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。

We just did the largest live deployment of interactive robots in history. Users online could ask the robot to do things it was never taught to do. A true foundation model, with 0 task-specific data and 0 task fine tuning. 1/10

Enigma

4,699 subscribers

100,748 次观看 • 1 天前 •via X (Twitter)

科学技术新闻政治

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

We’re excited to introduce Text-to-LoRA: a Hypernetwork that generates task-specific LLM adapters (LoRAs) based on a text description of the task. Catch our presentation at #ICML2025! Paper: Code: Biological systems are capable of rapid adaptation, given limited sensory cues. For example, our human visual system can quickly adapt and tune its light sensitivity to our surroundings. While modern LLMs exhibit a wide variety of capabilities and knowledge, they remain rigid when adding task-specific capabilities. Traditionally, customizing these models requires gathering large datasets and performing often expensive, time-consuming fine-tuning for specific applications. To bypass these limitations, Text-to-LoRA (T2L) meta-learns a “hypernetwork” that takes in a text description of a desired task, as a prompt, and generates a task-specific LoRA that performs well on the task. In our experiments, we show that T2L can encode hundreds of existing LoRA adapters. While the compression is lossy, T2L maintains the performance of task-specifically tuned LoRA adapters. We also show that T2L can even generalize to unseen tasks given a natural language description of the tasks. Importantly, Text-to-LoRA is parameter-efficient. It generates LoRAs in a single, inexpensive step, based solely on a simple text description of the task. This approach is a step towards dramatically lowering the technical and computational barriers, allowing non-technical users to specialize foundation models using plain language, rather than needing deep technical expertise or large compute resources.

We’re excited to introduce Text-to-LoRA: a Hypernetwork that generates task-specific LLM adapters (LoRAs) based on a text description of the task. Catch our presentation at #ICML2025! Paper: Code: Biological systems are capable of rapid adaptation, given limited sensory cues. For example, our human visual system can quickly adapt and tune its light sensitivity to our surroundings. While modern LLMs exhibit a wide variety of capabilities and knowledge, they remain rigid when adding task-specific capabilities. Traditionally, customizing these models requires gathering large datasets and performing often expensive, time-consuming fine-tuning for specific applications. To bypass these limitations, Text-to-LoRA (T2L) meta-learns a “hypernetwork” that takes in a text description of a desired task, as a prompt, and generates a task-specific LoRA that performs well on the task. In our experiments, we show that T2L can encode hundreds of existing LoRA adapters. While the compression is lossy, T2L maintains the performance of task-specifically tuned LoRA adapters. We also show that T2L can even generalize to unseen tasks given a natural language description of the tasks. Importantly, Text-to-LoRA is parameter-efficient. It generates LoRAs in a single, inexpensive step, based solely on a simple text description of the task. This approach is a step towards dramatically lowering the technical and computational barriers, allowing non-technical users to specialize foundation models using plain language, rather than needing deep technical expertise or large compute resources.

Sakana AI

403,159 次观看 • 1 年前

How can a visuomotor policy learn from internet videos? We introduce Dreamitate, where a robot uses a fine-tuned video diffusion model to dream the future (top) and imitate the dream to accomplish a task (bottom). website: paper:

How can a visuomotor policy learn from internet videos? We introduce Dreamitate, where a robot uses a fine-tuned video diffusion model to dream the future (top) and imitate the dream to accomplish a task (bottom). website: paper:

Ruoshi Liu

50,797 次观看 • 2 年前

Atomic Bot put Hermes and OpenClaw head-to-head on the exact same task, running the same model (Qwen 3.6 35B) with the same goal: analyzing GitHub history, mapping growth spikes, and shipping a live dashboard in the browser. Key metrics to watch for 👀 > Time to complete the task > Tokens spent > Quality of the final result (dashboard in browser)

Atomic Bot put Hermes and OpenClaw head-to-head on the exact same task, running the same model (Qwen 3.6 35B) with the same goal: analyzing GitHub history, mapping growth spikes, and shipping a live dashboard in the browser. Key metrics to watch for 👀 > Time to complete the task > Tokens spent > Quality of the final result (dashboard in browser)

🚨 AI News | TestingCatalog

142,275 次观看 • 2 个月前

Alright, now that we know *what* an agent is, how does it actually work? When you ask for help on a task, the agent plans a series of steps and executes them directly in the application on your behalf, using the tools it has access to. Say you are booking a local service or trying to organize your inbox (which typically takes multiple steps): the AI model first plans how to achieve the task using its existing knowledge and then interacts with your inbox to execute the task. The agent will continue until it is confident the task has been successfully completed.

