正在加载视频...

视频加载失败

返回首页

Meta AI researcher who created DINOv2 explains modern self-supervised learning! Also introduces a new approach called CAPI TimDarcet gave a brilliant talk about his research at the MedARC journal club last month. Very clear and information-dense, I learned a lot from his talk. Give it a watch!

iScienceLuvr's profile picture

Tanishq Mathew Abraham, Ph.D.

90,337 subscribers

37,803 次观看 • 21 天前 •via X (Twitter)

LIVE

Anya Rossi• Live Now

Private livecam show

0 条评论

暂无评论

原始帖子的评论将显示在这里

相关视频

I recently gave a short talk at the International Workshop on Reimagining Democracy. The first half focused on feeling the AGI. The second half briefly outlined a new research direction I'm very excited about: leveraging AI to build unprecedentedly trustworthy institutions.
8:36

I recently gave a short talk at the International Workshop on Reimagining Democracy. The first half focused on feeling the AGI. The second half briefly outlined a new research direction I'm very excited about: leveraging AI to build unprecedentedly trustworthy institutions.

Richard Ngo

40,435 次观看 • 1 年前

A Notre Dame fan's father died at the Notre Dame-Pitt football game. The family called it "the perfect death." Marcus Freeman learned about it and spent took time to talk about his team's obligation to give it their all for their fanbase.
1:55

A Notre Dame fan's father died at the Notre Dame-Pitt football game. The family called it "the perfect death." Marcus Freeman learned about it and spent took time to talk about his team's obligation to give it their all for their fanbase.

Front Office Sports

388,205 次观看 • 9 个月前

Everybody is talking about recursive self-improvement (RSI) and meta learning. Here is my old 2020 talk about this [1]. It has aged well. Example: humans still define the starts & ends of trials of many modern meta learners. My RSI systems since 1994 LEARN to (re)define them [2]! [1] Meta Learning Machines in a Single Lifelong Trial (talk for workshops at ICML 2020 and NeurIPS 2021, based on earlier talks since 1994). Abstract: the most widely used machine learning algorithms were designed by humans and thus are hindered by our cognitive biases and limitations. Can we also construct meta learning algorithms that can learn better learning algorithms so that our self-improving AIs have no limits other than those inherited from computability and physics? This question has been a main driver of my research since I wrote a thesis on it in 1987 [2]. Here I summarize our work on meta reinforcement learning with self-modifying policies in a single lifelong trial (since 1994), and mathematically optimal meta-learning through the self-referential Gödel Machine (since 2003). Many additional publications on meta-learning since 1987 can be found in the RSI overview [2]. [2] J. Schmidhuber (AI Blog, 2020-2025). 1/3 century anniversary of first publication on recursive self-improvement (RSI) and meta learning machines that learn to learn (1987). For its cover I drew a robot that bootstraps itself. 1992-: gradient descent-based neural meta learning. 1994-: meta reinforcement learning with self-modifying policies. 1997: meta RL plus artificial curiosity and intrinsic motivation. 2002-: asymptotically optimal meta learning for curriculum learning. 2003-: mathematically optimal Gödel Machine. 2020-: new stuff!
23:42

Everybody is talking about recursive self-improvement (RSI) and meta learning. Here is my old 2020 talk about this [1]. It has aged well. Example: humans still define the starts & ends of trials of many modern meta learners. My RSI systems since 1994 LEARN to (re)define them [2]! [1] Meta Learning Machines in a Single Lifelong Trial (talk for workshops at ICML 2020 and NeurIPS 2021, based on earlier talks since 1994). Abstract: the most widely used machine learning algorithms were designed by humans and thus are hindered by our cognitive biases and limitations. Can we also construct meta learning algorithms that can learn better learning algorithms so that our self-improving AIs have no limits other than those inherited from computability and physics? This question has been a main driver of my research since I wrote a thesis on it in 1987 [2]. Here I summarize our work on meta reinforcement learning with self-modifying policies in a single lifelong trial (since 1994), and mathematically optimal meta-learning through the self-referential Gödel Machine (since 2003). Many additional publications on meta-learning since 1987 can be found in the RSI overview [2]. [2] J. Schmidhuber (AI Blog, 2020-2025). 1/3 century anniversary of first publication on recursive self-improvement (RSI) and meta learning machines that learn to learn (1987). For its cover I drew a robot that bootstraps itself. 1992-: gradient descent-based neural meta learning. 1994-: meta reinforcement learning with self-modifying policies. 1997: meta RL plus artificial curiosity and intrinsic motivation. 2002-: asymptotically optimal meta learning for curriculum learning. 2003-: mathematically optimal Gödel Machine. 2020-: new stuff!

