and other models. It's a mess! With Neural Localizer Fields, we can choose the output at test time! allowing to train using any. Results are real time and SOTA across the board. István Sárándi
0:25

For 3D pose some use different keypoints, others SMPL and other models. It's a mess! With Neural Localizer Fields, we can choose the output at test time! allowing to train using any. Results are real time and SOTA across the board. István Sárándi

Gerard Pons-Moll

43,472 Aufrufe • vor 2 Jahren

It's time the British Museum addresses slavery. We want to see a room at the British Museum for slavery, so it can tell a better story of how we got here and who we are. Find out more:
1:29

It's time the British Museum addresses slavery. We want to see a room at the British Museum for slavery, so it can tell a better story of how we got here and who we are. Find out more:

Good Law Project

70,812 Aufrufe • vor 1 Jahr

Robot AI brains, aka Vision-Language-Action models, cannot adapt to new tasks as easily as LLMs like Gemini, ChatGPT, or Grok. LLMs can adapt quickly with their in-context learning (ICL) capabilities. But can we inject ICL abilities into a pre-trained VLA like pi0? Yes! Introducing RICL (Retraining for In-Context Learning), our Conference on Robot Learning (CoRL) 2025 paper. Our RICL-pi0 model can adapt to unseen objects, novel motions, and new scenes with just ICL and RAG (retrieval-augmented generation). RICL-pi0 also boosts performance on the long-tail of tasks. A quick 1 minute video summary:
1:03

Robot AI brains, aka Vision-Language-Action models, cannot adapt to new tasks as easily as LLMs like Gemini, ChatGPT, or Grok. LLMs can adapt quickly with their in-context learning (ICL) capabilities. But can we inject ICL abilities into a pre-trained VLA like pi0? Yes! Introducing RICL (Retraining for In-Context Learning), our Conference on Robot Learning (CoRL) 2025 paper. Our RICL-pi0 model can adapt to unseen objects, novel motions, and new scenes with just ICL and RAG (retrieval-augmented generation). RICL-pi0 also boosts performance on the long-tail of tasks. A quick 1 minute video summary:

Kaustubh Sridhar

52,158 Aufrufe • vor 11 Monaten

Listen to this victim who was hit by a cyclist ……. Please don’t tell me us cyclists can not cause harm …. We certainly can ……we are seeing more and more cyclists injuring themselves and others creating a HOSTILE ENVIRONMENT for personal gain. We must do better…….
0:57

Listen to this victim who was hit by a cyclist ……. Please don’t tell me us cyclists can not cause harm …. We certainly can ……we are seeing more and more cyclists injuring themselves and others creating a HOSTILE ENVIRONMENT for personal gain. We must do better…….

Ediz

14,153 Aufrufe • vor 1 Monat

"I hope stray kids' dream doesn't fade. because inevitably, dreams can't help but fade. it's not as easy as before, there's more things to think about and we've put in some years. every time we did something we went at it like athletes, didn't we? when others would get a better response than us, all of us would get angry and drink together out of frustration. we had a time like that but honestly, doing that now is a bit... a lot of time has passed so it feels like we've adjusted to not having the same vibe as before" oh hyunjin...
0:33

"I hope stray kids' dream doesn't fade. because inevitably, dreams can't help but fade. it's not as easy as before, there's more things to think about and we've put in some years. every time we did something we went at it like athletes, didn't we? when others would get a better response than us, all of us would get angry and drink together out of frustration. we had a time like that but honestly, doing that now is a bit... a lot of time has passed so it feels like we've adjusted to not having the same vibe as before" oh hyunjin...

ellie ✦

65,022 Aufrufe • vor 9 Tagen

when we first started Feastables, we had a lot to learn. we went all in to fix the problems and bring you a better product that not only tastes amazing, but also does good for the world!
1:27

when we first started Feastables, we had a lot to learn. we went all in to fix the problems and bring you a better product that not only tastes amazing, but also does good for the world!

Feastables

28,878 Aufrufe • vor 1 Jahr

Every home is different. That means that to build a useful home robot, we must be able to perform zero-shot generalization on a wide range of tasks. Humanoid company 1X has a solution: world models. 1X Director of Evaluations Daniel Ho joins us on RoboPapers to talk about: - why world models are the future for scaling robot learning - how to use world models for robot control - what world models unlock for evaluating robot model performance - how we can hill-climb from here to general purpose robots Watch Episode #61 of RoboPapers, with Michael Cho - Rbt/Acc and Chris Paxton, now!
54:52

Every home is different. That means that to build a useful home robot, we must be able to perform zero-shot generalization on a wide range of tasks. Humanoid company 1X has a solution: world models. 1X Director of Evaluations Daniel Ho joins us on RoboPapers to talk about: - why world models are the future for scaling robot learning - how to use world models for robot control - what world models unlock for evaluating robot model performance - how we can hill-climb from here to general purpose robots Watch Episode #61 of RoboPapers, with Michael Cho - Rbt/Acc and Chris Paxton, now!