Alright, now that we know what an agent is, how does it actually work? When you ask for help on a task, the agent plans a series of steps and executes them directly in the application on your behalf, using the tools it has access to. Say you are booking a local service or trying to organize your inbox (which typically takes multiple steps): the AI model first plans how to achieve the task using its existing knowledge and then interacts with your inbox to execute the task. The agent will continue until it is confident the task has been successfully completed.

Google AI

22,487 次观看 • 8 个月前

🇨🇳 ROBOT RELAY ON THE GREAT WALL? CHINA JUST DID THAT Robots just ran an energy baton relay on the Great Wall of China to open the World Humanoid Robot Games. Yes, actual robots. In uniforms. Running. Is it the future of sports, or just the nerdiest flex in history? Source: CGTN

🇨🇳 ROBOT RELAY ON THE GREAT WALL? CHINA JUST DID THAT Robots just ran an energy baton relay on the Great Wall of China to open the World Humanoid Robot Games. Yes, actual robots. In uniforms. Running. Is it the future of sports, or just the nerdiest flex in history? Source: CGTN

Mario Nawfal

82,407 次观看 • 1 年前

What if your model handled its own tuning? ⚙️ With NVIDIA TAO 7, just prompt your coding agent what you want in plain language and let it do the rest. • Agent skills plug into your coding agent and improve accuracy. • AutoML removes the guesswork from hyperparameters. LLM-guided tuning finds strong configs up to 2x faster. • Fine-tune any Hugging Face CV or VLM model on local NVIDIA GPUs. • Data enhanced fine-tuning helps your agent identify why the model fails, then fix it. Try TAO skills ➡️

What if your model handled its own tuning? ⚙️ With NVIDIA TAO 7, just prompt your coding agent what you want in plain language and let it do the rest. • Agent skills plug into your coding agent and improve accuracy. • AutoML removes the guesswork from hyperparameters. LLM-guided tuning finds strong configs up to 2x faster. • Fine-tune any Hugging Face CV or VLM model on local NVIDIA GPUs. • Data enhanced fine-tuning helps your agent identify why the model fails, then fix it. Try TAO skills ➡️

NVIDIA AI

41,902 次观看 • 1 个月前

GPT-5.1 is now live in Warp. It’s much faster (40% faster task completion on a subset of SWE-bench Verified) without compromising quality. GPT-5.1 is available to all Warp users, and is now the default model for all new users.

GPT-5.1 is now live in Warp. It’s much faster (40% faster task completion on a subset of SWE-bench Verified) without compromising quality. GPT-5.1 is available to all Warp users, and is now the default model for all new users.

Warp

80,683 次观看 • 9 个月前

Peeling a potato is trivial -- until you try to make a robot do it with a knife. This is actually one of the hardest problems in manipulation: contact-rich, force-sensitive, and success is subjective. We taught a robot arm to peel with >90% success :D

Peeling a potato is trivial -- until you try to make a robot do it with a knife. This is actually one of the hardest problems in manipulation: contact-rich, force-sensitive, and success is subjective. We taught a robot arm to peel with >90% success :D

Toru

33,109 次观看 • 5 个月前

We were able to embed all of Wikipedia in < 15 minutes on Modal Labs for ~ 17$. Here's what this actually means for an organization: 1. Unrestricted Rates: Eliminates the bottleneck of rate limits for large-scale operations. 2. Rapid Experimentation: Allows quick iterations of data chunking experiments, reducing wait times from hours to minutes. 3. Fine-tuning: Use of the same hf infrastructure for model fine-tuning, speeding up experiment cycles, and deployment. Reducing experiment timelines from hours to approximately to minutes is something that has compounding effects! take a look!

We were able to embed all of Wikipedia in < 15 minutes on Modal Labs for ~ 17$. Here's what this actually means for an organization: 1. Unrestricted Rates: Eliminates the bottleneck of rate limits for large-scale operations. 2. Rapid Experimentation: Allows quick iterations of data chunking experiments, reducing wait times from hours to minutes. 3. Fine-tuning: Use of the same hf infrastructure for model fine-tuning, speeding up experiment cycles, and deployment. Reducing experiment timelines from hours to approximately to minutes is something that has compounding effects! take a look!

jason

230,404 次观看 • 2 年前

The team just sent me this funny video of how AWFUL the days after the 2020 election were. People were kinda laughing at me, but it was miserable. And little did we know how bad it would truly get under the Biden and progressive retaliation machine. This was when we first locked in with the Charlie Kirk show and kept up real time updates and analysis where Arizona and a number of other states became central. I never want to go back to this feeling. Never. Charlie was adamant that we do everything we could possibly do to fix this in 2024. We DID IT and we must continue for 2026 and 2028 to have a shot at saving this country.