Jürgen Schmidhuber

232,164 次观看 • 5 个月前

Everything you have been told about climate change is false. Not spun. Not exaggerated. Mathematically false. I spent the last year proving it. Last month I delivered the whole case, uncut, to a room of scientists in Halle, Germany. The IPCC Circle of Lies: Five Papers Falsify the Entire Framework My co-author Willie Soon also gave a talk covering all the implications, from every angle, as only he can, in his talk: From Demon to Dinner: The Scientific Case for CO2 and Fossil Fuels. Watch my talk here. Willie's in my next post.
47:11

Everything you have been told about climate change is false. Not spun. Not exaggerated. Mathematically false. I spent the last year proving it. Last month I delivered the whole case, uncut, to a room of scientists in Halle, Germany. The IPCC Circle of Lies: Five Papers Falsify the Entire Framework My co-author Willie Soon also gave a talk covering all the implications, from every angle, as only he can, in his talk: From Demon to Dinner: The Scientific Case for CO2 and Fossil Fuels. Watch my talk here. Willie's in my next post.

Jonathan Cohler

77,779 次观看 • 1 个月前

Here is the 20 minute talk from Gwynne Shotwell and Michael Nicolls at the MWC26 with added subtitles. This talk includes a lot of new information about the future of Starlink Mobile. I highly recommend watching the full keynote!
21:29

Here is the 20 minute talk from Gwynne Shotwell and Michael Nicolls at the MWC26 with added subtitles. This talk includes a lot of new information about the future of Starlink Mobile. I highly recommend watching the full keynote!

Robin

201,903 次观看 • 6 个月前

Sirianni, Hurts and the players talk a lot about learning from past experiences These next 5 games will tell us how much they really learned from 2023 — and if they can prevent it from happening again
0:35

Sirianni, Hurts and the players talk a lot about learning from past experiences These next 5 games will tell us how much they really learned from 2023 — and if they can prevent it from happening again

Eliot Shorr-Parks

24,280 次观看 • 9 个月前

[ 🐰 i talk a lot! 👤 that’s why you’re the diction princess! diction queen! there are these sayings. your pronunciation is never wrong! who do you talk to the most? 🐰 i think i talk to my mom a lot. even to my dad! i also talk by myself but there’s this existence called hongsi, and i talk to it a lot! have you heard a cat’s sigh? 👤 i haven’t! 🐰 let’s say it’s lying down and looking at me like this, then.. (sigh) and it’ll go inside the room!
0:27

[ 🐰 i talk a lot! 👤 that’s why you’re the diction princess! diction queen! there are these sayings. your pronunciation is never wrong! who do you talk to the most? 🐰 i think i talk to my mom a lot. even to my dad! i also talk by myself but there’s this existence called hongsi, and i talk to it a lot! have you heard a cat’s sigh? 👤 i haven’t! 🐰 let’s say it’s lying down and looking at me like this, then.. (sigh) and it’ll go inside the room!

🎀

39,232 次观看 • 11 个月前

Jason Heyward hasn’t been in Houston long but he’s already made his mark as an Astros big time performer. He’s also learned a lot about this ball club. “They walk it like they talk it.” 🚶‍♂️🗣️🚶‍♂️🗣️
0:29

Jason Heyward hasn’t been in Houston long but he’s already made his mark as an Astros big time performer. He’s also learned a lot about this ball club. “They walk it like they talk it.” 🚶‍♂️🗣️🚶‍♂️🗣️

Will Kunkel

101,993 次观看 • 1 年前

Recently, Andrew Ng took part in a discussion at Stanford where he described the endless opportunities in AI. Watch his full talk and learn more about why AI is the new electricity:
1:45

Recently, Andrew Ng took part in a discussion at Stanford where he described the endless opportunities in AI. Watch his full talk and learn more about why AI is the new electricity:

DeepLearning.AI

31,899 次观看 • 2 年前

This is amazing research result! Xun Huang and team trained a bidirectional teacher model for motion control, and used Self-Forcing to make it autoregressive. Xun recently gave a great talk at the Real Time Video AI summit about this work, and a few other projects around video world models. Here is the recording.
29:48

This is amazing research result! Xun Huang and team trained a bidirectional teacher model for motion control, and used Self-Forcing to make it autoregressive. Xun recently gave a great talk at the Real Time Video AI summit about this work, and a few other projects around video world models. Here is the recording.