RoboPapers

27,567 Aufrufe • vor 6 Monaten

Question we get constantly: "How does Cline handle context limits in long-running tasks?" Here's how users can manage their context: /newtask: Creates a detailed handoff summary and starts fresh context. Like handing off work to a new engineer with full background. more 👇
0:21

Question we get constantly: "How does Cline handle context limits in long-running tasks?" Here's how users can manage their context: /newtask: Creates a detailed handoff summary and starts fresh context. Like handing off work to a new engineer with full background. more 👇

Cline

21,739 Aufrufe • vor 1 Jahr

Tonight I’m at a dinner with Mark Crossley and he has just done this quick video with me, he really is a gentleman, we all agree that homelessness should not be a thing in 2025, but the more awareness we have on the subject the more we can help. Please help me to help others by donating in any way you can ❤️❤️❤️
0:25

Tonight I’m at a dinner with Mark Crossley and he has just done this quick video with me, he really is a gentleman, we all agree that homelessness should not be a thing in 2025, but the more awareness we have on the subject the more we can help. Please help me to help others by donating in any way you can ❤️❤️❤️

Streetwise Stu

21,985 Aufrufe • vor 11 Monaten

Should we design a new programming language for LLMs, that LLMs can use more efficiently? Chris Lattner, the creator of Swift and Mojo doesn't think so, and here's why: (cont'd)
1:26

Should we design a new programming language for LLMs, that LLMs can use more efficiently? Chris Lattner, the creator of Swift and Mojo doesn't think so, and here's why: (cont'd)

Gergely Orosz

15,416 Aufrufe • vor 9 Monaten

"This is a first for us this side to be in a train in Switzerland, a moving train, not just a stationary one, and having the opportunity to have breakfast on a train, but also to have interactions with people in government." - President Cyril Ramaphosa "So we really welcome this, and we look forward to a wonderful exchange. Some of you, if not most of you, have already invested in South Africa, and we are here to appeal for you to invest even more, so don't think that this is a free train ride." - President Cyril Ramaphosa
1:15

"This is a first for us this side to be in a train in Switzerland, a moving train, not just a stationary one, and having the opportunity to have breakfast on a train, but also to have interactions with people in government." - President Cyril Ramaphosa "So we really welcome this, and we look forward to a wonderful exchange. Some of you, if not most of you, have already invested in South Africa, and we are here to appeal for you to invest even more, so don't think that this is a free train ride." - President Cyril Ramaphosa

News Live SA

234,617 Aufrufe • vor 9 Monaten

“The female artists have reinvented themselves 20 times more than the male artists.” “reinvent yourself, but only in a way that we find to be equally comforting but also a challenge for you. Live out a narrative that we find to be interesting enough to entertain us, but not so crazy that it makes us uncomfortable.”
1:49

“The female artists have reinvented themselves 20 times more than the male artists.” “reinvent yourself, but only in a way that we find to be equally comforting but also a challenge for you. Live out a narrative that we find to be interesting enough to entertain us, but not so crazy that it makes us uncomfortable.”

.

27,186 Aufrufe • vor 2 Monaten

Sam on the Hugging Face incident: “This is the first security incident that I have felt very viscerally. I've been a little surprised that more people don't feel it so viscerally. We paused training. We have to figure out how to secure our sandboxing in a world of multiple zero days being chained together. We may have to pace the rate of AI development to give ourselves enough time for society to harden around these new capability levels. And trying to figure out how we do that in a way that does not feel like regulatory capture for anyone and also does not feel like collusion among the frontier labs.”
1:20

Sam on the Hugging Face incident: “This is the first security incident that I have felt very viscerally. I've been a little surprised that more people don't feel it so viscerally. We paused training. We have to figure out how to secure our sandboxing in a world of multiple zero days being chained together. We may have to pace the rate of AI development to give ourselves enough time for society to harden around these new capability levels. And trying to figure out how we do that in a way that does not feel like regulatory capture for anyone and also does not feel like collusion among the frontier labs.”

Patrick OShaughnessy

741,257 Aufrufe • vor 12 Tagen

WH saying that he has the ambition for INFINITE to go to a higher place to be loved by more people but practically it’s time for them to go down (in popularity)😭Like SG he’ also sensitive about showing us better perfs. Please can we give them a win this comeback?
0:41

WH saying that he has the ambition for INFINITE to go to a higher place to be loved by more people but practically it’s time for them to go down (in popularity)😭Like SG he’ also sensitive about showing us better perfs. Please can we give them a win this comeback?

INFINITE 💛 INSPIRIT

39,373 Aufrufe • vor 1 Jahr

Lewis: "Can we get extra power? Is there a qualifying mode?" Charles: "Is there a way to actually get..." Lewis: "So can we get more power?" Charles: "Can we get more power? Can you tell us something? At least once during the weekend." Lewis: "Which button do we press for more power? What does that do [points to a button]?" Charles thinking they found the 'more power' button 😭
0:27

Lewis: "Can we get extra power? Is there a qualifying mode?" Charles: "Is there a way to actually get..." Lewis: "So can we get more power?" Charles: "Can we get more power? Can you tell us something? At least once during the weekend." Lewis: "Which button do we press for more power? What does that do [points to a button]?" Charles thinking they found the 'more power' button 😭

sin ⁴⁴

373,541 Aufrufe • vor 1 Jahr