The team just sent me this funny video of how AWFUL the days after the 2020 election were. People were kinda laughing at me, but it was miserable. And little did we know how bad it would truly get under the Biden and progressive retaliation machine. This was when we first locked in with the Charlie Kirk show and kept up real time updates and analysis where Arizona and a number of other states became central. I never want to go back to this feeling. Never. Charlie was adamant that we do everything we could possibly do to fix this in 2024. We DID IT and we must continue for 2026 and 2028 to have a shot at saving this country.

Tyler Bowyer

25,672 次观看 • 2 个月前

The interactive product page is now fully live on Shopify . Tested the integration of interactive animation directly within the online store. You can configure the product, and all data stays in sync through to checkout. 🎄 Test it:

The interactive product page is now fully live on Shopify . Tested the integration of interactive animation directly within the online store. You can configure the product, and all data stays in sync through to checkout. 🎄 Test it:

Daria Nevezhyna

291,483 次观看 • 7 个月前

MolmoAct2 is landing in LeRobot! Ai2's open Action Reasoning Model combines a Molmo2-ER vision-language backbone with a flow-matching continuous action expert to predict robot action chunks from images, language instructions, and proprioceptive state. An open robot foundation model built for real-world control, with strong out-of-the-box performance and easy fine-tuning in LeRobot. Pick-and-place inference running on NVIDIA DGX Spark! Blog: Paper: Thanks to Ai2 Jiafei Duan Haoquan Fang

MolmoAct2 is landing in LeRobot! Ai2's open Action Reasoning Model combines a Molmo2-ER vision-language backbone with a flow-matching continuous action expert to predict robot action chunks from images, language instructions, and proprioceptive state. An open robot foundation model built for real-world control, with strong out-of-the-box performance and easy fine-tuning in LeRobot. Pick-and-place inference running on NVIDIA DGX Spark! Blog: Paper: Thanks to Ai2 Jiafei Duan Haoquan Fang

LeRobot

24,727 次观看 • 2 个月前

Jalen Green hopped off the jet, said he was going to go 10-0, and then dropped off the 3rd best team in the league by himself to go 10-0 just like he said he would. We are witnessing the rise of a superstar.

Jalen Green hopped off the jet, said he was going to go 10-0, and then dropped off the 3rd best team in the league by himself to go 10-0 just like he said he would. We are witnessing the rise of a superstar.

ᵂᴵᴸᴸ

274,186 次观看 • 2 年前

Announcing our commercial partnership with Booster Robotics Booster builds humanoid robot hardware, OS, and developer tools to make humanoid robots more affordable, reliable, and practical. The partnership centers on using simulation to multiply the value of real-robot data—expanding teleoperation demonstrations into scalable training data across diverse tasks and scenes. This joint effort powers sim-real co-training and foundation-model development, accelerating progress from hardware iteration to deployable robot policies. Together, we're building simulation‑powered data infrastructure for Physical AI — making scalable training data accessible to model developers and the broader robotics ecosystem.

Announcing our commercial partnership with Booster Robotics Booster builds humanoid robot hardware, OS, and developer tools to make humanoid robots more affordable, reliable, and practical. The partnership centers on using simulation to multiply the value of real-robot data—expanding teleoperation demonstrations into scalable training data across diverse tasks and scenes. This joint effort powers sim-real co-training and foundation-model development, accelerating progress from hardware iteration to deployable robot policies. Together, we're building simulation‑powered data infrastructure for Physical AI — making scalable training data accessible to model developers and the broader robotics ecosystem.

Axis Robotics

62,691 次观看 • 6 天前

It was all so we could build homes with robots. All of our work in floorplan AI,and generative CAD modeling, we did all of that, just so that we could build homes with robots. Early on we talked to a robotics engineer from Katerra. It took them $750k to switch from one house to another house. It takes us 25 minutes, and that's if we had a big lunch. P.S. The soundtrack is hardcore punk, lyrics written by me, but everything else is udio.