Eric is Daydreaming

52,796 次观看 • 10 个月前

Async, automatic, non-blocking context compaction for long-running agents. Last week I gave a talk called Space Machine Sandboxes at the Daytona AI builders meetup about patterns for long-running agents. I work a lot on voice AI agents, which are fundamentally multi-turn, long-context loops. I also build lots of other AI agent stuff, often as part of bigger systems that include voice. One of the patterns I showed in the talk is non-blocking compaction. Here's a short clip.
1:46

Async, automatic, non-blocking context compaction for long-running agents. Last week I gave a talk called Space Machine Sandboxes at the Daytona AI builders meetup about patterns for long-running agents. I work a lot on voice AI agents, which are fundamentally multi-turn, long-context loops. I also build lots of other AI agent stuff, often as part of bigger systems that include voice. One of the patterns I showed in the talk is non-blocking compaction. Here's a short clip.

kwindla

25,877 次观看 • 7 个月前

Dana White says Josh Hokit "walked the talk" at #UFC327 and explains the quick booking vs. Derrick Lewis at #UFCWhiteHouse. He also won't criticize his gimmick again: "You won't hear a word from me from hereon out about him."
2:08

Dana White says Josh Hokit "walked the talk" at #UFC327 and explains the quick booking vs. Derrick Lewis at #UFCWhiteHouse. He also won't criticize his gimmick again: "You won't hear a word from me from hereon out about him."

MMA Junkie

234,875 次观看 • 4 个月前

THE NEVER-GIVE-UP ATTITUDE OF SURYAKUMAR YADAV. 🙌🫡 - Removed from T20I captaincy - dropped from the T20I team - and surrounded by a lot of negative talk about his form. And this very evening, in the T20 Mumbai match, Suryakumar Yadav came out in brilliant form and played a fantastic knock of 48 runs off just 24 balls.💪🏻🔥
0:59

THE NEVER-GIVE-UP ATTITUDE OF SURYAKUMAR YADAV. 🙌🫡 - Removed from T20I captaincy - dropped from the T20I team - and surrounded by a lot of negative talk about his form. And this very evening, in the T20 Mumbai match, Suryakumar Yadav came out in brilliant form and played a fantastic knock of 48 runs off just 24 balls.💪🏻🔥

𝐑𝐮𝐬𝐡𝐢𝐢𝐢⁴⁵

58,843 次观看 • 3 个月前

I had Brad Bradley on the show today to talk about his missing sister Amy. He shared some new theories and what the documentary missed. Hope you give it a listen.
2:54

I had Brad Bradley on the show today to talk about his missing sister Amy. He shared some new theories and what the documentary missed. Hope you give it a listen.

Eddie

251,746 次观看 • 1 年前

everyone: expecting quackity to talk about tommy and dream #quackity: promoting his new project, the live translation and his merch then saying “there’s this other thing that i wanted to talk about but i forgot” i have to give a standing ovation for this marketing move ngl lmao
0:30

everyone: expecting quackity to talk about tommy and dream #quackity: promoting his new project, the live translation and his merch then saying “there’s this other thing that i wanted to talk about but i forgot” i have to give a standing ovation for this marketing move ngl lmao

mari ❤️‍🔥 ⁷

164,583 次观看 • 1 年前

eric talked about the articles of him joining a new company for his individual activities! 🦄 actually we’ve been talking about this a lot for a long time, and i’ve talked about it a lot with the members too 🦄 i’ve also talked about it a lot with the people around us, the people at the company, the representatives, and so on 🦄 rather than rushing into something like this, we all had our own plans all along 🦄 so now i can talk about it with peace of mind, right? 🦄 i got a job! 💪⚡️
0:34

eric talked about the articles of him joining a new company for his individual activities! 🦄 actually we’ve been talking about this a lot for a long time, and i’ve talked about it a lot with the members too 🦄 i’ve also talked about it a lot with the people around us, the people at the company, the representatives, and so on 🦄 rather than rushing into something like this, we all had our own plans all along 🦄 so now i can talk about it with peace of mind, right? 🦄 i got a job! 💪⚡️

ga🅱️z

32,550 次观看 • 25 天前

“The grossest changeup I ever threw got hit 450 feet…He knew I was going to throw that pitch and he was committed to that approach”🤯 Listen to Eugene talk about a story he heard from Cole Hamels about Manny Ramirez sticking to his approach and about knowing who you are as a player🗣️ #the108way
1:04

“The grossest changeup I ever threw got hit 450 feet…He knew I was going to throw that pitch and he was committed to that approach”🤯 Listen to Eugene talk about a story he heard from Cole Hamels about Manny Ramirez sticking to his approach and about knowing who you are as a player🗣️ #the108way

108 Performance

70,533 次观看 • 3 年前

Rep. Burlison: New UFO Video is Remarkable “I received it anonymously in a way in which I’ve learned is called ‘the dead drop’. The video was scrubbed from all information… There’s nothing to trace back to and I have no idea who gave it to me.”
1:20

Rep. Burlison: New UFO Video is Remarkable “I received it anonymously in a way in which I’ve learned is called ‘the dead drop’. The video was scrubbed from all information… There’s nothing to trace back to and I have no idea who gave it to me.”

UAP James

192,999 次观看 • 11 个月前