It was all so we could build homes with robots. All of our work in floorplan AI,and generative CAD modeling, we did all of that, just so that we could build homes with robots. Early on we talked to a robotics engineer from Katerra. It took them $750k to switch from one house to another house. It takes us 25 minutes, and that's if we had a big lunch. P.S. The soundtrack is hardcore punk, lyrics written by me, but everything else is udio.

Barrett Ames

390,137 次观看 • 1 年前

RoboBallet: End of Manual Robot Programming Google DeepMind/Intrinsic/UCL just cracked multi-robot coordination. Published in Science Robotics. 🤖What it does: – 8 robots, 40 tasks, zero collisions - all planned in seconds – 25% better motion plans than expert-designed solutions – 60% faster task completion when scaling from 4 to 8 robots – Works on factory layouts, it's never seen 👩‍💻 The tech: – Graph Neural Networks map robots/obstacles as connected nodes – Reinforcement learning trains on millions of scenarios – No manual programming required - just give it CAD files and tasks ❓Question: The bottleneck to robot collaboration has always been manual prototyping. Now that we can automate the automation what will this unlock?

RoboBallet: End of Manual Robot Programming Google DeepMind/Intrinsic/UCL just cracked multi-robot coordination. Published in Science Robotics. 🤖What it does: – 8 robots, 40 tasks, zero collisions - all planned in seconds – 25% better motion plans than expert-designed solutions – 60% faster task completion when scaling from 4 to 8 robots – Works on factory layouts, it's never seen 👩‍💻 The tech: – Graph Neural Networks map robots/obstacles as connected nodes – Reinforcement learning trains on millions of scenarios – No manual programming required - just give it CAD files and tasks ❓Question: The bottleneck to robot collaboration has always been manual prototyping. Now that we can automate the automation what will this unlock?

Jack 🤖

41,633 次观看 • 11 个月前

New Orleans Dancin' Mikaylah Williams and LSU Women's Basketball run their record to 10-0 with 126-62 win at UNO. 1,588 was the official attendance, the second largest home crowd in UNO women's history. #LSU

New Orleans Dancin' Mikaylah Williams and LSU Women's Basketball run their record to 10-0 with 126-62 win at UNO. 1,588 was the official attendance, the second largest home crowd in UNO women's history. #LSU

Jacques Doucet

28,829 次观看 • 8 个月前

𝗜'𝘃𝗲 𝗵𝗲𝗮𝗿𝗱 𝘁𝗵𝗶𝘀 𝗮 𝗹𝗼𝘁 𝗿𝗲𝗰𝗲𝗻𝘁𝗹𝘆: "𝗪𝗲 𝘁𝗿𝗮𝗶𝗻𝗲𝗱 𝗼𝘂𝗿 𝗿𝗼𝗯𝗼𝘁 𝗼𝗻 𝗼𝗻𝗲 𝗼𝗯𝗷𝗲𝗰𝘁 𝗮𝗻𝗱 𝗶𝘁 𝗴𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝗲𝗱 𝘁𝗼 𝗮 𝗻𝗼𝘃𝗲𝗹 𝗼𝗯𝗷𝗲𝗰𝘁 - 𝘁𝗵𝗲𝘀𝗲 𝗻𝗲𝘄 𝗩𝗟𝗔 𝗺𝗼𝗱𝗲𝗹𝘀 𝗮𝗿𝗲 𝗰𝗿𝗮𝘇𝘆!" Let's talk about what's actually happening in that "A" (Action) part of your VLA model. The Vision and Language components? They're incredible. Pre-trained on internet-scale data, they understand objects, spatial relationships, and task instructions better than ever. But the Action component? That's still learned from scratch on your specific robot demonstrations. 𝗛𝗲𝗿𝗲'𝘀 𝘁𝗵𝗲 𝗿𝗲𝗮𝗹𝗶𝘁𝘆: Your VLA model has internet-scale understanding of what a screwdriver looks like and what "tighten the screw" means. But the actual motor pattern for "rotating wrist while applying downward pressure"? That comes from your 500 robot demos. 𝗪𝗵𝗮𝘁 𝘁𝗵𝗶𝘀 𝗺𝗲𝗮𝗻𝘀 𝗳𝗼𝗿 "𝗴𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝗮𝘁𝗶𝗼𝗻": • 𝗩𝗶𝘀𝗶𝗼𝗻 𝗴𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝗮𝘁𝗶𝗼𝗻: Recognises novel objects instantly (thanks to pre-training) • 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗴𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝗮𝘁𝗶𝗼𝗻: Understands new task instructions (thanks to pre-training) • 𝗔𝗰𝘁𝗶𝗼𝗻 𝗴𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝗮𝘁𝗶𝗼𝗻: Still limited to motor patterns seen during robot training Ask that same robot to "unscrew the bottle cap" and it fails because: • Vision: Recognises bottle and cap • Language: Understands "unscrew" • Action: Never learned the "twist while pulling" motor pattern 𝗧𝗵𝗲 𝗵𝗮𝗿𝗱 𝘁𝗿𝘂𝘁𝗵 𝗮𝗯𝗼𝘂𝘁 𝗩𝗟𝗔 𝗺𝗼𝗱𝗲𝗹𝘀: The "VL" gives you incredible zero-shot understanding. The "A" still requires task-specific demonstrations. We've cracked the perception and reasoning problem. We haven't cracked the motor generalisation problem.

𝗜'𝘃𝗲 𝗵𝗲𝗮𝗿𝗱 𝘁𝗵𝗶𝘀 𝗮 𝗹𝗼𝘁 𝗿𝗲𝗰𝗲𝗻𝘁𝗹𝘆: "𝗪𝗲 𝘁𝗿𝗮𝗶𝗻𝗲𝗱 𝗼𝘂𝗿 𝗿𝗼𝗯𝗼𝘁 𝗼𝗻 𝗼𝗻𝗲 𝗼𝗯𝗷𝗲𝗰𝘁 𝗮𝗻𝗱 𝗶𝘁 𝗴𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝗲𝗱 𝘁𝗼 𝗮 𝗻𝗼𝘃𝗲𝗹 𝗼𝗯𝗷𝗲𝗰𝘁 - 𝘁𝗵𝗲𝘀𝗲 𝗻𝗲𝘄 𝗩𝗟𝗔 𝗺𝗼𝗱𝗲𝗹𝘀 𝗮𝗿𝗲 𝗰𝗿𝗮𝘇𝘆!" Let's talk about what's actually happening in that "A" (Action) part of your VLA model. The Vision and Language components? They're incredible. Pre-trained on internet-scale data, they understand objects, spatial relationships, and task instructions better than ever. But the Action component? That's still learned from scratch on your specific robot demonstrations. 𝗛𝗲𝗿𝗲'𝘀 𝘁𝗵𝗲 𝗿𝗲𝗮𝗹𝗶𝘁𝘆: Your VLA model has internet-scale understanding of what a screwdriver looks like and what "tighten the screw" means. But the actual motor pattern for "rotating wrist while applying downward pressure"? That comes from your 500 robot demos. 𝗪𝗵𝗮𝘁 𝘁𝗵𝗶𝘀 𝗺𝗲𝗮𝗻𝘀 𝗳𝗼𝗿 "𝗴𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝗮𝘁𝗶𝗼𝗻": • 𝗩𝗶𝘀𝗶𝗼𝗻 𝗴𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝗮𝘁𝗶𝗼𝗻: Recognises novel objects instantly (thanks to pre-training) • 𝗟𝗮𝗻𝗴𝘂𝗮𝗴𝗲 𝗴𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝗮𝘁𝗶𝗼𝗻: Understands new task instructions (thanks to pre-training) • 𝗔𝗰𝘁𝗶𝗼𝗻 𝗴𝗲𝗻𝗲𝗿𝗮𝗹𝗶𝘀𝗮𝘁𝗶𝗼𝗻: Still limited to motor patterns seen during robot training Ask that same robot to "unscrew the bottle cap" and it fails because: • Vision: Recognises bottle and cap • Language: Understands "unscrew" • Action: Never learned the "twist while pulling" motor pattern 𝗧𝗵𝗲 𝗵𝗮𝗿𝗱 𝘁𝗿𝘂𝘁𝗵 𝗮𝗯𝗼𝘂𝘁 𝗩𝗟𝗔 𝗺𝗼𝗱𝗲𝗹𝘀: The "VL" gives you incredible zero-shot understanding. The "A" still requires task-specific demonstrations. We've cracked the perception and reasoning problem. We haven't cracked the motor generalisation problem.

Stephen James

51,356 次观看 • 11 个月前

PRESIDENT TRUMP on potential talks with Venezuelan President Nicolás Maduro: "If we can save lives, if we can do things the easy way, that's fine. And if we have to do it the hard way, that's fine too."

PRESIDENT TRUMP on potential talks with Venezuelan President Nicolás Maduro: "If we can save lives, if we can do things the easy way, that's fine. And if we have to do it the hard way, that's fine too."

Fox News

354,244 次观看 • 8 个月前